UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Glaciovolcanism in the Garibaldi volcanic belt Wilson, Alexander M. 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_november_wilson_alexander.pdf [ 37.96MB ]
Metadata
JSON: 24-1.0383250.json
JSON-LD: 24-1.0383250-ld.json
RDF/XML (Pretty): 24-1.0383250-rdf.xml
RDF/JSON: 24-1.0383250-rdf.json
Turtle: 24-1.0383250-turtle.txt
N-Triples: 24-1.0383250-rdf-ntriples.txt
Original Record: 24-1.0383250-source.json
Full Text
24-1.0383250-fulltext.txt
Citation
24-1.0383250.ris

Full Text

GLACIOVOLCANISM IN THE GARIBALDI VOLCANICBELTbyAlexander M. WilsonB.Sc., The University of Otago, 2012B.Mus., The University of Otago, 2012M.A., The University of Wales, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDoctor of PhilosophyinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Geological Sciences)The University of British Columbia(Vancouver)October 2019© Alexander M. Wilson, 2019The following individuals certify that they have read, and recommend to the Faculty ofGraduate and Postdoctoral Studies for acceptance, the thesis entitled:GLACIOVOLCANISM IN THE GARIBALDI VOLCANIC BELTsubmitted by Alexander M.Wilson in partial fulfillment of the requirements for the degreeof Doctor of Philosophy in Geological Sciences.Examining Committee:James K. Russell, University of British Columbia, Geological SciencesPrimary SupervisorBrent C. Ward, Simon Fraser University, Earth SciencesSupervisory Committee MemberStuart Sutherland, University of British Columbia, Geological SciencesUniversity ExaminerMichael A. Church, University of British Columbia, GeographyUniversity ExaminerJohn L. Smellie, University of Leicester, Geological SciencesExternal ExaminerAdditional Supervisory Committee Members:Greg M. Dipple, University of British Columbia, Geological SciencesSupervisory Committee MemberiiAbstractThis thesis investigates glaciovolcanism in the Garibaldi volcanic belt (GVB) of south-western British Columbia (SWBC), Canada. Field observations and modelling are used toinvestigate: i) the paleoenvironmental implications of glaciovolcanism in SWBC, ii) vol-canic eruption processes in glaciovolcanic environments, and iii) causal linkages betweenthe volcanic eruptions and the growth and decay of terrestrial ice masses. Throughout thePleistocene, the GVB has experienced multiple alpine and continental glaciations. TheGVB comprises >100 Quaternary volcanoes and much of the character of this volcanismis ascribed to the range of magma compositions (alkaline basalt to rhyolite), to the extremerelief of the landscape, and to interactions with ice. The three key findings are: i) The GVBvolcanoes are established as a powerful proxy for the local paleoenvironment. The volcanicdeposits are used in conjunction with a geometric model for mountain glacier growth andretreat to inform on the presence, thickness, and transient properties of the Cordilleran IceSheet over the last 1 Ma. ii) Studies of two effusive glaciovolcanoes (the Lillooet Glacierbasalts, and the Table) show that eruption style and deposit morphology are strongly influ-enced by the nature of heat exchange between the erupted lava and the ice. Specifically,meltwater drainage attending eruptions exerts a critical control on eruptive behaviour (i.e.,dictating the ephemeral presence of an englacial lake). Lava-dominated tuyas, may beconstructed from eruptions involving within-ice dike injection, steep, well-drained bedrocktopography and endogenous, englacial inflation of the massif. iii) Transient growth anddecay of terrestrial ice masses can influence the timing, size and distribution of eruptions.Specifically, glacier-induced deformation of topography may impart local, shallow crustalstresses which influence eruption frequency, eruption size and vent distribution, dependingon the rheology of the bedrock and the geometry of the topography. At the scale of thecrust, transient loading and unloading of ice sheets may act as a glacial pump, bending thecrust downwards during loading (causing a suppression in eruptions) and allowing the crustto rebound during unloading (causing an increase in eruptions).iiiLay SummaryGlaciovolcanism is the interaction between volcanoes and glaciers and is well expressed byvolcanoes in southwestern British Columbia, Canada. In this thesis, the volcanoes are usedto: i) Recover the thickness and extent of continental-scale ice sheets that have waxed andwaned over SWBC over the past 1 million years. ii) Investigate processes that control theeruption dynamics (e.g., the eruption environment and the nature of heat exchange betweenthe cooling lava and the surrounding ice). iii) Investigate the effect that waxing and waningglaciers may have on the timing, size and eruption style of volcanoes (i.e., could meltingglaciers cause eruptions?). A key finding of this thesis is that melting ice sheets may causemore eruptions as magma trapped within the crust during glacial build-up is rapidly releasedto erupt.ivPrefaceThis thesis comprises an introduction, a conclusion and six research chapters. The latter arepresented in manuscript format for publication in peer-reviewed international geosciencejournals. Chapters 2, 3, 4 and 5 are published. Chapters 6 and 7 are draft manuscripts thatare prepared for submission to international geoscience journals. I am the lead author on allmanuscripts. All manuscripts are also co-authored by my supervisor, Dr. Kelly Russell whoprovided guidance and editorial work on this entire dissertation. Several of the manuscriptsalso include an additional co-author. I have included list of contributions from each ofthese co-authors below. The research program that resulted in this thesis was identified anddesigned by me and the main ideas behind the research are my own. However, they wouldnot have come about without the continuous informal dialogue that took place betweenmyself and my supervisor, Dr. Russell.All of the chapters in this thesis are designed as stand-alone manuscripts, however,they are linked by several common themes, including the use of glaciovolcanic deposits aspaleoenvironmental indicators, the dynamics that control glaciovolcanic eruption processes,and, the causal linkages between volcanic eruptions and climate-related growth and decayof terrestrial ice masses. The chapters each build each on each other, but, because of theoverlap, some background and introductory ideas are repeated throughout.A version of chapter 2 is published as: Wilson, A.M., Russell, J.K., 2018. Quaternaryglaciovolcanism in the Canadian Cascade volcanic arc – paleoenvironmental implications.Geological Society of America Special Papers. 2538, 133–157. I conducted the reviewof Canadian and United States volcanoes and constructed the geologic maps in ArcGIS®.Dr. Russell guided interpretations regarding the paleoenvironmnetal implications of the de-posits and their use for reconstructing the thickness of the Cordilleran Ice Sheet. I wrote themanuscript with editorial work by Dr. Russell. The manuscript was peer-reviewed by Dr.Melanie Kelman (Geological Survey of Canada), Dr. Mariek Schmidt (Brock University),and Dr. Anita Grunder (Oregon State University). Integral to this thesis, and this chapterin particular, were two seasons of geological field mapping during the summers of 2015and 2016. During mapping, I made field observations at 54 of the ∼100 volcanoes in thevGaribaldi volcanic belt (GVB) (see Tables 2.1, S1 and S4). I produced a sample archivecontaining ∼530 field samples (see Table S1), ∼120 petrographic thin section descriptions(see Table S2) and ∼100 geochemical whole-rock and trace element analyses (see TableS3). Detailed field maps were created for 28 of the volcanoes. These maps incorporateextended lithological unit descriptions and facies analysis, aided by petrographic and geo-chemical data and a comprehensive collection of field photographs. The field maps havebeen divided into 9 map sheets which are currently accepted, following peer-review by Dr.Bram van Straaten, as OpenFile publications with the British Columbia Geological Survey,Victoria. The maps will be published as: Wilson, A.M., and Russell, J.K. 2020 Glaciovol-canism in the Garibaldi volcanic belt: nine geologic maps from southern British Columbia,Canada. Ministry of Energy and Mines, British Columbia Geological Survey, OpenFile2020-X, sheets 1 to 9.A version of chapter 3 is published as: Wilson, A.M., Russell, J.K., Ward, B., 2019.Paleo-glacier reconstruction in southwestern British Columbia, Canada: A glaciovolcanicmodel. Quaternary Science Reviews 218, 178–188. I conceptualised the geometric model,wrote the code in MATLAB®, and performed the analysis, guided by Dr. Russell. Dr.Brent Ward (Simon Fraser University) provided expertise regarding the implications of thegalciovolcanic deposits with the framework of Quaternary deposits within southwesternBritish Columbia. The manuscript was written by me with editorial work provided by Dr.Russell and Dr. Ward. The manuscript was reviewed by two anonymous peer-reviewers.A version of chapter 4 is published as: Wilson, A.M., Russell, J.K., 2017. LillooetGlacier basalts, southwestern British Columbia, Canada: products of Quaternary glaciovol-canism. Candian Journal of Earth Sciences. 54, 639–653. I conducted the field mapping,the lithostratigraphic, geochemical, petrographic analysis and the petrologic modelling,supported and guided by Dr. Russell. Dr. Mati Raudsepp (university of British Columbia)assisted in operation of the Cameca SX50 Scanning Electron Microprobe (EMP) at the Uni-versity of British Columbia, Canada. Discussions with Drs. John Clague, Brent Ward andGlyn Williams-Jones (Simon Fraser University) contributed during the analysis. I wrote themanuscript with editorial work by Dr. Russell. The manuscript was peer-reviewed by Dr.John Smellie (University of Leicester) and Dr. Benjamin Edwards (Dickinson College).A version of chapter 5 is published as: Wilson, A. M., Russell, J. K., and Quane, S. L.,2019. The Table, a flat-topped volcano in southern British Columbia: Revisited. AmericanJournal of Science. 319(1), 44–73. I conducted the field mapping and lithostratigraphicanalysis and photogrammetric modelling of the massif. I conceptualised the physical erup-tion model for emplacement and performed the thermodynamic calculations, guided by Dr.Russell. Dr. Steve Quane (Quest University) assisted with field analysis and provided fi-nancial support for helicopter-based data collection. I wrote the manuscript with editorialviwork provided by Dr. Russll and Dr. Quane. The manuscript was peer-reviewed by Dr.John Smellie, Dr. Magnu´s Guðmundsson (University of Iceland) and Dr. Thomas Sisson(United States Geological Survey).A version of chapter 6 is prepared for submission to a peer-reviewed geoscience jour-nal. The analogue experiments were conducted at the Laboratoire Magmas et Volcans,Clermont-Ferrand, France. The experiments were jointly conceptualised by me and Dr.Benjamin van Wyk de Vries (Universite´ Clermont Auvergne). I conducted the experiments,guided by Dr. van Wyk de Vries. I also wrote the manuscript with editorial work by Dr.Russell and Dr. van Wyk de Vries.A version of chapter 7 is prepared for submission to a peer-reviewed geoscience journal.I conceptualised the model, wrote the code in MATLAB®, and performed the analysis,guided by Dr. Russell. I wrote the manuscript with editorial work provided by Dr. Russell.viiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiList of Supplementary Material . . . . . . . . . . . . . . . . . . . . . . . . . . . xviAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Importance of glaciovolcanism . . . . . . . . . . . . . . . . . . . . . . . . 21.1.1 Glaciovolcanoes as a proxy for ancient ice . . . . . . . . . . . . . . 21.1.2 Causal linkages between volcanic eruptions and glaciers . . . . . . 31.2 Research goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.3 Scope of the thesis and outline . . . . . . . . . . . . . . . . . . . . . . . . 41.3.1 Paleoenvironmental implications of glaciovolcanism in the Garibaldivolcanic belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3.2 Physical controls on eruption dynamics of glaciovolcanoes in theGaribaldi volcanic belt . . . . . . . . . . . . . . . . . . . . . . . . 61.3.3 Climate triggers for volcanism . . . . . . . . . . . . . . . . . . . . 72 Quaternary glaciovolcanism in the Canadian Cascade volcanic arc — pale-oenvironmental implications . . . . . . . . . . . . . . . . . . . . . . . . . . . 9viii2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.2 Historical review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.1 The US Cascade volcanic arc . . . . . . . . . . . . . . . . . . . . 202.2.2 The Canadian Cascade volcanic arc . . . . . . . . . . . . . . . . . 212.3 Glaciovolcanism in the Garibaldi volcanic belt . . . . . . . . . . . . . . . . 222.3.1 Mount Garibaldi volcanic field . . . . . . . . . . . . . . . . . . . . 232.3.2 Garibaldi Lake volcanic field . . . . . . . . . . . . . . . . . . . . . 242.3.3 The Mount Cayley volcanic field . . . . . . . . . . . . . . . . . . . 292.3.4 The Mount Meager volcanic field . . . . . . . . . . . . . . . . . . 312.3.5 Salal Glacier volcanic field . . . . . . . . . . . . . . . . . . . . . . 312.3.6 Bridge River volcanic field . . . . . . . . . . . . . . . . . . . . . . 332.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.4.1 Diversity of glaciovolcanism in the Garibaldi volcanic belt . . . . . 342.4.2 Forensic recovery of paleoenvironment . . . . . . . . . . . . . . . 362.4.3 Paleo-ice height reconstruction and correlation with the Marine Iso-tope Stages timescale . . . . . . . . . . . . . . . . . . . . . . . . . 422.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493 Paleo-glacier reconstruction in southwestern British Columbia, Canada: Aglaciovolcanic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503.2 Glaciovolcanism in the Garibaldi volcanic belt . . . . . . . . . . . . . . . . 523.3 Glaciovolcanoes as recorders of paleoenvironment . . . . . . . . . . . . . . 523.4 Geometric model for paleo-glacier distributions . . . . . . . . . . . . . . . 533.4.1 Geometric model methods . . . . . . . . . . . . . . . . . . . . . . 553.4.2 Model caveats and assumptions . . . . . . . . . . . . . . . . . . . 593.5 Modelled paleo-glacier distributions . . . . . . . . . . . . . . . . . . . . . 633.6 Paleo-glacier distributions constrained by the glaciovolcanoes . . . . . . . 643.6.1 Ring Mountain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.6.2 The Cheakamus basalts . . . . . . . . . . . . . . . . . . . . . . . . 673.6.3 Tuber Hill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673.7 Challenges faced by paleo-glacier reconstruction using glaciovolcanoes . . 693.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 704 Lillooet Glacier basalts, southwestern British Columbia, Canada: Productsof Quaternary glaciovolcanism . . . . . . . . . . . . . . . . . . . . . . . . . . 724.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72ix4.2 Geological setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 744.3 Physical Volcanology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754.3.1 Lithofacies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 764.4 Petrography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804.5 Geochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 824.5.1 Whole-rock chemical compositions . . . . . . . . . . . . . . . . . 824.5.2 Mineral and glass compositions . . . . . . . . . . . . . . . . . . . 854.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 894.6.1 Origin of chemical variations within the lavas and glasses . . . . . 894.6.2 Physical–chemical conditions of storage, transport, and eruption . . 904.6.3 Glaciovolcanic origin of Lillooet Glacier lithofacies . . . . . . . . 954.6.4 Model for emplacement and implications for paleo-ice thickness . . 974.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995 The Table, a flat-topped volcano in southern British Columbia: Revisited . . 1005.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1005.2 Geographic setting and context . . . . . . . . . . . . . . . . . . . . . . . . 1015.3 Lava-dominated tuyas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.4 Previous studies of the Table . . . . . . . . . . . . . . . . . . . . . . . . . 1035.4.1 Mathews (1951) . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.4.2 Other literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045.5 Field observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055.5.1 Geomorphology and surface morphologies . . . . . . . . . . . . . 1055.5.2 Lithofacies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1085.5.3 Cooling fractures and jointing . . . . . . . . . . . . . . . . . . . . 1095.5.4 Geochronology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1125.6 Summary of field observations . . . . . . . . . . . . . . . . . . . . . . . . 1125.7 Conceptual model for emplacement . . . . . . . . . . . . . . . . . . . . . 1195.8 Analytical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205.8.1 Dike injection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205.8.2 Ice melting and meltwater drainage . . . . . . . . . . . . . . . . . 1225.8.3 Heat transfer calorimetry . . . . . . . . . . . . . . . . . . . . . . . 1255.8.4 Steady-state heat transfer across a breccia carapace . . . . . . . . . 1285.8.5 Efficiency of heat transfer . . . . . . . . . . . . . . . . . . . . . . 1315.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132x6 Climate Change, distressed mountains and volcanic eruptions: Analoguemodels of topographic and lithologic volcano–ice coupling . . . . . . . . . . . 1346.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1346.2 Analogue model setup and scaling . . . . . . . . . . . . . . . . . . . . . . 1366.2.1 The effect of geometry (i.e., the size and shape of the valleys andmountains) and dimensional scaling . . . . . . . . . . . . . . . . . 1376.2.2 The effect of rheology (i.e., effective rock strength) and temporalscaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1396.3 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1416.3.1 Initial ridge contraction . . . . . . . . . . . . . . . . . . . . . . . . 1436.3.2 Initial ridge extension . . . . . . . . . . . . . . . . . . . . . . . . 1436.4 Discussion: Topographic and lithologic volcano–ice coupling . . . . . . . . 1446.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1507 Glacial pumping of a magma-charged lithosphere . . . . . . . . . . . . . . . 1527.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1527.2 Causal linkages between arc volcanoes and glaciers . . . . . . . . . . . . . 1547.3 The Garibaldi volcanic belt, British Columbia, Canada . . . . . . . . . . . 1567.4 Monte Carlo simulations of glacial loading and unloading as a forcing func-tion for magma stall or rise . . . . . . . . . . . . . . . . . . . . . . . . . . 1577.4.1 Isostatic lithospheric plate flexure . . . . . . . . . . . . . . . . . . 1597.4.2 Crustal dike migration . . . . . . . . . . . . . . . . . . . . . . . . 1617.5 Model results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1627.6 Discussion: Glacial pumping of a magma-charged lithosphere . . . . . . . 1667.6.1 Eruption implications of glacial pumping . . . . . . . . . . . . . . 1687.6.2 Glacial asymmetry and the timing and magnitude of volcanic response1697.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1708 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1728.1 Paleoenvironmental implications of glaciovolcanism in the Garibaldi vol-canic belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1728.2 Physical controls on eruption dynamics of glaciovolcanoes in the Garibaldivolcanic belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1748.3 Climate triggers for volcanism: Causal linkages between volcanoes andglaciers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179xiList of TablesTable 2.1 Compilation of Quaternary volcanic deposits in the Garibaldi volcanic belt 15Table 2.2 Quaternary volcanic rocks in the Garibaldi volcanic belt that constrainthe spatial-temporal distribution of the Cordilleran Ice Sheet . . . . . . . 44Table 4.1 Whole-rock chemical compositions of Lillooet Glacier volcanic mapunits measured by X-ray fluorescence and calculated properties includ-ing DI, SI, and normative mineralogy. . . . . . . . . . . . . . . . . . . . 84Table 4.2 Trace and rare earth element contents of Lillooet Glacier lithofaciesmeasured by XRF and ICP–MS, respectively. . . . . . . . . . . . . . . . 87Table 4.3 Electron microprobe (EMP) major element analyses of Lillooet Glaciervolcanic glasses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88Table 5.1 Constants used in analytical models at the Table. . . . . . . . . . . . . . 121Table 5.2 Variables used in analytical models at the Table. . . . . . . . . . . . . . 122Table 6.1 Experimental parameters and measured vertical and horizontal displace-ments in all analogue experiments. . . . . . . . . . . . . . . . . . . . . 142xiiList of FiguresFigure 2.1 Map showing location and tectonic setting of the Cascade volcanic arcin western North America . . . . . . . . . . . . . . . . . . . . . . . . 10Figure 2.2 Maps showing distribution and type of Quaternary volcanic rocks in thesouthern Garibaldi volcanic belt . . . . . . . . . . . . . . . . . . . . . 13Figure 2.3 Maps showing distribution and type of Quaternary volcanic rocks in thenorthern Garibaldi volcanic belt . . . . . . . . . . . . . . . . . . . . . 29Figure 2.4 Field photographs showing lava-dominated tuyas in the Garibaldi vol-canic belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Figure 2.5 Field photographs showing ice-impounded lavas in the Garibaldi vol-canic belt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Figure 2.6 Field photographs showing the passage zone preserved at the Salal tindar 41Figure 2.7 Mid–Late Pleistocene ice-height reconstruction for southwest BC usingthe record afforded by the glaciovolcanic deposits . . . . . . . . . . . . 46Figure 3.1 Location and physiographic setting of southern British Columbia,Canada. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51Figure 3.2 Field photographs of volcanic edifices used to drive geometric ice dis-tribution scenarios. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54Figure 3.3 Cross-section diagram illustrating geometric cone method for growthand decay of glaciers in 2-D. . . . . . . . . . . . . . . . . . . . . . . . 57Figure 3.4 Block diagrams of an arbitrary example glacier (the Job Glacier, MountMeager), illustrating the geometric method in 3-D. . . . . . . . . . . . 59Figure 3.5 Cell-by-cell comparison of modelled present-day ice distribution ver-sus Global Land Ice Measurements from Space (GLIMS) inventory ofglaciers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Figure 3.6 Modelled geometric ice distributions corresponding to the phase 2 (”in-tense alpine”) and phase 3 (”mountain ice sheet”) glaciations. . . . . . 63Figure 3.7 Cross section showing surface profiles of modelled glaciers. . . . . . . 64xiiiFigure 3.8 Block diagrams illustrating the local and regional paleo-glacier impli-cations for three glaciovolcanoes. . . . . . . . . . . . . . . . . . . . . 67Figure 3.9 The Marine Isotope Stages (MIS) δ O18 record for the last 700 ka . . . 68Figure 4.1 Location of the Lillooet Glacier volcanic deposits. . . . . . . . . . . . 74Figure 4.2 Geological map of the Lillooet Glacier volcanic deposits. . . . . . . . . 75Figure 4.3 Cross-sections and graphical logs of the Lillooet Glacier volcanic de-posits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Figure 4.4 Field photographs of Lillooet Glacier volcanic deposits. . . . . . . . . 78Figure 4.5 Field photographs of pillow basalt and tubes feeding pillow basalt. . . . 79Figure 4.6 Field photograph and photomicrograph of hyaloclastite lithofacies. . . . 81Figure 4.7 Photomicrographs of Lillooet Glacier basaltic rocks in cross polarizedlight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82Figure 4.8 Major and trace element geochemistry of Lillooet Glacier lava and glass. 86Figure 4.9 Electron microprobe analyses of Lillooet Glacier mineral compositions. 92Figure 4.10 Pearce element ratio (PER) diagrams showing chemical variations inwhole-rock and glass analyses of Lillooet Glacier basalts. . . . . . . . 93Figure 4.11 Phase diagram for Lillooet Glacier magma as a function of pressure andwater content calculated using rhyolite-MELTS. . . . . . . . . . . . . . 94Figure 4.12 Phase diagram for Lillooet Glacier magma as a function of pressure andwater content calculated using rhyolite-MELTS. . . . . . . . . . . . . . 96Figure 4.13 Schematic sequence model of the Lillooet Glacier volcanic eruption. . . 98Figure 5.1 Location and geology of the Table, southwestern British Columbia,Canada. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102Figure 5.2 Field photographs showing morphology of the Table edifice. . . . . . . 105Figure 5.3 3-D photogrammetry model of the Table. . . . . . . . . . . . . . . . . 107Figure 5.4 High-resolution field photographs showing subvertical embayment andprotrusion structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . 108Figure 5.5 Line diagram of orthorectified image that was constructed using 3-Dphotogrammetry showing the distribution of the lithofacies. . . . . . . 110Figure 5.6 Field photographs showing lithofacies and fracture characteristics at theTable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111Figure 5.7 Field photograph and line diagram showing the distribution and orien-tation of coarsely-jointed layers (dikes and sills) on the northern side ofthe Table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113xivFigure 5.8 Summary of 40Ar/39Ar geochronology results for sample AW-15-139(unit T3). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Figure 5.9 Conceptual physical model for dike-injection and endogenous growthat the Table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119Figure 5.10 Results of time-dependent heat transfer calculations. . . . . . . . . . . 124Figure 5.11 Conceptual one-dimensional heat transfer model from lava to ice at theend of the eruption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129Figure 5.12 Two methods for determining the average thickness (L) of breccia cara-pace surrounding the Table at the end of the eruption. . . . . . . . . . . 130Figure 6.1 Geologic context and experimental setup showing valleys excavated insand-plaster mixture and silicone putty. . . . . . . . . . . . . . . . . . 138Figure 6.2 Measured horizontal displacements for all experiments using particleimaging velocimetry (PIV). . . . . . . . . . . . . . . . . . . . . . . . 140Figure 6.3 Dimensionless diagrams showing cumulative vertical and horizontalstrain for all experiments. . . . . . . . . . . . . . . . . . . . . . . . . . 144Figure 6.4 Representative experiments showing ridge compression and ridge ex-tension. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145Figure 6.5 Generalized conceptual model for climate-induced glacier loading/un-loading affecting volcanic eruptions. . . . . . . . . . . . . . . . . . . . 148Figure 7.1 Location and physiographic setting of the Garibaldi volcanic belt insouthern British Columbia, Canada and the last Cordilleran Ice Sheet. . 157Figure 7.2 Diagram (not to scale) showing cylindrical thin-plate bending methodand variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158Figure 7.3 Model loading curve of maximum crustal displacement and calculateddeviatoric stresses developed through the bent plate. . . . . . . . . . . 160Figure 7.4 Cartoon illustrating key parameters controlling the Monte Carlo simu-lation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163Figure 7.5 Randomly-sampled uniform distribution of input dikes and change inmean number of eruptions. . . . . . . . . . . . . . . . . . . . . . . . 164Figure 7.6 Model results over 40 ka loading/unloading cycle. . . . . . . . . . . . . 165Figure 7.7 Conceptual model of glacial pumping of a magma-charged crustallithosphere. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167xvList of Supplementary MaterialTable S1 Archive of all samples collected from the Garibaldi volcanic belt.Table S2 Archive of petrographic thin sections and descriptions of rocks col-lected from the Garibaldi volcanic belt.Table S3 Whole-rock and trace-element geochemistry of samples collected fromthe Garibaldi volcanic belt.Table S4 Compilation of Quaternary volcanic deposits in the Garibaldi volcanicbelt.Table S5 Helium pycnometry connected porosity measurements and propagateduncertainty.Table S6 ElectronMicroprobe (EMP) major element analyses (wt.%) of LillooetGlacier single crystal, glomerophenocryst, sieve-textured xenocryst,gabbroic xenolith and groundmass mineral phases.Table S7 ElectronMicroprobe (EMP) major element analyses (wt.%) of LillooetGlacier volcanic glass.xviAcknowledgmentsI would first like to thank my supervisor, Kelly Russell, for investing your time, energy andfinances supporting me over the past four and a half years. Your enthusiasm for new ideasis endless and your passion for volcanology is inspiring. It has been a privilege to be partof your laboratory.I would like to thank my field assistants, Nikolas Matysek and Nate Willett, whoworked tirelessly collecting samples and mapping the remote corners of southwesternBritish Columbia with me. I would also like to thank Mila Huebsch and Lucy Porritt foryour assistance in the field and the lab.A very appreciative thanks to all of my lab mates over the past few years: Ryan, Dave,Chuck, Luke, Marie, Rose, Mila, Nate, Nick, Mu and David, you’ve all helped make mytime at UBC some of the best years in my life. A special thanks to Amy, I am grateful forthe support, friendship and discussions we’ve had over the years.I am indebted to the assistance provided by British Columbia Parks, in particular, KatyChambers. I thank Tony Cailes at the Black Tusk snowmobile club, Murray Sovereign atValhalla sports, Reid in Goldbridge, and the great pilots at Blackcomb Aviation, Black TuskAviation and Nelson Helicopters.I would like to thank Glyn Williams-Jones, John Clague and Brent Ward at SimonFraser University, and Melanie Kelman at the Geologic Survey of Canada for the excellentdiscussions and support over the years.I am grateful to Benjamin van Wyk de Vries and the Laboratoire Magmas et Volcansgroup in France for welcoming me on two occasions to conduct experiments. Swetha,Valentine and Anne, thank you for making my trips to France so special.I would like to thank Steve Quane at Quest University Canada for being a constantsource of quality field support and friendship.I acknowledge manuscript reviewers: Drs. John Smellie, Magnu´s Guðmundsson, BenEdwards, Catherine Hickson, Melanie Kelman, Mariek Schmidt, Anita Grunder, ThomasSisson and Barb Duthrow. I am also grateful to Bram van Straaten and Larry Aspler at theBritish Columbia Geological Survey for critical comments on my geological maps.xviiI gratefully acknowledge funding for my Ph.D. project. For the most part, field supportand geochemical, petrographic and geochronological analyses were funded by a CanadianNatural Sciences and Engineering Research Council of Canada (NSERC) Discovery grantawarded to Dr. Russell (RGPIN-2018-03841). My salary was funded by an NSERC CanadaGraduate Scholarships Doctoral (CGSD) grant (PGSD3–489731–2016). Field work di-rectly contributing to this thesis (in particular, Chapters 2, 4 and 5) was also supported by aGeological Society of America (GSA) Jackson Award, a Thomas and Marguerite MacKayMemorial Scholarship awarded by the University of British Columbia, and an NSERC En-gage grant (EG 499046 2016) awarded to Dr. Russell. Travel and living expenses to supportlaboratory analysis (chapter 6) was supported by a Canada Graduate Scholarships –MichaelSmith Foreign Study Supplement (CGS-MSFSS) (6581-2016W).I would also like to thank Dominique Weis and Marina Martendale (University ofBritish Columbia) and Melanie Kelman (Geological Survey of Canada) for their gener-ous sharing of newly acquired, unpublished geochronological data (see Chapter 2). I alsothank Janet Gibites, Jeff Benowitz and Brian Jicha for your valiant attempts at dating thevolcanoes of southwestern BC.Finally, I would like to thank my friends and family, especially my mum, dad, sisterand brother, who have supported me in my various life pursuits. A special thank you to mypartner and co-Ph.D. buddy, Olenka Forde. Your support over the years has been incredible.I am so lucky to have shared this experience with you just down the hall.xviiiChapter 1IntroductionGlaciovolcanism is the interaction between volcanoes and ice in all of its forms, includingsnow, firn and meltwater (Kelman et al., 2002a; Russell et al., 2014; Smellie et al., 2006).Glaciovolcanic eruptions involve unique eruption processes which produce a diverse arrayof distinctive edifices and deposit lithofacies (Edwards et al., 2015a; Gudmundsson et al.,2008; Smellie, 2018; Smellie and Edwards, 2016; Smellie et al., 2013). The physical prop-erties of glaciovolcanoes (e.g. their shape, form and lithofacies architecture) directly reflecttheir eruption environment, which results in confinement or impoundment by ice, as well asaccelerated cooling by contact with ice, snow or melt water (Edwards et al., 2012; Kelmanet al., 2002a; Russell et al., 2014, 2007; Smellie et al., 2008, 2013; Wilson et al., 2013).Glaciovolcanoes are of interest to science and the public because they are a hazard tohuman life and economies (Gudmundsson et al., 2008, 2012; Smellie, 2018). However,these volcanoes also offer unique opportunities to reconstruct the ancient climate of theEarth (e.g., Edwards et al., 2011; Smellie et al., 2008, 2011) and to investigate possiblecausal linkages between the changing climate and past and future eruptions (e.g., Grove,1974; Jull and McKenzie, 1996; Maclennan et al., 2002; Mathews, 1958; Rampino et al.,1979; Rawson et al., 2016; Slater et al., 1998; Watt et al., 2013). This thesis investigatesglaciovolcanism in the Garibaldi volcanic belt (GVB) of southwestern British Columbia(BC), Canada. The GVB is a Quaternary volcanic province that has been subjected tonumerous continental and alpine glaciations over the past 2 Ma. During this time, the GVBhas been repeatedly enveloped by a large continental ice sheet, referred to as the CordilleranIce Sheet (CIS) (Blaise et al., 1990; Booth et al., 2003; Clague andWard, 2011; Fulton et al.,1986; Jackson and Clague, 1991; Ryder et al., 1991). Intermittent Quaternary volcanism,superimposed upon the waxing and waning of the glaciers has produced a large number ofedifices that show evidence for eruption in the presence of ice (Green et al., 1988; Kelmanet al., 2002a; Mathews, 1958; Russell et al., 2007).11.1 Importance of glaciovolcanismGlaciovolcanism is an emerging field of research that has seen exponential growth in thelast few decades (Russell et al., 2014). There are three main motivations for this growth.Firstly, glaciovolcanoes are a geological hazard. Glaciovolcanic eruptions present a risk tohuman life and cause significant regional and global economic and environmental disrup-tion (Gudmundsson et al., 2008, 2012; Hickson, 1994; Major and Newhall, 1989; Pagneuxet al., 2012). The 2010 eruption of Eyjafjallajo¨kull in Iceland, for example, produced dan-gerous, large-scale flooding by meltwater (jo¨kulhlaups) and emitted fine tephra (ash) thatdisrupted global air travel, causing an estimated CAD $2.1 billion in costs to the airlineindustry (Gudmundsson et al., 2012; Oddsson et al., 2016; Smellie, 2018).Secondly, glaciovolcanoes shape and reshape planetary surfaces (Head and Wilson,2002; Smellie et al., 2008, 2013; Wilson and Head, 2002; Wilson et al., 2013). The ge-ologic history of our planet, including areas such as Iceland, Antarctica, Canada, SouthAmerica and New Zealand cannot be fully understood without considering the role glacio-volcanoes play in landscape evolution (Cole et al., 2018; Conway et al., 2016; Edwards andRussell, 2002; Edwards et al., 2011; Lachowycz et al., 2015; Smellie, 2018; Smellie et al.,1993). Advances in remote sensing techniques have also called for improved terrestrialanalogues to support investigations into the presence of glaciovolcanoes on extraterrestrialplanets (namely Mars; e.g., Allen, 1979; Chapman and Tanaka, 2001, 2002; Head and Wil-son, 2002) in order to support understanding their geologic, climatic and biologic histories.Thirdly, the lithofacies and architecture of glaciovolcanoes provide a direct record ofinteractions with the cryosphere (Russell et al., 2014; Smellie, 2018). Where dated (typ-ically using radiometric techniques such as 40Ar/39Ar or K–Ar), glaciovolcanoes form anempirical paleoenvironmental database that can be used to reconstruct the paleo-presenceand properties of extinct glaciers. Although, historically, these data have been largely un-derutilized, glaciovolcanoes are now being rapidly developed as one of the most powerfuland most holistic methods for determining, quantitatively, a wide range of critical parame-ters of past ice (Smellie, 2018). These parameters include: the paleo-ice presence/absenceand age of the ice, ice thickness, ice distribution, the elevation of the coeval ice surface,and an estimate of the basal thermal regime of the ice (e.g., Kelman et al., 2002a; Mathews,1958; McGarvie et al., 2007; Smellie, 2018; Smellie and Edwards, 2016; Smellie et al.,2009; Smellie and Skilling, 1994; Smellie et al., 2013, 2018; Stevenson et al., 2009).1.1.1 Glaciovolcanoes as a proxy for ancient iceMost information regarding Earth’s ancient climate is established, by proxy, throughrecords of isotopic variation preserved in marine sedimentary sequences (e.g., the Marine2Isotope Stages record (MIS); Lisiecki and Raymo, 2005) or paleo-atmospheric composi-tions recovered from ice cores (e.g., The Greenland Ice Core Project (GRIP); Dansgaardet al., 1982; Stauffer, 1993). In general, most reconstructive ice sheet models (e.g., TheParallel Ice Sheet Model (PSIM); Winkelmann et al., 2011) rely on these proxies to drivethe simulations of past-glacier growth, decay and extent. By their nature, these proxiesprovide a globally-averaged record of the paleo-climate, thus, there is a need for robustempirical determinations of local paleo-glacier presence and properties in order to validatethe models (de Boer et al., 2017; Seguinot et al., 2016; Stokes et al., 2015). The terres-trial record provided by glaciovolcanoes is crucial for expanding and testing the integrity ofthese established and widely used paleo-climate proxies (Smellie, 2018).The nature of paleoenvironmental information recovered from glaciovolcanoes is influ-enced by the properties of the erupting magma (i.e., the composition, viscosity), the erup-tion rate and style (i.e., mass flux; explosive versus effusive), and, the environment (i.e.,the ice thickness, ice temperature and bedrock topography) (Edwards and Russell, 2002;Hickson, 2000; Russell et al., 2014; Smellie et al., 2006). Mafic glaciovolcanoes have beenstudied extensively in Iceland, Antarctica and northeastern British Columbia. Correspond-ingly, techniques used to recover paleoenvironment from these kinds of deposits (i.e., mafic)are advanced (e.g., Jakobsson and Gudmundsson, 2008; Jones, 1968, 1970; Mathews, 1947;Smellie et al., 2008; Smellie and Skilling, 1994; Smellie et al., 2013). Equivalent studies ofintermediate and felsic glaciovolcanoes (e.g., Cole et al., 2018; Lescinsky and Fink, 2000;McGarvie et al., 2007; Stevenson et al., 2011, 2006; Tuffen, 2007), especially those in areaswith extreme topography, are lacking (Kelman et al., 2002a). There is a need to thoroughlyinvestigate glaciovolcanoes of this type (i.e., intermediate–felsic) in order to understandtheir eruption processes and fully exploit their paleoenvironmental potential (e.g., The Cas-cade volcanic arc; Hickson, 1994; Hildreth, 2007).1.1.2 Causal linkages between volcanic eruptions and glaciersVolcanic eruptions can exert a major influence on global climates (e.g., Aubry et al., 2016;Baldini et al., 2015; Hansen et al., 1992; Kelly and Jones, 1996; Rampino et al., 1979;Sigl et al., 2015), however, it is unclear if the climate can also affect the rate, size anddistribution of eruptions. It has been suggested that changes in global climate (primarilythrough the growth and decay of continental ice sheets) may trigger volcanic eruptions(e.g., Grove, 1974; Hooper et al., 2011; Huybers and Langmuir, 2009; Nakada and Yokose,1992; Nyland et al., 2013; Sigvaldason et al., 1992; Watt et al., 2013). Uncovering thenature of these causal linkages will support improved eruption forecasting and mitigationof the hazards and costs associated with future eruptions (e.g., Roberti et al., 2018; Tuffen,32010). There is compelling evidence that surface unloading of glaciers in regions witha thin crustal lithosphere (e.g., Iceland) may promote magma production (decompressionmelting) in the mantle and therefore increase the frequency of eruptions at the surface (e.g.,Jull and McKenzie, 1996; Maclennan et al., 2002). However, the models for Iceland do notapply well to regions with a thick crust (e.g., continental volcanic arcs), where melting ismostly by dehydration processes (Grove et al., 2012). For these areas, linkages betweenglacier cycling and volcanic activity remain elusive (e.g., Jellinek et al., 2004; Kutterolfet al., 2013; Nakada and Yokose, 1992; Rawson et al., 2016; Watt et al., 2013).1.2 Research goalsThis thesis examines three broad issues: i) The paleoenvironmental significance of glacio-volcanism southwestern BC, Canada. ii) Glaciovolcanic eruption processes in regions ofextreme topography, and, iii) the causal linkages between volcanoes and the changing cli-mate in tectonic regions with a thick continental crust (e.g. continental volcanic arcs). Toaddress these issues, I investigate glaciovolcanoes in the Garibaldi volcanic belt (GVB) ofsouthwestern BC, Canada. This Quaternary volcanic province has >100 eruptive centres,most of which are <1 Ma in age (Green et al., 1988; Mathews, 1958; Souther, 1991). Morethan 50% of the volcanoes in the GVB interacted with ice during eruption (Andrews et al.,2014b; Hickson, 1994; Russell et al., 2007). The volcanic deposits are extremely diverse,reflecting a wide range of magma compositions, eruption styles and eruption environments(e.g., Mathews, 1951; Roddick and Souther, 1987). Much of the character of glaciovol-canism in the GVB is attributed to the wide range of magma compositions (alkaline basaltto rhyolite) and to the extreme relief of the landscape (Kelman et al., 2002a). Together,the volcanic rocks form a large, empirical dataset offering the opportunity to investigate aunique range of glaciovolcanic processes (e.g., Mathews, 1951) and informs on the pres-ence, thickness and transient properties of the CIS and local glaciers that have existed in thearea throughout the Pleistocene.1.3 Scope of the thesis and outlineI use the Pleistocene glaciovolcanoes of southwestern BC to answer three main questions:1. How can glaciovolcanoes in the GVB be used to reconstruct the presence, extent andtransient properties of the CIS over the past 1 Ma?2. How has the CIS helped to shape the pattern and nature of volcanism in the GVB?What physical eruption processes have led to the diversity of glaciovolcanism in the4GVB, and how do these processes influence the nature of paleoenvironmental infor-mation that is recorded?3. Is there a causal linkage between the volcanoes and climate-driven glacier loadingand unloading of continental volcanic arcs in space and time, and what is the natureof this connection?1.3.1 Paleoenvironmental implications of glaciovolcanism in the Garibaldivolcanic beltChapters 2 and 3 investigate the diversity and paleoenvironmental implications of glaciovol-canism in southwestern BC. The primary data for these chapters is derived from field map-ping paired with lithofacies identification and geochemical, petrographic and geochrono-logical analysis of the rocks.Chapter 2 provides an overview and database of all recognized volcanic (includingglaciovolcanic and non-glaciovolcanic) deposits in the GVB. In this chapter I summarizeall relevant literature concerning glaciovolcanism in the Cascade volcanic arc and compileexisting and new geochronological information of their ages. I discuss factors contributingto the wide volcanic diversity seen in the GVB and outline several methods (some of whichare unique to the volcanoes of SWBC) for extracting paleoenvironmental information fromthe deposits. Using these field constraints and geochronology, this chapter concludes witha simple reconstruction of the evolving surface elevation of the CIS through the past 1 Ma,derived entirely from glaciovolcanic data.Chapter 3 further explores the paleoenvironmental significance of glaciovolcanism inthe GVB. In this chapter, I recognise that, as a tool for paleo-glacier reconstruction, glacio-volcanoes are not well suited to constrain the absolute distribution of past glaciers. Toaddress this limitation, I develop a simple geometric model that extrapolates glaciers overmountainous regions using an assumption of the average surface angles of different types ofglaciers. The model relies on empirical data derived from three well-constrained glaciovol-canoes. Paleo-glaciers are constructed independently of climate forcing data (i.e., globally-averaged paleoenvironmental proxies). I use the model to build snapshots of the regionaldistribution and thickness of the CIS in southwestern BC at three times during the Pleis-tocene (at ∼50 ka, ∼140 ka and at ∼600 ka). I conclude this chapter with a discussion ofthe caveats and current limitations of using glaciovolcanism as a paleoenvironmental proxyand suggest that the lack of robust geochronology for GVB volcanoes presents the mostsignificant challenge in applying this model at a higher temporal resolution and over longertime periods.51.3.2 Physical controls on eruption dynamics of glaciovolcanoes in theGaribaldi volcanic beltMany of the glaciovolcanoes in the GVB are unique in terms of their lithofaces and mor-phology (Kelman et al., 2002a; Mathews, 1951, 1958). The second part of my thesis in-vestigates two edifices in detail (the Lillooet Glacier basalts and the Table). My goal inthese chapters is to explore the physical pre- and syn-eruptive processes responsible forthe eruptive behaviour and construction of these volcanoes. The two volcanoes contrastin terms of their magma composition (i.e., basalt versus andesite) and deposit style (i.e.,pillow-dominated tinder versus lava-dominated tuya). However, they are also similar; bothwere erupted effusively, beneath similar thicknesses of ice, and in similar topographic envi-ronments (i.e., high-relief bedrock). The detailed analyses of these two deposits are dividedinto two chapters.Chapter 4 gives a petrological and geochemical study of the Lillooet Glacier basalts,a package of basaltic pillow lava and hyaloclastite erupted into a sustained englacial lakeduring the Last Glacial Maximum (LGM). This chapter includes a detailed discussion ofthe geochemistry and petrographic character of the lavas and develops a geochemical andthermodynamic phase equilibria model using Pearce Element Ratio analysis and Rhyolite-MELTS. The model establishes co-magmatism in the erupted suite of lavas and a shallowpre-eruptive crustal storage zone (< 7.5 km).Chapter 5 explores the nature of intermediate volcanism in the GVB (i.e., andesitic),through a rare kind of glaciovolcano, a lava-dominated tuya. This study re-examines theTable, a unique and enigmatic glaciovolcano first described by Mathews (1951). I usefield mapping, 3-D photogrammetry, lithofacies analysis and 40Ar/39Ar geochronology tore-examine the emplacement origins of this unique edifice and to develop a new physicalmodel for the eruption involving within-ice dike injection and endogenous inflation of themassif. Field mapping indicates that the edifice is preserved near to its original form (i.e.,there has been only minor erosion since emplacement). I use the geometry of the massifto model the thermal exchange between the erupted lava and the enclosing paleo-glacialenvironment. This heat-transfer model suggests that the efficiency of heat transfer betweenthe lava and the ice at intermediate lava-dominated tuyas is extremely low when comparedwith all other glaciovolcanic systems (e.g., Gja´lp, Iceland). The low heat-transfer efficiencyis attributed to the eruption dynamics (i.e., dike injection and endogenous inflation), theeruption environment (i.e., a well-drained subglacial eruption cavity) and the nature of heatexchange across a thick carapace of insulating quench breccia.61.3.3 Climate triggers for volcanismChapters 6 and 7 examine the causal linkages between volcanoes and glaciers, specifically,volcanoes situated in thick-crusted continental arcs. These chapters use a combination ofanalogue models and numerical simulations to show that transient changes in crustal stressinduced by glacier loading/unloading can affect storage and transport of magmas differ-ently; at the shallow-crustal scale (i.e., within the first few km of the surface) and at thedeep-crustal scale (i.e., within 35 km of the surface).Chapter 6 investigates the effects of surface topography and the load induced by small,valley-filling glaciers on near-surface (i.e., within several km of the surface) transport andcapture of magma in the crust. I use analogue, sand-box models to track crustal deformationin response to glacier unloading and determine the effect on shallow stress fields, magmamigration, storage and eruptions. These experiments suggest that post-glacial mountaincompression is probably the most common result of climate warming and leads to ridge/-mountain compression and less, but more explosive volcanism after deglaciation. However,crustal lithology also plays a key role; in mountains where soft rocks can flow, the effect isreversed.Chapter 7 investigates how transient stress changes at the scale of the entire continen-tal crust (i.e., up to 35 km-thick) may modulate the transport and storage of magma to thesurface. I develop an idea initially proposed by Nakada and Yokose (1992), wherein load-ing/unloading of large, continental-scale ice sheets may cause down-warping (i.e., bending)of the crust during glaciation and elastic rebound of the crust following glacial retreat. Thiselastic flexure induces compressive and extensional deviatoric stress throughout the crust ata magnitude that could affect dike propagation, storage and eruptions. I model elastic defor-mation associated with the surface mass redistribution of ice and track the transient stresschanges that occur throughout the crust. To test the effects of transient stress change on dikemigration, I develop a stochastic, Monte-Carlo simulation that traces the migration of mag-mas through the crust during bending. The model shows that loading and unloading of icesheets may act as a glacial pump modulating volcanic eruptions; the crust is charged withmagma during loading (causing an overall suppression in eruptions) and then evacuated bytheir ascent and eruption during unloading (causing an overall increase in eruptions).The summary and conclusions chapter (see chapter 8) explores the wider implicationsof each of the themes addressed above and places them in a global context. Specifically,chapter 8 first discusses the major advantages and disadvantages of using glaciovolcanoesas a paleoenvironmental proxy, the future challenges that will be faced by this approachand where methods developed in the GVB may be applied elsewhere in the world. Next,chapter 8 compares the eruption dynamics involved in wet, versus the well-drained (dry)7glaciovolcanic eruptions (e.g., the Lillooet Glacier basalts versus the Table) and exploresthe relative importance of an englacial lake in constructing a lava-dominated tuya versusan unconstrained pillow mound. Finally, chapter 8 compares the results of analogue andnumerical models presented in chapters 6 and 7. I explore the relative importance of shallowcrustal versus deep crustal transient stress field changes in controlling eruption frequencyand distribution of volcanic eruptions and provide suggestions for future applications of thiswork.8Chapter 2Quaternary glaciovolcanism in theCanadian Cascade volcanic arc —paleoenvironmental implications12.1 IntroductionGlaciovolcanism encompasses the interaction of volcanism with ice in all of its forms, in-cluding by implication, any meltwater that is created during volcanic heating (Kelman et al.,2002a; Russell et al., 2014; Smellie, 2006; Smellie and Edwards, 2016). Volcanic eruptionsthat interact with the cryosphere (i.e., snow, ice, firn, permafrost and meltwater) display adiverse range of eruption styles that produce diagnostic edifice morphologies and character-istic deposit lithofacies (e.g., Edwards and Russell, 2002; Furnes et al., 1980; Jones, 1968,1970; Mathews, 1947; Russell et al., 2014, 2013; Smellie, 2013; Smellie and Edwards,2016; Smellie et al., 2008). Many physical aspects of glaciovolcanoes directly reflect theireruption environment, which involves physical confinement (impoundment) by ice, as wellas hydroclastic fragmentation and accelerated cooling by contact with ice, snow or melt-water (Edwards et al., 2011; Kelman et al., 2002a; Mathews, 1952a; Russell et al., 2014;Smellie, 2013; Tuffen, 2007; Watton et al., 2013). Glaciovolcanic deposits can indicate thepaleo-presence, distribution and thickness of ancient ice, and also, where dated, the timing(Guillou et al., 2010; McGarvie et al., 2007; Smellie, 2013; Smellie and Edwards, 2016;Smellie et al., 2009, 1993, 2008, 2011). Glaciovolcanoes, therefore, provide a means oftracking the waxing and waning of ancient ice sheets through space and time.1A version of this chapter has been published. Wilson, A.M., Russell, J.K., 2018. Quaternary glaciovolcan-ism in the Canadian Cascade volcanic arc – paleoenvironmental implications. Geological Society of AmericaSpecial Papers 2538, 133–157. https://doi.org/10.1130/2018.2538(06).9SilverthroneBridge RiverMount MeagerMount GaribaldiMountBakerGlacierPeakMountRainierMountAdamsMount St.HelensMountHoodMountJeffersonNewberrycalderaMount Mazama(Crater Lake)MountShastaMedicineLake volcanoMountLassenFranklin GlacierGaribaldi LakeMount CayleyGoatRocksThree Sisters & Little BrotherBrokenTopCappyMountainRainbowMountainSnowMountainMaidu volcaniccenterDittmarvolcanic centerSalal GlacierLone Butte & Crazy HillsHayrick Butte & Hogg Rock120° W125° W125° W130° W130° W50° N50° N45° N45° N40° NKulshanCalderaMattheiu Lake Fissure SystemKokostick ButteQuaternary Volcanic CentersGlaciovolcanic Non-glaciovolcanic0 10050KilometersDistribution of Quaternary volcanic rocksIce Extent (LGM ~ 17 ka)Juan de Fuca PlateGorda PlateExplorer PlatePacific PlateCanadaUSVancouver IslandBritish ColumbiaWashingtonOregonCaliforniaFigure 2.1: Map showing location and tectonic setting of the Cascade volcanic arcin western North America. The Garibaldi volcanic belt (GVB) is the northern(Canadian) portion of the arc. The distribution of Quaternary volcanic rocks,and locations of major volcanic centres are modified from Hildreth (2007), andthe locations of volcanic and glaciovolcanic occurrences are compiled from theliterature. The stippled section outlines the maximum extent of the most recent(Fraser) Cordilleran ice sheet (CIS) at approximately 17–14 ka (after Booth et al.,2003).The Cascade volcanic arc is a chain of volcanoes extending ∼1250 km from south-west British Columbia (BC), Canada to northern California in the United States (US) (Fig-ure 2.1). Quaternary magmatism was a response to northeasterly subduction of remnants ofthe Farallon plate (Juan de Fuca, Explorer and Gorda plates) beneath the North Americanplate (Hildreth, 2007; Mullen et al., 2017; Rogers, 1985). Volcanism has produced morethan 2300 individual vents, of which 22 are major volcanic structures (Hildreth, 2007). Thevolcanoes define a band of predominantly calc-alkaline, intermediate composition strato-volcanoes that are situated ∼200 km inland from the coast (Wood and Kienle, 1990).10Glaciovolcanism in the Cascade volcanic arc occurs in two main forms. The firstinvolves small-scale interactions between volcanoes and well-drained, permeable alpineglaciers, and/or seasonal snowpack. Most of the major stratovolcanoes in the Cascade vol-canic arc are, or have until recently been, snow-clad. Consequently, they are increasinglyrecognized for their hazard potential involving large-volume lahars and floods (Driedgerand Kennard, 1986; Lescinsky and Fink, 2000; Major and Newhall, 1989).The second involves volcanoes that interacted with the now-diminished Cordilleran icesheet (CIS). Throughout the Quaternary, the Pacific Northwest has been repeatedly en-veloped by a large continental-scale ice sheet that has waxed and waned in thickness andextent (Blaise et al., 1990; Clague, 1980; Clague and Ward, 2011; Fulton, 1991; Jacksonand Clague, 1991; Ryder et al., 1991). Locally, the last major glacial event in southwestBC is called the Fraser glaciation (Clague and Ward, 2011). At its maximum extent (i.e.,the last glacial maximum; LGM; ∼17 ka), the Fraser ice sheet covered the entire westernCordillera, overwhelming central Washington, the Canadian Rocky Mountains and Van-couver Island (Figure 2.1). The ice was thickest over the spine of the Coastal mountains,where its upper surface resided at 2500–3000 m in elevation (Booth et al., 2003; Kovanenand Easterbrook, 2001; Ryder and Maynard, 1991).The Garibaldi volcanic belt (GVB) is the northern (Canadian) segment of the Cascadevolcanic arc (Green et al., 1988; Hickson, 1994; Kelman et al., 2002a; Mathews, 1958;Russell et al., 2007) (Figure 2.2A). It extends from deposits located at Watts Point at thehead of the Howe Sound, to the Bridge River volcanic field approximately 150 km to thenorth (Figure 2.2A). Some classifications (e.g., Green et al., 1988; Souther, 1991) alsoinclude the Silverthrone and Franklin Glacier volcanic fields which are situated a further∼200 km northwest (Figure 2.1). For comprehensive purposes, we have included thesevolcanic fields in our database (Table 2.1), but do not discuss them further.11PembertonSquamishWhistler123° W123° W124° W52° N51° N51° N50° N50° N80060040020060040020010008001400 1200 80018001600140012001000140012001000800140012001000100060012004001600 1800100012001400160080010001200123° W123° W123.2° W50° N50° N49.8° N 49.8° N1000800600400Glaciovolcanicnon-GlaciovolcanicUnknownRiverOpen WaterPermanent Ice0 - 1,000 m1,000 - 2,000 m2,000 - 3,000 m100 5Kilometers0 2 41KilometersNon-GlaciovolcanicSubaerial TephraConeUndifferentiatedDepositSubaerial LavaSubaerial PyroclasticDepositGlaciovolcanicSubglacial TephraConeSubglacial Domesand BrecciaSubglacial LavaIce-Impounded LavaLava-dominated TuyaMount Garibaldivolcanic fieldGaribaldi lakevolcanic fieldMount Cayley volcanic fieldMount Meager volcanic fieldBridge rivervolcanic fieldSalal Glaciervolcanic fieldLillooet RiverHowe SoundElaho RiverDaisyLakeCheekye RiverHowe SoundA BEBCBYCBOMGBAWPMCOCRCENYENOPRPRCPGDBADBAMGMGBAWGVBRGPTM TETSPXMPCLCBEBWBMBBTCCOCCYHCFDVFLLJLCBOCBYTSEMBEGaribaldiLakeCheakamusLakeUpperCracked Mountain Lillooet Glacierbasalts200012001400160012Figure 2.2: Maps showing distribution and type of Quaternary volcanic rocks in thesouthern Garibaldi volcanic belt. A) Map showing Glaciovolcanic and non-Glaciovolcanic rocks across the GVB (excluding the Silverthrone and FranklinGlacier volcanic fields). Volcanic rocks of unknown affinity are displayed inyellow. The GVB is organized into six, geographically-defined volcanic fieldsdelineated by black rectangles corresponding to the mapped domains in Figures2.2B and 2.3A–2.3E. B) Map showing the volcanic deposits within the MountGaribaldi and Garibaldi Lake volcanic fields. The distributions are based onnew field mapping and the existing literature (Green, 1977; Green et al., 1988;Mathews, 1948, 1958). See Table 2.1 for the edifice names and acronyms thatare used on this map. The base map hill shade relief and 200 m topographiccontours are from the terrain resource information management (TRIM) digitalelevation model (DEM).Volcanism in the belt spans the entire Quaternary Period and is expressed by more than100 eruptive centres with deposits ranging in composition from alkaline basalt to rhyo-lite (Green, 1981b; Green et al., 1988; Hickson, 1994; Lawrence et al., 1984; Mathews,1957; Mullen and Weis, 2013, 2015; Stasiuk and Russell, 1989). The belt hosts three majorlong-lived stratovolcanoes (Mount Garibaldi, Mount Cayley and Mount Meager) in addi-tion to a plethora of smaller, ancillary volcanic centres (Hildreth, 2007; Souther, 1991).The three stratovolcanoes consist of porphyritic andesite to rhyolite and contain some ofthe oldest Quaternary volcanic rocks within the belt (Green et al., 1988). The most re-cent volcanic activity was a late Holocene (2360 BP), sub-Plinian to Vulcanian eruption atMount Meager (Hickson et al., 1999; Read, 1990). The smaller, ancillary centres are well-distributed throughout the belt and are predominantly monogenetic or short-lived. Theycomprise mainly andesite, rare dacite, calc-alkaline basalt and alkaline basalt, and haveages spanning the entire Mid–Late Pleistocene. Moderate volumes of alkaline basalt are re-stricted to the northern end of the belt in the Mount Meager, Bridge River and Salal Glaciervolcanic fields (Lawrence et al., 1984; Mullen and Weis, 2013; Roddick and Souther, 1987;Stasiuk and Russell, 1989). The origin of these basaltic rocks is widely debated, with stud-ies suggesting that they derive from either low-degree partial melting of the mantle wedge(Green, 2006) or from mantle upwelling along the margin of the subducting Juan de Fucaand Explorer plates (e.g., Bacon et al., 1997; Mullen and Weis, 2013, 2015; Nelson andCarmichael, 1984).The GVB is of interest due to its widespread evidence for glaciovolcanism. Approx-imately 50% of the volcanoes record eruptions in the presence of ice and of these, 75%show evidence for interaction with ice on a continental scale (i.e., the CIS) (Figure 2.2A).We present a database of all known Quaternary volcanic centers in the GVB and discuss13the substantial diversity in the glaciovolcanic deposits. We then demonstrate forensic vol-canological methods that can be used to map the paleo-presence (or absence), thickness, andtransient properties of the ice sheet at the time of each eruption, and use these data to createa preliminary, terrestrial-based reconstruction of the Late Quaternary paleoenvironment forsouthwest BC.14Table 2.1: Compilation of Quaternary volcanic deposits in the Garibaldi volcanic belt.VolcanicfieldEdifice∗ Abv. Class† Type¶ Comp.⋆Location Elevation (m asl) SurfaceArea (m3)∗∗Method‡ Age (ka) 1σ MIS Source#Lat◦N Long◦W Low HighWatts Point volcanic center ⋄ WP G SDB D 49.6497 -123.2167 0 240 7.90E+05 A 21.9 3.35 2 1, 2, 3, 18Monmouth Creek volcaniccomplex •MC G SDB A 49.6931 -123.1906 10 800 2.91E+06 - - - - 1, 4, 5, 6Opal Cone ⋄ OC S TC D 49.8239 -122.9756 1550 1680 8.12E+05 C 10 0.7 1 7Ring Creek Flow ⋄ RC S L D 49.7277 -123.0518 370 1685 3.01E+07 C 10 0.7 1 7Columnar Peak CP G SDB D 49.8003 -123.0053 1200 1860 2.68E+06 K 220 220 ? 1, 2, 6The Gargoyles (formerly: Lavapeak)G G SDB D 49.8058 -122.9981 1600 1800 9.87E+05 - - - - 2, 8Glacier Pikes GP GA SDB D 49.8794 -122.9789 1930 2120 1.60E+06 - - - - 2, 6MountGaribaldiEnostuck Flows (Y) ⋄ ENY G SL BA 49.7986 -122.9292 1080 1240 1.42E+05 - - - - 1, 2, 9Enostuck Flows (O) ⋄ ENO S L A 49.778 -122.9333 1050 1140 6.44E+05 - - - - 1, 2, 9Mount Garibaldi ⋄ MG SG PL D 49.8506 -123.0047 1600 2655 4.29E+06 G 14 0 2 9, 10, 11Diamond Head (Block & Ash) ⋄ DBA S P D 49.825 -123.0083 1200 2100 1.74E+07 G 14 0 2 1, 11Mount Garibaldi (Block & Ash) ⋄ MGBA S P D 49.7573 -123.1303 60 300 3.64E+06 C 11.7 0.425 1 1, 11Warren Glacier volcanics (RoundMountain complex) •WGV S PL B 49.8612 -123.0237 1750 220 1.55E+05 - - - - 1Brohm Ridge volcanics (RoundMountain complex) •BR S PL A 49.8536 -123.0553 1550 1930 1.71E+07 K 460 20 12? 1, 2Paul Ridge (Round Mountaincomplex) ⋄PR S PL A 49.7648 -123.0241 340 1650 1.99E+07 - - - - 1, 2The Table • T G LDT A 49.8953 -123.0147 1550 2010 3.43E+05 A 100 12 5 1, 2, 12, 13The Table East • TE GA SDB A 49.8917 -123.0061 1700 1890 3.84E+05 - - - - 1, 2, 12, 13GaribaldiLakeTable Meadows • TM GA SDB A 49.8806 -122.9956 1620 1960 8.25E+05 - - - - 1, 2, 12, 13Sphinx Moraine SPX GA SDB BA 49.9286 -122.9814 1530 1570 5.25E+04 - - - - 2, 6Mount Price ⋄ MP S PL A 49.918 -123.0353 1470 2050 3.87E+06 K 300 200 ? 2, 6, 9Clinker Peak ⋄ CL S TC D 49.9147 -123.0439 1790 1920 2.97E+05 G 13 0 2 1, 2, 9Continued on next page...15Table2.1– Continued from previous pageVolcanicfieldEdifice∗ Abv. Class† Type¶ Comp.⋆Location Elevation (m asl) SurfaceArea (m3)∗∗Method‡ Age (ka) 1σ MIS Source#Lat◦N Long◦W Low HighThe Barrier ⋄ B S I D 49.9444 -123.0832 1100 1920 1.29E+07 G 13 0 2 2, 6, 11Culliton Creek flow C S I D 49.8907 -123.0968 1050 1920 7.43E+06 G 13 0 2 2, 6, 11Cinder Cone (Y) • CCY S TC B 49.9717 -123.0072 1860 1900 1.40E+05 K 40 40 3 1, 2, 9Cinder Cone (O) • CCO GA STC BA 49.9717 -123.0072 1780 1860 7.97E+05 - - - - 1, 2, 9Helm Creek Flow • HCF S L B 49.9796 -123.0197 1518 1760 1.78E+06 - - - - 1, 2, 6, 9Desolation Valley Flows • DVF S I BA 50.0325 -122.9799 850 1780 6.65E+06 K 110 30 5e 1, 2, 6, 9GaribaldiLakeThe Black Tusk • BT G LDT A 49.9753 -123.0428 1820 2300 4.38E+05 A Unpub. - 6 1, 2, 6, 9,17West Bluff • WB G SDB A 49.9718 -123.0605 1940 2020 3.09E+05 - - - - 1, 2, 6, 9East Bluff • EB S F A 49.9697 -123.0487 1960 2110 8.48E+05 K 1300 100 40? 1, 2, 6, 9Microwave Bluff • MB GA SDB A 49.9844 -123.0527 1800 1950 3.59E+05 - - - - 1, 2, 6, 9Loggers Lake (formerly:Cheakamus River) ⋄LL G SDB A 50.0628 -123.0389 620 900 1.97E+06 - - - - 1, 2Jane Lakes JL S L A 50.0443 -123.0728 940 970 9.58E+05 - - - - 6Cheakamus Basalts (Y) ⋄ CBY G SL B 50.0365 -123.1173 470 530 1.72E+05 C 34.2 0.8 2 2, 13, 14,15Cheakamus Basalts (O) ⋄ CBO S L B 50.0628 -123.1092 360 510 1.73E+07 A 140.8 12.9 5c 2, 13, 14,15, 18Tricouni Southeast TSE S I A, D 49.9922 -123.2194 500 1600 6.11E+06 - - - - 15Tricouni Southwest TSW S I BA 49.9903 -123.2308 820 1500 2.33E+06 - - - - 15Mount Brew East MBE GA SDB A 50.0369 -123.1786 1200 1660 8.73E+05 A 189.2 11.8 6 15, 16, 18Mount Brew West MBW GA SDB A 50.0394 -123.2064 1600 1720 2.47E+05 - - - - 15, 16MountCayleyEmber Ridge North ERN G SDB A 50.0767 -123.2389 1700 1920 2.41E+05 - - - - 2, 8, 15, 16Ember Ridge Northeast ERNE G SDB A 50.0722 -123.2153 1500 1600 3.50E+04 - - - - 2, 8, 15, 16Ember Ridge Northwest ERNW G SDB A 50.0758 -123.2564 1800 1950 3.09E+05 A Unpub. - 12 2, 8, 15, 16Ember Ridge Southeast ERSE G SDB A 50.0464 -123.2219 1400 1650 2.39E+05 - - - - 2, 8, 15, 16Ember Ridge Southwest ERSW G SDB A 50.0458 -123.2522 1800 1940 4.09E+05 - - - - 2, 8, 15, 16Ember Ridge West ERW G SDB A 50.0672 -123.2614 1720 1860 3.50E+04 - - - - 2, 8, 15, 16Continued on next page...16Table2.1– Continued from previous pageVolcanicfieldEdifice∗ Abv. Class† Type¶ Comp.⋆Location Elevation (m asl) SurfaceArea (m3)∗∗Method‡ Age (ka) 1σ MIS Source#Lat◦N Long◦W Low HighMount Fee ⋄ F S PL D 50.0836 -123.2447 1200 2160 1.96E+06 A Unpub. - 6 15, 17Shovelnose Stage (Mount Cayley) S S PL D 50.1107 -123.2947 400 1900 3.53E+06 K 310 50 9 2, 15, 16Vulcan’s Thumb Stage (MountCayley)V S PL D 50.1108 -123.2958 500 2350 7.31E+06 K 2700 700 ? 2, 15, 16Mount Cayley Stage (MountCayley)MC S PL D, R 50.1203 -123.2894 1000 2390 3.06E+06 - - - - 15, 16MountCayleyPauli Dome East ⋄ PDE G SDB A 50.1356 -123.2692 1600 2300 1.93E+06 A Unpub. - 6 1, 15, 17Pauli Dome West PDW GA SDB A 50.1369 -123.3097 1850 2200 1.57E+06 A Unpub. - 6 1, 15, 17Upper Cauldron Dome ⋄ UCD G LDT A 50.1556 -123.3194 1860 2170 1.88E+06 - - - - 15Lower Cauldron Dome LCD S I A 50.1451 -123.3295 1350 1900 1.95E+06 A Unpub. - 14? 1, 2, 15, 17Slag Hill SH GA PL A 50.1897 -123.3069 1700 2100 2.89E+06 A Unpub. - 17 1, 2, 15, 17Slag Hill Tuya ⋄ SHT G LDT A 50.2028 -123.2803 1940 2060 1.75E+05 - - - - 15Ring Mountain ⋄ RM G LDT A 50.2222 -123.3039 1600 2192 3.12E+06 A 49.1 5.5 4 15Ring Mountain Northwest RMN G SL BA 50.231 -123.3159 940 1380 4.71E+05 - - - - 15Little Ring Mountain ⋄ LRM G LDT A 50.2797 -123.315 1800 2147 5.00E+05 - - - - 15Unnamed 1 U1 Unk Unk - 50.1782 -123.4729 1580 1760 1.05E+06 - - - - 26Unnamed 2 U2 Unk Unk - 50.4391 -123.4645 2000 2250 2.76E+06 - - - - 26Elaho Valley Basalts ⋄ EV S L B 50.45 -123.5817 500 1000 1.94E+07 K 140 100 ? 2, 6Cracked Mountain • CM G STC AB 50.5497 -123.5565 1450 1600 1.29E+06 - - - - 1, 19Unnamed 3 ⋄ U3 S L AB 50.6993 -123.5127 850 1350 5.03E+06 - - - - 1, 2, 19Mosaic Ridge ⋄ MR S PL AB 50.6518 -123.5979 1600 1800 4.52E+05 K 90 60 ? 1, 2, 19MountMeagerCapricorn Assemblage (MountMeager)CA S PL RD 50.6272 -123.5319 1600 2500 9.33E+06 K 100 20 5 2, 19, 20Job Assembalge (Mount Meager) ⋄ JA S PL RD 50.6175 -123.5428 1900 2100 1.89E+06 - - - - 2, 19, 20Pylon Assemblage (MountMeager) ⋄PYA S PL A 50.5975 -123.5197 1100 2300 3.02E+07 K 900 200 ? 2, 19, 20Continued on next page...17Table2.1– Continued from previous pageVolcanicfieldEdifice∗ Abv. Class† Type¶ Comp.⋆Location Elevation (m asl) SurfaceArea (m3)∗∗Method‡ Age (ka) 1σ MIS Source#Lat◦N Long◦W Low HighDevastator Assemblage (MountMeager) ⋄DA S PL RD 50.5931 -123.5322 1200 2100 4.17E+06 K 1900 200 ? 2, 19, 20Plinth Assemblage (MountMeager) ⋄PP SG P D 50.6608 -123.4408 600 650 5.99E+05 A 75 9 5 1, 2, 19, 20MountMeagerPlinth Peak (O) (Mount Meager) ⋄ PPO S PL D 50.6453 -123.5117 800 2677 1.49E+07 K 200 140 ? 1, 20Plinth Peak (Y) (Mount Meager) ⋄ PPY S PL D 50.6453 -123.5117 800 2677 1.53E+06 K 57 78 ? 1, 20Pebble Creek Formation ⋄ PBF S PL RD 50.6603 -123.5072 420 1400 1.24E+07 C 2.34 0.05 1 19, 21, 28Lillooet Glacier Basalts • LG G TN AB 50.747 -123.7226 820 1400 1.01E+06 G 17 4 2 1, 22Unnamed 4 ⋄ U4 S L B 50.7377 -123.6278 1940 2000 8.60E+04 - - - - 1Northwest Volcanic Remnant • NWVR G STC AB 50.8103 -123.4506 2200 2430 2.21E+05 - - - - 1, 23Salal Mountain Volcano • SM G SDB A 50.7635 -123.4084 2100 2370 7.94E+05 - - - - 1, 23Logan Ridge Tindar • LR G TN AB 50.8125 -123.4083 2140 2340 8.31E+04 - - - - 1, 23Salal Tindar (O) • STO S L AB 50.7839 50.7839 2240 2340 3.00E+05 K 970 50 26? 1, 23SalalGlacierSalal Tindar (Y) • STY G TN AB 50.7926 -123.3854 2260 2360 5.81E+05 K 590 50 15 1, 23Mud Lake Flow • ML S L AB 50.8047 -123.3786 2180 2260 6.41E+03 - - - - 1, 23Ochre Mountain Tindar • OM G TN AB 50.7978 -123.3653 2000 2140 1.48E+05 - - - - 1, 23Unnamed 5 U5 S PL R, RD 50.7624 -123.4794 1920 2160 3.24E+06 - - - - 1, 23Unnamed 6 U6 Unk Unk - 50.8036 -123.5168 2080 1650 8.08E+06 - - - - 1Sham Hill • SHM G SDB AB 50.9047 -123.5076 1860 2000 7.06E+04 K 752 9 18 1, 24Sham Plateau Volcanics • SPV G SL AB 50.9099 -123.495 1820 1950 5.69E+05 - - - - 1Sham Plateau Lava • SMP G SL AB 50.9174 -123.5015 1970 2010 2.77E+05 - - - - 1, 24BridgeRiverSham Ridge Basalts • SRB S L AB 50.9244 -123.4951 1880 2100 1.74E+05 - - - - 1, 24Sham Ridge Volcanics • SR S L D 50.9308 -123.506 2020 2070 5.69E+05 - - - - 1Nichols Valley Flows • NVF S L AB 50.9506 -123.3772 1550 2000 4.02E+06 K 405 11.5 11 1, 24Tuber Hill (O) • THO G STC AB 50.9233 -123.4519 1700 2180 2.95E+06 K 731 13.5 18 1, 24Tuber Hill (Y) • THY G T AB 50.9103 -123.4578 1900 2010 6.25E+06 K 598 7.5 15 1, 24Continued on next page...18Table2.1– Continued from previous pageVolcanicfieldEdifice∗ Abv. Class† Type¶ Comp.⋆Location Elevation (m asl) SurfaceArea (m3)∗∗Method‡ Age (ka) 1σ MIS Source#Lat◦N Long◦W Low HighThunder Creek Volcano • TCV G STC AB 50.9108 -123.4146 1600 1950 8.59E+05 - - - - 1Arrowhead Glacier • AG G STC AB 50.9503 -123.5187 2150 2170 5.25E+05 - - - - 1Tretheway Lake Volcano TLV Unk Unk B 49.6247 -122.2868 1370 1480 1.53E+06 - - - - 1, 27FranklinGlacierFranklin Glacier (O) FGO S L D 51.313 -125.4538 2000 2400 2.33E+07 K 4000 100 ? 2Franklin Glacier (Y) FGY S L R 51.313 -125.4538 2000 2400 K 2200 100 ? 2Silver-throneSilverthrone Caldera (O) SCO S PL R 51.4408 -126.3017 1500 2000 1.30E+08 K 770 80 17-19 2Silverthrone Caldera (Y) SCY S L R 51.4408 -126.3017 1500 2000 K 400 100 11? 2Silverthrone Caldera Valley Flow SCVF S L BA 51.5381 -126.329 90 900 8.76E+06 C 12.2 0.14 1 25Note: Abv-Abbreviation; Comp-Composition; MIS- Marine Isotope Stages; Unk-Unknown; Unpub-Unpublished age (see text).∗ Some edifices display long-lived activity: O-older age constraint; Y-younger age constraint. Edifices marked with a ⋄ were visited during this study. Edifices marked with a •were mapped in detail (see preface).† Deposit classification. G-glaciovolcanic; S-subaerial; SG-supraglacial; GA-glaciovolcanic-alpine.¶ LDT-lava-dominated tuya; SDB-subglacial domes and breccia; L-subaerial lava; P-subaerial pyroclastic deposit; PL-undifferentiated subaerial deposit; I-ice-impounded lava;SL-subglacial lava; T-tuya; STC-subglacial tephra cone; T-tindar; TC-subaerial tephra cone.⋆ D-dacite; A-andesite; BA-basaltic andesite; B-basalt; R-rhyolite; RD-rhyodacite; AB-alkaline basalt.∗∗ Surface areas calculated from mapped distributions.‡ Geochronology method: A-40Ar/39Ar; C-C14; K-potassium-argon (K-Ar); G-geomorphology. “Unpublished” ages are given in the text.#1: This thesis; 2: Green et al. (1988); 3: Bye et al. (2000); 4: Wilson et al. (2016); 5: Green (1994); 6: Mathews (1958); 7: Brooks and Friele (1992); 8: Kelman et al. (2002a);9: Mathews (1948); 10: Mathews (1952a); 11: Friele and Clague (2009); 12: Mathews (1951); 13: Green (1977); 14: Green (1981b); 15: Kelman (2005); 16: Souther (1980);17: M. Martendale and D. Weis, pers. comm (2017); 18: M. Kelman, pers. comm. (2017); 19: Read (1990); 20: Read (1977); 21: Hickson et al. (1999); 22: see chapter 4; 23:Lawrence et al. (1984); 24: Roddick and Souther (1987); 25: Hickson (1994); 26: P. Adam, pers. comm (2017); 27: N. Drader, pers. comm (2017)192.2 Historical reviewGlaciovolcanism is an emerging field of research that has seen exponential growth in the lastfew decades, especially in studies involving the paleoenvironment (Edwards et al., 2009;Russell et al., 2014; Smellie, 2018; Smellie and Edwards, 2016). Although accumulationsof ice and snow have covered significant portions of the planet in both space and time,new glacial advances tend to destroy evidence of previous ones (i.e., till and moraines).Thus, constraining the absolute timings of ancient ice advances and retreats can be notori-ously difficult (e.g., Clague, 1981; Mathewes et al., 2015; Richmond and Fullerton, 1986).Early work recognized that glaciovolcanic deposits are physically robust and can preservedetailed information on syn-eruptive glacier extent and thickness (Mathews, 1947; Noe-Nygaard, 1940; Peacock, 1926; Pjetursson, 1900). By combining these data with high pre-cision geochronology, modern studies use glaciovolcanoes as a proxy for paleo-climate thatcan document ancient glacial fluctuations and the nature of local paleoenvironments (Car-rivick et al., 2009; Cole et al., 2018; Conway et al., 2016, 2015; Edwards et al., 2011; Guil-lou et al., 2010; Kelman et al., 2002a; McGarvie et al., 2007; Owen et al., 2012; Schmidtand Grunder, 2009; Smellie, 2018; Smellie et al., 2009, 2008, 2011; Stevenson et al., 2009;Tuffen and Castro, 2009).Most glaciovolcanic research has focused on volcanoes located in Iceland, Antarcticaand Canada. Exceptional overviews of glaciovolcanism for Iceland are given by Jakob-sson and Gudmundsson (2008), and McGarvie (2009), for the Antarctic by Smellie andSkilling (1994), Smellie et al. (1993, 2008), and Smellie (2008), for north-central BC byMathews (1947), Edwards and Russell (1999), Edwards and Russell (2002); Edwards et al.(2003, 2009), and Hickson (1986, 1994, 2000), and for southwestern BC byMathews (1948,1958), Green et al. (1988), Hickson (1994, 2000), Kelman et al. (2002a) and Andrews et al.(2014a). A recently compiled comprehensive summary of glaciovolcanism on Earth andMars is given by Smellie and Edwards (2016). Below we summarize the existing litera-ture on glaciovolcanism in the Cascade volcanic arc. Firstly, we review deposits located inWashington and Oregon (US), and then move north to describe those situated in the GVB(Canada).2.2.1 The US Cascade volcanic arcCompared with Canada, studies of glaciovolcanism in the US are relatively few. Most havefocused on lava-ice interactions on stratovolcanoes in Washington (Mount Rainier, MountBaker) and Oregon (North Sister, South Sister, Kokostick Butte andMount Jefferson). Workby Lodge and Lescinsky (2009), Lescinsky and Fink (2000) and Lescinsky and Sisson(1998), describe intermediate composition effusive deposits resulting from volumetrically20small, ice-impounded lavas and subglacial flows and domes. Lescinsky and Fink (2000)give a short compilation of effusive glaciovolcanic deposits in the Cascade volcanic arc, aswell as a highly detailed analysis of cooling joint and fracture morphologies.Geologic mapping and geochronology in the Mount Mazama and Crater Lake re-gions of Oregon by Bacon and Lanphere (2006) have identified at least nine mid–latePleistocene intermediate to felsic deposits consisting of ice-impounded lavas (RedcloudCliff, Grotto Cave, Pumice Point, Dutton Cliff, Mundson Ridge, Grouse Hill, RoundTop) and tuyas (Bear Bluff, Arant Point). Schmidt and Grunder (2009) provide abundantevidence for late Pleistocene volcano-ice interactions in central Oregon, which includesice-marginal/confined (ice-impounded) lavas, pillow-margined lavas, lava-dominated tuyas(Matthieu Lake Fissure system), and substantial accumulations of palagonitized tuff andpillowed palagonitized tuff breccia (North Sister and Little Brother volcano).Studies have also identified several purported tuyas throughoutWashington and Oregon.The Lone Butte and Crazy Hills complexes described by Hammond (1987) consist of twoadjacent accumulations of palagonitized basaltic hyaloclastite (e.g., White and Houghton,2006), pillow lava and subaerial capping lavas that are interpreted to have been emplacedbetween ∼190 and 130 ka. Walupt Lake volcano (part of the Goat Rocks volcanic centerin southern Washington) displays a ∼500 m-thick sequence of dipping hyaloclastite bedsand sub-horizontal basaltic lava flows (60–90 m thick). The activity at Walupt Lake isinterpreted to have occurred at ∼140 ka (Hammond, 1980; Swanson, 1996). Farther south,in central Oregon, there exist two small, flat-topped andesitic volcanoes (Hayrick Butteand Hogg Rock) which are probably lava-dominated tuyas (Conrey et al., 2002; Wood andKienle, 1990). A potassium-argon (K-Ar) date from Hogg Rock gives an age of 80± 20 ka.Lastly, Pleistocene phreatomagmatic activity and associated supraglacial tephra depositionare described at Kulshan caldera in northern Washington (Hildreth, 1996; Westgate et al.,1987). The water involved in these eruptions is interpreted to be glacial in origin.2.2.2 The Canadian Cascade volcanic arcThe GVB has fostered several seminal studies of glaciovolcanism. Early work by Math-ews (1948), provided initial recognition of glaciovolcanism in the belt, and include the firstaccounts of intermediate-composition lava-ice interactions. Mathews identified and de-scribed: an ice-impounded lava flow (The Barrier; Mathews, 1952b), a lava-dominated tuya(The Table; Mathews, 1951), subglacial lava flows exploiting pre-existing drainage chan-nels (The Cheakamus basalts; Mathews, 1958), numerous small, subglacial lava domes andbreccias (Mathews, 1948, 1958), and a major stratovolcano (Mount Garibaldi), interactingwith an ice-dominated environment (Mathews, 1952b). The work also established the ini-21tial framework for organizing the GVB volcanoes into a series of geographically definedvolcanic fields (e.g., Figure 2.2A).Geothermal reconnaissance at Mount Cayley (Souther, 1980) identified fifteen an-desitic and dacitic glaciovolcanoes including subglacial domes and breccias (e.g.,EmberRidge, Pali Dome and Mount Brew) and lava-dominated tuyas (Ring Mountain and LittleRing Mountain) (Figure 2.2A). Subsequent studies by Kelman (2005) and Kelman et al.(2002a,b) provided further insight into the nature of glaciovolcanism in the Mount Cay-ley volcanic field, and place important constraints on the eruptive conditions for effusiveintermediate glaciovolcanic deposits in general.Other studies have focused on the alkaline basalts in the northern GVB, where a di-verse array of glaciovolcanic products occur in the Salal Glacier volcanic field (Lawrenceet al., 1984) and the Bridge River volcanic field (formerly: Bridge River Cones; Roddickand Souther, 1987) (Figure 2.2A; Table 2.1). Green (1977, 1990) and Green et al. (1988)provided the first detailed petrological, geochemical, and geochronological study of Qua-ternary volcanism in the GVB. More recent site-specific studies include the Watts Pointvolcanic center (Bye et al., 2000), the Monmouth Creek volcanic complex (Wilson et al.,2016) and the Lillooet Glacier basalts (Chapter 3) (Figure 2.2A, 2.2B).The glaciovolcanoes of the GVB have also provided a basis for studies of hazard po-tential. Moore and Mathews (1978) for example, examined landslides that originated fromthe Barrier in 1855–1856 (Figure 2.2A, 2.2B). Hickson (1994) gives an overview of thehazards (e.g., air-fall tephra, debris flow etc.) associated with Quaternary volcanism in theCascade volcanic arc with a specific focus on the northern portion of the arc (Canada andWashington).2.3 Glaciovolcanism in the Garibaldi volcanic beltFor the purposes of this discussion, we have organized the GVB into 6 geographically de-fined volcanic fields which builds on the existing framework of Mathews (1948, 1958),Green (1977) and Green et al. (1988). The six volcanic fields are named: Mount Garibaldi,Garibaldi Lake, Mount Cayley, Mount Meager, Salal Glacier, and Bridge River. Ta-ble 2.1 lists all Quaternary volcanic deposits in the GVB (both glaciovolcanic and non-glaciovolcanic) and their physical and geographic properties. The glaciovolcanoes are clas-sified descriptively using a GVB-modified scheme that is based on Russell et al. (2014),Smellie (2013) and Smellie and Edwards (2016). Specifically, we use the terms ”tuya” and”tindar” morphologically, to refer to flat-topped (e.g., Mathews, 1947; Smellie, 2013) andelongate (e.g., Jakobsson and Gudmundsson, 2008) glaciovolcanic edifices, respectively.However, we distinguish ”lava-dominated tuyas” based on the lithological modifier of Rus-22sell et al. (2014). We also include ”ice-impounded lavas” to denote lavas erupted from asubaerial vents that flowed downslope to encounter ice (e.g., Mathews, 1952a) and, ”sub-glacial lavas” to refer to lavas emplaced within a sub-glacial environment (e.g., Mathews,1958). Below, we provide an overview of the glaciovolcanic occurrences within the GVBmoving from the south to the north, and, within groupings, from the oldest deposit to theyoungest.2.3.1 Mount Garibaldi volcanic fieldWatts Point volcanic center and the Monmouth Creek volcanic complexThe Watts Point volcanic center is situated at the head of Howe Sound, at the southern endof the GVB (Figure 2.2B). It consists of a pile of glassy hornblende, pyroxene-phyric dacitelavas and minor associated hyaloclastite breccia (Bye et al., 2000; Green et al., 1988). Thelavas are arranged in at least 3 dome shaped structures that are characterized by intenseradial columnar jointing and carapaces of glassy, monolithologic hyaloclastite. Marginalcolumnar joints are fine-scale (Bye et al., 2000), suggesting enhanced cooling in contactwith glacial ice. Recent 40Ar/39Ar analysis indicates an age of 21.9 ± 3.35 ka (MelanieKelman pers. comm., 2017) (Table 2.1).West across the valley, the Monmouth Creek volcanic complex displays a ∼200 mthick package of aphyric to sparsely pyroxene-phyric basaltic andesite and andesite form-ing radial (i.e., fanning) fine-scale columnar jointed lavas, domes and lobes, and partiallypalagonitized hyaloclastite breccia (Wilson et al., 2016) (Figure 2.2B). The complex is cutby a series of en echelon hornblende, pyroxene-phyric dacite dikes that crop out as a se-ries of steep-sided pinnacles and spires (Green, 1994; Mathews, 1958; Wilson et al., 2016).Indicators for lava-ice/water interaction, including fine-scale jointing and hyaloclastite, arepreserved up to an elevation of ∼800 m, and suggest that the eruption took place under,or adjacent to, a large valley-filling glacier. A single sample was found to be too young todate by 40Ar/39Ar methods, suggesting that, as with Watts Point, the eruption was probablyconcurrent with the last glaciation (Mathews, 1958; Wilson et al., 2016).Mount Garibaldi volcanic complexThe Mount Garibaldi volcanic complex encompasses the volcanic rocks constructing andsurrounding Mount Garibaldi, a highly dissected composite stratovolcano in the southernGVB (Figure 2.2B). Little is known about the oldest portions of the volcano which con-sist of basaltic to dacitic lavas and pyroclastic rocks that show inconclusive evidence forinteraction with ice (Table 2.1). Collectively, these rocks are named the Round Mountain23complex by Green et al. (1988). The youngest parts of Mount Garibaldi were built in aseries of Pele´an, supraglacial eruptions that deposited a thick apron of unconsolidated ma-terial on top of the adjacent (waning) ice sheet which was occupying the nearby Cheekyevalley Mathews (1952b). As the ice receded, the western flank of the edifice collapsed,leaving behind steep (∼600 m high) southwest-facing cliffs. Quaternary sediments indicatethat the peak of Mount Garibaldi was emplaced shortly before 13.5 ka (Friele and Clague,2009). In addition, primary in situ, block and ash deposits that are located ∼16 km south-west of the main summit (Mount Garibaldi block and ash; Table 2.1; Figure 2.2B) returneda radiocarbon age of 11,700 ± 475 years BP (Friele and Clague, 2009). These rocks weredeposited subaerially, probably along the margin of a valley-filling glacier, and attest to thearea (at ∼60 m elevation) being ice-free at the time of eruption.Columnar Peak, Glacier Pikes and Eenostuck volcanic complexesHornblende-phyric dacite lavas constitute several dome-shaped masses that are distributedaround the fringes of Mount Garibaldi. In the south, the Columnar Peak volcanic complexforms a group of partially eroded lava domes that are surrounded by a thick mantle of talus(Mathews, 1958). In the north, the Glacier Pikes volcanic complex is a small collection oflava domes and fins of lava with precipitous margins. Both of these volcanic complexesdisplay radiating aggregates of fine-scale (10–40 cm diameter) columnar joints that areconsistent with eruption into ice (Green et al., 1988; Mathews, 1958).The Eenostuck volcanic complex is a small collection of andesite and basaltic andesitelavas. Only the youngest, northern basaltic andesite flow appears to be glaciovolcanic,displaying an elongate profile and radial, fine-scale marginal jointing (Green et al., 1988;Mathews, 1948, 1958).2.3.2 Garibaldi Lake volcanic fieldThe Table, Sphinx Moraine and the Mount Price volcanic complexesThe Table is probably the best known glaciovolcanic deposit in the GVB (Green, 1977;Green et al., 1988; Hickson, 1994; Kelman et al., 2002a; Mathews, 1951). The Table isenigmatic and defines the type-example of a flat-topped, lava-dominated tuya (Kelmanet al., 2002a; Russell et al., 2014; Smellie et al., 2013). The edifice is elongated in thewest–east direction, and displays sheer margins that are coated in fine-scale, glassy, hacklyjoints (e.g., Lodge and Lescinsky, 2009). It consists of a massive package of andesite lava,and, contrary to previous analysis, does not display evidence for successive upward growth(c.f., Mathews, 1951). All workers agree that the Table erupted beneath glacial ice (e.g.,24Mathews, 1951). Recent 40Ar/39Ar analysis (100 ± 12 ka; Table 2.1) indicates that theedifice is considerably older than previously thought (i.e., older than Fraser glaciation; seechapter 5).The Sphinx Moraine volcano displays several glacially-scoured basaltic andesite lavadomes at the terminus of Sphinx Glacier (Green, 1977; Green et al., 1988; Mathews, 1958)(Figure 2.2B). Fine-scale radial columnar joints suggest enhanced cooling in an environ-ment consistent with ice or water, however the proximity to an established mountain glacierdoes not necessitate the CIS.The Garibaldi Lake volcanic field also hosts one of the first described occurrences of anice-impounded lava flow (Mathews, 1952a). Part of the Mount Price volcanic complex, theBarrier forms a 250 m-high, sheer dacite cliff at the head of Rubble Creek (Figure 2.2B). Itoriginated subaerially at the nearby tephra cone, Clinker Peak (Figure 2.2B), which sourcedtwo major lava flows, one going west (down Rubble Creek), and the other to the south(down Culliton Creek). Both lavas were buttressed by valley-filling glacial ice in the formof the waning CIS, which, at the time, resided in the adjacent Cheakamus valley (Mathews,1952a; Moore and Mathews, 1978). The Barrier displays numerous ice-contact featuresincluding an over-thickened profile (e.g., Harder and Russell, 2006), irregular, radiatingand horizontal columnar joints, and pseudo-pillow jointing (e.g., Forbes et al., 2012; Lodgeand Lescinsky, 2009).The Black Tusk volcanic complexFarther north, the Black Tusk volcanic complex shows abundant evidence for glaciovolcan-ism (Green et al., 1988; Mathews, 1952a) (Figure 2.2B). In the west, Microwave Bluff isan andesitic lava dome that has radially-jointed margins that are coated in glassy, hackly-jointed material. West Bluff is also an andesite lava dome that displays similar fine-scale,fanning radial joints that terminate in hackly margins (Mathews, 1958). The ages of theMicrowave Bluff and West Bluff edifices are unconstrained. Recent mapping of the cen-tral Black Tusk portion of the complex suggests that it shares a number of characteristicswith the Table: it is elongated and displays fine-scale marginal columnar joints that fan to-wards horizontal and terminate in hackly and pseudo-pillow joints (e.g., Forbes et al., 2012;Lodge and Lescinsky, 2009). Although the Black Tusk is apparently more eroded than theTable on its northern, western and eastern sides, these similarities suggest that it is a lava-dominated tuya. A recent 40Ar/39Ar analysis returned an age of 177 ± 16 ka (preliminarydate, Martindale and Weis, pers. comm., 2017) (Table 2.1). The nature of the local topogra-phy precludes the possibility that the Black Tusk was erupted into a thin alpine glacier thatmay have existed during a relatively warm climatic period (i.e., the Black Tusk is the local25high point and there are no major surrounding peaks to source such a glacier). The edificetherefore records an early (i.e., pre-Fraser) iteration of the CIS.The Cinder Cone volcanic complexThe Cinder Cone volcanic complex comprises a pair of overlapping tephra cones and as-sociated lava flows in the nearby Helm valley (Green, 1977; Green et al., 1988; Mathews,1948, 1958). The younger cone is constructed of oxidized scoria with spindle bombs (sub-aerial). The older cone, however, is a broad ring of well-bedded, indurated basaltic andesitetuff and tuff breccia. Infrequent cross-bedding, and minor palagonite-rimmed clasts suggestintermittent phreatomagmatic activity. The older cone was also the source of the Desolationvalley basaltic andesite lava flows that are exposed ∼6 km to the north at the base of theHelm valley (Green et al., 1988). These lavas form steep, radially-jointed tubes and lobesthat are up to 30 m high, suggesting impoundment against valley filling ice. A K-Ar dateindicates that their age is 110 ± 30 ka (Green et al., 1988).Loggers Lake volcanic complex and the Cheakamus basaltsThe Loggers Lake volcanic complex (formerly: the Cheakamus River complex; Green et al.,1988), is a small, highly dissected dome of hornblende, biotite-phyric dacite that is locatedon the southern bank of the upper Cheakamus River (Figure 2.2B). The complex is ex-tensively eroded, glacially-scoured and covered by glacial debris. Recent mapping has re-vealed three indicators for glaciovolcanism. Firstly, the western margin (located in a quarry)displays lavas with joints that change in orientation and spacing over short distances. Sec-ondly, the eastern margin displays fanning radial joints that are situated ∼30 m above thesurrounding valley floor. Finally, minor accumulations of glassy breccia in the central com-plex probably indicate eruption in association with water. The complex must predate theend of the last (Fraser) glaciation, however further geochronological information is unavail-able.The Cheakamus basalts are a ∼70 m-thick package of lavas occupying a 1–2 km wide,22 km long belt in the Cheakamus River and Callahan Creek valleys (Figure 2.2B). Thelower three units were erupted subaerially, and consist of ∼14 stacked lavas that range inthickness from 2 to 10 m (Green, 1977). A recent 40Ar/39Ar analysis of a subaerial lavaexposed at the junction of Callahan Creek and the Cheakamus River valley gives an age of141 ± 12.9 ka (Melanie Kelman, pers. comm. (2017) (Table 2.1).The youngest Cheakamus basalts are characterized by a series of well-preserved, anas-tomosing, channelized lavas that feature hemispherical cross-sections and fanning columnarjoints (Mathews, 1958). Minor accumulations of pillow lava, hyaloclastite and glaciofluvial26sediment are preserved beneath the lavas. Carbonized wood derived from this sediment isradiocarbon dated at 34.2 ± 0.8 ka, and provides a maximum age for the overlying lavas(Green, 1977, 1981b). Mathews (1958) suggested that the lavas were confined to subglacialdrainage channels developed beneath a waning glacier in the Squamish valley at the end ofthe last glaciation (i.e., ∼13 ka).271000800 40020016001400120010006001200100060040014001200800600400180018002000100010001800 2000100012001400160016001800100080010001200Callaghan Lake123.2° W123.2° W50.2° N50.2° N50° N50° N800200019001800210022002300240025002300220021001800170015002200210020001400123.35° W123.35° W123.4° W123.4° W123.45° W123.45° W50.8° N50.8° N1600190020002400230022002100200019001800150014001900170016001700 1600130019002000210022002300123.4° W123.4° W123.45° W123.45° W123.5° W123.5° W50.9° N50.9° N50.95° N50.95° N24001700180019002000210022002300240016001500140018001300 1200 1100 100020001900180017001500140090014001500160017001800SiltLake123.75° W123.75° W123.7° W50.75° N 50.75° N1200110010001400130015001600FishLake123.55° W123.55° W123.6° W123.6° W50.55° NCMSHMSPFSRVSRBTHYTHONVFTCVTCVTCVTHO?SPVAGAGLGSMVSTYSTOOMMLLRNWVREBCBOCBYTSETSWMBEMBWERSEERSWERWERNW ERNEREFSV MCPDW PDWUCDLCDSHSHTRMRMNLRMCBYCBOCBOTHOTHOEDCBDaisy LakeArrowhead RiverBridge RiverArrowheadGlacierLillooet GlacierAthenley PassSalal GlacierASMVTHY0 10.5Kilometers0 10.5Kilometers0 1 20.5Kilometers0 1 20.5Kilometers0 2 41KilometersGlaciovolcanicTindarTuyaSubglacial Tephra ConeSubglacial Domes andBrecciaSubglacial LavaIce-Impounded LavaLava-dominated TuyaNon-GlaciovolcanicSubaerial Tephra ConeUndifferentiated DepositSubaerial LavaSubaerial Pyroclastic Deposit28Figure 2.3: Maps showing distribution and type of Quaternary volcanic rocks in thenorthern Garibaldi volcanic belt. A) Map showing the volcanic deposits withinthe Mount Cayley volcanic field based on Kelman et al. (2002b) and Kelman(2005). Legend applies to all figure panels. B) Cracked Mountain. C) LillooetGlacier. D) Salal Glacier volcanic field. The distributions are based on Lawrenceet al. (1984), and supplemented by new field mapping. E) Map (oriented E–W)showing the volcanic deposits in the Bridge River volcanic field based Roddickand Souther (1987) and supplemented by new field mapping. See Table 2.1 forthe key to the edifice names and acronyms used on this map. Base maps are asin Figure 2.2.2.3.3 The Mount Cayley volcanic fieldThe Mount Cayley volcanic field contains numerous glaciovolcanic deposits that range incomposition from basaltic andesite to rhyodacite (Figure 2.3A). The Mount Cayley strato-volcano (comprising the Shovelnose, Vulcan’s Thumb, andMount Cayley stages; Table 2.1)is a highly eroded composite pile of overlapping lavas and pyroclastic rocks that, alongwith Mount Fee, do not appear glaciovolcanic. Surrounding Mount Cayley, however, are alarge number of small-volume centers that display a rich variety of effusive, intermediate-composition glaciovolcanic interactions (Kelman, 2005; Kelman et al., 2002a,b; Souther,1980).The Ember Ridge, Mount Brew, Pali Dome and Tricouni complexesAt Ember Ridge and Mount Brew, eight discrete piles of porphyritic andesite form jagged,bulbous, domes that are coated in hackly, fine-scale columnar joints (Green et al., 1988;Kelman et al., 2002b; Souther, 1980). An abundance of small-scale spires, knobs, gulliesand ridges with fine-scale marginal jointing suggests that erosion of these features has beenminimal. Two 40Ar/39Ar ages indicate that the activity is older than suggested by Kelman(2005)s. The older, Ember Ridge NorthWest is 477± 26.5 ka (preliminary date, MartindaleandWeis, pers. comm., 2017), and the younger Mount Brew East is 189± 11.8 ka (MelanieKelman, pers. comm., 2017) (Table 2.1).Farther north, the Pali Dome complex contains andesite lavas and minor matrix-supported hyaloclastite that are arranged in two dome-shaped piles (Figure 2.3A). Thedomes terminate in steep, glassy cliffs that are >100 m high, and show bulbous profileswith fine-scale hackly joints (Kelman, 2005). Two 40Ar/39Ar analyses indicate that the twoeruptions were roughly contemporaneous. Pali Dome West was erupted at 172 ± 9.5 kaand Pali Dome East at 178 ± 13 ka (preliminary dates, Martindale and Weis, pers. comm.,2017) (Table 2.1).29The Tricouni complex includes andesite and dacite lavas that have minor quantities ofassociated autobreccia and hyaloclastite. Kelman (2005) reported fine-scale jointing alongthe eastern and southern margins, suggesting impoundment against waning ice likely duringthe last (Fraser) glaciation (Figure 2.3A).Slag Hill, Cauldron Dome, Slag Hill Tuya, Ring Mountain and Little Ring MountainSlag Hill is a pile of andesite lava flows that display fine-scale jointing, bulbous marginalprofiles and minor hyaloclastite (Figure 2.3A; Kelman, 2005). The edifice is dated using40Ar/39Ar geochronology at 691 ± 17 ka (preliminary date, Martindale and Weis, pers.comm., 2017) (Table 2.1).The adjacent, undated Slag Hill Tuya is a flat-topped mass of andesite that is ∼120 mhigh. Like Slag Hill, the tuya displays jointing that is characteristic of ice confinement.The upper portions of the lava are rich in rounded xenoliths of varying lithology, suggestingtheir origin as ice-rafted cargo melted from the overlying CIS and incorporated into the lava(Kelman, 2005).Cauldron Dome is a flat-topped, elliptical, ∼400 m high package of stacked andesitelava and volcaniclastic rock (Kelman, 2005) (Figure 2.3A). Fine-scale marginal columnarjointing and poorly developed pseudopillow jointing at the northern end, and the overall flat-topped edifice morphology suggest that the eruption occurred beneath ice. Like the Table,the Black Tusk and Slag Hill Tuya, Upper Cauldron Dome is probably a lava-dominatedtuya. The age of the edifice is constrained by 40Ar/39Ar analysis at 540± 75 ka (preliminarydate, Martindale and Weis, pers. comm., 2017) (Table 2.1).At the northern end of the Mount Cayley volcanic field there are two additional lava-dominated tuyas (Figure 2.3A). Ring Mountain comprises porphyritic andesite lava that isarranged in a horse-shoe-shaped, flat-topped cylinder ∼2.3 km in diameter and ∼500 mhigh (Kelman, 2005; Kelman et al., 2002a). The upper surface contains oxidized scoria andcoarse columnar joints (presumably exposed subaerially). The margins display fine-scalejointing that is consistent with ice confinement (Kelman, 2005). 40Ar/39Ar analysis indi-cates that Ring Mountain is 49 ± 5.5 ka (Melanie Kelman, pers. comm., 2017) (Table 2.1).Little Ring Mountain is ∼900 m in diameter, flat-topped and ∼240 m high. It is composedof a stack of at least 3 separate andesite lavas. A talus slope covers the majority of the edi-fice, so it is likely that more lava is currently buried. Fine-scale radial joints are pervasive,and diagnostic of interaction with ice. Similar to Slag Hill Tuya, the lavas exposed nearthe upper surface of Little Ring Mountain are rich in melt-out xenoliths presumably meltedfrom the overlying CIS and incorporated into the lavas (Kelman, 2005).302.3.4 The Mount Meager volcanic fieldMount Meager is a highly dissected stratovolcano that was last active in the late Holocene(Hickson et al., 1999; Read, 1990). Although glaciovolcanic features have not been reportedfrom the stratovolcano, several small alkaline basalt centers located on the periphery ofthe massif (e.g., the Mosaic Assemblage; Read, 1990; Stasiuk and Russell, 1989) recordinteractions with the CIS.Cracked Mountain volcano and the Lillooet Glacier basaltsNear the headwaters of Meager Creek, Cracked Mountain volcano is a thick package ofpalagonitized tephra and intrusive dike-fed pillow lava that is arranged in an elliptical semi-flat-topped mound that is ∼100 m thick (Figure 2.3C). Recent mapping indicates that thelower portion is predominantly pillow lava and the upper zone is crudely bedded, palagoni-tized tuff, lapilli tuff, tuff breccia and pillow breccia. More than 50 basaltic dikes (0.5 to 2.5m width) cut the massif. Locally, the dikes terminate in large, pillowed pods that were ap-parently injected into the crudely bedded, highly unconsolidated and water-saturated tephra.The surface is strewn with abundant glacial debris and is extensively scoured, indicatingpost-eruptive glacial overriding and an age that pre-dates the LGM.Retreat of Lillooet Glacier has exposed a succession of pillow lava and minor hyalo-clastite. The edifice forms an elongate, pillow-dominated tindar that is ∼150 m thick (Fig-ure 2.3B). Hydrostatic constraints suggest that the eruption occurred under ice that was>645 m thick, and produced an ephemeral englacial lake that was>150 m deep (see Chap-ter 4). This correlates with attempted 40Ar/39Ar dating, which revealed the edifice as tooyoung to date, and suggests that the eruption took place at the peak of, or in the waningstages of, the last (Fraser) glaciation (i.e., 13–17 ka; Table 2.1).2.3.5 Salal Glacier volcanic fieldThe Salal Glacier volcanic field consists of a collection of six, isolated, small-volume vol-canic centers (Figure 2.3D). Five of the centers are alkaline basalt in composition (Lawrenceet al., 1984; Mullen and Weis, 2013) and one is composed of calc-alkaline andesite. Alledifices except the Mud Lake flow show evidence for glaciovolcanism. The lithofacies pre-served in the Salal Glacier volcanic field were formed through extensive lava-water inter-actions and subaqueous depositional processes (see below). Given the local physiographicsetting of the volcanic field (i.e., steep topography), thick glacial ice in the form of the CISis the only plausible mechanism for damming and sustaining the large water bodies that arerequired (Lawrence et al., 1984).31Salal tindar, Logan Ridge tindar, the Northwest volcanic remnant, Ochre Mountaintindar and Salal Mountain volcanoSalal tindar (formerly: Salal Glacier volcanic center; Lawrence et al., 1984) is an elon-gate glaciovolcanic edifice in the northern GVB (e.g., Jakobsson and Gudmundsson, 2008).Salal tindar displays a lower subaerial, plagioclase-phyric basaltic lava that is K-Ar datedby Lawrence et al. (1984) at 970 ± 50 ka. Unconformably overlying this lava is an elon-gate, ∼40 m thick accumulation of massive to bedded, normally graded, palagonitized tuff,lapilli tuff and tuff breccia. The deposit is poorly to moderately sorted, and contains juve-nile clasts of vesicular, subrounded basalt and a matrix of palagonitized vitric ash. It alsocontains up to 20% angular, <20 cm diameter accessory monzonite lithics, suggesting sig-nificant explosivity. Overlying this, at an elevation of 2360 m, the deposit transitions intoa ∼25 m-thick cone of loose, oxidized, scoria and lava with spindle bombs. The eruptionculminated in an effusive lava (∼10 m thick), erupting subaerially from near the top of thetephra cone, and flowing southward to terminate in an over-thickened, ∼60 m-high bluffwith fine-scale, radial marginal columnar joints. These features are classic indicators ofice-impoundment. The radially jointed lava is dated using K-Ar geochronology at 590 ±50 ka (Lawrence et al., 1984) (Table 2.1).Logan Ridge tindar (formerly: Logan Ridge volcanic center; Lawrence et al., 1984) isa ∼70 m thick sequence of dike-fed basaltic pillow lava, pillow breccia and radially jointedlava lobes that are overlain by an ∼8 m thick accumulation of laminated, cross-beddedpalagonitized tuff and lapilli tuff.The Northwest volcanic remnant forms a similar dike-fed tephra cone that is perchedhigh on an alpine ridge. It comprises a <100 m thick sequence of massive, jointed palago-nitized tuff breccia that is intruded by dikes and overlain by pillow lava and further palago-nitized tuff breccia.The Ochre mountain tindar is volumetrically small and distributed as a <15 m-thickveneer trending northeast–southwest along the ridge above the Athenley Pass (Figure 2.3D).The edifice consists of a radially-jointed basaltic lava flow that is underlain by a sequenceof pillow lava and hyaloclastite.Salal Mountain volcano is a∼200 m thick package of andesitic lavas and volcaniclasticrocks that overlie glacial deposits. The basal lavas are ∼50 m thick, show multiple individ-ual flow units, and display colonnade columnar jointing and fanning entablature. Fartherup-sequence, the edifice is increasingly dominated by fragments and comprises monolitho-logic, poorly consolidated, poorly sorted, dense, glassy breccia with intrusive pods of co-herent lava (e.g., Tuffen et al., 2002).322.3.6 Bridge River volcanic fieldThe Bridge River volcanic field (formerly: Bridge River Cones; Roddick and Souther,1987) (Figure 2.3E) is located at the northern end of the GVB, and is dominated by lavasof alkaline basalt, hawaiite and mugearite (Mullen and Weis, 2013, 2015; Roddick andSouther, 1987). Almost every portion of the volcanic field shows evidence for glaciovol-canism apart from Nichols Valley flows; a package of coarsely jointed basaltic lavas thatoverlie glaciofluvial conglomerate (see below: Tuber Hill). Roddick and Souther (1987)date the Nichols Valley flows using K-Ar geochronology at 405 ± 11.5 and 374 ± 11.5 ka(Table 2.1).The Sham Plateau volcanics, Tuber Hill, the Arrowhead Glacier volcanics and theThunder Creek volcanicsAt the western margin of the Bridge River volcanic field, the Sham plateau volcanics con-tain two glaciovolcanic centers (Figure 2.3E). The Sham Plateau lava flow is ∼1 km longand has fine-scale radially-jointed margins. Sham Hill consists of a dome-shaped mass ofmugearite with an undulating, glassy lower margin that indicates emplacement onto a wet,unconsolidated substrate. At the southern edge of Sham Hill, fine-scale horizontal colum-nar joints point into the open air above the Bridge River valley. Sham Hill is K-Ar dated at752 ± 9 ka (Roddick and Souther, 1987).The center of the field is dominated by Tuber Hill, a large polygenetic, complex tuya(Roddick and Souther, 1987). The core consists of an older, ∼300 m-thick hawaiite tephracone containing crudely bedded to massive, coarsely-jointed, indurated, palagonitized tuff,lapilli tuff and tuff breccia. The tephra has angular to subrounded, highly-vesicular pyro-clasts that are set in a matrix of palagonitized vitric ash. An abundance of large (<30 cmdiameter), angular, basement accessory lithic clasts suggests that this eruptive phase washighly explosive. A geochemically identical basaltic lava that was subaerially erupted fromthe peak of the cone, is K-Ar dated at 731 ± 13.5 ka (Roddick and Souther, 1987) (Ta-ble 2.1). A stage of volcanic quiescence (∼100 ka) is recorded by a 10–20 m-thick packageof bedded fluvial conglomerates that contain abundant, locally derived volcanic material.The physiographic (high-altitude, mountainous) setting suggest that their origin may beglaciofluvial (i.e., ice-dammed). The sediments are overlain by a ∼20 m-thick sequence ofalkaline basaltic pillow lava, pillow breccia and hyaloclastite that is in turn, overlain, via apassage zone (Russell et al., 2014), by stacked subaerial lava flows that are 10–20 m thick.These lavas are the youngest portion of Tuber Hill, and are K-Ar dated by Roddick andSouther (1987) at 598 ± 7.5 ka (Table 2.1). The Tuber Hill deposits may also extend lat-erally across the adjacent Arrowhead and Thunder Creek valleys (Figure 2.3E). Mugearite33lavas blanket the plateau to the east, and overlie a similar package of glaciofluvial conglom-erate. Glaciofluvial conglomerates are also found overlying an ∼20 m thick accumulationof bedded palagonitized tuff and lapilli tuff on the west side of Sham Plateau (these arepossibly derived from the older Tuber Hill tephra cone). Further analysis is required to thegenetically link these deposits, however, the consequence is that volcanism at Tuber Hillmay pre-date the development of the 500 m-deep bounding glacial valleys.The recently exposed Arrowhead Glacier volcanics comprise a small collection ofmugearite hyaloclastite, pillow lava, and jointed lavas of unknown age. Given their prox-imity to a currently established mountain glacier and their low profile, these glaciovolcanicdeposits do not necessitate involving the CIS in their eruption.The Thunder Creek volcanics (∼600 m-thick) stretch from Thunder Creek (∼1400 min elevation), eastward to ∼910 m in elevation. The lower part of the deposit displaysmugearite lava lobes that are surrounded by glassy, clast-supported hyaloclastite (e.g.,White and Houghton, 2006; Wilson et al., 2016). The upper part of the deposit (∼1750m elevation) forms a crude cone of bedded, well-indurated tuff breccia. No geochronol-ogy exists, so we infer that these rocks are substantially younger than those constructingthe adjacent Tuber Hill. Future geochronology may provide an age for the Thunder Creeksequence and a minimum time frame for the development of the Thunder Creek valley (i.e.,it is eroded into Tuber Hill sequence rocks).2.4 Discussion2.4.1 Diversity of glaciovolcanism in the Garibaldi volcanic beltWhen compared with the glaciovolcanoes in Iceland, Antarctica and northcentral BC, thosein the GVB display a diverse range of deposit types, reflecting a rich array of emplacementprocesses (Andrews et al., 2014a; Hickson, 1994; Kelman et al., 2002a). Over its relativelyshort length (∼150 km), the GVB encompasses tuyas, tindars and subglacial tephra cones,in addition to effusive subglacial domes and breccias, subglacial lavas, ice-impounded lavasand lava-dominated tuyas.Two primary factors contribute to this diversity: magma composition and topography.Both can be ascribed to the unique tectonic and physiographic setting of the GVB. Theregion overlies a subduction zone, leading to the production mainly of calc-alkaline inter-mediate to felsic composition magmas (Hickson, 1994; Hildreth, 2007; Rogers, 1985). De-posits in Iceland, Antarctica and northcentral BC are dominantly mafic (e.g., Edwards andRussell, 1999; Hickson, 1986; Jakobsson and Gudmundsson, 2008; Smellie and Edwards,2016), and intermediate to felsic volcanic deposits are subordinate to rare (e.g., Edwards34and Russell, 2002; McGarvie, 2009; Smellie et al., 2011; Stevenson et al., 2011, 2006,2009; Tuffen et al., 2002). Composition, including volatile content, controls the physicalproperties of magma (liquidus temperature, viscosity, density, glass transition temperature)and, in turn, these properties influence eruption style (Giordano et al., 2005; Ho¨skuldssonand Sparks, 1997; Kelman et al., 2002a).Topography, however, is probably the most important control on the nature of glacio-volcanism in the GVB. The belt is situated in a mountainous region with a relief of >3200m (Figure 2.2A). Volcanism is also contemporaneous with a rich history of advancing andretreating Pleistocene ice sheets. As a consequence, the GVB eruptions must have encoun-tered a wide range of ice thicknesses and glacier styles. Topography controls the thicknessof ice overlying a vent. During periods of glacial maxima (e.g., the LGM), ice is thickestin the valleys and is thinnest at high elevations (Clague, 1981; Fulton, 1991). This con-trols both the availability of ice to be melted (Kelman et al., 2002a), and the overburden(i.e., isostatic) pressure acting on the magma. In turn, ice thickness influences volatile ex-solution and explosivity potential (Smellie and Skilling, 1994; Tuffen, 2007; Tuffen andBetts, 2010). Finally, ice thickness can also affect the rate at which a buoyancy-driven dikecan propagate upwards, thereby influencing the eruption dynamics at the ice-rock interface(Wilson and Head, 2002).Topography and ice thickness also control the glacio-hydraulics of a volcanic system.The temperate alpine glaciers found in mountainous areas maintain extensive subglacialdrainage networks that are highly efficient at moving large volumes of meltwater (e.g.,Hickson, 2000; Kelman et al., 2002a; Smellie, 2018; Smellie et al., 2006). Enhanceddrainage influences the deposit lithofacies (i.e., subaqueous versus subaerial), and the avail-ability of water for quenching and hydroclastic fragmentation (Gudmundsson, 2003; Gud-mundsson et al., 2004; Hickson, 2000; Smellie and Skilling, 1994; Stevenson et al., 2011;Tuffen, 2007; Tuffen et al., 2001). Thick overlying ice can suppress the discharge of waterfrom an established englacial lake until the encircling ice is floated and releases a meltwaterflood (i.e., jo¨kulhlaups; Bjo¨rnsson, 2003; Gudmundsson et al., 1997, 2004; Russell et al.,2013; Tweed and Russell, 1999). Therefore, eruptions occurring during periods of exten-sive ice cover and thickness (i.e., glacial maxima) should be more capable of maintainingenglacial lakes in areas of extreme topography.Topography and magma composition operate together to contribute to the diversity ofglaciovolcanism in the GVB (Hickson, 2000; Kelman et al., 2002a). One possible ex-ample of this involves the heat transfer dissimilarity between mafic and felsic magmas.High temperature basalts provide a larger thermal budget for ice melting and, particularlywhere explosive, are highly efficient at exchanging heat (Allen, 1980; Gudmundsson, 2003;Ho¨skuldsson and Sparks, 1997; Kelman et al., 2002a; Tuffen, 2007; Tuffen et al., 2002).35Therefore, basaltic volcanoes may be more capable of producing and maintaining englaciallakes due to accelerated ice melting rates when in competition with topography-enhanceddrainage rates. The GVB contains many intermediate magmas that were erupted in areas ofextreme topography. This combination reduces the potential for water storage at eruptionsites, and suggests that many GVB volcanoes were erupted under well-drained conditions(Hickson, 1994, 2000; Kelman et al., 2002a; Russell et al., 2014).2.4.2 Forensic recovery of paleoenvironmentWe use examples from the GVB to illustrate a variety of forensic methods that are used toextract detailed paleoenvironmental information from glaciovolcanic deposits.Lava-dominated tuyasIntermediate composition lava-dominated tuyas (LDT) are a style of glaciovolcanism that isunique to the GVB. Like all tuyas, their diagnostic feature is a flat upper surface, however,unlike other tuyas, LDTs show no fragmental lithofacies (c.f., Jones, 1968, 1970; Mathews,1947; Russell et al., 2014; Werner et al., 1996). Intermediate composition LDTs in the GVBlack the typical tuya lithofacies sequence of subaerial capping lava overlying subaqueouspillow lava, pillow breccia and hyaloclastite. Instead, they are composed of massive orstacked, effusive lavas with a narrow compositional SiO2 range of 57 to 62.8 wt.% (Slag HillTuya and Ring Mountain respectively; Kelman, 2005). Fine-scale, radial columnar joints,glassy lavas and sporadic pseudo-pillow jointing indicate ice/water-lava interactions andsheer-sided edifice morphologies suggest physical ice impoundment (Kelman et al., 2002a;Mathews, 1951; Russell et al., 2014; Smellie, 2013; Smellie and Edwards, 2016). The Tablein the southern GVB embodies this type of glaciovolcanic landform (Figure 2.4A).A ”passage zone” is a surface that marks the transition between a subaqueous and sub-aerial deposition environment. They are present in many tuyas, and are significant becausethey unequivocally preserve the height and depth of the paleo-englacial lake that existedat the time of eruption, and therefore the minimum height of surrounding ice (Allen et al.,1982; Jones, 1968, 1970; Russell et al., 2014; Smellie, 2006). LDTs in the GVB do notpreserve typical passage zones. However, they offer a unique way of placing constraints onthe maximum height of enclosing ice through an apparent subglacial to subaerial transition.Previous workers have suggested that LDTs form in well-drained subglacial environmentsthat are open to the atmosphere (Kelman et al., 2002a; Mathews, 1951; Russell et al., 2014).Both the Table and Ring Mountain corroborate this with upper surfaces of oxidized lava,scoria and autobreccia (Figure 2.3B, Figure 2.3C). At both of these volcanoes, lavas at theupper outer edge show evidence for enhanced cooling and buttressing by ice (e.g., fine-scale36AWE ~100 mOxidized lava~ 30 mBC~ 300 mDSN ~100 mFigure 2.4: Field photographs showing lava-dominated tuyas in the GVB. A) Mosaicimage derived from a 3D photogrammetry model of the Table (looking north-east). B) Photograph looking west at the upper, oxidized portion of the Table. C)Photograph of Ring Mountain in the Mount Cayley volcanic field (photo cour-tesy of Paul Adam). D) Photograph of the Black Tusk in the Garibaldi Lakevolcanic field.37jointing, glassy lava and sheer margins). This implies that the surface of the ice sheet wasat least as high as the top of the edifice, but did not fully enclose the edifice. The actualheight of surrounding ice above the edifice margin is speculative, but we suggest, given theoxidized upper surfaces and the relatively small size of the edifices, that it was probably nomore than ∼100 m (e.g., Smellie, 2013). LDTs are therefore rich sources of paleoenviron-mental information, providing a measure of the minimum enclosing ice sheet height (i.e.,upper elevation of the edifice) and a rough estimate of the maximum enclosing ice sheetheight (i.e., the minimum elevation plus ∼100 m). During the eruption at the Table, weestimate that the surface of the ice sheet was, at minimum, located at ∼2010 m. Similarly,the Black Tusk (Figure 2.3D) preserves a record of ice at ∼2300 m, and Ring Mountain at∼2190 m (Figure 2.3C).Ice-impounded lavasThe GVB also hosts several subaerially-erupted lavas that flowed downslope to become im-pounded against valley filling glacial ice (Andrews et al., 2014b; Kelman, 2005; Mathews,1952a). These edifices are significant because they record both the paleo-presence of valleyglaciers, and the absolute elevation of the glaciers’ outer margin at the time of eruption.In the Cordillera, the growth of large ice sheets is initiated by expanding alpine glaciersadvancing into the surrounding valley systems. Eventually, these valley-filling glaciers be-come thick enough to coalesce and form an expansive, flat-topped sheet (Clague and Ward,2011; Fulton, 1991; Jackson and Clague, 1991). During periods of ice retreat, the oppositeprocess occurs, and valley walls and some alpine areas become ice-free faster than the highpeaks and lowland valleys (Fulton, 1991). Lavas that are erupted in these ‘ice-free win-dows’, can flow downslope and become impounded. Thus, their deposits can be used tofingerprint important climatic transitions.The GVB hosts the first recognized example of an ice-impounded lava. The RubbleCreek flow is a dacite lava that originated subaerially at the nearby scoria cone, ClinkerPeak. It terminates in a highly unstable, over-thickened cliff (the Barrier) that displays radialcolumnar, pseudo-pillow and fine-scale hackly jointing (Hickson, 1994; Mathews, 1952a;Moore and Mathews, 1978) (Figure 2.2B). The Rubble Creek flow (Figure 2.5A) eruptedwhile the alpine area around Clinker Peak was ice-free, and a glacier (most likely in a stateof retreat) resided within the lower Cheakamus valley. Although no direct geochronologyexists, an age of ∼13 ka can be inferred through the local geomorphology and Quaternaryglacial sediments (Friele and Clague, 2009; Mathews, 1952a).A similar feature occurs ∼12 km to the northeast at the base of the Helm valley wherethe Desolation valley flows terminate in a sheer-margined, 2 km-wide, fan-shaped deposit38AThe Black TuskBedrock~ 100 mFlow directionDacite lavaCollapsed lava~ 10 mCB~ 10 mTo Clinker PeakFigure 2.5: Field photographs showing ice-impounded lavas in the GVB. A) Pho-tograph looking northeast at the Barrier, a sheer, ∼100 m-high cliff of in-tensely jointed dacite lava. B) Photograph looking south showing a∼20 m-high,radially-jointed basaltic andesite lava tube from the Desolation valley lava flowsexposed at the base of Helm valley. C) Photograph looking east showing a cross-sectional exposure of the steep, corrugated lava bluffs at the base of at the baseof Helm valley. The long axes of columnar joints are indicated with white lines.39(Figure 2.2B). Geochemical and petrographic fingerprinting suggests that the lavas origi-nated from the lower portion of the Cinder Cone complex which is located∼6 km up valleyto the south (Green et al., 1988; Mathews, 1948, 1958). At high elevations (above ∼800m), the lavas exhibit coarse joint morphologies that are typical of a subaerial environment.At the base of the Helm valley, however, they form sheer ∼30–50 m high bluffs consistingof irregular, corrugated bluffs and tubes of fine-scale, radial to hackly-jointed lava (Fig-ure 2.5B). Cross-sectional faces show horizontal columnar joints that point outwards intothe open air >70 m above the adjacent upper Cheakamus River valley floor (Figure 2.5C).Although their source (the lower Cinder Cone sequence) preserves subordinate evidence forminor water interaction (e.g., sporadic palagonite-rimmed clasts and minor surge bedding),the Cinder Cone deposits indicate an ice-free vent (Mathews, 1948). The Desolation valleyflows are dated at 110 ± 30 ka (Green et al., 1988), and we interpret that they preserve arecord of glacial ice occupying the upper Cheakamus River valley during this time. It ismost likely that this ice was associated with the onset (i.e., advance) of the last major (i.e.,Wisconsin) glaciation.Passage zonesIn the northern GVB, Salal tindar is a mid-Pleistocene glaciovolcanic edifice that preservesa passage zone (Lawrence et al., 1984). The volcano occupies a narrow, alpine ridge, andis bounded on all sides by steep-sided glacial valleys (Figure 2.3D). At the base, the litho-facies consist of a ∼40 m thick package of massive to bedded, highly palagonitized tephrathat shows normal grading, subrounded vesicular clasts and moderate sorting within discon-tinuous lenticular accumulations. These deposits also comprise rare pillow lavas and dikeswith irregular, undulating pillowed margins. These lithofacies indicate subaqueous deposi-tion, and their thickness requires a standing body of water that was at least 40 m deep. Thelocal topography precludes the possibility that a non-glacier-related body of water existedin the area, implying that their origin must be contemporaneous with a period of extensiveice cover (Lawrence et al., 1984). At ∼2360 m elevation, the deposit transitions into a pet-rographically and geochemically identical, ∼25 m-thick, subaerial tephra cone of oxidizedscoria, spindle bombs and lava (Figure 2.6A). The eruption culminated with a subaeriallava, that flowed south from the summit of the cinder cone, to become impounded againsta surrounding ice sheet at ∼2260 m in elevation. The lava front forms an over-thickenedbluff that is∼60 m high, and is composed of fine-scale, radially-jointed basalt (Figure 2.6B)(Lawrence et al., 1984). The stratigraphy records the growth of a tephra-dominated tindarand preserves a passage zone that records the height of the paleo-englacial lake into whichthe eruption took place (Russell et al., 2013). It is not entirely clear if the cone became40~2360 mOxidized scoria and subaerial lavaAB~ 60 mFlow directionBedded palagonite tuff, lapilli tuff, tuff breccia,  rare pillow lava and pillow-margined dikes.CFigure 2.6: Field photographs showing the passage zone preserved at the Salal tin-dar. A) Photograph looking west of passage zone contact (dashed white line)between subaerial and subaqueous pyroclastic volcanic deposits at ∼2360 m(person circled for scale). B) Photograph showing glassy, basaltic pillow frag-ment contained within palagonitzed tuff breccia (subaqueous lithofacies at Salaltindar). The ice axe head is ∼20 cm long. C) Photograph looking west showingan intensely-jointed, ice-impounded lava that is ∼60 m high.41emergent, or the lake was drained, however a lack of pillowed lava around the base andmargins of the subaerial flow suggests the latter interpretation. The subaqueous lithofaciesat Salal tindar are distributed over an elevation range of∼90 m (2280–2360 m). This eleva-tion change translates into a hydrostatic pressure that must be balanced by the surroundingice load to prevent drainage of the lake via ice lifting (e.g., Bjo¨rnsson, 2003; Gudmundssonet al., 1997, 2004; Russell et al., 2013; Tweed and Russell, 1999). We calculate that a mini-mum of ∼90 m of overlying ice would be required to sustain a lake of this size, suggestingthat, at the time of eruption, the surrounding ice sheet surface was located at >2400 m inelevation (Russell et al., 2013).2.4.3 Paleo-ice height reconstruction and correlation with the MarineIsotope Stages timescaleGlaciovolcanic deposits have long been recognized for their significance as proxies of localand global paleoenvironments (e.g., Mathews, 1947; Noe-Nygaard, 1940; Peacock, 1926;Pjetursson, 1900). Similarly, non-glaciovolcanic rocks record periods of ice absence, andare therefore equally important for these studies. Although numerous previous workershave correlated the ages of glaciovolcanic eruptions with established periods of ancient ice(e.g., Edwards et al., 2011; Guillou et al., 2010; Kelman et al., 2002a; Lachowycz et al.,2015; McGarvie et al., 2007; Schmidt and Grunder, 2009; Smellie, 2008; Smellie et al.,2009, 2008, 2011), these studies have largely focused on individual, short-lived centres,thereby limiting their potential for longer-term reconstructions. The GVB supports a sus-tained record of well-distributed, diverse and intermittent Quaternary volcanic activity thatis contemporaneous with a protracted history of waxing and waning ice sheets. Using de-tailed forensic techniques and precise geochronology, the volcanoes of the GVB, capture acomprehensive picture of the spatial-temporal distribution of ice throughout the region.The most widely recognized record of the global paleoclimate is afforded by the ra-tios between 18O and 16O isotopes within benthic marine sediments. These Marine Iso-tope Stages (MIS) provide a proxy for global ice volume and ocean temperature changes,and thereby an averaged, high-resolution record of the global paleoclimate (Lisiecki andRaymo, 2005). As the MIS is a globally averaged record, it is unable to identify the lo-cations of ice across the globe or the absolute volume of terrestrial ice, relative to globalsea level (e.g., Martinson et al., 1987; Raymo et al., 2006; Shackleton, 1987). Individ-ual glaciers respond to climate forcing at a scale that can be independent of the averageglobal temperature, and thus the growth and decay of various continental ice sheets are notsynchronous (Martinson et al., 1987; Ward et al., 2007). To estimate the average volumeof continental ice, several studies have also paired the marine isotope record with ocean42temperature-sensitive indicators such as the Mg/Ca proxy (e.g., Martin et al., 2002) andthen subtract this signal from the co-occurring δ18O record (Shakun et al., 2015).Table 2.2 compiles GVB volcanoes that preserve demonstrable evidence for a glacio-volcanic or non-glaciovolcanic origin and have robust (i.e., precise) radiometric age con-straints. For each edifice, the minimum or maximum height (m above sea level) of enclosingice is measured from the mapped elevations of the deposits. Glaciovolcanic deposits (G) re-quire that the minimum height of enclosing ice must reside above their highest elevation de-posit. Subaerial (S) volcanic deposits indicate that the maximum height of the surroundingice sheet (if present) must be at a lower elevation than the mapped deposit. Glaciovolcanic-alpine (GA) edifices show glaciovolcanic features, however they lack irrefutable evidenceof interaction with extensive ice cover (i.e., the CIS). For these volcanoes, the height ofthe enclosing ice sheet (if present) may be either above or below their mapped elevationrange. The Pali Dome West and Slag Hill deposits are examples of this type. Several ofthe younger glaciovolcanic edifices (The Barrier, Lillooet Glacier basalts and the youngCheakamus basalts) constrain the distribution and height of the CIS, however have agesthat are interpreted using the local geomorphology and Quaternary sedimentary record.In some cases, the absolute height of the ice sheet surface can be more accurately con-strained. Lava-dominated tuyas record the minimum surrounding ice sheet surface, and, iftheir upper surface lithofacies are oxidized, can also provide an estimate of the maximumice sheet surface. Ice-impounded lavas were buttressed along the margin of valley-fillingglaciers, and thus provide an absolute measurement the glacier surface at the time of theinteraction. Finally, subaqueous glaciovolcanic lithofacies and passage zones constrain theheight and depth of ephemeral englacial lakes, thereby placing hydrostatic constraints onthe minimum ice thickness required to suppress lake drainage.Our method can recover the growth and decay of ice sheets using a single elevation-based variable that is based on the record afforded by glaciovolcanic deposits. For thisstudy, we have opted to omit the paleoenvironmental information that is hosted in the Qua-ternary sedimentary record in southwest BC. For detailed studies of these deposits we directthe reader to: Armstrong (1981); Armstrong et al. (1965), Clague (1976, 1980, 1981), Hic-ock and Armstrong (1983), Mathews and Rouse (1986), Ryder et al. (1991), Jackson andClague (1991), Booth et al. (2003), Menounos et al. (2009) and Clague and Ward (2011)and references therein. We assume a simplified ice advance and retreat model: expand-ing alpine glaciers advance into the surrounding valley system and eventually coalesce andthicken to form a flat-topped ice sheet (Clague and Ward, 2011; Fulton, 1991; Jackson andClague, 1991). We also assume that each individual ice-height marker represents the heightof the ice sheet across the entire GVB (i.e., we do not correct for parabolic curvature ofthe ice sheet surface or the influence of bedrock topography; Livingstone et al., 2013; Nye,43Table 2.2: Quaternary volcanic rocks in the Garibaldi volcanic belt that constrain thespatial-temporal distribution of the Cordilleran Ice SheetVolcanic Edifice∗ Abv. Class† Age (ka)¶ 1σ Height of Ice (m asl)#Watts Point volcanic center WP G 21.9 3.35 240Ring Creek Flow RC S 10 0.7 370Mount Garibaldi (Block & Ash) MGBA S 11.7 0.425 300The Table T G 100 12 2010Rubble Creek Flow (The Barrier) B S 13∗∗ 0 1100Cinder Cone (Y) CCY S 40 40 1860Desolation Valley Flows DVF S 110 30 850The Black Tusk BT G Unpub. - 2300Cheakamus Basalts (Y) CBY S 34.2∗∗ 0.8 530Cheakamus Basalts (O) CBO S 140.8 12.9 360Mount Brew East MBE GA 189.2 11.8 1660Ember Ridge Northwest ERNW G Unpub. - 1950Mount Fee F S Unpub. - 1200Shovelnose Stage (Mount Cayley) S S 310 50 400Pali Dome East PDE G Unpub. - 2300Pali Dome West PDW GA Unpub. - 1850Slag Hill SH GA Unpub. - 2100Ring Mountain RM G 49.1 5.5 2192Lillooet Glacier Basalts LG G 17** 4 1400Salal Tindar (O) STO S 970 50 2240Salal Tindar (Y) STY G 590 50 2335Sham Hill SHM G 752 9 2000Nichols Valley Flow NVF S 405 11.5 1550Tuber Hill (O) THO G 731 13.5 2180Tuber Hill (Y) THY G 598 7.5 2010Note: Abv-Abbreviation; MIS-Marine Isotope Stages; Unpub-Unpublished age (see text)∗Some edifices display long-lived activity: O, older; Y, younger.†G, glaciovolcanic; S, subaerial; GA, glaciovolcanic-alpine (see text for explanation).#Minimum or maximum height of the ice sheet in southwest BC (i.e. the CIS).∗∗Inferred age (see text for explanation)¶Source for age listed in Table 2.1.1952a,b; Russell et al., 2013). We consider this approximation to be appropriate when con-sidering the restricted size of the study area and the consistency of the local physiography.Figure 2.7A and Figure 2.7B show the reconstructed height of waxing and waning ice acrossthe GVB during the last 1 Ma. The reconstruction is juxtaposed against the MIS timescale(Lisiecki and Raymo, 2005), where periods of high global ice volume are represented byhigh δ 18O (‰) values and an even stage number. We find a number of positive correlationsand several important discordances with the MIS record.44050010001500200025003000Elevation of CIS (m asl)WPRCMGBATBCCYDVFBTCBY (sediment)CBOMBEFPDEPDWRMLG0102030405060708090100110120130140150160170180190200210220230240250Time (10 k.y. step)33.5 44.5 5δ18O( )12345a5b5c5d5e567???????CBY‰-0.5 00.5δ18O(sw)‰?ERNWSSHSTOSTYSHMNVFTHOTHY3003504004505005506006507007508008509009501000Time (50 k.y. step)89101112131415161718192021222324252627Global ice volume – Modified from Shakun et al., (2015)Global ice volume and ocean temperature – modified from Lisiecki and Raymo (2005)LegendGlaciovolcanicNon-GlaciovolcanicGlaciovolcanic-alpineAccurate Ice HeightInterpreted Ice HeightABGlaciovolcanic-inferred age12345a5b5c5d5e567891011121314151617181945Figure 2.7: Mid–Late Pleistocene ice-height reconstruction for southwest BC usingthe record afforded by the glaciovolcanic deposits. Glaciovolcanic edifices re-quiring extensive ice cover (i.e., the Cordilleran Ice Sheet; CIS) give a minimumice height at the time eruption. Subaerial edifices also provide a limit (maxi-mum) for the height of the CIS at the time of eruption. Glaciovolcanic-alpineedifices preserve glaciovolcanic interactions that do not necessitate the presenceof the CIS. Some of the younger edifices record the height of the CIS, howevertheir age is inferred from the local geomorphology. The black arrow indicates theactual age of the youngest Cheakamus basalts, and the marker denotes the ageof the underlying glaciofluvial sediments. The thick solid lines are drawn wherethere is a minimum and/or maximum estimate of ice-height; the thick dottedline denotes interpreted sections that connect clusters of edifices. The horizontallines surrounding the data points show the elevation range of each deposit, andthe vertical lines show 1σ uncertainty on the geochronology. Our reconstructionis compared with the MIS timescale (Lisiecki and Raymo, 2005) and a correctedglobal ice volume curve (δ 18Osw) constructed by Shakun et al. (2015). Highδ 18O (‰) values, and an even MIS number (e.g., 2, 4, 6) indicate periods wherethe average global climate was colder than today. The stage numbers are readfrom top to bottom, and only the beginning (oldest time) of each stage labeled.Figure 2.7A denotes the last 250 ka, and Figure 2.7B denotes 250–1000 ka. TheY-axis step changes from 10 ka in Figure 2.7A, to 50 ka in Figure 2.7B.Positive correlations1. The youngest GVB deposits record exclusively subaerial activity (Figure 2.7A). TheMount Garibaldi (block and ash) and Ring Creek lavas are both accurately dated,and attest to the low elevation areas surrounding Squamish being ice free by ∼12ka, following the demise of the last glaciation (MIS 2, Fraser). This conclusion cor-roborates the local glacial sedimentary record (Brooks and Friele, 1992; Friele andClague, 2002, 2009; Friele et al., 1999) and is in agreement with the MIS record(Lisiecki and Raymo, 2005).2. The youngest Cheakamus basalt lavas were confined to pre-existing subglacialdrainage channels during the waning stages of the last glaciation (Mathews, 1958)(Figure 2.7A). Importantly, these deposits overlie and preserve glaciofluvial sedi-ment with carbonized wood that is dated at 34.2 ± 0.8 ka (Green, 1981b). Whilethe carbonized wood gives a maximum age for the basalts, it also indicates that theSquamish valley was ice free at ∼500 m elevation during MIS 3 (Olympia intersta-dial). These glaciovolcanic deposits preserve a record of the local climate, endorsingboth the Quaternary sedimentary record and the MIS timescale (Hebda et al., 2016;Lisiecki and Raymo, 2005).463. Lava-dominated tuyas in the GVB provide a method for estimating the absolute sur-face height of the local enclosing ice sheet. The MIS timescale suggests that the vol-ume of ice in the GVB during MIS 3 should be∼75% of that experienced during MIS2 (LGM), when the ice over the southwest BC Cordillera reached ∼2500–3000 m inelevation (Clague and Ward, 2011; Ryder et al., 1991). Ring Mountain, is dated at49.1± 5.5 ka, and displays ice-confined margins and a subaerial upper surface whichindicate an ice surface height of∼2190 m. The volcanism corresponds with late MIS4, or early MIS 3, and indicates a substantial accumulation of ice in SWBC at thistime (Figure 2.7A). This record of ice is under-represented by the MIS time-scale andsuggests that the MIS 4 glacial advance may have been as, or more extensive than theMIS 2 glacial period in SWBC (see chapter 3).4. The oldest Cheakamus basalts indicate a period of no ice in the Cheakamus valleyat 140.8 ± 12.9 ka. This age is within uncertainty of MIS 5e, a warm interglacialperiod.5. The uncertainty on the radiometric age of the ice-impounded Desolation valley flowsprecludes accurate correlation with the MIS. In reality, the DVF may correspondwith any of the climatic fluctuations during MIS 5. We suggest that DVF may recorda period of ice advance leading up to MIS 5d (Figure 2.7A).6. Indicators for glaciovolcanism at the Black Tusk suggest that it is a high-altitude lava-dominated tuya that records an ice sheet surface at ∼2300 m. Recent geochronologycorrelates the volcanism with MIS 6, and suggests that at this time, the ice volume insouthwest BC was roughly equivalent with MIS 2 (i.e., the LGM). It is plausable thatthe Mount Fee (subaerial) deposits are highly eroded and were actually formed in ahigh-alpine glacial environment (see below).7. The mid-Pleistocene glaciovolcanic record is sparse based on the available data. Thisis partly due to a lack of high-resolution geochronological constraints on many GVBdeposits, and partly due to an apparent bias towards higher-elevation edifices. Theimportance of the record during this time is increased, as glaciovolcanoes, and vol-canic deposits that geochronologically constrain glacial sediments, provide the onlyphysical record of the mid-Pleistocene CIS in southwest BC (Ryder et al., 1991).Within this timeframe, however, the passage zone preserved at Salal tindar (STY)records a paleo-englacial lake and a minimum ice surface of 2360 m (probably atleast ∼2400 m). Although the uncertainty on this age is significant, it correspondswith a short period of ice advance in the middle of MIS 15 (Figure 2.7B).478. The passage zone at Tuber Hill records a minimum ice surface of ∼2000 m duringearly MIS 15. These rocks align closely with a small period of increased global icevolume between MIS 15 and 14. Similarly, older subaqueous deposits at Tuber Hilltransect an elevation range of 480 m (up to 2180 m elevation) and suggest that theice surface resided at >2413 m during MIS 18. This is roughly equivalent with themagnitude of ice cover during the LGM.Discordances1. Combinations of temporally overlapping glaciovolcanoes can be used to highlightthe inability of the MIS technique to accurately represent the spatial distribution ofice across the planet. As discussed earlier, the young Cheakamus basalts preserve arecord of MIS 3, the Olympia interstadial. At this time, the lower Cheakamus Val-ley must have been ice free at ∼500 m or more. Ring Mountain was also eruptedduring MIS 3, and records extensive ice cover up to 2190 m. The deposits are ∼24km apart, yet they define a change in ice-surface elevation of ∼1690 m. This vari-ance cannot be easily explained using existing corrections for ice-sheet profiles inmountainous regions (i.e., parabolic ice sheet decay; Livingstone et al., 2013; Nye,1952b). Evaluated together, the glaciovolcanoes demonstrate a powerful ability todefine location-specific, paleo-ice sheet extents at a resolution that is invisible to theMIS timescale.2. The Table also presents a discordance with the MIS. Recent 40Ar/39Ar geochronol-ogy correlates the Table with MIS 5d–5b, a period of globally fluctuating ice levelsthat followed the relatively warm MIS 5e (Sangamonian) interglacial stage (Clagueet al., 1992; Hicock and Armstrong, 1983). The MIS indicates that the ice volume inthe GVB at this time should be∼50% of what was experienced during the LGM (i.e.,MIS 2). In other words, the MIS suggests that the ice sheet surface covering the GVBshould reside at ∼1250–1500 m in elevation. The Table, however, demonstrates iceat ∼2010 m elevation (and more likely at 2110 m; see previous sections). This is∼65–80% of the LGM level, suggesting that at this time, the paleoclimate in south-west BC was somewhat colder and/or wetter than the global average (Figure 2.7A).3. The Mount Fee deposits appear to be subaerial, which suggests that there was noice at ∼1200 m elevation during their eruption. Their age overlaps with the BlackTusk and Pali Dome East, both of which record high-elevation ice up to 2300 m.Two explanations exist. Either the Mount Fee deposits are glaciovolcanic and areseverely eroded, or they correspond with a minor period of ice retreat during MIS486 (Figure 2.7B). If the Mount Fee deposits are indeed subaerial, this discrepancy isvery similar to the one that is recorded by Ring Mountain and the young Cheakamusbasalts. The ice sheet elevation change that is demonstrated by Mount Fee and theBlack Tusk (∼1100 m) appears to be under-represented in the MIS record.2.5 ConclusionsQuaternary volcanism in southwest BC is coupled to a dynamic history of waxing andwaning continental-scale ice sheets. As a consequence, the GVB hosts a diverse array ofglaciovolcanic deposits. The unique tectonic and physiographic setting of the belt controlsthe glaciovolcanic diversity, and each deposit preserves highly detailed and specialized in-formation concerning eruption environment. Our volcano database incorporates geospatial,lithological and geochronological information, and provides a comprehensive summary ofglaciovolcanism within the belt. We use several GVB-specific examples to demonstrateforensic, field-based techniques for extracting detailed paleoenvironmental information.These data, combined with precise geochronology, are used to develop an ice height re-construction for the GVB that spans the last 1 Ma. We compare this glaciovolcanic recordto the establishedMIS timescale, and highlight a number of positive correlations and discor-dances. The MIS is a high-resolution proxy for global ice volume. Our method, however,shows that it does not accurately record the local paleoenvironment experienced in south-west BC. We show that the glaciovolcanic record can document paleoenvironmental changethat is sometimes invisible to the MIS. Our analysis also provides an independent physicalrecord of numerous previously undocumented, Mid–Late Pleistocene glaciations in the re-gion, contributing significantly to existing studies of Quaternary sediments. We suggestthat all volcanoes (both glaciovolcanic and subaerial) are equally important for paleoen-vironmental reconstructive purposes. While demonstrating the powerful potential for thismethod, our reconstruction utilizes only∼25% of the available Quaternary volcanic centersin the GVB. Future geochronology and field-based studies will serve to continually refineand improve this approach.49Chapter 3Paleo-glacier reconstruction insouthwestern British Columbia,Canada: A glaciovolcanic model23.1 IntroductionLarge ice sheets in the Northern Hemisphere have waxed and waned in thickness and ex-tent since the Pliocene (Fulton, 1989; Richmond and Fullerton, 1986; Shackleton, 1987).During this time, the Cordillera of southwestern British Columbia (SWBC), Canada, hasbeen repeatedly enveloped by a large continental ice sheet, referred to as the CordilleranIce Sheet (CIS) (Clague, 2011; Jackson and Clague, 1991). Although the extent of thisice sheet during the Last Glacial Maximum (LGM) is relatively well understood (e.g., Fig-ure 3.1A), our knowledge of its previous incarnations is inhibited by a lack of reliable andquantifiable field data. Successive glacial advances commonly obscure the physical recordsleft by their predecessors and, thus, reconstructing older glaciations depends on alternativemethods (Stokes et al., 2015).One approach has been to use computational models to simulate interactions be-tween paleoclimate, glacier mass-balance and glacier dynamics (Jackson and Clague, 1991;Robert, 1991; Seguinot et al., 2014, 2016). These numerical models mostly use time-dependent, globally-averaged climate proxies such as ice core and/or marine isotope data todrive the simulations (e.g., Dansgaard et al., 1982; Lisiecki and Raymo, 2005). The integrityof these models is then evaluated against independent physical indicators for local/regional2A version of this chapter has been published. Wilson, A.M., Russell, J.K., Ward, B.C., 2019. Paleo-glacier reconstruction in southwestern British Columbia, Canada: A glaciovolcanic model. Quaternary ScienceReviews. 218, 178–188. https://doi.org/10.1016/j.quascirev.2019.06.024.50 47.5° N   50.0° N   52.5° N   115.0° W 117.5° W 120.0° W 122.5° W 125.0° W 127.5 ° W4.4 4.6 4.8 5 5.2 5.45.485.55.525.545.565.585.65.625.645.660 1000 2000 3000 4000-1000-2000-3000x easting (m × 105) y northing (m × 10 6) Pacific OceanModel RegionInterior PlateauCanadaUSAWashington IdahoMontanaCanadianPrairiesElevation (m asl)UTM Projection: Zone 10 NRing MountainCheakamus basaltsTuber HillQuaternary Volcanic RocksLGM IceVancouverLast Glacial Maximum (~17 ka) ice distributionTowards Fraser LowlandsFigure 3.1: Location and physiographic setting of southern British Columbia, Canada.The distribution of Quaternary volcanic rocks is outlined in red. The locations ofspecific volcanoes used in this study are indicated. The yellow line correspondsto the Last Glacial Maximum Cordilleran Ice Sheet extent at∼17–14 ka. (Boothet al., 2003; Clague, 2011). The region considered is within the white box.ice, including glacial sedimentary sequences, glacial trim lines, geomorphological indica-tors, or biological, ecological and paleontological records. In general, for reconstructionsthat are aimed prior to the LGM (i.e., >17 ka), evaluating such numerical models is limitedby a paucity of physical evidence (Bowen et al., 1986; Clague, 2011; Evans et al., 2005;Kleman et al., 2010; Rutter et al., 2012; Smellie, 2018).Glaciovolcanic deposits provide a direct, datable record of ancient ice (Smellie, 2018;Smellie and Edwards, 2016). Glaciovolcanic rock sequences are distinctive and can be usedto establish: i) whether ice was present or absent during eruption (e.g., chapter 2), ii) theage of that ice, by radiometric dating of the volcanic rocks (e.g., Smellie, 2018; Smellieet al., 2009, 1993), and iii) the minimum, maximum or absolute thickness of overlying ice(i.e., ice surface elevation; Edwards et al., 2011; Russell et al., 2014). Conversely, sub-aerial (non-glaciovolcanic) rocks erupted into a landscape periodically occupied by ice areequally powerful in that they record times of ice-free environments. As paleoenvironmentalindicators, volcanic rocks offer significant advantages over other records of glaciation (e.g.,tills) because they are resistant to glacial and other erosional processes (Jakobsson and Gud-mundsson, 2008; Kleman, 1994; Smellie and Edwards, 2016) and are generally amenable to51radiometric dating. Furthermore, glaciovolcanic rocks are able to provide a chronology pastthe range of radiocarbon-dated deposits (i.e., >50 ka). In this manner, glaciovolcanic de-posits are a powerful, empirical record of global paleoenvironments (Smellie, 2018; Smellieet al., 2008, 2011). In this study, we use the collective term ’glaciovolcanoes’ to refer to allvolcanic rocks that record information pertaining to the distribution of paleo-glaciers.Our goal is to reconstruct paleo-glacier distributions for the Coast Mountains of SWBCusing data from glaciovolcanoes. The primary data are collected from three glaciovolcanoesand include: geographic position, age, and the elevation of the coeval ice surface that theyrecord. Our study uses a simple geometric model for the growth and coalescence of glacierswhich we use to create a relative sequence of regional ice distributions. Integrating the harddata from the glaciovolcanoes into the geometric ice model provides snapshots of paleo-glacier distributions across SWBC at specific times of ∼50, ∼140 and ∼600 ka.3.2 Glaciovolcanism in the Garibaldi volcanic beltQuaternary magmatism in SWBC has produced more than 100 eruptive centers (Hildreth,2007). The deposits are mostly <1 Ma in age and comprise the Garibaldi volcanic belt, thenorthernmost (Canadian) portion of the Cascade volcanic arc of the western United States(Figure 3.1) (Green et al., 1988; Mathews, 1958; Souther, 1991). The region has a diversearray of volcanic deposits, reflecting a wide range of magma compositions, eruption stylesand eruption environments, and, more than 50% of the volcanoes interacted with ice duringeruption(see chapter 2). The volcanic rocks in the Garibaldi volcanic belt provide a large,empirical dataset that informs on the presence, thickness and transient properties of the CISand other ice masses over the last ∼1 Ma (Green et al., 1988; Mathews, 1958).3.3 Glaciovolcanoes as recorders of paleoenvironmentGlaciovolcanic deposits have long been recognised for their significance as indicators of lo-cal and global paleoenvironments (e.g., Allen et al., 1982; Grove, 1974; Jones, 1968; Math-ews, 1947; Noe-Nygaard, 1940; Pjetursson, 1900). They offer windows into the state ofancient ice and a means to validate computational paleo-glacier reconstructions with directgeologic field evidence (e.g., Smellie et al., 2006, 2011). In British Columbia, individualglaciovolcanoes have been used to establish the presence of paleo-ice and to estimate localice distributions and thicknesses (Edwards et al., 2011, 2009; Hickson et al., 1995; Huscroftet al., 2004; Jackson et al., 2012; Lawrence et al., 1984; Mathews and Rouse, 1986; Ryderet al., 1991).Here, we use paleoenvironmental data collected from three volcanoes in the Garibaldivolcanic belt of SWBC to provide constraints on paleo-glacier distributions at times cor-52responding to the ages of the three eruptions. The volcanic localities include: i) RingMountain, ii) the Cheakamus basalts, and, iii) Tuber Hill (Figure 3.1B). Below, we providea brief summary of the physical volcanology and paleoenvironmental implications of eachdeposit. For detailed lithologic and stratigraphic descriptions we refer the reader to: Green(1981b); Lawrence et al. (1984); Roddick and Souther (1987); Kelman et al. (2002a); Kel-man (2005) and chapter 2 of this thesis. To remain consistent with chapter 2, we report theuncertainty on all geochronometric age determinations as one standard deviation from themean age.Ring Mountain (Figure 3.2A) is an andesitic, lava-dominated tuya located in the MountCayley volcanic field (Kelman et al., 2002a). Ring Mountain is situated between the Elahoand Callaghan river valleys and forms a prominent, flat-topped, isolated topographic fea-ture (see chapter 2). The volcano has an ice-confined lower sequence and an oxidised,subaerially-emplaced upper surface that constrains the absolute surrounding glacier surfaceat the time of the eruption to be ∼2190 m asl (Kelman et al., 2002a; Russell et al., 2014).A sample collected from the northern side of the edifice has an age of 49.1 ± 5.5 ka (seechapter 2).The Cheakamus basalts comprise a ∼70 m-thick package of at least 14 individual lavasthat occupy a 1 to 2 km-wide, 22 km-long section at the base of the Callaghan and Cheaka-mus river valleys (Figure 3.2B) (Green, 1981b; Mathews, 1958). One of the stratigraphi-cally oldest lavas has an age of 141± 12.9 ka (see chapter 2). These lavas are demonstrablysubaerial and did not interact with ice or water during their eruption or emplacement. Theyrecord the presence of an ice-free corridor over their entire distribution, from 510 to 360 masl (see chapter 2).Tuber Hill (Figure 3.2C) is a large, polygenetic tuya that comprises a thick sequenceof basaltic pillow lava, pillow breccia and hyaloclastite that is overlain, via a passage zone(Russell et al., 2014), by subaerially-emplaced basalt (Roddick and Souther, 1987). Thepassage zone records the presence of an ephemeral lake surface that existed at∼2010 m aslat 598 ± 7.5 ka (see chapter 2). The deposits are located at high-elevation, in a rugged areaat the northern end of the Garibaldi volcanic belt (the Bridge River volcanic field). In thislocation, there is no plausible mechanism, other than ice, for damming and maintaining anextensive and required (englacial) body of water. The deposits preserve a minimum glaciersurface of 2010 m asl (see chapter 2).3.4 Geometric model for paleo-glacier distributionsOur goal, is to explore the larger implications of each of the three volcanoes for paleo-glacier distributions across SWBC. Specifically, we ask how isolated determinations of53ABC~400 m~1 kmFigure 3.2: Field photographs of volcanic edifices used to drive geometric ice distri-bution scenarios. A) Ring Mountain, a lava-dominated tuya (photograph look-ing north). B) Columnar-jointed basaltic lava flow from the Cheakamus basalts(photograph looking west). Person circled for scale. C) Tuber Hill, a polygeneticbasaltic tuya (photograph looking north).54paleo-glacier presence and elevation (i.e., determined from the glaciovolcanoes) can beused to evaluate the distribution of ice across a wider region. To address this question,we develop a simple geometric model for the relative sequential growth or decay of ice inmountainous regions. The model is geometric in that: i) the distributions by themselveshave no time connotation, ii) the model is not based on ice-flow physics, and iii) the modeldistributions of ice are independent of climate forcing (e.g., globally-averaged paleoclimateclimate proxies). Most computational models for paleo-glacial ice sheets rely on data repre-senting transient climate records as inputs to drive the models. We have chosen a geometricapproach because it facilitates simple and rapid estimations of ice distribution that are inde-pendent of climate forcing and can be applied over longer paleo-times to include the agesof older glaciovolcanic successions (e.g., in this case, up to ∼1 Ma).Our model first replicates the present-day glacier distribution. Then, using a simple setof geometric rules, we extend the ice distributions by growing the glaciers until they meetthe local constraints imposed by the glaciovolcanic rocks (i.e., the location and surfaceelevation of a paleo-glacier).3.4.1 Geometric model methodsIn the mountains, ice sheets consist of a complex, coalescent system of cirque glaciers, val-ley glaciers and piedmont glaciers (Booth et al., 2003; Clague, 2011; Jackson and Clague,1991). Previous workers have outlined a four-phase model for glacier expansion to form theCIS (Davis and Mathews, 1944; Fulton, 1991; Kerr, 1934; Ryder et al., 1991; Stumpf et al.,2000). In this model, small alpine glaciers (phase 1, “the alpine phase”) grow and coa-lesce into a system of piedmont glaciers and ice fields (phase 2, “the intense alpine phase”).Eventually, the glaciers form a large, confluent system of glaciers (phase 3, “the mountainice sheet phase”). Infrequently, further glacial expansion develops a larger dome of ice witha single centre of accumulation whose flow is unconstrained by basal topography (phase4, “the continental ice sheet phase”). Our geometric model is based on this conceptualidea, coupled to the idea that the majority of glaciers in each of the phases can be broadlyrepresented by average surface angle (e.g., Haeberli and Hoelzle, 1995; Raup et al., 2007).We adopt the terminology ”phase 1” and ”phase 2” etc. to refer to these sequential glacierexpansions. Our modelling does not extend to phase 4.The surface angle of a glacier is a direct reflection of the depth of glacier ice in com-parison to the relief of the surrounding land surface (Davis and Mathews, 1944). Duringphase 1, the majority of the glaciers in the Cordillera are steeply-dipping cirque glacierswith average surface angles of ∼30◦ (Davis and Mathews, 1944; Kerr, 1934; Raup et al.,2007). As these glaciers grow and coalesce (i.e., into phase 2), the alpine cirque glaciers55develop more moderate surface angles of∼15◦, forming large valley glaciers, ice fields andice caps (Clague, 2011; Davis and Mathews, 1944; Stumpf et al., 2000). During phase 3,when a confluent system of glaciers has formed (i.e., a coalesced ice cap or ice sheet), theglaciers have shallow surface angles of ∼5◦ or less) (Mathews, 1958, 1974; Sugden andJohn, 1976). During the first three phases of glacier expansion, the glacier distribution isintimately tied to the topography (Clague, 2011; Davis and Mathews, 1944; Fulton, 1991;Jackson and Clague, 1991; Stumpf et al., 2000). We suggest that the glacier surfaces can berepresented by linear-sided cones of ice projected from nodal positions on the topography(Figure 3.4). Growth or decay of the glaciers can be achieved by systematically loweringor raising the surface angles of the linear-sided cones of ice, respectively.To initiate the geometric model, we use the Global Land Ice Measurement from Space(GLIMS) inventory of existing glacier outlines (Raup et al., 2007) in SWBC and a DigitalElevation Model (DEM) of SWBC resampled to a grid-cell-size of 1000-by-1000 m grid-cell-size (resolution) (Figure 3.3A and Figure 3.3B). We extract the regular DEM griddedpoints (i.e., including the easting, northing and elevation) that lie within the confines of theexisting glaciers (Figure 3.3B). The extracted grid cells serve as nodal positions from whichlinear-sided cones of ice are projected to reproduce the glacier surfaces (Figure 3.3C). Thecone apexes are located at the same elevation as the original DEM surface. We assign asteep, initial cone angle of 30◦ to all nodes representing steep alpine glaciers of phase 1.A value of 30◦ was chosen as a starting angle because, at the DEM resolution (1000 m),this angle is just steep enough to not project significantly beyond the edges of each grid-ded cell in most topographic situations, thereby preserving the input GLIMS distribution ofglaciers (Figure 3.3C). Steeper starting angles are less realistic because alpine glaciers arerarely steeper than 30◦, and operationally, we found that angles steeper than 30◦ yieldedno glacier growth during the initial stages whilst shallower angles over-represent the inputglacier distribution. We then compute the ice distributions over 30 model scenarios, ex-panding the glaciers systematically by reducing the surface angle of the cones of ice in 1◦increments (Figure 3.3D). A maximum model ice distribution is achieved where all nodesproject cones of ice at an angle of 1◦. Portions of overlapping cones are merged to cre-ate a single glacier surface and portions of cones that lie beneath the current topographyare removed (Figure 3.3C and Figure 3.3D). We also remove any portions of cones thatare physically unconnected from nodal sources (i.e., portions of cones that are projectedthrough a hill/ridge and re-emerge in an adjacent valley; Figure 3.3D). Figure 3.4 shows, in3-D, how the geometric method may be applied to reconstruct the former extent of a singleglacier, in this case, the Job Glacier on Mount Meager. In this example (i.e., Job Glacier),we have used a DEM with a 100 m resolution (as opposed to 1000 m resolution) becauseof its small size and extent.56High-resolution topography (DEM)GLIMSGlaciersElevation1000 m resolutiontopography (DEM)GLIMS glacier ‘nodes’1000 mABCDPortion of overlapping conesModelled glacier surface 30°“Cones”“Cones”10°Portion of cones underlying topographgyPortion of cones projected through topographyModelled glacier surface (merged cones)Surface angleSurface angleFigure 3.3: Cross-section diagram illustrating geometric cone method for growth anddecay of glaciers in 2-D. A) High-resolution topography is derived from a digitalelevation model (DEM). The real distribution of existing glaciers in the modelregion is derived from the Global Land Ice Measurements from Space (GLIMS)inventory of glaciers (Raup et al., 2007). B) Resampled DEM and glacier dis-tribution at a resolution of 1000 m. The extracted GLIMS glacier “nodes” areshown using red markers. C) Linear-sided “cones” representing ice (blue shadedregion) projected at each of the GLIMS glacier “nodes”. The cones have a steepinitial surface angle of 30◦. Portions of cones that are removed are indicated withdashed lines. The modelled glacier surface is shown using blue markers and ablue line. D) Simulated growth of glaciers using shallow 10◦ surface angle conesprojected at each of the GLIMS glacier “nodes”. The modelled glacier surface(blue markers and blue line) is created by merging cones that project above theDEM surface. The two nodal positions responsible for the glacier surface areindicated using yellow stars.574.584.594.64.614.624.634.644.654.665.6055.615.6155001000150020002500elevation (m asl)BCDx easting (m × 10 5) y northing (m × 10 6) A58Figure 3.4: Block diagrams of an arbitrary example glacier (the Job Glacier, MountMeager), illustrating the geometric method in 3-D. A) Global Land Ice Measure-ments from Space (GLIMS; Raup et al., 2007) outline of the Job Glacier (yellowline), and extracted nodal positions (red dots). Note: this example glacier isconstructed at a 100 m grid-cell-size (resolution). B) Present-day glacier distri-bution (modelled glacier is the blue surface) derived from linear-sided cones ofice at 30◦, projected at each of the nodal positions representing Job Glacier. C)Arbitrary paleo-glacier distribution derived from nodal positions with cone sur-face angles of 16◦. D) Arbitrary paleo-glacier distribution derived from all nodalpositions extracted from all of the glaciers in the example model region (note:each glacier, including the Job Glacier, is outlined using yellow lines). The nodalpositions are projecting cones surfaces of 13◦, creating a single glacier surface.Our geometric method closely replicates the present-day (phase 1; alpine phase) glacierdistribution (Figure 3.5). There are only minor cases of model overestimation involvingsome extremely steep-sided alpine glaciers (>30◦) where our model erroneously projectscones of ice into the surrounding valleys (Figure 3.5). Below, we outline the main caveatsand assumptions of our geometric approach.3.4.2 Model caveats and assumptions1. The present-day glacier distribution is assumed to be the minimum amount of ice overthe model timescale. This is broadly appropriate for SWBC (e.g., Bolch et al., 2010;Clague andWard, 2011; Schiefer et al., 2007), although, the last interglacial transition(Marine Isotope Stages; MIS 5e) was probably slightly warmer (and therefore mayhave had slightly less ice) than today (Clague et al., 1992; Hicock and Armstrong,1983).2. The paleo-glacier distributions are constructed on top of the current topography,which includes existing glaciers. There is no topographical change (i.e., erosion)over the model lifespan. This is reasonable at 1000 m resolution for SWBC, wheremost major topographic features are considered to be at least several million yearsold (Clague, 2011; Densmore et al., 2007; Gascoyne et al., 2006; Tribe, 2005). Allelevations are given relative to current sea level. Isostatic depression or rebound, sealevel change and regional tectonic uplift are not considered.3. All glacier surfaces are constructed as linear-sided cones. We favor a linear approxi-mation over more complex glacier surface profiles (e.g., parabolic; Benn and Hulton,2010; Mathews, 1974; Nye, 1952a,b) because a linear method allows for a contin-59uum of glaciers to be evenly modelled using a single equation (i.e., from steep cirqueglaciers to shallower ice fields and ice caps).4. All nodal positions initially project ice cones starting at 30◦ after which they aredecreased in even increments of 1◦. This assumes that all glaciers have the samesurface profile to begin with and all grow equally. This is probably the biggest caveatof our geometric approach, because, in reality the glaciers that exist today in themodel region have a range of surface angles that vary from steep (e.g., cirque) toshallow (e.g., ice fields) and also change over a single the glacier. Our model onlyprovides a broad representation of average glacier surface slope. However, becauseour method combines the surfaces of overlapping cones (Figure 3.3C), this allowslarger, shallow-surfaced ice masses to be well-represented during early model sce-narios, thus, we were able to approximate the distribution of most present-day valleyglaciers and ice fields reasonably well at the resolution and scale considered (e.g.,Figure 3.5).5. No new nodal positions are created during glacier growth. In reality, this is not thecase as there are numerous cirques with no ice at present that would have becomeglacierized. However, given the resolution and scale of the model, this is not a sig-nificant drawback.603.5 4 4.5 5 5.5 65.455.55.555.65.655.7Model ice ≠Real icex easting (m × 10 5) y northing (m × 10 6) VancouverIce surface height (m asl)0 1000 2000 3000 4000Elevation (m asl)Figure 3.5: Cell-by-cell comparison of modelled present-day ice distribution versusGlobal Land Ice Measurements from Space (GLIMS; Raup et al., 2007) inven-tory of glaciers. The modelled present-day ice distribution (blue surface) hasnodal positions that project ice cones with steep, 30◦ surface angles. Portionsof the modelled distribution where the cones extend beyond the real present-dayglacier distribution (i.e., GLIMS inventory) are shown in orange.615.455.55.555.65.655.75.455.55.555.65.655.7B3.5 4 4.5 5 5.5 65.455.55.555.65.655.7Cx easting (m × 10 5) y northing (m × 10 6) y northing (m × 10 6) y northing (m × 10 6) AIce surface height (m asl)0 1000 2000 3000 4000Elevation (m asl)aa’aa’aa’VancouverVancouverVancouverScenario 15(phase 1-2)Scenario 25(phase 2-3)Scenario 30(maximum configuration)62Figure 3.6: Modelled geometric ice distributions corresponding to the phase 2 (”in-tense alpine”) and phase 3 (”mountain ice sheet”) glaciations (Davis and Math-ews, 1944; Fulton, 1991; Ryder et al., 1991; Stumpf et al., 2000): A) “intensealpine” phase 2 glaciation (model scenario 15 with glacier cone angles of 15◦),B) the “mountain ice sheet” phase 3 glaciation (model scenario 25, minimumglacier angle of 5◦), and C), the maximum geometric ice distribution scenario 30(glacier cone angle of 1◦). The coloured cross section lines (red, green and blue)correspond with the coloured glacier profiles drawn in Figure 3.7.3.5 Modelled paleo-glacier distributionsOur modelled paleo-glacier distributions consist of 30 glacier configurations that are createdby sequential changing of the angle of the cones of ice. We have shown three exampleglacier configurations in Figure 3.6. The three examples correspond to the glaciation phasetransitions of phase 1 to phase 2 (Figure 3.6A), phase 2 to phase 3 (Figure 3.6B) and themaximum modelled glacier configuration (Figure 3.6C). The cross-sectional profile of allmodel scenarios is shown in Figure 3.7.A phase 2 glaciation is represented by our model when the nodal positions project coneswith an angle of ∼15◦ (scenario 15; Figure 3.5A and Figure 3.6). Near-complete mountainglacier coalescence (i.e., into phase 3) occurs when the nodal positions project cones withan angle of ∼5◦ (scenario 25; Figure 3.6B). At this stage, a well-developed ice cap coversthe Cordillera and large outlet glaciers are present in most of the fjords and sounds alongthe western coastal margin (Figure 3.6B). At the maximum model configuration (i.e., thenodal positions project cones with an angle of 1◦) the ice has overwhelmed present-dayVancouver and has coalesced with glaciers located on Vancouver Island (Figure 3.6C). Thisstage is represented by a broad dome of ice with a divide centered over the axis of the CoastMountains (Figure 3.7). The maximummodel configuration is not equivalent to phase 4 anddoes not completely reproduce the extent and thickness of the CIS at its maximum extent(Clague and Ward, 2011; Ryder et al., 1991; Seguinot et al., 2016). This is because: i)the geometric model does not reflect glacier contributions from outside of the model region(e.g., the Rocky Mountains, which are located∼150 km to the east of the region boundary),and ii) the glacier distribution in our geometric model is intimately tied to the topography.For this reason, the northeastern portions of our simulations (Figure 3.6) are consideredextreme minima.633.8 4 4.2 4.4 4.6 4.8 5 5.201000200030004000Elevation (m asl) 8x vertical exaggeration51015202530Scenario #> Phase 3Phase 2-3Phase 1-2x easting (m × 10 5) a a’Figure 3.7: Cross section showing surface profiles of modelled glaciers. The trace (a-a’) of this cross section is shown in Figure 3.6A, Figure 3.6B and Figure 3.6C.The glacier surface profiles show all 30 model scenarios and are coloured usingred (scenarios 1–15), green (scenarios 15-25) and blue (scenarios 25-30), cor-responding with phase 1-2, phase 2-3 and >phase 3 glaciations, respectively.The darkest shade of each color corresponds with the scenarios represented inFigure 3.6A, Figure 3.6B and Figure 3.6C.3.6 Paleo-glacier distributions constrained by theglaciovolcanoesWe selected the three glaciovolcanoes based on: i) the quality of the geochronometric agedetermination, ii) the confidence and unambiguousness of the physical volcanological in-terpretation of each deposit, and, iii) the temporal and spatial importance of the deposit forresolving important climatic transitions or previously undocumented glacial events. (i.e.,absolute glacier surface height, absence of ice and minimum glacier surface height) and,thus, illustrate different ways in which the geometric model can be used to reconstructpaleo-glaciers. The deposits record the glacial conditions in SWBC during: i) the MIS 4 to3 transition, ii) the MIS 6 to 5e transition, and iii) the middle of MIS 15. The MIS 4 to 3 and6 to 5e transitions both illustrate deglaciation events, when a major ice sheet was rapidlydisappearing from SWBC. The exact timings and magnitudes of each of these deglaciationsis a topic of ongoing debate (e.g., Armstrong et al., 1965; Clague and Ward, 2011; Cosmaet al., 2008; Hebda et al., 2016; Hicock and Armstrong, 1983). MIS 15 refers to a relativelyminor interglacial period during the Mid-Pleistocene that also hosted a short advance inglobal ice volume (Lisiecki and Raymo, 2005). The Tuber Hill deposits provide the firstphysical evidence of the MIS 15 global ice advance manifesting as a major glaciation inSWBC. In the following sections we discuss, in detail, the modelled paleoenvironmentalimplications of each deposit.643.6.1 Ring MountainThe eruption of Ring Mountain required an enclosing glacier with a surface elevation of2190 m asl. This glacial extent is represented by our geometric scenario 28, where thenodal positions are projecting cones with surface angles of 2◦. (Figure 3.8A). This stage isequivalent to a phase 3 glaciation. This suggest that, within the timespan bracketed by theuncertainty on the age of Ring Mountain (i.e., 49.1 ± 5.5 ka), the Coast Mountains werecovered by a substantial, coalesced ice cap. The areas surrounding present day Vancouverand the southern part of Vancouver Island were probably ice free, but most of the intermon-tane valleys to the north were glaciated. Major outlet glaciers were maintained in the fjordsof the lower mainland (e.g., the Howe Sound).Physical evidence and numerical models both suggest that MIS 4 (Figure 3.9) produceda well-coalesced CIS in SWBC, although, the temporal extent and magnitude of this icesheet are still unresolved (Clague, 2011; Cosma et al., 2008; Hicock and Armstrong, 1983;Seguinot et al., 2016; Troost, 2016; Ward et al., 2007). Although there is compelling phys-ical evidence to suggest that the MIS 4 glaciation was extensive in the Yukon (Ward et al.,2007), the deposits at Ring Mountain provide the first physical, terrestrial evidence indicat-ing that the MIS 4 glaciation was also widespread and evident in the southern Cordillera.The age of Ring Mountain does not align with the peak of MIS 4, instead, it is within un-certainty of the end of this glacial stage. These rocks probably llustrate a significant accu-mulation of ice in SWBC during the waning stages of MIS 4, or possibly during early MIS3 (the Olympia Interstade; Hebda et al., 2016). This eruption probably corresponded withthe transition between MIS 4 and MIS 3 (i.e., a deglaciation; Figure 3.9). Although majorglaciers may have existed in the mountains, the Fraser lowlands were likely to have beenice free (e.g., Hebda et al., 2016). Our glacier configuration is supported by: i) ice-rafteddebris (dated at ∼47 14C ka; at the limit of radiocarbon) located offshore of VancouverIsland, indicating deglaciation of a large MIS 4 ice sheet that reached the continental shelf(Cosma et al., 2008), and ii) numerical simulations of the CIS during the MIS 4 to MIS 3transition (Seguinot et al., 2016), which persistently show a coalesced ice cap formed overthe southern Coast Mountains during this time.65x easting (m × 10 5 ) y northing (m × 10 6) x easting (m × 10 5) y northing (m × 10 6) x easting (m × 10 5) y northing (m × 10 6) x easting (m × 10 5) y northing (m × 10 6) 4.654.74.754.84.855.545.5455.555.5555.565.5650100030003.5 4 4.5 5 5.55.455.55.555.65.655.7x easting (m × 10 5 ) y northing (m × 10 6) 4.754.84.854.94.9555.055.5355.545.5455.555.5555.565.5655.570100030003.5 4 4.5 5 5.55.455.55.555.65.655.7x easting (m × 10 5 ) y northing (m × 10 6) 4.554.64.654.74.754.85.635.6355.645.6455.655.6550100030003.5 4 4.5 5 5.55.455.55.555.65.655.7ABC= SCENARIO 28Age: 49.1 ± 5.5 ka (MIS 3)Ice surface elevation: = 2190 m aslAge: 140.8 ± 12.9 ka (MIS 6–5e)Ice surface < 510 m to 360 m asl Age: 598 ± 7.5 ka (MIS 15)Ice surface elevation: >2010 m aslWhistler< SCENARIO 15> SCENARIO 26Cheakamus BasaltsTuber HillRing Mountainelevation (m)elevation (m)elevation (m)66Figure 3.8: Block diagrams illustrating the local and regional paleo-glacier implica-tions for three glaciovolcanoes. A) Ring Mountain; absolute ice surface of 2190m asl at 49.1 ± 5.5 ka which corresponding with scenario 28. B) The Cheaka-mus basalts; ice-free environment (maximum ice extent) at 140.8 ± 12.9 kacorresponding with scenario 15 or less. C) Tuber Hill; minimum ice surface of2010 m at 598 ± 7.5 ka corresponding with scenario 26 or greater. The yellowboxes show the aerial extent of the smaller 3-D block diagram regions.3.6.2 The Cheakamus basaltsSediments on southwestern Vancouver Island indicate that the local climate during MIS 5e(Sangamonian stage) may have been warmer and drier than today (Armstrong et al., 1965;Clague et al., 1992; Clague and Ward, 2011; Hicock and Armstrong, 1983) (Figure 3.9).The Cheakamus basalts also contribute towards resolving glacier distributions during thisperiod. These subaerial lavas did not interact with ice or water during their eruption. Theyindicate, at a minimum, an ice-free corridor in the Callaghan and Cheakamus river valleys.This eruption occurred during a phase 1, or a phase 2 glaciation, as represented by ourgeometric model at a maximum configuration of scenario 15 (where the nodal positions areprojecting cones with surface angles of 15 or more) (Figure 3.8B).Sporadic evidence exists for a well-developed MIS 6 (Illinoian) glaciation in the Fraserlowlands of SWBC (e.g., the Westlynn drift; Clague andWard, 2011) (Figure 3.9). Previousworkers have suggested that the late MIS 6 to 5e transition (a deglaciation) was probablyrapid and similar to the last deglaciation (e.g., Hicock and Armstrong, 1983). The volcanicrecord corroborates this, although our interpretation is hampered by the large uncertainty onthe age of our sample. We suggest that the age of the Cheakamus basalts is probably closerto the lower limit of the 40Ar/39Ar age (∼127.9 ka) if correlation is to be made with MIS5e (Figure 3.9). At this time, ice covering the Coast Mountains of SWBC had degradedleaving ice-free corridors in many of the larger valleys.3.6.3 Tuber HillThe Mid to Early Pleistocene glacial record in SWBC is unclear because: i) this timeframeextends beyond the limits of many of the paleoclimate proxies that have been used to informcomputational ice models of this region (e.g., Seguinot et al., 2016), and ii) there is a lackof glacial sedimentary evidence (e.g., tills) due to erosion and burial corresponding to thisperiod (Clague, 2011). In southern British Columbia, the deposits at Tuber Hill providea physical record of ancient ice during this time and record a minimum enclosing glaciersurface elevation of at least 2010 m asl. The deposits, dated at 598 ± 7.5 ka (Roddick and673 3.5 4 4.5 5 5.518O( )0100200300400500600700Time (ka)234567891011121314151617‰5e Realistic ageTuber HillCheakamus BasaltsRing MountainLess ice More iceLast Glacial MaximumSangamonianOlympia InterstadeIllinoian GlaciationFigure 3.9: The Marine Isotope Stages (MIS) δ O18 record for the last 700 ka (after.,Lisiecki and Raymo, 2005), a proxy for global ice volume. Individual glacialstages segregations and indicated with horizontal grey dotted lines and numbers.Glacial stages are numbered using even numbers, and interglacial stages withodd numbers. The ages of glaciovolcanoes: Ring Mountain (49.1 ± 5.5 ka), theCheakamus basalts (140.8 ± 12.9 ka) and Tuber Hill (598 ± 7.5 ka) are shownusing red points. The most realistic age of the Cheakamus basalts is indicatedwith a green point. The 1σ uncertainty on the age of the deposits are shown withyellow bands.68Souther, 1987), are represented by our geometric glacier scenario 26 or greater, where thenodal positions are projecting cones with surface angles of 4◦ or less (Figure 3.8C).These rocks provide evidence for one of the oldest recorded glaciations in the southernCordillera (Bowen et al., 1986; Clague, 2011; Richmond and Fullerton, 1986). Their agealigns with MIS 15, technically an interglacial period that also had a period in the middlethat indicates a moderate increase in global ice volume (Figure 3.9). Like Ring Moun-tain, the deposits at Tuber Hill also correspond with a phase 3 glaciation, however, as theserocks only constrain the minimum surrounding glacier surface height, they may also recorda phase 4 glaciation (i.e., a fully coalesced CIS). In support of our interpretation, we notethat the Tuber Hill deposits overlap temporally with nearby glaciovolcanic rocks erupted atSalal tindar (590 ± 50 ka; see Lawrence et al., 1984). The field evidence from the Salalglaciovolcanic centre is consistent with a similar geometric-model scenario (i.e., scenario25 or greater). In this case, the two deposits work together to indicate that even during this(presumably warm) interglacial, SWBC hosted a substantial glaciation, likely correspond-ing to the period in the middle of MIS 15 (see Lisiecki and Raymo (2005), their Figure 4).The minimum ice configuration for this time included a well-coalesced ice cap covering theCoast Mountains with outlet glaciers extending to the present-day shoreline in most of thelarger fjords.3.7 Challenges faced by paleo-glacier reconstruction usingglaciovolcanoesDetermining ice distributions prior to the last glaciation is difficult due to the lack of robustempirical data. Glaciovolcanoes offer a means to address this gap in information by provid-ing a physical record of ancient terrestrial glaciers. However, by themselves, glaciovolcanicdeposits only offer isolated windows into the state of local ice. Our pilot study addressesthis by using a geometric model paired with data from glaciovolcanoes to provide estima-tions of regional glacier distributions. The main limitation to our approach is the inherentuncertainty in the age determinations for the deposits that we use to inform the geometricmodel. Currently, even the best 40Ar/39Ar and K-Ar geochronometric determinations pro-duce uncertainties (1σ ) of at least several thousand years. In some situations, the magnitudeof uncertainty in the radiometric age is acceptable when compared against high-resolution,precise paleo-climate proxies (e.g., the MIS record). For example, the Tuber Hill deposits(598 ± 7.5 ka), to within 1σ analytical uncertainty, align well with the mid-MIS 15 globalglacial advance (Figure 3.9). However, in many instances the relative imprecision of ra-diometric dating of young (i.e., <250 ka) poses a significant challenge and ambiguity. Forexample, the subaerial Cheakamus basalts (140.8 ± 12.9 ka) allow for a 1σ range in age69of 153.7 to 127.9 ka (Figure 3.9). If these lavas are >140 ka they would correspond toMIS 6 which is a peak global glaciation (Illinoian) and we would conclude that, either: i)the MIS 6 glaciation was not widely developed in SWBC, or ii) deglaciation of the MIS 6ice sheet was extremely rapid and preceded the global trend. Both of these scenarios areunlikely. However, the youngest allowable age (∼128 ka) shows an excellent correspon-dence between these subaerially erupted lavas and the earliest part of MIS 5e Sangamonianinterglacial (Figure 3.9).Glaciovolcanoes still hold a key advantage over many traditional methods because theirresistance to erosion enables them to survive subsequent glaciations and provide physical,terrestrial evidence for ancient paleo-glaciers and to provide chronology past the range ofradiocarbon-dated deposits. In SWBC, more than 100 volcanic edifices with ages spanningthe last 2 Ma are known. Many of these volcanoes (e.g., the Black Tusk, the Table and theSalal Glacier and the Bridge River volcanic fields etc.) demonstrate major glacial and non-glacial intervals that have existed prior to the last glaciation (see chapter 2). In many cases,the volcanic rocks provide the only known physical evidence for ancient ice in the NorthAmerican Cordillera (e.g., Edwards et al., 2011; Hickson et al., 1995; Roddick and Souther,1987) and can provide a powerful complement to global ice records such as the MIS. Ourpaleoenvironmental reconstructions for SWBC based on data recovered from glaciovolca-noes will advance substantially as additional geochronometric age determinations are madeand as techniques used to date volcanic rocks continue to improve (e.g., Conway et al.,2016).3.8 ConclusionWe investigate the paleoenvironmental implications of Pleistocene glaciovolcanic depositsin the Garibaldi volcanic belt of SWBC, Canada. These deposits offer physical terrestrialevidence for a number of major, ancient glaciations in the southern Canadian Cordilleraover the last ∼1 Ma. However, by themselves, glaciovolcanic deposits can preserve onlylocal determinations of paleo-ice presence and thickness. We address this limitation bypairing the data with a simple geometric model that provides regional ice-distribution esti-mations from localised data points. We use three volcanoes, in conjunction with our modelto recover: i) the existence of a coalesced mountain ice sheet during late MIS 4, and pos-sibly into MIS 3 (up to 49.1 ± 5.5 ka), ii) a major deglaciation (or absence of) glaciersin the southern Cordillera during the late MIS 6 to 5e transition (between 153.7 and 127.9ka), and, iii) a significant, coalesced mountain ice sheet, or continental ice sheet that ex-isted during the middle of MIS 15 (at 598 ± 7.5 ka). Our pilot study tests the use of ageometric model for reconstructing local glaciers and ice sheets using only three glaciovol-70canoes. We find that the deposits resulting from glaciovolcanic eruptions, when paired withthe model, have significant potential to be developed as a powerful, holistic, terrestrial pa-leoenvironmental proxy. Future field and geochronometric development of the database ofglaciovolcanoes, supplemented by geometric modelling at a range of spatial and temporalscales will thoroughly assess the advantages and disadvantages of the geometric approach.Ultimately, these studies will produce a high-resolution map of Pleistocene ice in SWBCthat could be used to independently test and refine currently-established local and globalclimate records.71Chapter 4Lillooet Glacier basalts,southwestern British Columbia,Canada: Products of Quaternaryglaciovolcanism34.1 IntroductionQuaternary volcanic rocks in the Garibaldi volcanic belt (GVB) (Canadian Cascade arc)are dominated by calc-alkaline intermediate compositions typical of continental arc envi-ronments (Figure 4.1A, Figure 4.1B) (Green et al., 1988; Hickson, 1994; Kelman et al.,2002a; Mathews, 1952a, 1958; Wilson et al., 2016). However, small volumes of alkalinebasalt having distinctly ocean island basalt (OIB) signatures, including the Bridge River(BR) (Roddick and Souther, 1987) and Salal Glacier (SG) (Lawrence et al., 1984) volcanicfields, and theMeagerMosaic Assemblage (MMA) (Read, 1977; Stasiuk and Russell, 1989)are distributed across the northern GVB (e.g., Friedman et al., 2016; Stasiuk and Russell,1989) (Figure 4.1C). The origin of these basaltic rocks has fostered a number of petrologicand geochemical studies suggesting that they derive from low degrees of partial melting ofthe mantle wedge (Green, 2006) or from mantle upwelling along the margin of the sub-ducting Juan de Fuca and Explorer plates (e.g., Bacon et al., 1997; Mullen and Weis, 2013,2015; Nelson and Carmichael, 1984) (Figure 4.1A).3A version of this chapter has been published. Wilson, A.M., Russell, J.K., 2017. Lillooet Glacier basalts,southwestern British Columbia, Canada: Products of Quaternary glaciovolcanism. Canadian Journal of EarthSciences. 54, 639–653. https://doi.org/10.1139/cjes-2016-0201.72114°0'0"W120°0'0"W126°0'0"W132°0'0"W53°0'0"N53°0'0"N46°30'0"N46°30'0"N40°0'0"N40°0'0"NQuaternary Volcanic RocksCanadaU.S.AVolcanic CentersSubduction Zone2200160040010001600220022002200220022002200123°14'0"W123°40'0"W50°45'0"N50°45'0"N50°30'0"N50°30'0"N123°0'0"W51°0'0"N50°0'0"N0 10050Kilometers0 2010Kilometers0 105KilometersJuan de Fuca PlateGorda PlateExplorer PlateWashingtonOregonCaliforniaAlaskaNootka FaultBritish ColumbiaABCHigh CascadesSilverthroneBridge River Mt. MeagerMt. CayleyMt. GaribaldiMt. BakerGlacier PeakMt. RanierMt. St. HelensMt. AdamsMt. HoodMt. JeffersonNewberryCrater LakeMt. ShastaMedicine LakeMt. LassenMt. Garibaldi Volcanic FieldMt. Cayley Volcanic FieldElaho VolcanicsMt. MeagerSGBRLGGlaciovolcanicNon-GlaciovolcanicIce FieldRiverBRSGMMALGMt. MeagerLillooet RiverPacific Plate73Figure 4.1: Location of the Lillooet Glacier (LG) volcanic deposits. A) Map of theCascade volcanic arc and Garibaldi volcanic belt (GVB) (modified from Mullenand Weis, 2013) showing distribution of Quaternary volcanic rocks, major vol-canic centres and convergent subduction setting. Outlined area is enlarged in B.B) Distribution of Quaternary volcanic rocks of the GVB. Outlined area is en-larged in C. C) Distribution of glaciovolcanic vs. non-glaciovolcanic centres inthe northern GVB including the Bridge River (BR), Salal Glacier (SG) volcanicfields, and Meager Mosaic Assemblage (MMA).Southwestern British Columbia also has a long and complex history of encroachingand retreating Pleistocene glaciations (e.g., Clague, 2009; Clague and Ward, 2011). Asa consequence, a disproportionate number of volcanic deposits within the GVB displayevidence for eruption under or against ice (Hickson, 1994; Kelman et al., 2002a; Mathews,1958). These glaciovolcanic edifices serve as an important tool for forensic reconstructionof regional and global paleoclimates and offer new insights into the nature of glaciovolcanicprocesses (e.g., Edwards and Russell, 2002; Edwards et al., 2009; Kelman et al., 2002a,b;Mathews, 1947; Pollock et al., 2014; Schopka et al., 2006; Smellie et al., 2008; Stevensonet al., 2009; Tuffen and Betts, 2010).Recent (<100 years) retreat of Lillooet Glacier (LG) has exposed a basaltic pillow-dominated tindar (e.g., Edwards et al., 2009; Jakobsson and Gudmundsson, 2008; Jones,1968; Russell et al., 2014; Smellie, 2013; Smellie and Skilling, 1994) that is associatedwith subordinate amounts of breccia and bedded hyaloclastite. Our mapping of these newlyexposed basalts establishes a glaciovolcanic origin associated with the waning of the Wis-consinan (Fraser) glaciation (17–13 ka) (Clague andWard, 2011). We also use petrographicand geochemical data combined with petrologic and thermochemical modeling to establishthe depth of magma storage and the pre-eruptive H2O content. Lastly, we present a physicalmodel for the glaciovolcanic emplacement of the LG basalts including an estimate for thedepth of the ephemeral englacial lake and the thickness of the enclosing ice sheet.4.2 Geological settingLG is located in the northern GVB, ∼20 km northwest of Mt. Meager (Figure 4.1). Theglacier is situated in a steep-sided mountain valley and is bound by lateral moraines (Reyesand Clague, 2004). Since the peak of the Little Ice Age (AD∼1900; Menounos et al., 2009),the glacier has retreated ∼5 km up valley (Reyes and Clague, 2004; Walker, 2003). Glacialretreat has exposed an irregular valley floor comprising Late Cretaceous granodiorite andEarly Cretaceous metamorphic rocks of the Gambier Assemblage (Monger and Journeay,1994; Reyes and Clague, 2004).74QlQvpQvpQgmtQgmtQgmtQgmtQgbtQgbtQgbtKb1400130011001500140013001100150014001400100090013009001500111512 10100101005510551015515510302025101520152051010158580867080889085889080808575515352040307065101551010QvtQvtQvtQvh123°43'0"W123°44'0"W123°45'0"W50°45'0"N50°44'0"NQvtQvpQvpQvh 10150151255510105851510905025758810251011601150114011201130107010601130112011101110110050 10.5Kilometers0 200100MetersA’BB’Qvp3AABSection 1Section 2Section 3Section 3SYMBOLSPillow lava (long axis)Bedding (Strike & Dip)Contact certainContact obscured2525Foliation (Strike & Dip)25Jointing (Strike & Dip)25LEGENDKbBASEMENT ROCKSGLACIAL DEPOSITSSEDIMENTARY DEPOSITSVOLCANIC DEPOSITSQvtQgbt Glacial: basal tillQlUndivided Cretaceous greenschist & granodioriteGlacial: lateral moraineQgmtQvpBasaltic sheets, lobes & tubesBasaltic pillow lavaAlluvial & gravel, sand & silt Scale: 1:20,000UTM : Zone 10NContour Interval: 20 mTNMN+ 15Contour Interval: 10 mPillow breccia & hyaloclastiteQvhBluffVentStratigraphic SectionKbKbKbSilt LakeFigure 4.2: Geological map of the LG volcanic deposits. A) Geological map show-ing the distribution of Lillooet Glacier (LG) lithofacies. B) Inset (outlined in A)showing detailed lithofacies distributions near the terminus of LG. UTM, Uni-versal Transverse Mercator.4.3 Physical VolcanologyGeological field mapping at 1:5,000 used a 1:20,000 Terrain Resource InformationManage-ment (TRIM) digital elevation model, supported by global positioning system for enhancedaccuracy. The LG volcanic deposits comprise basaltic pillow lava, lava tubes, and minorvolumes of pillow breccia and bedded hyaloclastite. The deposits are distributed across thevalley floor near the terminus of the glacier and occupy an area of 0.95 km2. The minimumvolume of exposed volcanic rock is estimated to be 0.078 km3.75QvtQvpLillooet GlacierQgmtKbKb100 300 500 700 900 1100 1300 1500 1700 1900100011001200130014001500QvtQvpQgmtQgbtKb500 1000 1500 2000 2500 3000 35001000110012001300140015001200124012801320136011601200124012801060108011001400A A’B’Distance (m)Elevation a.s.l (m)Elevation a.s.l (m)No vertical exaggerationSection BendABBDistance (m)CKbQvhQvpQgmtQvtKbQvpQgbtQgmtKbQvhQvtElevation a.s.l (m)Section 1Section 2Section 3Section 1Section 1Section 2Figure 4.3: Cross-sections and graphical logs of the Lillooet Glacier (LG) volcanicdeposits as indicated in Figure 4.2. A) Cross-section perpendicular to valleyshowing interpreted thickness of the northern pillow pile. B) Cross section par-allel to valley showing thinning volcanic stratigraphy towards the southeast. Sec-tion lines are outlined in Figure 4.2A. C) Graphic logs showing volcanic stratig-raphy at various locations within the map area (locations are outlined in Fig-ure 4.2A and Figure 4.2B). Kb, Cretaceous greenschist and granodiorite; Qgbt,basal till; Qgmt, lateral moraine till; Qvh, bedded hyaloclastite; Qvp, basalticpillow lava; Qvt, basaltic sheets, lobes, and tubes.4.3.1 LithofaciesGlacial depositsGlacial deposits unconformably overlie the Quaternary volcanic rocks exposed at the north-ern margin and terminus of LG (Figure 4.3C). The youngest of these deposits is a 50 m-thick lateral moraine (Qgmt) comprising crudely bedded, boulder-rich till hosting layers ofwoody debris (Figure 4.4A, Figure 4.4C).Qgmt is underlain by a southward thickening lens of dark grey, dense, pervasivelysheared diamictite (Qgbt). The nature of the contact between these units appears sharpand unconformable (Figure 4.4A, Figure 4.4C). The unit contains cobble-sized clasts ofsubrounded vesicular basalt and minor bedrock. Petrographic analysis of the matrix revealsa high proportion of subrounded volcanic fragments, including fresh sideromelane and min-eral grains liberated from the volcanic deposits (i.e., olivine and plagioclase). Reyes andClague (2004) interpreted unit Qgbt as a basal till.Radiocarbon (14C) dating of a wood fragment from the base of the lateral moraine(Qgmt) established a radiocarbon age of 3034± 42 14C years BP (Reyes and Clague, 2004)and sets the minimum age of volcanism. Reyes and Clague (2004) suggest that the basal76till (Qgbt) underlying Qgmt resulted from an earlier phase of late Fraser alpine glacierexpansion (Menounos et al., 2009; Reyes and Clague, 2004), which would indicate thatvolcanism is at least as old as the Fraser glaciation. The volcanic deposits rest directly onglacially worn bedrock with no till at the interface.Basaltic pillow lavaOlivine plagioclase-phyric pillow basalt (Qvp) is the most abundant lithofacies. Two majoroutcrop morphologies are observed. To the northwest of the map area, the facies is ex-pressed by densely packed pillows ranging in diameter from 0.3 to 1.5 m. The pillows forman elongate, 1.8 km-long, 150 m-thick pile that is interspersed rarely with large (2–8 m),radially jointed lava tubes (Figure 4.4). To the south (Figure 4.2A), the pillow basalts forma 10–20 m-thick veneer covering the valley floor.All pillow lavas have thick, well-developed glassy rims (<1.5 cm), are subrounded inshape, and have internal radial fracture sets. The pillows are vesicular; vesicles are nearspherical and 0.5–2 mm in size and commonly form multiple concentric bands (<2 cm).Connected vesicularity, measured by helium pycnometry, varies from 25% to 35% (seeTable S5). Intrapillow voids are filled sporadically with minor accumulations of moder-ately sorted vitric ash and lapilli tuff. Pillows are typically elongate and oriented towardsthe southeast. Near the toe of the glacier, however, the pillows follow the local bedrocktopography and plunge towards the northwest (Figure 4.5B).Basaltic sheets, lobes, and tubesOlivine plagioclase-phyric lava tubes (Qvt) occur within the northern pillow pile as discon-tinuous (<20 m wide), radially jointed masses Figure 4.4C). Farther south, lobes, tubes,and sheets form a large, branching network that feeds the pillow lavas (Figure 4.5C). Lobesand tubes are elongate in the local downhill direction and are typically equidimensional incross-section. Tubes are up to 15 m wide and they commonly terminate in abrupt, perva-sively radial-jointed toes (Figure 4.5D). The Qvt facies is typically micro-vesicular (<0.1mm-diameter), with a connected porosity of 15–20%.One sample of holocrystalline basaltic lava (Qvt) was dated by 40Ar/39Ar geochronom-etry at the University of Fairbanks, Fairbanks, Alaska; the sample is too young to record anage (i.e., plateau and isochron ages ∼ zero age) because of a high atmospheric Ar contentand a low radiogenic component (i.e., low K).77Glacial: lateral moraine tillGlacial: basal tillBasaltic pillow lavaBasaltic lobes, sheets & tubesPillow breccia & hyaloclastiteScreeBasement rocksLEGENDCADB100 m (approx.)10 m (approx.) 10 m (approx.)Figure 4.4: Field photographs of Lillooet Glacier (LG) volcanic deposits. A) Com-posite panoramic image of stacked pillow basalts (Qvp) situated in the northernpart of the map. Light grey material overlying the pillow lava is moraine till(Qgmt). Boxed areas denote enlargements shown in C and D. B) Line diagramof the panorama showing contact relationships between stratigraphic units. C)Enlarged portion of A showing large, radial jointed lava tube surrounded by pil-low lava situated at the top of the stacked pillow basalts. Photograph courtesyof John Clague. D) Enlarged portion of A showing discontinuous, southwardthickening lens of dense, dark grey diamictite (Qgbt).78QvpQvtQvh10 m (approx.)NACDQvp (covered)BFigure 4.5: Field photographs of pillow basalt and tubes feeding pillow basalt. A)Densely packed pillow lava exposed near the terminus of Lillooet Glacier (LG).B) Elongate pillow tube exposed near the terminus of LG. Curved white arrowdenotes the flow direction. C) Pillow accumulation overlying a thin, discontin-uous lens of bedded hyaloclastite (Qvh). D) Large, radially jointed lobe of lava(person for scale). Qvt, holocrystalline basaltic lava.79Pillow breccia and bedded hyaloclastiteMassive to crudely bedded pillow breccia and well-stratified hyaloclastite form a volumet-rically minor (5%) portion of the LG lithofacies. The unit forms discontinuous lens-shapeddeposits that separate successive pillow lava accumulations throughout the map area. Typ-ically, these lenses are thin (<2 m) and comprise massive, poorly sorted, clast-supportedpillow breccia with angular clasts that range from <1 to 20 cm in diameter. Pillow brec-cia layers grade rapidly into finely bedded, well-sorted, poorly consolidated ash and lapillivitric tuff (e.g., Figure 4.5C). Within the fine layers (e.g., Figure 4.6A), the vitric clasts areblocky, angular, poorly vesicular and densely packed, and are typically surrounded by athin (<0.1 mm) rim of dark orange palagonite (Figure 4.6B). Rare lithic clasts (subroundedto rounded metamorphic and plutonic rocks) up to 2 cm in diameter comprise <1% of theentire unit.Poorly sorted pillow breccia accumulations likely derive from gravitational collapseof pillow lavas propagating downhill. The fine, blocky ash and lapilli-sized fragmentsresult from quench fragmentation and spalling of the glassy pillow margins and weresyn-eruptively transported by subglacial melt water and deposited in discontinuous, well-stratified lenses and within intrapillow void spaces. Rare, accidental lithic clasts of graniteand metamorphic rocks likely originated as cargo in the overlying ice sheet and were in-corporated into the deposit during volcanic heating (melting). Because of the dominance ofvitric fragments and the lack of magmatic fragmentation indicators, we interpret the ash andlapilli vitric tuff as resedimented (bedded) hyaloclastite (e.g., White and Houghton, 2006).4.4 PetrographyThe LG volcanics are porphyritic, with phenocrysts of olivine (10 vol.%) and plagioclase(30 vol.%). Olivine phenocrysts are subhedral, range from 0.1 to 1 mm-diameter, and showweak chemical zoning. Olivine phenocrysts commonly host small (<0.01 mm) equant in-clusions of spinel (Figure 4.7A). Plagioclase phenocrysts are subhedral and can be stronglyzoned chemically. Glomerophenocrysts of olivine and plagioclase up to 2 mm in diameterare common. The basalt lavas also contain minor quantities of subhedral, sieve-texturedgrains of colourless to pale green augite (0.1–0.5 mm). These appear to be derived from thebreakup of cognate inclusions (see later in the text). All lavas contain a fine- grained ground-mass comprising glass, plagioclase, olivine, augite (commonly Ti–augite), and Fe–Ti ox-ides.All LG lavas carry a population (<10 vol.%) of sieve-textured, euhedral 1–3 mm di-ameter grains of plagioclase. The plagioclase grains show intense chemical zoning andfeature dusty, embayed cores and recrystallized margins (Figure 4.7B). These textures in-80AB1 mmFigure 4.6: Field photograph and photomicrograph of hyaloclastite lithofacies. A)Bedded, poorly sorted, discontinuous lens of hyaloclastite (Qvh) that is both un-derlain and overlain by accumulations of massive pillow lava (Qvp). Photographcourtesy of John Clague. Arrow denotes location of sample used for thin section.B) Photomicrograph of thin section of moderately sorted, partially palagonitized,resedimented hyaloclastite comprising fine to coarse, ash-sized, blocky, weaklyvesiculated, vitric particles.dicate chemical disequilibrium between these plagioclase crystals and the melt, suggestingthe possibility of plagioclase-recycling processes (e.g., Davidson and Frank J. Tepley, 1997;Nixon and Pearce, 1987; Tepley, 1999; Tsuchiyama, 1985). One sample of lava (sample 59;Table 4.1) contains a large 3 cm-diameter gabbroic inclusion (Figure 4.7C, Figure 4.7D).The gabbro has a coarse (1–3 mm) groundmass texture comprising olivine (20 vol.%), pla-gioclase (35 vol.%), and augite (20 vol.%) and a 25% intracrystalline void space filled bymicrocrystalline to quenched glass. The glass is vesiculated, and the vesicles appear to haveexpanded and deformed the partly crystalline matrix. These textural relationships suggest81OlPlA BC D0.5 mm1 cm 1 mmFigure 4.7: Photomicrographs of Lillooet Glacier (LG) basaltic rocks in cross po-larized light. A) Photomicrograph showing glomerophenocryst of subhedralolivine. B) Sieve-textured plagioclase crystals within a fine, subglassy ground-mass. C) Two centimetre long, cognate, gabbroic inclusion hosted within sample59. The clot displays a coarse, equigranular texture. D) Enlarged portion of Cshowing interlocking gabbroic texture. Cpx, clinopyroxene; Ol, olivine; Pl, pla-gioclase.that the gabbroic inclusion is cognate. This material also manifests as disaggregated augite,plagioclase, and olivine grains within the rocks (see later in the text).4.5 Geochemistry4.5.1 Whole-rock chemical compositionsMajor and trace element contents of six samples of lava (Qvp, Qvt) were analyzed by X-ray fluorescence and inductively coupled plasma – mass spectrometry (Table 4.1 and Ta-ble 4.2, respectively). Replicate major element analyses of samples 62 and 69 performed82at a separate laboratory are also reported (labeled 62b and 69b, respectively). Two samplesof coarsely crushed (to <2 mm), hand-picked glass from the quenched margins of pillowlobes (samples 62 and 70) were also analyzed.LG lavas range from basalt to basaltic andesite and display transitional to alkaline chem-ical affinity (closed black circles, Figure 4.8A) (Le Bas et al., 1986; MacDonald, 1968).Values of Mg/(Mg + Fe) vary from 0.51 to 0.57, and all samples are hypersthene normative,however mineralogically, they lack Ca-poor pyroxene. Whole-rock SiO2 concentrationsof 50.8-53% make the LG volcanic rocks the highest-silica-content Quaternary mafic vol-canics in the GVB. Although the Lillooet lavas are distinctly more siliceous than other GVBbasalts, their total alkali contents (4.5-5%) match those of lavas from the Bridge River andthe Salal Glacier volcanic fields.Trace element contents of the LG basalts have an OIB affinity, display little variation(Table 4.2; Figure 4.8B), and are similar to other alkaline GVB basalts. Notably, they lackthe Nb–Ta depletion typically associated with subduction-related products.83Table 4.1: Whole-rock chemical compositions (wt.%) of Lillooet Glacier volcanic map units measured by X-ray fluorescence∗, andcalculated properties including DI, SI, and normative mineralogy.Label† 59 62 ¶62b 65 69 ¶69b 70Unit Qvt Qvp Qvp Qvt Qvt Qvt QvpRock Type Lava tube Pillow glass Pillow centre Lava tube Lava tube Lava tube Pillow glassLat◦N 50.74340059 50.74264705 50.74264705 50.74214724 50.74733916 50.74733916 50.74712458Long◦W -123.7298373 -123.7282305 -123.7282305 -123.7251255 -123.7275395 -123.7275395 -123.7255798SiO2 52.19 52.04 51.70 53.26 50.79 50.36 51.81TiO2 1.59 1.71 1.68 1.59 1.69 1.66 1.57Al2O3 15.94 15.95 15.85 15.85 15.39 15.18 15.62FeO⋆ 9.43 9.93 10.78 9.56 10.31 11.18 9.67MnO 0.14 0.15 0.15 0.15 0.16 0.15 0.14MgO 6.01 5.90 5.48 6.08 7.52 7.40 6.80CaO 8.17 8.24 8.20 7.98 8.64 8.42 8.20Na2O 4.01 3.97 3.73 4.09 3.72 3.66 3.83K2O 0.98 0.96 1.10 1.05 0.88 0.86 0.97P2O5 0.30 0.32 0.33 0.30 0.30 0.30 0.29Totals 98.76 99.16 100.05 99.91 99.38 99.63 98.89LOI 0.77 0.76 0.70 0.00 0.60 0.08 0.35Mg # 53.17 51.44 47.54 53.15 56.53 54.13 55.61D.I. 62.48 62.29 61.48 62.78 59.51 58.66 60.87S.I. 41.94 40.81 37.99 41.79 46.72 45.07 44.93Normative Mineralogy (wt. %)Orthoclase 5.80 5.65 6.50 6.18 5.17 5.08 5.70Albite 33.90 33.57 31.56 34.65 31.45 30.97 32.43Anorthite 22.61 22.90 23.26 21.77 22.72 22.45 22.56Diopside 13.21 13.18 12.75 13.12 14.98 14.37 13.41Hypersthene 8.79 8.76 10.72 11.04 2.92 3.14 8.44Olivine 10.75 11.13 10.28 9.45 18.24 19.33 12.71Ilmenite 3.02 3.25 3.19 3.01 3.21 3.15 2.98Apatite 0.70 0.74 0.76 0.71 0.70 0.70 0.67Note: DI, differentiation index; LOI, loss on ignition; Mg# = 100MgO/(MgO + FeO) in mol%; SI, solidification index.∗X-ray fluorescence chemical analyses measured by the Peter Hooper GeoAnalytical Laboratory, Washington State University, Pullman, Washington, and ALSLaboratories, Vancouver, British Columbia.†All samples contain prefix “AW-15-0”.¶Replicate analyses performed at ALS Laboratories, Vancouver, British Columbia.⋆Total Fe reported as FeO.844.5.2 Mineral and glass compositionsMajor element concentrations of phenocrysts, possible xenocrysts, groundmass phases, andglass were measured using a Cameca SX50 scanning electron microprobe (EMP) at TheUniversity of British Columbia, Vancouver, British Columbia. Analytical conditions in-clude an acceleration voltage of 15 keV, a beam current of 20 nA, and a beam diameterof 30 µm. Groundmass Ti-augite was analyzed using a beam diameter of 5 µm. A fullcompilation of EMP mineral analyses aregiven in Tables S6 and S7. Olivine compositionsrange from forsterite (Fo) Fo84 to Fo74 and contain low CaO (<0.31 wt.%) and NiO (<0.23wt.%). Olivine compositions define two groups (Figure 4.9A) corresponding to the glom-erophenocrysts and to the gabbroic inclusion. Glomerophenocrysts are the most forsteritic,whereas olivine from the gabbroic inclusion are the least forsteritic and have the lowest NiOcontents. Individual olivine phenocrysts plot in both groups, suggesting that some grainsderive from breakup of the cognate gabbroic inclusions.Plagioclase compositions are derived from single crystals (phenocrysts), sieve-texturedcrystals, and the gabbroic inclusion (Figure 4.9B). Anorthite (An) compositions of the cog-nate inclusion occupy a narrow range (An0.60–An0.64), however, they are unresolvable fromall other phases. Sieve-textured crystals were analyzed to investigate possible sources ofcontamination (e.g., granitic basement rock). Although a rare number of analyses are an-desine (An30–An50) in composition, most have compositions consistent with crystalliza-tion from the LG magma (i.e., cognate). Augite analyses are divided into single crystals(xenocrysts), cognate gabbroic inclusion groundmass, and groundmass Ti–Aug. The inclu-sion and single crystal (xenocryst) phases are indistinguishable, suggesting that the singlecrystal xenocrysts represent disaggregated inclusion material. Acicular groundmass Ti–Augdisplays high TiO2 (average of 3.01 wt.%), in conjunction with low Mg/Mg + Fe values.Sideromelane glass compositions are consistently 2 wt.% higher in silica (52.0–53.8 wt.%;basaltic andesite, Figure 4.8A) than are the corresponding whole-rock compositions (Fig-ure 4.8A and Table 4.2).85CsRbBa100101102103Sample / N-MORBThUNbTaLaCePbPrSrNdSmZrHfEuGdTbDyYHoErTmYbLuBRSGMMACBEVLG44 46 48 50 52 54SiO2 (wt. %)234567Na2O + K2O (wt.%)basalticandesitebasalthawaiitemugearitebasaniteSGBRMMACBLGEVsub-alkalicalkalicABFigure 4.8: Major and trace element geochemistry of Lillooet Glacier (LG) lava andglass. A) Total alkalis versus silica diagram (after Le Bas et al., 1986) show-ing whole-rock (closed black circles) and microprobe glass (open grey circles)analyses. Additional analyses of GVB basalts (after Mullen and Weis, 2015)are shown for comparative purposes. These analyses include the Bridge River(BR) and Salal Glacier (SG) volcanic fields, the Mt. Meager Mosaic Assem-blage (MMA), the Elaho volcanics (EV), and the Cheakamus basalts (CB).LG volcanic rocks are dominantly basalt and basaltic andesite and show transi-tional alkalinity (MacDonald, 1968). Glass analyses comprise basaltic andesite,mugearite, and minor hawaiite. B) Trace element compositions of the LG basaltsshow very little variation within the suite and lack distinctive subduction-relatedsignatures. N-MORB, normal mid-ocean ridge basalt.86Table 4.2: Trace and rare earth element contents (ppm) of Lillooet Glacierlithofacies measured by XRF∗ and ICP–MS, respectively.Label† 59 62 65 69 70Unit Qvt Qvp Qvt Qvt QvpRock Type Lava sheet Pillow glass Lava lobe Lava tube Pillow glassLat◦N 50.74340059 50.74264705 50.74214724 50.74733916 50.74712458Long◦W -123.7298373 -123.7282305 -123.7251255 -123.7275395 -123.7255798XRF Trace Elements (ppm)Ni 74.9 69.0 74.5 110.5 93.4Cr 130.5 137.3 131.5 227.9 210.6V 178.4 185.8 176.2 188.8 175.2Nb 13.9 14.6 14.2 15.0 14.1Ga 19.7 21.1 19.5 20.1 20.9Cu 33.0 39.0 33.6 46.2 48.0Zn 96.0 104.6 98.0 100.9 96.5La 10.1 17.2 13.2 13.3 18.1Ce 33.0 31.1 33.9 30.8 30.3Nd 18.2 19.2 16.8 17.2 18.9ICP-MS Trace Elements (ppm)La 14.76 15.26 15.26 13.97 14.24Ce 31.72 32.85 32.45 30.22 30.70Pr 4.19 4.42 4.31 4.05 4.07Nd 18.08 18.75 18.38 17.74 17.37Sm 4.51 4.79 4.57 4.53 4.47Eu 1.59 1.67 1.60 1.60 1.59Gd 4.60 4.80 4.59 4.51 4.34Tb 0.70 0.73 0.70 0.72 0.66Dy 3.97 4.11 3.98 3.97 3.87Ho 0.75 0.77 0.76 0.75 0.72Er 1.85 1.98 1.88 1.85 1.76Tm 0.25 0.27 0.25 0.25 0.24Yb 1.51 1.56 1.54 1.47 1.42Lu 0.22 0.24 0.23 0.22 0.21Ba 326 320 345 256 296Th 1.53 1.50 1.63 1.41 1.59Nb 13.65 14.65 13.76 14.36 13.74Y 18.42 19.63 18.67 18.29 17.68Hf 3.31 3.36 3.38 3.05 3.10Ta 0.88 0.93 0.89 0.93 0.88U 0.68 0.57 0.62 0.51 0.59Pb 2.58 2.53 2.70 2.07 2.51Rb 12.00 11.48 12.41 10.04 11.62Cs 0.17 0.16 0.19 0.15 0.18Sr 608 596 605 566 571Sc 18.87 19.18 18.68 20.55 19.11Zr 133.2 137.3 137.6 123.4 125.1Note: ICP–MS, inductively coupled plasma–mass spectrometry; XRF, X-ray fluorescence.∗XRF and LA-ICPMS chemical analyses measured by the Peter Hooper GeoAnalytical Laboratory,Washington State University, Pullman, Washington.†All samples contain prefix “AW-15-0”.87Table 4.3: Electron microprobe (EMP) major element analyses (wt.%) of Lillooet Glacier volcanic glasses.Label∗ LOD SD† 073-1-1 073-3-4 073-4-1 073-5-3 055-12-1 055-12-4 055-13-5 055-13-3 070-10-5 070-12-1 070-13-1 070-14-5 070-15-4 070-16-4Unit Qgbt Qgbt Qgbt Qgbt Qvp Qvp Qvp Qvp Qvp Qvp Qvp Qvp Qvp QvpRock Type Till Till Till Till Pillow Pillow Pillow Pillow Pillow Pillow Pillow Pillow Pillow PillowLat◦N 50.7444645 - - - 50.74357753 - - - 50.74712458 - - - - -Long◦W -123.720825 - - - -123.7297911 - - - -123.7255798 - - - - -SiO2 0.042 0.208 52.00 52.26 52.69 53.16 52.83 53.26 52.75 52.52 53.75 54.28 53.05 53.59 53.49 52.87TiO2 0.045 0.057 1.71 1.83 1.69 1.82 1.85 1.93 1.98 1.99 1.92 2.14 1.96 1.97 1.97 1.87Al2O3 0.041 0.111 15.29 15.41 15.52 15.00 15.25 15.23 15.34 15.05 15.05 14.90 14.92 14.99 14.72 14.89Cr2O3 0.081 0.049 0.00 bdl 0.01 0.02 0.04 0.03 0.01 0.01 bdl bdl bdl 0.00 0.03 0.00FeO 0.090 0.240 9.12 9.25 9.32 9.16 9.84 9.85 9.92 9.81 9.60 8.81 9.52 9.36 9.38 9.39MgO 0.041 0.077 4.48 4.47 4.66 4.25 4.30 4.18 4.36 4.25 4.03 3.59 4.21 4.09 4.15 4.21MnO 0.085 0.060 0.21 0.14 0.17 0.16 0.17 0.15 0.16 0.18 0.11 0.10 0.18 0.14 0.17 0.16CaO 0.041 0.128 7.99 8.12 8.03 7.77 7.64 7.66 7.79 7.74 7.87 7.62 7.93 7.90 7.92 7.96Na2O 0.059 0.117 3.79 3.80 3.78 3.65 4.03 4.02 4.26 4.06 3.87 3.90 4.06 4.12 4.05 4.13K2O 0.043 0.062 1.09 1.14 1.12 1.29 1.18 1.20 1.22 1.13 1.27 1.27 1.24 1.29 1.25 1.24P2O¶5 0.064 0.054 n/a n/a n/a n/a n/a n/a n/a n/a 0.35 0.45 0.35 0.37 0.43 0.28Totals 95.68 96.42 97.00 96.28 97.15 97.51 97.78 96.73 97.83 97.06 97.42 97.82 97.55 97.00Mg # 0.47 0.46 0.47 0.45 0.44 0.43 0.44 0.44 0.43 0.42 0.44 0.44 0.44 0.44Note: The highest and lowest MgO values within each analyzed sample are given. bdl, below detection limit; LOD, limit of detection; n/a, not available; SD,standard deviation.∗All samples contain prefix “AW-15-”. Format: “sample-grain-analysis”.†Analytical uncertainty given as 1σ .¶P2O5 was measured in one sample only (070).884.6 Discussion4.6.1 Origin of chemical variations within the lavas and glassesThe LG whole-rock and glass analyses define a 4 wt.% variation in SiO2 content. We in-vestigate the nature of differentiation attending storage, transport, and eruption of thesebasaltic and basaltic andesite rocks. Pearce element ratios (PERs) offer a robust and effi-cacious means of testing petrologic hypotheses (Pearce, 1968; Russell and Nicholls, 1988;Russell and Stanley, 1990; Stanley and Russell, 1989). Because they preserve geochemicalinformation in absolute, stoichiometrically defined units, PERs remove the distorting effectof closure (i.e., constant sum problem Chayes, 1960; Skala, 1979) and, thus, help recoverthe absolute changes in geochemical composition attending magmatic processes (Russelland Nicholls, 1988). Element ratio diagrams can be tailored to test for certain mineralphases that are involved in magmatic sorting. For a PER to be effective, the denomina-tor constituent must remain conserved by processes affecting the chemical evolution of thesystem (Stanley and Russell, 1989).For our purposes, we have chosen K as a conserved denominator. K is well measuredand conserved relative to the observed phenocryst assemblage. Ratios of incompatible el-ements and K show that K is conserved to within intra-laboratory precision (e.g., Halleranand Russell, 1990). Conservation of K across the rock suite strongly supports these lavasbeing comagmatic (Nicholls and Russell, 1991).Figure 4.10A presents tests for crystal-sorting processes involving only olivine usingwhole-rock and glass analyses. Variation due to crystal sorting processes involving olivinewill cause the data to parallel a model line with a slope of 1. Sample 69 was collectedfrom the base of the northern pillow pile and is the most primitive within the suite. Glassanalyses comprise the most evolved constituent of the LG volcanics. The model line isdrawn through the most primitive data value (69). The plot indicates that olivine sortingprocesses alone cannot be responsible for the chemical variation with the LG lavas. Self-mixing processes involving the end-member compositions, shown as a line connecting themost primitive and evolved compositions, could contribute to the chemical variations. Allsamples fall within the field delimited by the lines for olivine sorting and self-mixing limits(e.g., Russell and Stanley, 1990).Figure 4.10B presents tests of a two-phase fractionation or sorting model involving thephenocryst assemblage olivine ± plagioclase. A model line with a slope of 1 is drawnthrough sample 69. The data (rocks and glasses) all fall within 2σ uncertainty of this line,indicating that the model explains the observed chemical variations. There is no discerniblechemical signal indicative of differentiation involving augite. Furthermore, crustal assim-89ilation processes have not imprinted a distinctive chemical signal on the major elementcompositions of the LG lavas.4.6.2 Physical–chemical conditions of storage, transport, and eruptionIsobaric equilibrium crystallization during storageWe used rhyolite-MELTS (Ghiorso and Gualda, 2015; Gualda et al., 2012) to model theconditions of storage, crystallization pressure, ascent path, and pre-eruptive volatile con-tent of the magma. The mineralogy of the cognate inclusion recovered from sample 59 isused to constrain the pre-eruptive storage history of the magma that ultimately fed the LGvolcanic eruption. The cognate inclusion comprises an equigranular mosaic of subequalproportions of olivine and augite (20 vol.%), plagioclase (35 vol.%), and intracrystallinemicrocrystalline glass (25 vol.%). The textures are consistent with crystallization form-ing a 75% solid and 25% melt crystal mush. Although the lavas feature only olivine andplagioclase phenocrysts, augite is a major and critical component of the cognate inclusion.We constructed a phase diagram showing the stable equilibrium mineral assemblagesthat would crystallize from a parental melt (sample 69) as a function of depth (i.e., pres-sure) and dissolved water content (Figure 4.11. After 75% crystallization, all assemblagescontain plagioclase, augite, and spinel. However, the presence of olivine and orthopyrox-ene vary as a function of pressure and H2O content. The absence of orthopyroxene in thecognate inclusion restricts the storage and crystallization conditions to <2 kbar (∼7.5 km)under dry conditions; with increasing H2O content, the storage pressure could be as low as1.5 kbar (i.e., 5 km). On the basis of these model phase relationships, we suggest that theLG magma was stored at 2 kbar long enough to promote sidewall crystallization of plagio-clase + olivine + augite gabbroic material. Subsequent ascent and eruption of the magmaentrained fragments of the partially crystallized crystal mush. Vesiculation of the intracrys-talline melt within the cognate inclusion occurred during ascent and locally deformed themush to cause partial alignment of the crystalline framework.90Single crystalsCognate inclusionGlomerophenocrystsAn %0.2 0.3 0.4 0.5 0.6 0.7N - Analyses0510152025Single crystalsCognate inclusionSieve-textured crystals: coresBSieve-textured crystals: rimsCr   + Al   + Ti0.6 0.65 0.7 0.75 0.8CMg #3+3+4+102030405060Fo % NiO (ppm)A0.72 0.74 0.76 0.78 0.8 0.82 0.84 0.8611.522.533.544.555.56Single crystalsCognate inclusionGroundmass Ti-augite91Figure 4.9: Electron microprobe analyses of Lillooet Glacier mineral compositions.A) Olivine compositions derived from a single cognate inclusion (Figure 4.7C),glomerophenocrysts, and single olivine crystals. Olivine from the cognate inclu-sion has the lowest Fo% content. Single grains of olivine span the gap betweenthe glomerophenocrysts and the cognate inclusion, suggesting a mixed origin. B)Anorthite composition of plagioclase split into single crystals, cognate inclusion,and sieve-textured crystals. The cores of sieve-textured crystals are the most al-bitic; however, they cannot be distinguished from the single plagioclase crystalsor the cognate inclusion analyses. C) Augite analyses resolved into single augitecrystals (xenocrysts), cognate inclusion groundmass, and groundmass Ti–augite.Cognate inclusion analyses are indistinguishable from the single augite crystalsin the rocks, suggesting a common origin. Groundmass Ti–augite shows in-creased Ti concentration.Ascent- and eruption-driven fractional crystallizationWe also used rhyolite-MELTS to model two-phase fractional crystallization of olivine + pla-gioclase (Figure 4.10) attending ascent and eruption. We use sample 69 as a proxy startingmagma composition. Phases that are not significantly involved in the process (clinopyrox-ene, orthopyroxene, fluid, etc.) are suppressed. Isobaric fractional crystallization is mod-elled at a minimum pressure of 1 kbar and for a range of dissolved water contents (0–1.5wt.%).During the initial stages of the rhyolite-MELTS simulations, the olivine liquidus is metat 1205 ◦C (1 wt.% H2O simulation) (Figure 4.12A). At this time, olivine is the only phasebeing crystallized; the result is that the liquid line of descent (LLD) has a slope of 1 thatparallels the model PER line. When the liquid composition reaches the olivine–plagioclasecotectic, the onset of plagioclase crystallization causes a deviation along a shallow vectorwith a slope of 0.2 (reflecting the average ratio of olivine/olivine + plagioclase). Increasingwater content suppresses the onset of plagioclase crystallization, lowering or raising theposition of the cotectic in element ratio space.Glass analyses show a small but measurable chemical variation and represent the liquidcomposition. Thus, we require our model to (i) display a trend subparallel to the mea-sured glass chemistry and (ii) intersect and transect the field of measured glass analyses.On this basis, we can establish a first-order constraint on the pre-eruptive H2O content ofthe magma. Our simulations indicate that between 0.5 and 1 wt.% H2O is required to sup-press the onset of plagioclase crystallization to a sufficient degree for the evolving liquidcomposition to intersect the analyzed glasses. Higher pressures (i.e., 2 kbar) necessitate anelevated H2O content of up to 2 wt.%. A further test of our hypothesis uses axes designedto test for sorting processes involving olivine versus plagioclase (Figure 4.12B).9225 30 35 40 45 50Si / K  345678910110.5 (Fe + Mg) / K  059062065069070062b069bLimits of 'self-mixing'30 35 40 45 50Si + 0.33 Ca / K  25303540451.17 Al + 0.5 (Fe + Mg) + 1.83 Na / K  059062065069070062b069bOlPlCpxPlCpxOlABm = 1Figure 4.10: Pearce element ratio (PER) diagrams showing chemical variations inwhole-rock and glass analyses of Lillooet Glacier basalts. A) PER with axesdesigned to test for crystal-sorting processes involving olivine (Ol). The datado not lie on a model line with a slope of 1, indicating that Ol sorting alonecannot be responsible for the observed chemical variation. 2σ uncertainty el-lipses are shown for the whole-rock analyses only. B) PER designed to testfor crystal sorting processes involving Ol + plagioclase (Pl). The data define aclear trend and all lie (within uncertainty) on the model line with a slope of 1,which means that all chemical variation within the LG basalts can be ascribedto sorting processes involving Ol + Pl. Mineral response vectors showing theeffect of removal of 1 mol of Ol, Pl, and clinopyroxene (Cpx) are shown in bothdiagrams.930 0.5 1 1.5 2H2O wt.%01234Pressure (kbar)Opx + Ol + Pl + Aug ± SplOl + Pl + Aug ± Spl051015Depth (km)Exsolution of fluidOpx + Pl + Aug ± SplFigure 4.11: Phase diagram for Lillooet Glacier (LG) magma as a function of pres-sure and water content calculated using rhyolite-MELTS (Ghiorso and Gualda,2015; Gualda et al., 2012). Mineral assemblages are calculated for a magmacomposition represented by sample 69 and 75% equilibrium crystallization (seetext). Phase portrait derives from 20 simulations run at variable pressures andH2O contents. The phase diagram comprises three pressure-dependent assem-blages: (olivine (Ol) + plagioclase (Pl) + augite (Aug) + spinel (Spl)), (Ol + Pl+ orthopyroxene (Opx) + Aug + Spl), and (Pl + Opx + Aug + Spl). The absenceof Opx within the cognate inclusion (Figure 4.7C) indicates that magma stor-age was at <2 kbar (<7.5 km). The shaded field denotes the P-H2O conditionsindicative of an exsolved fluid (i.e., oversaturation).94Displacement vectors are at a high angle (90◦), providing the most rigorous test of thedata. Again, the simulated LLD maintains a horizontal trend until it reaches the olivine-plagioclase cotectic. Between 0.5 and 1 wt.% H2O is required to intersect the measuredglasses. At these low pressures, the effect of vesiculation-induced crystallization (e.g.,Russell, 1987) could further refine the thermo-dynamic model. Continued fractionationof the fluid phase, during vesiculation and ascent, will further encourage crystallizationof plagioclase, causing a subtle deflection upwards within the element ratio space. Thisprocess would favor a slightly higher H2O content (i.e., 1 wt.%). When PERs are used toportray thermodynamic simulations, the space immediately surrounding the LLD can belabeled to display the temperatures of the fractionated melt. Under these conditions, oursimulations indicate that the most evolved measured glass composition will be reached at1045 ◦C. This value serves as a potential proxy for the eruption temperature (Figure 4.12A,Figure 4.12B).4.6.3 Glaciovolcanic origin of Lillooet Glacier lithofaciesOur mapping and analysis of the lithofacies are consistent with a glaciovolcanic origin onthe basis of the following evidence:1. Pillow lava is the dominant lithofacies, indicating a subaqueous eruption. Sub-rounded pillow morphologies, intense radial jointing, well-developed glassy rims,and the presence of rare syn-eruptive vitric sediments within intrapillow voids meanthat these deposits had to have been emplaced subaqueously. LG is situated in ahigh-alpine region and is bounded on two sides by steep topography. The currentvalley profile does not form a closed catchment, excluding the possibility that a non-glacier-related water body existed in the area. The pillow pile reaches a maximumthickness of 150 m, suggesting a water body in excess of 150 m deep. Glacial iceis the only plausible mechanism for damming and maintaining a water body of thisdepth. Thus, LG, at a thickness >150 m greater than that of the present day, musthave enclosed an englacial lake into which the eruption took place.2. The pillow pile sequence has no discernible breaks (i.e., passage zones; Jones, 1968,1970; Pollock et al., 2014; Russell et al., 2013), indicating that the eruption did notexperience any major hiatus or water level changes. The englacial lake was sustainedthroughout the duration of volcanism.3. Discontinuous lenses of massive to crudely bedded, variably sorted pillow brecciaand bedded hyaloclastite accompany the pillow lavas in several places. These crudelyreworked deposits indicate local, aqueous transportation. The discontinuity of these9525 30 35 40 45 50Si / K  3456789100.5 (Fe + Mg) / K  12051045106510851105 0.5% H2O0 % H2O1 % H2O1.5% H2OPlCpxOl100.5 (FM) / K  161820222426283032Al + 2 Na / K  069a12051105108510651045PlOl1.5% H2O1 % H2O0.5% H2O0 % H2OAB069amodel line 118511651145112511851165114511253 4 5 6 7 8 9Figure 4.12: Phase diagram for Lillooet Glacier (LG) magma as a function of pres-sure and water content calculated using rhyolite-MELTS. Fractional crystalliza-tion simulations calculated with rhyolite-MELTS (Ghiorso and Gualda, 2015;Gualda et al., 2012), using sample 69 as a proxy magma composition. Dia-grams are formatted as Pearce Element Ratios, with axes designed to test forsorting processes involving (A) olivine (Ol) and (B) Ol versus plagioclase (Pl).Simulations are run at 1 kbar, under H2O contents ranging from 0 to 1.5 wt.%.The exsolution of H2O at these low pressures was suppressed manually duringsimulations. Whole-rock compositions are shown as closed circles, and glassanalyses are shown as open circles. Analytical 2σ error ellipses on whole-rockanalyses are shown in light grey. Simulations suggest that at 1 kbar, between 0.5and 1 wt.% H2O is required to suppress the onset of plagioclase crystallizationto a sufficient degree to allow the liquid line of descent to intersect the mea-sured glass compositions. The PER space immediately surrounding the data islabelled to display the temperature of the modelled liquid at 20 ◦C intervals.Cpx, clinopyroxene.96lenses suggests that they may have been constrained laterally and concentrated withinlocal depressions in the accumulating pillow pile.4. Lava tubes (Qvt) commonly terminate in large, partially intact lobes featuring radialcolumnar joints (e.g., Figure 4.5C, Figure 4.5D). These regular but highly variablejointing patterns are common in water or ice-cooled environments (e.g., Kelman et al.,2002a,b; Lescinsky and Fink, 2000; Lodge and Lescinsky, 2009; Mathews, 1952b,1958) and suggest the possibility of emplacement into a confined space, such as acavern melted beneath a glacier or a water-saturated hyaloclastite pile (Stevensonet al., 2009).4.6.4 Model for emplacement and implications for paleo-ice thicknessThe initial subglacial activity occurred in the northern map area. It is likely that the vent,now obscured beneath the large pillow pile, was a northwest-southeast-trending fissure,possibly exploiting basement structural elements (e.g., Figure 4.2). The volcano rapidlymelted the overlying ice, forming a stable body of water.Initial effusive subglacial eruptions developed an elongate tindar (e.g., Jakobsson andGudmundsson, 2008; Russell et al., 2014) deposit comprising densely stacked pillows andminor pillow breccia and hyaloclastite (Figure 4.13A). Continued eruption fed lava down-slope of the northern pillow pile, where it burrowed beneath the ice by melting, and ac-cumulated as thin sheets across the valley floor (Figure 4.13B) (Hungerford et al., 2014;Oddsson et al., 2016). The irregular basement topography exerted significant control overthe distribution of these lavas, concentrating their distribution in topographically low areas(e.g., Figure 4.13B, Figure 4.13C).Volcanic heating formed an englacial lake that was impounded against the northernvalley wall. At its maximum extent, the lake was ≥150 m deep, with its surface located atan elevation ≥1400 m above sea level. Nearly continuous exposure of pillow lava extendsfrom the northern pillow pile to the margin of Silt Lake (820 m above sea level). Thiselevation change (∆580 m) translates into a hydrostatic pressure that must be balancedby the ice load to prevent drainage of the lake via ice-lifting events (i.e., a hydrostaticpressure of ∼5.7 MPa (580 m×1000 kg m−3×9.81 m s−2) must be maintained throughoutthe eruption) (i.e., jo¨kulhlaups; Bjo¨rnsson, 2003; Clague and Evans, 2000; Oddsson et al.,2016; Tweed and Russell, 1999). We calculate (5.7 MPa ÷ 900 kg m−3÷9.81 m s−2) that>645 m of additional overlying ice would be required to sustain a lake of this size (Russellet al., 2013).On the basis of erosion profiles of the local mountainous terrain, Lawrence et al. (1984)estimated that the maximum elevation reached by the Fraser ice sheet was 2300 m above sea97150 mSection 1Section 1Section 2Section 3Section 2Section 3Basement hillsSyn-eruptivehyaloclastiteVent areaQvpQvpQvpQvpQvpQvtPoorly-bedded pillow breccia and hyaloclastiteSyn-eruptive HyaloclastiteNWCross Sections:ABC(Qvh)Qvh3D Models:WaterWaterWaterIceIceIceVertical pipe melted into ice sheetFigure 4.13: Schematic sequence model of the Lillooet Glacier (LG) volcanic erup-tion. A) The onset of effusive volcanism at LG melts an elongate, vertical pipe(cavern) into the ice and produces a pillow pile that is accompanied by syn-eruptive hyaloclastite and pillow breccia. B) Lava eventually breaches the pileand flows downslope, concentrated within topographically low regions of thebedrock. C) Further activity allows lava to breach the growing pillow pile to thenorth of a large basement hill. Lava forms thin, sheet-like pillow accumulationsdistributed across the valley floor. Significant stages of the eruption are illus-trated in cross sections (outlines given by white boxes in the three-dimensional(3D) model). Lithofacies colours follow the same convention as those in asFigure 4.298level. Given this constraint, our calculated minimum ice thickness (>645 m) and ice surfaceelevation (>1895 m above sea level) serve as a conservative estimate. Thermodynamic heattransfer calculations based on the methods and physical constants used by Ho¨skuldsson andSparks (1997) and Gudmundsson (2003) and Kelman et al. (2002b) suggest that the maxi-mum thickness of LG basalt (150 m) would be capable of melting 720 m of overlying ice.Our calculations consider the total sensible heat available from an eruption temperature of1165 ◦C to the glass transition temperature (700 ◦C), which roughly accords with a thermaltransfer efficiency of 45% (i.e., effusive basaltic eruption; Gudmundsson, 2003). We donot account for heating of the melt water because we consider heating of the englacial laketo operate on a longer time scale than the melting of the overlying ice to create the initialcavern. The thickness of ice (720 m) that could be melted by the thickest accumulation ofbasalt is less than the maximum thickness of ice (1050 m) found in the Lillooet Valley at thepeak of Fraser glaciation. A potential consequence of this is that if the eruption took placeduring peak glaciation, it may have had little surface expression other than a depression inthe surface of the ice. Given the quality of preservation of these deposits, and their locationat the base of a major glacially carved valley, we suggest that the eruption likely took placesyn- to post-peak Fraser glaciation (i.e., 17–13 ka; Clague, 2009; Clague and Ward, 2011;Menounos et al., 2009).4.7 SummarySouthwestern British Columbia is host to a large number of Quaternary glaciovolcanic cen-ters. Each offers information pertinent to paleoclimate studies and glaciovolcanic processes.Here, we provide, to the best of our knowledge, the first volcanological description of theLG basalts, a succession of pillow lava, pillow breccia, and hyaloclastite exposed by re-cent glacier retreat. By establishing a glaciovolcanic origin for the deposit, we elucidatethe presence of a sustained, ≥150 m deep englacial lake and calculate a minimum paleo-ice surface elevation of 1895 m above sea level. Our petrochemical analysis of the LGbasalts necessitates that all LG lavas be comagmatic and that all chemical variation can beascribed to crystal-sorting processes involving olivine + plagioclase. In addition, we usethermochemical modelling to provide first-order constraints on the conditions of storageand ascent. We establish possible shallow (<2 kbar or 7.5 km) storage and crystallizationunder 0.5–1 wt.% pre-eruptive H2O content conditions. Our model could be further testedand refined through melt inclusion analyses, thereby providing an independent estimate ofstorage conditions and volatile content.99Chapter 5The Table, a flat-topped volcano insouthern British Columbia:Revisited45.1 IntroductionGlaciovolcanism incorporates the interactions between volcanoes and ice. Glaciovolcaniceruptions commonly involve physical impoundment by ice, accelerated cooling by contactwith ice, snow and meltwater, hydrovolcanic fragmentation, and subaqueous eruption anddeposition (Smellie, 2018; Smellie et al., 2013). Interest in glaciovolcanism and glacio-volcanoes has increased in recent decades for two main reasons: the recognition of uniquehazards associated with glaciovolcanic eruptions, and their importance for paleoenviron-mental reconstructions (Edwards et al., 2009; Russell et al., 2014; Smellie and Edwards,2016).Large-scale melting of ice during volcanic eruptions can initiate lahars, floods andjo¨kulhlaups (Bjo¨rnsson, 2003; Gudmundsson et al., 1997; Major and Newhall, 1989;To´masson, 1996). Basaltic subglacial eruptions in Iceland, for example the 1918 Katlaeruption, have produced jo¨kulhlaups with discharge rates of 3× 105 m3 s−1 (To´masson,1996). The heat transfer processes attending these basaltic eruptions have been well-studied(e.g., Edwards et al., 2015b, 2012; Gudmundsson, 2003; Gudmundsson et al., 1997, 2004;Ho¨skuldsson and Sparks, 1997; Wilson and Head, 2002), but opportunities for equivalentstudies of intermediate and felsic glaciovolcanic eruptions are rare (e.g., Kelman et al.,4A version of this chapter has been published. Wilson, A.M., Russell, J.K., Quane, S.L., 2019. The Table,a flat-topped volcano in southern British Columbia: Revisited. American Journal of Science. 319, 44–73.https://doi.org/10.2475/01.2019.02.1002002a; Oddsson et al., 2016; Tuffen and Castro, 2009; Tuffen et al., 2002). Consequently,our understanding of many critical aspects of these systems, such as heat exchange effi-ciency and rates of ice melting, is minimal.Glaciovolcanism can produce a diverse range of volcanic lithofacies (e.g., Mathews,1947; Russell et al., 2014; Smellie and Edwards, 2016; Smellie et al., 2013) which indicatethe paleo-presence of ice and can constrain ice sheet surface elevation, ice thickness, andthe extent and nature of subglacial hydrology (e.g., Hubbard et al., 2006; Russell et al.,2013; Smellie, 2013, 2018; Smellie and Skilling, 1994; Walker, 1965). Commonly, glacio-volcanoes are the only preserved terrestrial record of ancient ice (Ryder et al., 1991; Smellieet al., 2008).The Table was one of the first recognized examples of intermediate-composition glacio-volcanism (Mathews, 1951). It is widely recognized as the type-example of a lava-dominated tuya, namely, a glaciovolcanic edifice that was predominantly constructed byeffusive eruption of lavas (Hickson, 2000; Kelman et al., 2002b; Mathews, 1951; Russellet al., 2014; Smellie, 2013; Smellie and Edwards, 2016). Lava-dominated tuyas are one ofnine categories of glaciovolcano recognized globally (Russell et al., 2014). Following thepioneering work of Mathews (1951), we use field observations and analytical modellingto re-examine this iconic volcano and reconstruct the volcanic processes attending its em-placement and growth.5.2 Geographic setting and contextThe Garibaldi volcanic belt is the northernmost (Canadian) portion of the Cascade volcanicarc of the western United States (Andrews et al., 2014b; Green et al., 1988; Kelman et al.,2002a) (Figure 5.1A). Quaternary volcanism has produced a 450 km-long band of 100 calc-alkaline to alkaline basaltic volcanoes (Bacon et al., 1997; Green et al., 1988; Green andHarry, 1999; Green and Sinha, 2005; Hildreth, 2007). Many of these volcanoes interactedwith large-scale ice masses and local glaciers (Hildreth, 2007; Kelman et al., 2002a).The Table is located in a high-relief area in Garibaldi Provincial Park (Green et al., 1988;Mathews, 1951, 1958) (Figure 5.1B). It is situated on a steeply sloping ridge of plutonicbedrock and rises to an elevation of 2010 m above the adjacent valley floors (1500 m)(Figure 5.1B and Figure 5.1C). The Table unconformably overlies the Table Meadows suite(Green and others, 1988) and minor accumulations of unrelated polymictic lithic brecciaand conglomerate (Figure 5.1B).1012000180017001600150018001700160017001600150018001900160015004985 00m.E4985 00m.E9909904995 00m.E4995 00m.E5526500m. N2702755528000m. NMt. HoodMt. AdamsMt. BakerLassen PeakMt. ShastaGlacier PeakMt. RainierMt. JeffersonNewberry Mt. St. HelensCrater LakeMt. CayleyMt. MeagerMt. SilverthroneMt. Garibaldi Franklin Glacier120°0'0"W120°0'0"W125°0'0"W125°0'0"W50°0'0"N45°0'0"N40°0'0"N0 200 400 600 800 1000 1200 1400 1600 1800 2000Distance (m)14001600180020002200Elevation (m asl)AJuan de Fuca PlateCanadaQuaternary volcanic rocksT1T1T2T4TmTmTmGGGCCBBBGaribaldi LakeWarren Valley100 0 10050KilometersBUSLegendPorphyritic andesiteOxidized, porphyritic andesite Porphyrtic andesite (Warren valley lobe)Table Meadows suite Glacial depositsLithic conglomerate and brecciaPlutonic rocks LakeRiverT1T2T4TmGCB0 0.50.25KilometersCCAA’A’AWarren Valley Garibaldi LakeThe TableThe Table The Table rocksOther rocksLegendGlacial striationFigure 5.1: Location and geology of the Table, southwestern British Columbia,Canada. A) Map of the Cascade volcanic arc along the western coast of NorthAmerica. The Garibaldi volcanic belt is the Canadian part of the arc. The Tableis indicated with a white star. B) Geologic map of the Table volcanic complex(units T1, T2 and T4). Unit T3 is not visible in map view at this scale. C) Crosssection (no vertical exaggeration) A to A’ (extent shown in 1B), showing thedistribution of Table rocks situated on a steep, alpine ridge..1025.3 Lava-dominated tuyasLava-dominated tuyas are a relatively uncommon glaciovolcanic landform (Kelman et al.,2002a; McGarvie, 2009; Russell et al., 2014; Smellie et al., 2013; Stevenson et al., 2011).They are constructed almost entirely of lava and lack the subaqueous and volcaniclasticlithofacies that typify the classical (mafic) tuyas (for example, pillow lavas, pillow breccia,lava-fed delta sequences etc.) (Jones, 1968, 1970; Kelman et al., 2002b; Mathews, 1947;Russell et al., 2014, 2013; Smellie and Edwards, 2016; Smellie et al., 2013; Werner et al.,1996). Broadly, lava-dominated tuyas are positive relief, flat topped, steep sided volcanoescomprising massive or stacked, intermediate-to-felsic lavas (Kelman et al., 2002a; Russellet al., 2014; Smellie and Edwards, 2016).Quaternary intermediate-to-felsic volcanic deposits are common throughout the Garibaldivolcanic belt, and five of these volcanoes are recognized as lava-dominated tuyas, includ-ing: Ring Mountain, Little Ring Mountain, Cauldron Dome and Slag Hill Tuya in theMount Cayley volcanic field (Kelman et al., 2002a), and the Table in the Garibaldi Lakevolcanic field (Green et al., 1988; Mathews, 1951). The Black Tusk (also in the GaribaldiLake volcanic field) is probably a highly eroded, sixth lava-dominated tuya (see chapter 2).Each of these volcanoes are almost entirely composed of coherent andesite and all displayunequivocal evidence for eruption within enclosing ice. This evidence includes: high re-lief and high aspect ratio morphologies reflecting physical confinement, jointing texturesthat suggest rapid cooling and quenching of the outer margins, and a flat upper surface thatapparently breached the enclosing ice (Kelman et al., 2002b; Russell et al., 2014; Smellie,2013). Within the literature, there is one description of a felsic (that is, rhyolitic) lava-dominated tuya in Prestahnu´kur, Iceland (McGarvie et al., 2007). It is demonstrably effu-sive, and ice quenched but lacks a flat upper surface and clear evidence for ice breaching(that is, oxidation of the top). Because Prestahnu´kur was emplaced subglacially, it mayrepresent a subglacial dome (Smellie and Edwards, 2016).5.4 Previous studies of the Table5.4.1 Mathews (1951)Mathews (1951) was the first to propose a glaciovolcanic origin for the Table. He describedthe edifice as an elongate “pillar of black dacite” with sheer marginal cliffs and a flat topthat showed striking superficial morphological resemblance to basaltic tuyas in northernBritish Columbia. Mathews made three key observations:1031. the Table is composed of alternating layers of “stoutly” (coarsely) and “slender”(finely) columnar jointed andesite lava flows;2. there are no primary volcaniclastic lithofacies associated with the edifice;3. and, the stacked andesite lava flows appear to bow sharply downward around theperiphery of the volcano (denoted by subhorizontal columnar jointing).Mathews suggested that the differentially-jointed andesite rocks were emplaced as in-dividual, layered lavas. Each layer consisted of a slow-cooled “stoutly jointed basal colon-nade” overlain by a thick zone of “slender jointed entablature”. His conceptual modelenvisaged lavas erupted sequentially beneath an ice sheet causing melting of a large cylin-drical pipe in the ice. Successive eruptions produced layers of lava that flooded the floor ofthe cylindrical cavity, spreading laterally until they came into contact with the surroundingwalls of ice. During the final stage of the eruption, molten lava erupted from near the top ofthe edifice broke through the solidified annulus of quenched lava and flowed down the gapthawed between the cooling massif and the surrounding ice. Mathews proposed a vent nearthe upper eastern margin and suggested that the prominent east-northeast to west-southwestelliptical shape of the massif may be due to migration of the ice cavity in response to co-eval east-northeast to west-southwest ice flow (Figure 5.1B and Figure 5.2). He argued thatthe volume of talus surrounding the edifice was erosional debris from the margins. Thisinterpretation has two consequences: i) the original form of the edifice was a semi-circular,vertical-walled cylinder, and ii) the eroded talus fan has remained essentially intact sincethe eruption. For this reason, he assumed that the eruption was probably younger than theLast Glacial Maximum (i.e., 17 ka; Clague and Ward, 2011).5.4.2 Other literatureGreen (1977, 1981a,b) provided a geochemical and petrological study of the Garibaldi Lakevolcanic field, and included the Table. Lescinsky and Fink (2000) include the Table in astudy of intermediate lava-ice interactions at stratovolcanoes in the Cascade volcanic arcand provide a morphometric classification of cooling fractures and a list of idealized land-forms. The Table is also referred to by Kelman et al. (2002a) and Kelman (2005) whomapped and described the lava-dominated tuyas (termed, flow-dominated tuyas by theseauthors) in the Mount Cayley volcanic field. These authors suggest that lava-dominatedtuyas are constructed from massive to stacked andesite lava flows, the uppermost of whichmay be separated by thin zones of oxidized autobreccia and scoria. Kelman et al. (2002a)highlight the importance of extreme topography and magma composition working togetherto control the eruptive behavior of effusive intermediate volcanoes in the Garibaldi volcanic104ABC330 m330 m470 m150 mBasement2010 m 2004 m160 m?Table Meadows suite260 mWarren valley lobeFigure 5.2: Field photographs showing morphology of the Table edifice. Views of theTable A) looking south, B) looking northwest, C) looking east. The dashed linesindicate inferred geologic contacts..belt. Russell et al. (2014) use the Table as an example in a continuum of nine different stylesof “tuya” that are descriptively named and classified using morphological and lithologicalmodifiers. Using this scheme, the Table is considered a flat-topped, lava-dominated tuya.Finally, the Table is discussed in chapter 2 of this thesis, where the edifice is used to dis-cuss the effectiveness of lava-dominated tuyas as paleoenvironmental indicators. This workdemonstrates how lava-dominated tuyas can constrain both the minimum and the maximumheight of the surrounding ice mass at the time of eruption. Other studies that mention theTable include: Ryder et al. (1991), Hickson (2000), Smellie (2013), Fillmore and Coulson(2013), Andrews et al. (2014a), and Smellie and Edwards (2016).5.5 Field observations5.5.1 Geomorphology and surface morphologiesThe Table is a steep-sided, flat-topped mass of coherent, porphyritic andesite (Figure 5.2).The upper surface is elliptical in shape (330 m long; 150 m wide) and has a long axis105trending approximately east-northeast to west-southwest. The northern and southern wallsare sub-vertical and are 160 and 260 m in height, respectively (Figure 5.2). The volcanicdeposits span a total relief of 470 m (Figure 5.2B) and the upper surface has an elevationof 2010 m (east) to 2004 m (west). The flat upper surface of the Table dips at ∼ 2◦ towardsthe west. It supports a thin mantle of soil and vegetation with only minor, frost-heavedisturbed rock exposed. There are no indicators of post-eruptive glacial overriding (forexample, glacial striations and erratic boulders) on the upper surface or on the accessible,lateral outer margins. Looking east-northeast, the Table has a height-to-width aspect ratioof 1. Looking south-southeast, the aspect ratio is 0.5. There is a single lobe of andesitethat extends to a lower elevation located in Warren valley, southwest of the main edifice(Figure 5.2B).The Table is surrounded by an apron of unconsolidated talus (Figure 5.2A). On thenorthern side, the talus pile comprises decimeter-sized fragments of andesite, dips outwardat an average angle of 35◦, and extends up to 100 m from the margin of the edifice. Alongthe southern side, the andesite talus fragments are finer (centimeter to decimeter-sized)and form a thin and steeper (average dip of 40◦) veneer overlying coherent andesite. Theandesite talus fragments are aphanitic. They are angular and have shapes ranging fromrecognizable individual columns to hackly fragments (e.g., Lescinsky and Fink, 2000). Thetalus exposed on the apron surface closely resembles the material making up the margins ofthe edifice.The walls of the Table are corrugated and form a series of prominent embayments andprotrusions. Along the western and eastern margins, these structures are vertically ori-ented, are 2–20 m across and extend over the entire vertical height (160–260 m) of theedifice (Figure 5.3 and Figure 5.4). Along the northern margin, the embayments and pro-trusions form large, irregular lobate structures (e.g., Figure 5.3A). The surface featuresmeet at both smooth, concave (U-shaped) and sharp, (V-shaped) convex intersections (Fig-ure 5.3C). Concave intersections predominate along the northern wall and near the uppermargin of the edifice (Figure 5.3B). Along the southern wall, the corrugations define a se-ries of east-northeast to west-southwest trending lineaments that form several subvertical,subparallel, planar bodies (Figure 5.3 and Figure 5.3).We used three-dimensional (3-D) photogrammetry to accurately measure the geometryof the Table. The photogrammetry model was created from 125 aerial images and wasgeoreferenced using 5 ground-control points (3 m) with a cumulative uncertainty of ±15.3m. The 3-D model constrains the minimum volume of the edifice to be 107 m3 (Figure 5.3).106~100 m~100 mNNNNNABDEFWarren valley lobeConvex intersectionsConcave intersectionsPlanar, sheet-like bodiesSurface structures extend across the vertical height of the edificeSub-parallel, sub-vertical structures2010 m2004 mConvex intersectionsCLobate structures“V”-Convex “U” -ConcaveProtrusions Embayment Figure 5.3: 3-D photogrammetry model of the Table. Model views A) and B) are atthe same scale (scales shown) and approximate a cross section (looking north-east) and plan view, respectively. C) is a plan view cartoon showing convex andconcave embayment and protrusion structures. D) to F) highlight the orienta-tions of sub-parallel, sub-vertical structures identified using the 3-D models. Allmodels show a white star at a shared point of reference. North orientation isindicated with an arrow..107BA10 m10 mBrecciaFigure 5.4: High-resolution field photographs showing subvertical embayment andprotrusion structures. A) View looking east. Enlargement shows vertically ori-ented structures meeting at sharp, convex intersections. B) View looking west.Enlargement shows the inner portion of a major subvertical structure exposingbreccia..5.5.2 LithofaciesThe Table is composed entirely of coherent, porphyritic andesite; volcaniclastic rocks arepractically absent. There is little chemical or petrographic variation within the massif: SiO2and Al2O3 contents range between 57.7 to 58.6 wt.%, and 18.1 to 18.4 wt.%, respectively.The andesite contains phenocrysts (0.5 to 2 mm) of plagioclase (20%), magnetite-replacedhornblende (10%), and microphenocrysts of pyroxene (2%). Between samples, the ground-mass varies from medium grained to glassy. Some samples exhibit a trachytic texture de-fined by flow-aligned microlitic plagioclase. For the purpose of this study, we have dividedthe lithofacies into four units that are listed by relative age (oldest to youngest) and definedon the basis of geometry, fracture morphology and groundmass texture:1081. T1: sparsely vesicular (10%), aphanitic to glassy, columnar and hackly jointed an-desite. These rocks are dark gray and form the bulk of the massif (Figure 5.5 andFigure 5.6A).2. T2: sparsely vesicular, aphanitic to glassy, variably oxidized, hackly jointed andesite.This lithofacies forms a sub-horizontal layer that is 10 to 30 m-thick at the top of theedifice (Figure 5.5 and Figure 5.6C). At the edges of the upper surface, the andesiteshows incipient surficial oxidization giving a reddish color (Figure 5.6C and Fig-ure 5.6D). The contact between units T1 and T2 is gradational. There are also rareexposures of scoriaceous, pervasively-oxidized autobreccia on the top of the massif.3. T3: sparsely vesicular, medium grained, holocrystalline, coarsely columnar jointedandesite dikes and sills. This lithofacies is light gray in color and forms numer-ous, thin sheets (2–5 m thick) that cross-cut the edifice (Figure 5.7). The layersare predominantly sub-horizontally oriented. They have gradational upper and lowercontacts with the enclosing rocks (unit T1). The contacts lack glass or breccia.4. T4: sparsely vesicular, subtly flow banded, fine-to-medium grained, columnar jointedandesite. This lithofacies defines a large lobe of andesite that is draped across thenorthern wall of Warren valley. The andesite has a 2 to 5 m-thick, coarsely jointedlower colonnade that is overlain by a 10 to 15 m-thick, fanning upper entablature(Figure 5.5 and Figure 5.6B).5.5.3 Cooling fractures and jointingThere are two main styles of cooling and contraction fractures developed within the Tablerocks: i) columnar fractures where adjacent fractures intersect to define regular columns,and ii) hackly fractures where arcuate, chaotically-oriented fractures form angular frag-ments with a range of sizes (Lescinsky and Fink, 2000; Lodge and Lescinsky, 2009).Pseudo-pillow fractures are rarely present (e.g., Forbes et al., 2012, 2014). In general,all fractures are closed, fracture spacing is closer and the fractures better developed in rockswith a finer grain size (Goehring and Morris, 2008; Grossenbacher and McDuffie, 1995).All fractures are primary and related to syn-emplacement cooling. The fractures at the Ta-ble are devoid of infilling material such as palagonite, glassy shards or secondary minerals(c.f., Lescinsky and Fink, 2000). The distribution of hackly fractures and columnar jointson the northern and southern walls of the Table is shown in Figure 5.5.Fine (10–20 cm joint spacing) columnar joints are well-developed around the northernbase of the edifice (T1) and in the upper part of the lobe of andesite that occupies Warren109?AB100 m100 mView: 168º (South)View: 358º (North)W EE WPorphrytic andesitePorphrytic dikes and sillsTalusVegetationMajor surface structuresVisible columnar jointsLEGENDFig. 5.6DFig. 5.6AFig. 5.6EFig. 5.6FFig. 5.6BFig. 5.6CPorphrytic andesite (Warren lobe)Oxidized porphrytic andesiteGeochronology sample:AW-15-139T1T2T2T1T1T1T3T3T3T3T4T1T2T3T4VVVVFigure 5.5: Line diagram of orthorectified image that was constructed using 3-D pho-togrammetry showing the distribution of the lithofacies. Views of the Table A)looking south, B) looking north. The locations of field photographs (fig. 6) areindicated. The orientations of visible columnar joints are indicated using finelines. All other areas show indistinguishable or hackly fractures. Larger surfacestructures are illustrated with heavy lines..110A BCE F~30 m0.5 m~50 m10 cm1970 m aslOxidized (unit T2)Non-oxidized (unit T1)Coarse colonnadeFanning entablatureUnit T4Unit T1?~1 mDFigure 5.6: Field photographs showing lithofacies and fracture characteristics at theTable. The location of each photograph is indicated in figure 5.5. A) Finecolumnar joints exposed near the base of the edifice (unit T1). B) Photographof columnar-jointed lobe of andesite draped over the edge of Warren valley (unitT4). C) Photograph looking west showing the gradational transition from non-oxidized, hackly fractured andesite (unit T1) into an incipiently oxidized equiv-alent (unit T2). D) Image looking north showing hackly-fractured, incipientlyoxidized upper outer margin of the Table. E) Image showing convoluted uppermargin of the Table (unit T2). Persons circled for scale. The white arrow indi-cates the location of figure 5.6F. F) Close-up image showing typical, aphanitic,hackly fractured andesite at the Table (unit T2). These hackly fractures are ex-posure within the inner portion of a 8 m-deep surface embayment..111valley (T4) (Figure 5.6A and Figure 5.5B). The columnar joints are irregularly oriented andcommonly define radially-jointed bulbous masses. Coarser columnar joints (up to 2 m jointspacing) are exposed in the basal colonnade of the Warren valley lobe (T4) and within thethin subhorizontally oriented dike and sill layers (lithofacies T3) (Figure 5.7C).Fine, hackly fractures (10–15 cm joint spacing) (e.g., Figure 5.6D, Figure 5.6E andFigure 5.6F) characterize the vast majority of the outer walls of the edifice (80%). Hacklyfractures also cover the entire surfaces of all embayment and protrusion structures (e.g.,Figure 5.4, Figure 5.6E and Figure 5.6F). The embayments do not expose coarser columnarjoints in their interiors. Finally, the upper outer edges of the edifice are composed of hackly-fractured, fine-grained andesite (unit T2). These rocks show incipient, surficial oxidation(Figure 5.6D).5.5.4 GeochronologyTo constrain the age of the eruption, a sample (AW-15-139) of holocrystalline andesite fromlithofacies T3 was analyzed using 40Ar/39Ar geochronology. The sample was obtained froman intrusive sill located on the northern side of the edifice. The location of the sample isindicated in Figure 5.5A. Because it is an intrusive lithofacies, T3 is amongst the latest lavato be emplaced. There is no evidence for long-lived activity at the Table and thus, sampleAW-15-139 is representative of the age of eruption for the entire volcano. A bulk separatewas sieved, washed in acetone and hand-picked for phenocryst free whole-rock groundmasschips. The monitor mineral TCR–2 with an age of 28.619 Ma (Kuiper et al., 2008) was usedto monitor neutron flux and calculate the irradiation parameter (J). The sample and standardwere wrapped in aluminum foil and loaded into aluminum cans of 2.5 cm diameter and 6cm height. The samples were irradiated in the research reactor of McMaster Universityin Hamilton, Ontario, Canada for 20 megawatt- hours. Upon return, the sample was step-heated, and analyzed using a 6-watt argon-ion laser following the technique described inYork et al. (1981), Layer et al. (1987), and Benowitz et al. (2014) using a VG–3600 massspectrometer at the Geophysical Institute, University of Alaska Fairbanks. Figure 5.8 sum-marizes the 40Ar/39Ar heating steps. The integrated and plateau age (100 ± 12 ka) andisochron age (97 ± 20 ka) (1σ analytical uncertainty) are within error. Due to the higherprecision we prefer the plateau age of 100 ± 12 ka.5.6 Summary of field observationsMathews (1951) recognized the enigma of relatively low-viscosity lavas forming a steep-sided edifice rather than thin sheets or broad channels. He concluded that their physicalbehavior could only be explained by the eruption environment, specifically, physical con-112CABAndesiteTalus/snow Fine columnar joints (T1) Coarse columnar joints (T3)Hacky fractures~10 m~2 mFig. 7CFig. 7C(T1)T1T1T1T3T3T3LEGENDFigure 5.7: Field photograph A) and line diagram B) showing the distribution and ori-entation of coarsely-jointed layers (dikes and sills) on the northern side of the Ta-ble. C) Enlarged image (the location is indicated in Figure 5.7A and Figure 5.7B)showing coarsely, columnar-jointed, sub-horizontal layer with gradational upperand lower contacts that lack glass and breccia..113 0 100200300 400Fraction of 39Ar Released0.0 0.2 0.4 0.6 0.8 1.0Age  (ka) 0  0.05  0.1  0.15  0.239Ar/40Ar36Ar/40ArSample: AW-15-139A BAge: 100 ± 12 ka.002.003.004Figure 5.8: Summary of 40Ar/39Ar geochronology results for sample AW-15-139(unit T3). A) Plateau diagram showing fraction of 39Ar versus age for the sam-ple. Eight out of eight fractions released 100% 39Ar, and the mean squaredweighted deviation (MSWD) is equal to 0.46. The sample shows a plateau at100 ka with a calculated 1 σ uncertainty of 12 ka. B) Inverse isochron plotshowing 8 heating steps. Ellipses denote analytical 1 σ uncertainty.finement of the lavas until they solidified. Our field observations and analysis corroboratethis interpretation.The outer surfaces of the Table feature aphanitic andesites that are hackly fractured tofinely columnar jointed. Both the aphanitic nature of these rocks and the finely-spacedfractures sets imply rapid cooling and quenching (Kelman et al., 2002a; Lescinsky andFink, 2000; Lodge and Lescinsky, 2009; Tuffen et al., 2001). The columnar joints aretypically oriented perpendicular to the sub-vertical outer surface or form radiating aggre-gates. These orientations indicate that the external surface of the Table cooled against avertical surface. We suggest that the corrugations are indicative of the original morphol-ogy of the cooling surface. Convex corrugation intersections probably result from overlap-ping lobes of outwardly expanding lava (e.g., Figure 5.3C). The more narrow, sub-verticalcorrugations that characterize the eastern and western sides broadly define a series of sub-parallel, sub-vertical sheet-like structures (e.g., Figure 5.3B). Importantly, all of these sur-face structures are coated by hackly fractures including the inner parts of the embayments(e.g., Figure 5.6E). Hackly fractures develop in response to rapid cooling/quenching. Forintermediate-composition lava-ice contacts, these fractures only form within the first meterof the lava surface (Lescinsky and Fink, 2000). The presence of hackly fractures coveringalmost the entire outer surface of the Table suggests that this surface is primary and largelyunmodified by erosion (c.f., Mathews, 1951). Portions of the surface that expose coarsely-jointed sills and dikes (unit T3) likely indicate regions of local surface collapse/erosion.The Table’s surface features, morphology and local physiography support ice as theagent of impoundment (Kelman et al., 2002a; Mathews, 1951). Furthermore, they suggestan eruption into a closely-confining englacial cavity having steep, sub-vertical walls of ice114(Lescinsky and Fink, 2000; Lodge and Lescinsky, 2009; Stevenson et al., 2009). There areseveral pieces of evidence that suggest that the glacier flow rate was slow relative to theemplacement time of the edifice:1. Well-defined ice-contact features populate all vertical surfaces. If the glacier wasflowing rapidly relative to the rate of edifice growth, there would be little opportunityto preserve ice-contact features on the lee side of the edifice.2. The vertical character of the ice-contact surfaces, representing the original edificewalls, suggests close ice confinement throughout the eruption.3. The geographical location of the edifice (that is, astride an alpine ridge; Figure 5.1C)implies a slow-moving local glacier, compared with the glaciers in the adjoining val-leys (i.e., glaciers move faster in valleys than on ridges; Sugden and John, 1976).There are some field indicators of paleo-ice flow directions around the Table (glacialstriations; Figure 5.1B), however these striations are probably associated with the LastGlacial Maximum (i.e., 17 ka; c.f., Clague, 2011; Mathews, 1951). The Table is datedat 100 ± 12 ka (Figure 5.8), corresponding to a period of time before the Last GlacialMaximum. There are no reliable indicators of syn-eruption ice flow direction.Oxidation of the lithofacies exposed in the upper 30 m of the Table (Figure 5.6C) in-dicates exposure of these lavas to the atmosphere and is consistent with breaching of theenclosing ice mass. This constrains the glacier surface to an elevation of 2010 to 2110m (see chapter 2). Rare occurrences of pervasively oxidized, slower-cooled autobreccia onthe upper surface also support a subaerial emplacement and some outward flow of the uppersurface lavas. On the outer, upper edges of the Table, the rocks are incipiently oxidized anddemonstrably quenched (Figure 5.6D). This combination suggests the erupted lava did notovertop the surrounding ice sheet to erupt supra-glacially (c.f., Smellie et al., 2013). Thesepaleo-glaciological constraints indicate the presence of a substantial Cordilleran Ice Sheetat the time of the eruption (100 ± 12 ka).The Table is volumetrically dominated by units T1, T2 and T3, however, the strati-graphic relationship of the unit T4 andesite is unconstrained. The T4 lobe has a coarselycolumnar-jointed lower colonnade that is overlain by a fanning entablature (e.g., Fig-ure 5.6B). The geometry of the unit (i.e., draped over the wall of Warren valley) suggestsflow downhill. Combined, these features imply a late eruption of lava that found and ex-ploited a subglacial meltwater drainage pathway that was probably developed during theearlier phases of the eruption.Mathews (1951) described coarse (“stout”) columnar jointed lavas forming discontin-uous, thin, sub-horizontal layers throughout the edifice. He interpreted these layers as the115lower colonnade of a series of stacked lava flows, an idea that formed the basis for his modelinvolving progressive, layered growth of the edifice. Mathews’ observations are valid, butincomplete, leading to an erroneous interpretation. Our more detailed field study indicatesthat, while periodically horizontal, these layers (unit T3) commonly cross-cut the edificeat a high angle (Figure 5.7A and Figure 5.7B). Furthermore, the bounding contacts withthe surrounding andesite (unit T1) are gradational. The contacts lack breccia and chilledmargins (fig. Figure 5.7C). The layers strongly resemble a network of dikes and sills, aninterpretation that is supported by their coarser jointing and holocrystalline groundmass tex-ture. This lithofacies probably represents bodies of late-to-synchronous magma injectionsas dikes into the growing massif.The Table lacks any evidence for layering of lavas or successive upward growth. Thereis also a lack of indicators for long-lived activity in general (for example, stratigraphic dis-continuities). The overall elongate morphology of the Table implies that the eruption wasinitiated and controlled by injections of dikes into the overlying ice. The linear, planar,subvertical sheet-like bodies that are exposed along the southern margin of the Table (Fig-ure 5.3B) resemble a series of stacked, partially coalesced sub-vertical dikes. These featuresextend from the base to the top of the massif, suggesting that the dikes were injected to anelevation of 2010 m into the ice. The maximum length of the dikes is constrained by thelength of the massif (330 m). We suggest that sequential injections of vertical sheets of lavaled to endogenous inflation and lateral expansion of the edifice in the shape of a continually-widening elliptic cylinder. The sub-vertical, linear, planar sheet-like bodies that are exposedalong the southern margin of the Table may be partially coalesced vertical dikes emplacedalong the margins of the main (inflating) massif. The lack of explosive fragmental litho-facies suggests that the Table lavas did not erupt explosively either beneath the ice (e.g.,Stevenson et al., 2009) or upon penetrating the ice surface (e.g., Wilson and Head, 2007).The andesite lavas are predominantly dense-to-sparsely vesicular implying that the eruptingmagma had a low volatile content (Edwards et al., 2015b; Kelman et al., 2002a).The Table also lacks lithofacies indicative of subaqueous eruption. In particular, thereare no pillow lavas, pillow breccia nor associated quench-fragmented breccia (e.g., hyalo-clastite). There are also no basal exposures of quench-fragmented material preserved bylater lavas (c.f., Smellie et al., 2011). This scarcity of primary explosive and hydrovolcaniclithofacies has been observed at other effusive intermediate glaciovolcanoes throughout theGaribaldi volcanic belt (Kelman, 2005; Kelman et al., 2002a,b) and in some effusive, felsicglaciovolcanic domes in Antarctica (Smellie et al., 2011). At the Table, this is strong evi-dence for a lack of meltwater impoundment and/or a sustained englacial lake. The limiteddirect water-lava interactions, suggested by the absence of subaqueous lithofacies, may be116a consequence of endogenous growth, low rates of meltwater production and/or high ratesof meltwater drainage.Direct lava-ice interactions commonly produce carapaces of loose quench breccia sur-rounding coherent lava cores, even under well-drained (dry) conditions (Bye et al., 2000;Edwards and Russell, 2002; Edwards et al., 2002; Kelman et al., 2002a,b; Stevenson et al.,2006, 2009; Wilson et al., 2016). At the Table, quench breccia is present, but sparse, and isintimately related to the vertical walls of lava, where relict pockets of breccia are preservedlocally between sub-parallel vertical sheet-like bodies along the southern side of the edifice(Figure 5.4B).117~250 mIce surface: ~2110 (?) mPlutonic basementIce massDikeLavaBrecciaIceN SA B Cracks in ice (?)~2010 mIncipient oxidation of lavaInflating edificeEncasing breccia carapaceCInflating edificeDMarginal dike injectionInjecting dikes and sills Marginal dikes coalesce with main edificeE FTalusTalusPreserved breccia?Preserved brecciaMinor erosion of edifice wallsHackly fractures on surfaceHeight reached by dike injecting into ice(2010 m)Advancing lava wallsLavaBrecciaIceProtrusions coalescePlan ViewCoarse internal jointing?118Figure 5.9: Conceptual physical model for dike-injection and endogenous growth atthe Table. The individual stages are outlined in the text (A to F). Meltwaterdrainage is indicated with double-headed arrows. The stage B enlargement de-picts rapid quenching and breccia carapace formation at the margin of the dike.The enlargement in panel 9D is in plan view and shows a mechanism for creatingconvex and concave surface intersections.5.7 Conceptual model for emplacementFigure 5.9 presents a conceptual model for emplacement of the Table volcano based onfield analysis alone. There are two basic components to the model. Firstly, the onset was afissure eruption involving injection of a dike into the overlying ice. Secondly, the growth ofthe edifice was by endogenous inflation of the dike system and continued injection of dikes.1. At the onset of eruption, the ascending dike propagates a vertical fracture to a heightof 250 m within the ice (2010 m elevation) (Figure 5.9A).2. Ice melting and lava quenching produces a carapace of unconsolidated, permeablebreccia. The breccia facilitates efficient meltwater drainage and adds thermal insula-tion (Figure 5.9B).3. Ongoing eruption causes outward, endogenous inflation of the dike. The volume ofinjected lava is approximately matched by the volume of ice melted (Figure 5.9C).The high-relief and steep surrounding slopes facilitated efficient melt water drainagefrom the eruption site.4. The sub-vertical walls advance forming the bulbous, embayed and protruded surfacestructures. Sharp, V-shaped (convex) intersections form where adjacent lobes of lavaoverlap (Figure 5.9D).5. Further dike injections along the southern margin form subvertical sheet-like bodies(Figure 5.9D). Endogenous inflation causes these sheets to partially coalesce withthe main edifice and preserve relic pockets of quench breccia between the sheets(Figure 5.9E).6. Near the end of the eruption, a network of dikes and sills is injected into the cooling,partially molten massif (Figure 5.9E).7. Following the eruption, final cooling and contraction fracture sets develop (Fig-ure 5.9F). Mass wasting of the unconsolidated carapace of breccia attends glacialretreat.1195.8 Analytical modellingField evidence provides the basis for an emplacement model involving within-ice dike in-jection and endogenous inflation in a closely-confined ice cavity. Our model assumes that:i) the eruption rate and dike ascent velocity (driven by magma buoyancy) had the capac-ity to crack the ice and ascend to 250 m within the ice mass, and that ii) melting of theice on the timescale of the injection approximates the effusion rate of the lava. The con-stants and variables used in the following calculations are given in Table 5.1 and Table 5.2,respectively.5.8.1 Dike injectionWilson and Head (2002) used simple models to show conceptually, how basaltic magmaswith high rise rates could crack the overlying ice and inject as dikes. However, they alsosuggested a low preservation potential for these ice-hosted dikes. Their hypothesis is sup-ported by recent field evidence from Iceland where ice radar soundings indicate a laterallypropagating dike into the ice cap at Ba´rðarbunga in 2014 (Reynolds et al., 2017). Addi-tionally, the crevasse pattern developed over the southern part of the 1996 Gja´lp eruption isattributed to within-ice dike injection and fracturing of the ice to the surface (Gudmundssonet al., 2004). Our model for the Table is of interest because: i) the within-ice dike injectioninvolves intermediate magma having higher viscosity and lower rise rates than the Icelandicexamples, ii) the injection rose 250 m into the ice mass (i.e., to the upper surface of the ed-ifice), and iii) the volcanic edifice is well preserved and records this process. To completethe dike injection calculation requires an estimate of magma buoyancy (driven by densitycontrasts), and strain rate imposed on the ice. The latter is a function of the initial effusionrate and the dike geometry, which controls magma rise rate at the base of the ice.Magma effusion/eruption rates and the rates of heat transfer to the ice control the erup-tion style, lithofacies, and edifice morphology (Gudmundsson, 2003; Gudmundsson et al.,1997; Harris et al., 2007; Hickson, 2000; Kelman et al., 2002a; Tuffen, 2007; Wilson andHead, 2002). We have used observations from active volcanic systems to estimate appro-priate effusion rates for the Table. The Table is a calc-alkaline volcano resulting fromsubduction-driven magmatism. Magma discharge rates observed at compositionally similarvolcanoes include: Bagana, New Guinea (0.3 to 1.8 m3 s−1) (Wadge et al., 2012) and Are-nal, Costa Rica (0.3 to 3 m3 s−1) (Borgia et al., 1983). The effusion rates for calc-alkalinedacite and andesite lava domes and spines give lower limit values: 0.25 to 2 m3 s−1 (Finkand Griffiths, 1998; Nakada et al., 1999; Watts et al., 2002). In our model, we use effusionrates (F) of 0.5 and 2 m3 s−1 which, given the volume of the Table, correspond to eruptiondurations of ∼2 and ∼8 months.120Table 5.1: Constants used in analytical models.Symbol Legend Units ReferenceF Magma eruption/effusion rate 0.5–2 m3 s−1(Borgia et al., 1983; Fink and Griffiths,1998; Nakada et al., 1999; Wadgeet al., 2012; Watts et al., 2002)w Dike width 1 m(Gudmundsson, 1984; Tuffen andCastro, 2009; Wilson et al., 2016)l Dike length 330 m -a Semi-major axis (half-length) 150 m -b Semi-minor axis (half-width) 65 m -z Height of dike injection into ice 300 m -ρm Density of magma in dike 2400 kg m3 (Oddsson et al., 2016)ρc Density of crustal rocks 2745 kg m3 (Bustin et al., 2013)ρi Density of overlying ice 920 kg m3 (Smellie and Edwards, 2016)ρw Density of water 1000 kg m3 -g Acceleration due to gravity 9.81 m s−1 -Tm Magma eruption temperature 1300 K (Oddsson et al., 2016)TK Reference temperature 273.15 K -Ti Temperature of glacier ice 268.15 K (Cuffey and Paterson, 2010)T f emperature of fusion of ice 273.15 K -Tw Exit temperature of meltwater 283.15 K(Gudmundsson, 2003; Gudmundssonet al., 1997; Major and Newhall, 1989;Oddsson et al., 2016)H f Heat of fusion of ice 3.33 × 10 5 J kg−1 -Cρm Heat capacity of magma 1000 J kg−1 K−1 (Oddsson et al., 2016)Cρi Heat capacity of ice 2108 J kg−1 K−1 -Cρw Heat capacity of water 4187 J kg−1 K−1 -The andesitic magma is positively buoyant relative to the local country rock (Table 5.1).This ensures that the magma-filled dike could rise at least to the base of the overlying icemass. Equating the combined lithostatic and cryostatic driving pressures to the magmastaticpressure of the intruding dike and allowing 10 MPa for the strength of the local host rockssupports dike injection to a height of 250 m (that is, an elevation of 2010 m) (Edwards et al.,2002; Wilson and Head, 2002; Wilson et al., 2013). We find that injection to 250 m withinthe overlying ice is feasible for any magma stored at depths greater than 4.4 km.Our model assumes that the eruption was fed by a fissure with a length (l) of 330 m.Dike aspect ratio (length:width) is a function of host rock strength, magma viscosity anddike overpressure (Kusumoto et al., 2013). We apply a dike width (w) of∼1 m based on theaspect ratios of∼300:1 for dikes observed in Iceland and British Columbia (Gudmundsson,1983, 1984; Tuffen and Castro, 2009; Wilson et al., 2016). The assumed range of eruptionrates (0.5 to 2 m3 s−1) and assumed dike geometry suggest magma ascent velocities of6 to 22 m hr−1. These rise velocities imply corresponding strain rates near the dike tipthat are ∼2 orders of magnitude greater than the relaxation timescale of ice (i.e., ∼10−3121Table 5.2: Variables used in analytical models.Symbol Legend Unitsvm Dike rise velocity m hr−1Ve Volume of elliptical cylinder m3Vi Volume of ice melted m3Vw Volume of melt water produced m3Vbx Volume of breccia carapace m3t Time shm Enthalpy of magma J m−3hi Enthalpy required to melt ice and heat water J m−3H˙e Rate of heat increases in the edifice J s−1H˙i Rate of heat consumption from melting ice and heating water J s−1f Efficiency of heat transfer -∆Hm Total heat supplied by the magma J∆Hi Total heat required to melt the hole in the ice J∆Hbx Total heat contained within the breccia carapace J∆Hm∗ Total heat remaining within the breccia carapace JAl Lateral surface area of elliptical cylinder m2qi Heat flux out of elliptical cylinder J s−1 m-2U Overall heat transfer coefficient J m-2 s−1 K−1L Length of the heat exchanger wall mK Thermal conductivity of heat exchanger wall J s−1 m−1 K−1versus 10−5 s−1; Glen, 1952; Wilson and Head, 2002). The implication is that even theseintermediate magmas would see the ice as a solid material similar to rock and overshootthe ice-rock interface causing brittle failure of the ice and leading to injection of the lava aspredicted by Wilson and Head (2002). The relatively low assumed eruption rates make therise of the dike into the ice accordingly slow (5.5 and 22 m hr−1). This implies that coolingmay hamper the initial rise of such a slow-moving dike, as presumably, the water drainagepathways may not establish themselves right away (e.g., Oddsson et al., 2016). In reality,the effusion rate during most eruptions of mafic to intermediate magmas peaks at the startof the eruption and then quickly declines, often waxing and waning in intensity with time(Harris et al., 2007). Higher initial effusion rates would further encourage within-ice dikeinjection (for example, if the initial effusion rates were 1 to 4 m3 s−1, the initial dike ascentvelocities would be 11 to 44 m hr−1).5.8.2 Ice melting and meltwater drainageRates of ice melting and meltwater production exert a primary control on the lithofacies andedifice morphologies formed at erupting vents (Edwards et al., 2013; Kelman et al., 2002a;Tuffen et al., 2002). Our field observations at the Table provide several key constraints onheat transfer from the lava to the ice:1221. The edifice is essentially preserved in its original form. The measured volume (∼107m3) is a reasonable approximation for the minimum volume of erupted lava (minus acarapace of quench breccia).2. The vertical quench-surfaced walls suggest that lava was emplaced within a closely-confined ice cavity and that the rate of ice-wall melt-back was equal to the rate oflava-wall advance (that is, the erupted lava was fully supported by ice through theduration of the eruption). The size of the edifice is approximately equivalent to thesize of the hole melted within the ice.3. The inflating lava mass was coated in a thin carapace of permeable quench brecciawhich modulated marginal heat transfer and provided an efficient drainage pathwayfor removing meltwater.4. The eruption was initiated by a dike injected directly into the overlying ice mass to aheight of ∼250 m. Growth primarily occurred by endogenous inflation of the dike.5. The lack of stratigraphic breaks indicates a relatively steady eruption rate.6. A lack of indicators for a subaqueous eruption indicate that the meltwater createdduring volcanic heating was efficiently and rapidly drained from the eruption site.This assertion is supported by the steep local relief (to facilitate drainage).The elliptical cylindrical portion of the volcano (i.e. excluding the Warren valley lobe:unit T4) has a volume (Ve) approximated as:Ve = piabz, (5.1)where, a is the semi-major axis (half-length; 165 m), b is the semi-minor axis (half-width; 65 m) and z is the height (250 m). This modelled volume (8.5× 106 m3) captures94% of the measured volume of the Table (minus the approximate volume of unit T4;9.0× 106 m3). We assume that the volume of ice melted (Vi) matched the volume of theedifice (Ve):Vi =Ve. (5.2)The eruption rate (F) therefore dictates the rate of edifice growth and ice melting:F =dVedt=dVidt=d(piabz)dt. (5.3)Using our assumed dike dimensions (Table 5.1), the dike inflates endogenously out-wards such that parameters a and z (Equation 5.3) are constant and volumetric growth is1230 50 100 150 200 t (days)02468 Ve (m3)1060 50 100 150 200 t (days)11.021.041.061.08 Al / Al (initial) (m2)10-1100101102103 Log t  (days)050100150200250 qi (J m2 s-1)10-1100101102103 Log t  (days)00.050.10.150.20.250.3 U (J m-2 s-1 K-1)0.5 m3 s-12 m3 s-1Meltwater Edifice 0.5 m3 s-12 m3 s-10.5 m3 s-12 m3 s-10.5 m3 s-12 m3 s-1A BC DFigure 5.10: Results of time-dependent heat transfer calculations. A) Volume of ed-ifice (Ve) and volume of meltwater (Vw). B) Increase in lateral surface area(Al/Al (initial)). C) Heat fluxes out of the surface, and, D) overall heat transfercoefficients (U).fully accommodated by an increase in b Figure 5.10. Rearranging and integrating Equa-tion 5.3 yields an expression for the transient growth of the ellipse (that is increasing b):b(t) =Ftpiaz+C. (5.4)The integration constant, C, is equal to w2 which is the dike half-width at t = 0. Thus,the volume of the edifice and the ice cauldron through time (Figure 5.10A) is:Ve (t) = Vi (t) = azpiFtpiaz+w2, (5.5)and the volume of water (Vw) produced from melting of the ice is:Vw (t) = 0.91 Ve (t). (5.6)Model meltwater production rates are low (0.5 to 1.8 m3 s−1; Figure 5.10A) and similarto melt rates observed at temperate alpine glaciers. For example, meltwater flow beneaththe Argentie`re Glacier in France, increases from 0.1 to 1.5 m3 s−1 in winter to 10 to 11 m3124s−1 in summer (Sugden and John, 1976). Considering the steep-relief setting of the Tableit is reasonable to assume that the meltwater was continuously and efficiently drained by anetwork of preexisting subglacial pathways (Kelman et al., 2002a; Smellie et al., 2006).Most glaciovolcanic eruptions exhibit much higher rates of ice melting (Gudmunds-son, 2003; Gudmundsson et al., 1997; Tuffen, 2007). As an example, an intermediate-composition lava flow emplaced beneath the Gı´gjo¨kull glacier during the 2010 Eyjafjal-lajo¨kull, Iceland eruption produced a near-continuous meltwater discharge of ∼50 to 300m3 s−1 over a 17 day period (Oddsson et al., 2016). In many cases, the rate of meltwa-ter production can exceed the rate of drainage from the vent causing meltwater accumu-lation and rapid meltwater discharge as jo¨kulhlaups (e.g., Grı´msvo¨tn and Katla, Iceland;Bjo¨rnsson, 2003; Gudmundsson, 2003; To´masson, 1996). We propose that, at the Table,the endogenous growth style limited direct lava-ice interactions, thereby producing rela-tively low meltwater rates. The eruption environment ensured that there was little to nocapacity for accumulation and storage of the meltwater and therefore limited potential forjo¨kulhlaups.5.8.3 Heat transfer calorimetryThe calculated volume of ice melted throughout the eruption is used as a large-scalecalorimeter that constrains surface heat flux of the erupted material (e.g., Gudmundsson,2003; Gudmundsson et al., 1997, 2004; Reynolds et al., 2017). For our analysis the Table istreated as a closed system where the heat loss through the sides of the edifice is used to meltthe ice and heat the meltwater. Heat loss through the growing upper surface of the edificeis subordinate and does not contribute to melt back of the ice walls, thus, it does not con-tribute to nor affect our calorimetric model. Treating the upper surface as a semi-infiniteslab with a convective boundary condition (Turcotte and Schubert, 2002), we calculatedthat the total heat loss through upper surface over the timescale of emplacement represents∼1.5% of the total enthalpy of the erupted material. In our thermal model there is potentialfor underestimating heat loss by not accounting for other important heat sinks. The mainsources of heat loss not considered in our model include: i) heating of pre-existing meltwa-ter, and ii) melting of ice volumes greater than the edifice volume. For example, englacialmeltwater flowing beneath or stored at the base of the glacier would represent a significantheat sink. However, given the physiographic setting of the table Figure 5.1C) it is unlikelythat significant bodies of water were stored or flowed through the vent area. We treated thesurrounding ice as essentially static over the timescale of the eruption. This limits the vol-ume of ice melting to a minimum and to be approximately equal to the volume of magmaerupted. If the ice flow was high, it would require more ice melting to accommodate the125growing edifice and, thus, greater heat loss. However, there is no evidence to support signif-icant ice flow during growth of the Table. Instead, the evidence suggests that the Table wasclosely confined on all sides throughout the duration of the eruption, implying the ice flowwas slow relative to the emplacement time. Conversely, there is the potential for underesti-mating the total heat in the system because the volume of erupted material is underestimatedor because we ignored important heat sinks. The volume of magma erupted is a minimumand does not include the portion of the edifice obscured by talus nor the unknown volume ofcollapsed breccia. This minimum volume of magma sets the volume of ice melted (that is aminimum). However, there is strong evidence for the edifice remaining more or less intactsince the eruption. Our model also ignores latent heats of crystallization as a potential heatsource. This is justified by the fact that on the timescale of emplacement most of the edificeremains at high temperature. Furthermore, rapid chilling of the interface quenches lava tobelow glass transition temperatures thereby suppressing crystallization and the release ofheats of crystallization (Kelman et al., 2002a).The volumetric heat content (hm) of the andesite lava at the eruption point is:hm = ρmCpm(Tm−TK), (5.7)where,Cpm is the specific heat capacity of the lava, Tm is the lava eruption temperatureand TK is a reference temperature. The rate of increasing heat content into the edifice is:H˙e = hmF. (5.8)Magmatic heat is dissipated by ice heating, ice melting, and heating of the meltwater(Gudmundsson, 2003). The combined heat loss (hi) is therefore:hi =Cpiρi(Tf −Ti)+H f +0.91[Cpwρw(Tw−Tf )], (5.9)where, Cpi and Cpw are the heat capacities of ice and water respectively, Tf is thetemperature of fusion of ice, H f is the heat of fusion of ice, ρw is the density of water ρi isthe density of ice, Ti is the ambient temperature of the ice, and Tw is the temperature of themeltwater exiting the system. Subsurface ice temperature (Ti) largely depends on whetherthe ice is cold-based, temperate, or polythermal. Many large, cold-based ice caps and icesheets have internal temperatures of around -20◦C, and temperate glaciers commonly resideat close to 0◦C (Cuffey and Paterson, 2010). We assume that the temperature of the ice was-5◦C (268.15 K) (Table 5.1).The energy used to warm the escaping meltwater represents a significant heat sink (Gud-mundsson, 2003; Wilson and Head, 2007). At the extreme end, the explosive eruption of126Gja´lp in 1996, produced meltwater with a mean residual temperature of 20 ◦C (Gudmunds-son, 2003; Gudmundsson et al., 1997); the eruption of Hekla in 1947 produced lahars andfloods with temperatures between 40 and 50 ◦C less than 5 km from the source (Major andNewhall, 1989) and an intermediate effusive lava emplaced at Eyjafjallajo¨kull in 2010 pro-duced meltwater temperatures of up to 17 ◦C measured ∼18 km from the vent (Oddssonet al., 2016). Here, we adopt a conservative meltwater temperature of ∼10 ◦C to reflect thefact that meltwater did not accumulate around the erupting vent (Table 5.1). The rate oftotal heat loss due to melting ice and warming meltwater is:H˙i = hiF (5.10)All heat is dissipated thorough the lateral margins of the edifice where the lateral surfacearea is:Al = 4zpi√a2+ b2, (5.11)and increase in the lateral surface area with respect to time (Figure 5.10B) is modelledas:Al (t) = 4zpi√a2+(Ftpiaz+w2)2. (5.12)The corresponding transient heat flow through that surface (qi) is:qi (t) =hiF4zpi√a2+(Ftpiaz +w2)2 . (5.13)Calculated heat fluxes are 43 to 185 W m−2 for lava extrusion rates of 0.5 and 2 m3s−1, respectively (Figure 5.10C). The geometry of the expanding elliptical cylinder causesthese fluxes to decrease slightly over the course of eruption. Effusion rate (F) causes thesurface heat flux to increase by ∼50 W m−2 for every additional 1 m3 s−1 increase in F .For comparison, our heat fluxes are 3 to 4 orders of magnitude lower than those observedfor both explosive basaltic glaciovolcanoes (e.g., Gja´lp, Iceland, 5 to 6×105 Wm−2; Gud-mundsson et al., 2004) and the explosive/effusive eruption at Eyjafjallajo¨kull in 2010 (1.3to 3.1× 105 W m−2; Oddsson et al., 2016). The heat flux values for the Table are moresimilar to conductive heat loss through the walls and roofs of mature (i.e., >2 weeks old)basaltic lava tubes (Witter and Harris, 2007).The overall heat transfer coefficient (U) for the system is a standard measure of theoverall ability of material barriers (in this case, breccia) to transfer heat. This value is127governed by the geometry of the barrier or interface (that is, the thickness) and its materialproperties (that is, thermal conductivity). The overall heat transfer coefficient is related tothe total heat flux, regardless of the specific energy exchange process (that is, conductive,convective or radiative heat transfer), by:U(t) =Hi(t)Al(t)∆T, (5.14)where, ∆T is the temperature difference between the erupted magma (Tm) and the sur-rounding ice (Ti). The calculated overall heat transfer coefficients for the Table eruptionare 0.04 to 0.18 W m−2 K−1 (for F = 0.5 and 2 m3 s−1, respectively) and decrease slightlyover the course of the eruption (Figure 5.10D). Low heat transfer coefficients can be takenas an indication of conductive heat transfer dominating over convective or radiative heattransfer at the lava-ice interface. For comparison, these calculated rates are an order ofmagnitude lower than the heat transfer coefficients associated with conductive cooling ofvesicular volcanic clasts (Stroberg et al., 2010) and are broadly similar to conductive heatloss from building walls constructed of stone, mud or brick (Mesda, 2012).5.8.4 Steady-state heat transfer across a breccia carapaceLavas in contact with ice cool rapidly to form quench breccia (Kelman et al., 2002a; Lescin-sky and Fink, 2000; Lescinsky and Sisson, 1998; Wilson and Head, 2002). At the Table,this is evidenced by relic pockets of breccia preserved around the margins of the volcano(Figure 5.4B). The breccia would provide thermal insulation and retard and modulate ratesof heat transfer (Edwards et al., 2013; Ho¨skuldsson and Sparks, 1997). We model thesteady-state heat transfer across the breccia at the lava-ice interface assuming a constantdifference in temperature between the erupted lava (Tm) and the ice (Ti), and the breccia tohave thickness (L) and thermal conductivity (K) (Figure 5.11).We provide two methods for estimating the thickness and properties of breccia sur-rounding the edifice. The first method is based on the model values of U and assuming allheat is transferred across a vertical barrier having a thickness (L) and a thermal conductiv-ity (K). As a model, the core of the edifice is kept constant at eruption temperature (Tm),assuming the lava loses only a fraction of its heat on the time scale of eruption and coolingof the internal portion of the edifice would be partially buffered by the latent heat of crystal-lization (e.g., Tuffen and Castro, 2009). Thermal conductivity (K) of the breccia is assumedto be an unknown constant. For a vertical heat exchanger, the relationship between K, LandU(t) is:U(t) =KL. (5.15)128Temperature (K)4006008001000200IceBrecciaLava120014001300 K: isothermal erupted lava (Tm)268.15 K: ambient ice temperature (Ti)Scenario 1:(All heat transferred to ice) Scenario 2:(Linear heat gradient) L* not to scaleFigure 5.11: Conceptual one-dimensional heat transfer model from lava to ice at theend of the eruption. The interface is a breccia layer of thickness (L) and thermalconductivity (K). Two plausible thermal gradient scenarios exist: i) all heat islost to the surrounding ice and the breccia carapace resides at the ambient icetemperature (Ti) and, ii) a linear heat gradient exists between the lava and thebreccia.Assuming K to be constant, any decrease in the magnitude ofU with respect to time (t)dictates an increase in the thickness of the breccia carapace (L):L=KU (t). (5.16)The magnitude of L is strongly controlled by U , which itself is primarily influencedby the eruption rate (Figure 5.12A). For an arbitrary thermal conductivity of 0.5 W s−1m−1 K−1 (a reasonable value for well-drained (dry) gravel; Dalla Santa et al., 2017) thecorresponding breccia thicknesses range from 11 to 12 and 2.7 to 3 m for effusion rates of0.5 and 2 m3 s−1, respectively. During the eruption the increases in breccia thickness are 1m and 0.3 m, respectively (Figure 5.12B).The second method is based on the geometric properties of a cylindrical veneer of en-closing breccia at the lava-ice interface. In this approach, we assume all heat for the icemelting derives from that veneer. Assuming that all heat was transferred through the lateralouter walls, the total enthalpy debt of the eruption (that is, enthalpy required to melt the icehole and heat the meltwater; ∆Hi) is equivalent to the enthalpy stored in the breccia (∆Hbx):∆Hi = ∆Hbx. (5.17)We envisage two scenarios: i) all heat is extracted from the breccia to the ice and thebreccia resides at the ambient ice temperature (Ti), or ii) the breccia preserves a linear1290 10 20 30 40 50 L (m)00.511.52 K (J s-1 m-2 K-1)0 2 4 6 8 10 L (m)101210131014101510161017 Log HbxR (J) He Hi0 50 100 150 200 t (days)02.557.51012.515 L (m)ABCArbitary thermal conductivity = 0.5  0.5 m3 s-12 m3 s-1Calculated breccia thicknesses (1.3 & 2.5 m)Scenario 1: all heat transferred to iceScenario 2: linear thermal gradient0.5 m3 s-12 m3 s-1Figure 5.12: Two methods for determining the average thickness (L) of breccia cara-pace surrounding the Table at the end of the eruption. A) U as a function ofL and K under steady-state conditions. The broad lines represent changing Uwith time. B) For arbitrary value of K = 0.5 J s−1 m−1 K−1, calculated increasein L through the duration of the eruption. C) Thickness (L) of breccia deter-mined through geometric properties of the breccia carapace and assuming allheat used for ice melting and meltwater heating derives from the breccia. Thetwo heat gradient scenarios that are shown in Figure 5.11 are denoted using agrey line (scenario 1) and a black line (scenario 2). The average thickness ofthe breccia (L) occurs at the intersection of ∆Hi and the scenario (model) line.thermal gradient from Tm to Ti (Figure 5.11). The volume of the breccia carapace, in termsof its thickness (L), is:Vbx = piazL, (5.18)hence, ∆Hbx can be expressed as:∆Hbx = ρmCpmpiaz(Tm−Ti)L, (5.19)where ρm is the density of the lava,Cpm is the heat capacity of the erupted lava, a and zare the half-length and height of the elliptic cylinder, respectively and (Tm−Ti) is the changein temperature across the breccia wall. For this scenario (all of the heat contained in thebreccia is transferred to the ice) the average thickness of the surrounding breccia carapaceis given by:L=∆H∗bxρmCpmpiaz(Tm−Ti). (5.20)130The thickness (L) of the breccia veneer would be 1.3 m (Figure 5.12C). However, ifwe assume that the breccia preserves a linear temperature gradient (from Tm to Ti), thenthe heat remaining (∆H∗bx) in the breccia is equivalent to 0.5(∆Hi) and the thickness of thebreccia is double (2.5 m). Given values of U (0.05 to 0.2 W m−2 K−1), this implies Kvalues of 0.1 to 0.4 W m−1 K−1 (for F = 0.5 and 2 m3 s−1, respectively). A 2.5 m thick,unconsolidated carapace would be susceptible to rapid erosion and would readily deformduring endogenous inflation, supporting our physical model for convoluted wall advance(e.g., Figure 5.3C and Figure 5.9D). This estimate is also in line with the scarcity of frag-mental lithofacies observed at other intermediate effusive deposits in the Garibaldi volcanicbelt (Kelman et al., 2002a), and suggests that, like the Table, they too once supported thincarapaces of breccia that have since been removed.5.8.5 Efficiency of heat transferThe efficiency ( f ) of heat transfer is the ratio of the rate of heat transfer from the eruptedmagma to its surroundings for melting of ice and heating of meltwater (Gudmundsson,2003; Gudmundsson et al., 2004; Ho¨skuldsson and Sparks, 1997). Thus,f =H˙iH˙e. (5.21)Our model effusion rates for the Table suggest rates of heat increase into the edifice of1.2 to 4.9× 109 J s−1 and rates of heat loss for melting of the ice and heating meltwaterof 2.4 to 9.6× 107 J s−1 (for F = 0.5 and 2 m3 s−1, respectively). This implies values ofefficiency of heat transfer ( f ) of ∼2 %.Gudmundsson (2003) showed that the efficiency of heat exchange for sequential erup-tion of basaltic pillow lava layers into a sustained englacial lake should be ∼10 to 45%,where the lowest values occur at high effusions (when the pillows are progressively cov-ered and spend comparatively little time exposed to the surrounding meltwater). In con-trast, explosive basaltic eruptions into the same environment will be much more efficientat transferring heat to the surrounding ice (commonly ∼70 to 80%) (Gudmundsson, 2003).Observed systems include the 1996 eruption at Gja´lp, Iceland (predominantly subglacial,basaltic and explosive), which displayed an overall heat transfer efficiency of 63 to 77%(Gudmundsson et al., 2004) and an effusive intermediate lava emplaced at Eyjafjallajo¨kull,Iceland in 2010, which delivered ∼45% of its total energy to the ice (Oddsson et al., 2016).The efficiency of heat transfer at the Table on the timescale of emplacement is thelowest (calculated or observed) of any known glaciovolcano and shows that the eruptiondelivered only a fraction (∼2%) of its total energy to the ice (an additional small fraction(∼1.5%) of the supplied heat would have been lost through the upper surface). This glacio-131volcanic edifice shares similar heat-transfer efficiencies to well-insulated, effusive subaeriallava flows. For example, basaltic lavas travelling in tubes lose ∼1 ◦C per km of distancetravelled suggesting an overall heat transfer efficiency (from lava to the surrounding wallrock) of <1% per km (Cashman et al., 1998; Harris and Rowland, 2009). Similarly, daciteblock-lavas featuring 2 to 4 m-thick, insulating blocky carapaces have extremely low ratesof internal cooling (∼0.08 ◦C h−1), indicative of very low heat transfer efficiencies (Har-ris et al., 2002). The extreme inefficiency exhibited by the Table is a consequence of loweffusion rates coupled with endogenous growth of a dike within a well-drained, closely-confined cavity. Endogenous growth limited the effective lava surface area available forheat exchange, and the subvertical, well-drained breccia carapace prevented convection ofmeltwater and/or steam (e.g., Ho¨skuldsson and Sparks, 1997; Tuffen, 2007; Tuffen et al.,2002). Higher efficiency at the Table could be achieved by increasing the energy cost as-sociated with melting (for example, increasing Tw to 20◦C would increase f to 3.7% andreducing Ti to -15◦C would increase f to ∼2.9%). Similarly, reducing the overall enthalpyinput into the system would also increase efficiency (e.g., reducing Tm to 1100 K would in-crease the efficiency to 2.3%). Also, if the upper surface (with a calculated surface area of∼ 3.4×104 m2) did not breach the ice surface, melting of the overlying ice would increasethe overall heat transfer efficiency marginally.Reasonable variations in the physical eruption parameters used in the models produceonly trivial increases in the calculated efficiency at the Table. This highlights the importanceof a sustained body of meltwater (englacial lake) surrounding the edifice during eruptionto facilitate rapid heat transfer. At the Table, the ratio of lava volume to the volume of theice cavern is equivalent to ∼1 (that is, a ratio of ∼1:1). In contrast, other thermally well-constrained glaciovolcanic eruptions such as Gja´lp (1996) and Eyjafjallajo¨kull (2010) showvolumetric ratios of ∼1:9.2 and ∼1:4, respectively (Gudmundsson et al., 2004; Oddssonet al., 2016). These ratios reflect fundamental differences between the mechanisms of heatexchange (that is, sustained convection versus conduction), and the effective surface areaavailable for heat transfer (that is, explosive versus effusive). For the Table to create a melthole that had an equivalent ratio to these edifices (that is, a volumetric ratio of ∼1:10) theefficiency ( f ) would need to be ∼21%; in line with typical estimates for effusive eruptionsinto sustained englacial lakes (e.g., Gudmundsson, 2003).5.9 ConclusionThe Table is an iconic, type-example of a lava-dominated tuya which we have revis-ited using field mapping, lithofacies identification, 3-D photogrammetry and 40Ar/39Argeochronology. We confirm a glaciovolcanic origin and propose a new model for em-132placement that involves within-ice dike injection and subsequent endogenous growth insidea well-drained, closely confined englacial melt cavity. The lithofacies and edifice mor-phology constrain the volume of the englacial melt cavity and the sequential growth ofthe edifice. We then explored the nature of heat transfer around the edifice margins. Theeruption produced a thin carapace of unconsolidated quench breccia. Thermal modellingsuggests that this breccia carapace was up to 2.5 m thick and had a thermal conductivityless than 0.5 W m−1 K−1. The overall efficiency of heat transfer from the Table massifto the ice was ∼2%; the lowest of any known glaciovolcanic system, reflecting the dom-inance of conductive (over convective) heat transfer. The Table, as the type example ofa lava-dominated tuya, provides new insights into glaciovolcanic eruption processes thatmay be relevant to a significant number of other volcanoes. Specifically, our results shouldbe pertinent to other glaciovolcanic eruptions including: lava-dominated tuyas, subglaciallava domes, ice-impounded flows, supra-glacial lavas, or lavas erupted from high-elevation,ice-clad vents.133Chapter 6Climate Change, distressedmountains and volcanic eruptions:Analogue models of topographic andlithologic volcano–ice coupling6.1 IntroductionUnderstanding feedbacks between changing climates and volcanoes is a critical and unre-solved issue in the Earth Sciences. Deglaciation and ice-sheet removal have been shownto enhance mantle magma flux, leading to higher rates of volcanism following deglaciationin Iceland (Hardarson and Fitton, 1991; Jull and McKenzie, 1996; Maclennan et al., 2002;Slater et al., 1998). The frequency, style and distribution of volcanic activity, however,are also influenced by the state of stress in the crust (Glazner et al., 1999; Jellinek et al.,2004; Maccaferri et al., 2015; Watt et al., 2013). Crustal stresses may affect the transportand storage of magma through the crust by modulating dike propagation (orientation andlikelihood of initiation) and changing the critical stress regime required for magma stor-age (Gonnermann and Manga, 2013; Gudmundsson, 2012; Kavanagh, 2018; Muller et al.,2001). Dike initiation and propagation has been shown to respond to stress induced by load-ing and unloading of major ice sheets (e.g., Glazner et al., 1999; Jellinek et al., 2004; Rubin,1995; Watt et al., 2013), transient changes in topography from tectonic activity (e.g., nor-mal faulting; Gudmundsson, 1984, 2012; Kavanagh, 2018; Maccaferri et al., 2015; Tibaldiet al., 2014) and to growth and collapse of volcanic edifices (Albino et al., 2010; Borgia and134van Wyk de Vries, 2003; Le Corvec et al., 2018; Mathieu et al., 2008; Muller et al., 2001;Pinel and Jaupart, 2005; Sigmundsson et al., 2012; Tibaldi et al., 2014).The question arises as to the extent to which loading and unloading of valley-scale,alpine glaciers in local, high-relief, landscapes surrounding volcanoes may modulate vol-canism in terms of eruption frequency, size, style and the distribution of vents. This ques-tion applies directly to both eruptions at large and eroded volcanic structures that host, orhave previously hosted glaciers (e.g., stratovolcanoes such as Mount Meager, southwest-ern British Columbia, Canada), and discrete volcanic events (i.e., monogenetic eruptions)occurring in regions of alpine-glaciated topography (e.g., the Cordillera of southwesternBritish Columbia, Canada).The premise of our study is that valley-filling glaciers will exert a non-uniform overbur-den pressure on the topography (i.e., the glaciers are thickest in the valleys) (Figure 6.1A)(Augustinus, 1995; Kinakin and Stead, 2005; Sternai et al., 2016). This overburden pres-sure will counteract and partially equilibrate the uneven loading pressure exerted by themountains and valleys (Savage and Swolfs, 1986), thereby slowing the rate of gravity-induced deformation in the rocks. Glacier retreat (and unloading of the valley topography)will rapidly alter the near-surface stress field promoting gravity-induced relaxation of thetopography (e.g., Kinakin and Stead, 2005; Savage and Swolfs, 1986). We discuss, qual-itatively, how the style of response of the topography (i.e., affecting the magnitude anddistribution of compressional or extensional stresses developed in the near-surface crustduring relaxation) may affect volcanism by influencing subsurface magma storage and thelocation, orientation and likelihood of propagating dikes.We use a series of fifteen exploratory analogue experiments to simulate glacial loadingand unloading on topography and observe deformation (and hence, stress) in the shallowcrust (i.e., within several km of the surface). Our experimental configuration models a highrelief, topographic landscape that is defined by the height of a volcanic edifice to the bot-tom of a broad valley which can be inundated by ice. The analogue models are designedto allow for variations in the near-surface crustal architecture. We allow for variations inthe local topography by changing the local relief and the breadth of valleys in our experi-ments. Variations in the rheology (i.e. the overall strength of the near-surface crustal rocks)are controlled using a two layer model, a classic analogue modelling set up (e.g., Byrneet al., 2013; Galland et al., 2014; Merle and Borgia, 1996). An upper layer of sand-plastermixture is used to simulate a stronger, more competent crustal layer and a layer of siliconeputty is used to simulate less competent, soft rocks. Overall rheological changes (i.e., ap-proximating a transition from predominantly hard, to softer rocks) are achieved by varyingthe relative thicknesses of these two layers. We do not vary the physical properties of thetwo materials. Changing the thicknesses of these two layers has the effect of promoting135or inhibiting ductile flow of the viscous layer, simulating the response from topographycomprising soft and hard rocks, respectively.The experiments provide a continuum representing relatively strong, competent sur-face geology versus weaker rocks. For application to volcanic landscapes, mechanicallystrong rocks may comprise coherent lavas or sub-volcanic intrusive deposits (e.g., Heapet al., 2016; Schaefer et al., 2015), while soft rocks may comprise unconsolidated pyroclas-tic material or strongly hydrothermally altered rocks (e.g., Merle and Borgia, 1996). Theexperiments track deformation of the surface of the models as they deform in response tosimulated glacial unloading of the valleys. Our preliminary experimental results suggestthat the rheology of the near-surface crustal rocks (i.e., the effective mechanical strength ofthe rocks) is a strong contributor to the overall style of deformation and dictates whetherthe high relief volcanic structures are in compression or extension during glacier retreat.We find that the topography (i.e., the geometry of the mountains and valleys) mostly affectsthe magnitude of deformation. We conclude by developing a conceptual understanding ofhow the glacial loading and unloading of valley-filling glaciers on volcanic structures couldaffect the distribution, size and frequency of magmatic intrusions and volcanic eruptions.6.2 Analogue model setup and scalingThe analogue experiments were made in a large rectangular container that had an under-lying layer of silicone putty and an overlying layer of sand-and-plaster mixture. The sil-icone putty responds to stresses by deforming in a dominantly ductile manner, while thesand-and-plaster mixture responds to stresses by deforming in a dominantly brittle man-ner (Figure 6.1B and Figure 6.1C). At the start of each experiment, a pair of dimensionallyidentical valleys was excavated from the sand/plaster/silicone and a thin layer of plaster wassprinkled over the top to preserve delicate structures. The excavation of the valleys simu-lated removal of a glacier load from the valleys (i.e., a rapid deglaciation). The excavationsused a shovel and a knife to rapidly cut through the silicone and sand-plaster as quicklyas possible, before any significant deformation could occur. Using our values of temporalmodel scaling (outlined in scaling section below) we set a threshold excavation time of 1minute (representing, at minimum, 1.9 years in nature; see below). In practice, this simu-lated a rapid deglaciation (i.e., removal of surface load) from the valley, occurring over anessentially instantaneous time frame. All valleys were constructed in less than 1 minute, orthe experiment was rejected. Because our experiments excavated sand-and-plaster mixturefrom the models (i.e., representing rock with approximately twice the density of glacialice), the observed stress-strain relationships probably represent maximum case scenarios136(i.e., deformation associated with removal of ice would be less than deformation associatedwith removal of rock).Using a two-layer brittle-ductile set up is standard for testing deformation resulting frommagmatic related topography in mountain ranges and volcanoes (Delcamp et al., 2012;Merle and Vendeville, 1995), and deformation associated with volcanic loading (Byrneet al., 2013; Merle and Borgia, 1996). The set up allows the shallow crustal behaviour tobe simulated in a simple way on the timescale of laboratory experiments, while preservingthe major salient features of the system. The experimental materials were identical to thoseused by Byrne et al. (2013), and as described in Galland et al. (2014). The lower boundarycondition of our models is the base of the rectangular container. To verify that this bound-ary did not impart any significant impacts on the major salient features of our models weconducted several similar experiments in a large container with a 1.2 m-deep silicone layer(e.g., Byrne et al., 2013) (see below for the results of these models).6.2.1 The effect of geometry (i.e., the size and shape of the valleys andmountains) and dimensional scalingTo test the effects of varied topography on the style and magnitude of deformation wechanged the size and shape of the valleys and mountains in the experiments. The topog-raphy is simplified to represent mountains and valleys as regular sequences of flat-toppedridges and flat-based valleys (Figure 6.1B). This allowed a variety of topographic configu-rations to be modelled in a repetitive sequence, thereby minimizing boundary effects thatmay be created by the walls of the container. All horizontal displacement measurementswere conducted on the central ridge (i.e., farthest from the walls of the container), and allvertical measurements were taken by averaging displacement of the two valleys against theheight of the central ridge (Table 6.1). For geometric scaling, we defined two dimension-less parameters that characterize the geometry of the topography. Valley aspect ratio (⊓1 =Vw/Vd) relates the overall dimensions of the valley, incorporating the valley top width (Vw)and the depth of the valley (Vd) at the onset of the experiment. We varied ⊓1 from 1.67to 4.50 to represent the broad range of narrow (i.e., deep, U-shaped valleys such as Athen-ley valley in southwestern British Columbia) to moderately-wide valleys (such as Lillooetvalley in southwestern British Columbia), found in most mountainous regions. We alsochanged the slope of the valley walls (Vsw), characterized by the valley-side-slope parame-ter (⊓2 = Vd /Vsw), which, combined with parameter ⊓1, enables the overall shape and sizeof most natural scenarios to be quantified. We varied ⊓2 from 0.44 (a broad valley with a24◦ side angle) to 1 (very steep-walled valley with a 45◦ side angle). Vertical displacement(i.e., strain) of the valleys (defined as [Vd(start) – Vd(end)]/Vd(start)) was measured after 30minutes. The Vd measurements were referenced to the top of the container (Table 6.1). To137DtBtVwVw- 2(Vsw) VswRw0.5(Rw)VdBrittle layerDuctile layerBC20 cmValley bulge Central ridge used for PIVAFigure 6.1: A) Photograph looking north at Lillooet Glacier at the head of the Lillooetvalley, a major glacierized feature in southwestern British Columbia, Canada.B) Experimental setup showing valleys excavated in sand-plaster mixture andsilicone putty. The area of the ridge used for particle imaging velocimetry (PIV)analysis is outlined with a dashed line. The dotted lines indicate twin basalthrusts developed around valley bulges. C) Cross section showing geometricparameters of all experiments: Bt = brittle layer thickness; Dt = ductile layerthickness; Vw = valley width; Vsw = valley side width; Vd = valley depth; Rw =ridge width.138quantify model displacement and displacement rate in the horizontal dimension, we usedparticle imaging velocimetry (PIV; Thielicke and Stamhuis, 2014) on images looking downon the model, taken at 1 minute intervals. These images were cropped to the central ridgepart of the models and the resulting vector displacement fields were averaged across thisarea (Table 6.1; area indicated in Figure 6.1B).6.2.2 The effect of rheology (i.e., effective rock strength) and temporalscalingTo simulate a wide range of shallow-crustal rock rheologies (i.e., simulating volcanoescomposed mainly of hard, competent rocks and those composed mainly of softer rocks),we varied the relative thicknesses of the two model layers. We did not vary the absolutephysical properties of the two model materials used in the experiments (i.e., sand-plastermixture and silicone putty). We quantify rheology using two dimensionless parameters.Firstly, the layer-thickness ratio (⊓3 = Bt /Dt), which relates the thickness of overlying brittlematerial (Bt) to the thickness of underlying ductile material (Dt). ⊓3 was varied from 0.22to 2.33. Secondly, the valley-depth-lithology ratio (⊓4 = Bt /Vd) describes the lithologiesthat are exposed by the valley wall. ⊓4 ranges from 0.33 to 1.75. Stress developed in theelastic portion of the model (i.e., the upper sand-plaster mixture layer) is scaled using thecohesion of the sand-plaster mixture (100 Pa; Galland et al., 2014) and a representativerock mass cohesion of 10 MPa (Byrne et al., 2013) (giving a cohesion scaling ratio of10−5). The Young’s Modulus value of the model sand-plaster is 5× 106 Pa (Byrne et al.,2013) and a value of 7.5× 1010 Pa is taken for natural rocks (giving a Young’s Modulusscaling ratio of 1.33× 104; Byrne et al., 2013). The viscosity of natural rock varies overmany orders of magnitude, from soft clays (∼ 1013 Pa·s) to stiff igneous rocks (∼ 1022 Pa·s).The viscosity of the silicone is about 105 Pa·s (Galland et al., 2014) and a natural viscosityof soft sediments is 1016 Pa·s (e.g., Makhnenko and Podladchikov, 2018). Dividing thesevalues obtains scaling ratios of 10−10 for soft sediments to 10−17 for stiff igenous rocks.Multiplying the viscosity scaling ratio by the cohesion scaling ratio yields a time scalingratio. For the soft rock case, 1 minute in the model equals 1.9 years, while, if much highervalue is use (e.g. a coherent lava at 1022 Pa·s) then one minute in model time would be 1.9million years.This temporal scaling shows that topography composed predominantly of soft rocks(i.e., represented by a model mainly composed of silicone putty) will deform viscouslya large amount, quickly after valley excavation, while valleys composed of harder rock(i.e. represented by a model composed mainly of sand-plaster mixture with a much higherviscosity and strength) will respond more slowly. We note that while competent rocks (e.g.coherent lavas) have a very high viscosity, the effective viscosity (and overall strength)13910-610-410-210 15 20 25 30-3-2-10110-4Ridge divergenceTime (minutes)Cumulative ridge strain0 50123410- 4Ridge strain rate (s-1) ExtensionContractionABStrain measured at 1 minuteModels showing initial ridge extensionModels showing initial ridge extensionCFigure 6.2: Measured horizontal displacements for all experiments using particleimaging velocimetry (PIV). Results are, A) the average of ridge strain rate, B)average of cumulative ridge strain, and, C) average ridge divergence field. PIVanalysis was conducted on the outlined (white dashed line) region in Figure 6.1.The red and blue gradients in panel C illustrate qualitative overall extension orcontraction, respectively.of fractured rock is lower, especially if the rock is altered, which is common in volcanicregions. Therefore 1022 Pa·s is a very high bound for long term deformation. In addition,soft rocks with low viscosities on the order of 1016 Pa·s, are less likely to support high-relieftopographies, and thus, this is also a very low bound for long term deformation. Glacialvalley lifetimes may extend over many hundreds of thousands of years, so even topographycomposed of mostly hard rocks may eventually accumulate viscous deformation.1406.3 Experimental resultsOur experiments simulate valley deglaciation by removing material to create a valley in thetopographic profile (i.e., valley excavation; Figure 6.1A). After the valleys were excavated,the analogue model undergoes viscous flow thereby relaxing the stresses induced by rapidunloading. On the timescale of our experiments, valley excavation is taken to be semi-instantaneous and is analogous to rapid deglaciation (i.e. less than 1 minute, see temporalscaling section above). We found that the rate of displacement was fastest at the onset ofeach experiment and then exponentially dissipated with time (Figure 6.2A). In all experi-ments, the horizontal displacement rate approached zero within 30-minutes, indicating thatviscous deformation had largely ceased and the stresses were fully relaxed by this time(Figure 6.2A). In all experiments we found that a relatively large vertical displacement alsocorresponded to a relatively large horizontal displacement (Table 6.1; Figure 6.3). Thirteenof the experiments showed contraction of the ridge at the onset of the experiment, while twoof the experiments showed initial ridge extension (Table 6.1). We conducted two additionalexperiments in a large container with a 1.2 m-deep silicone layer (e.g., Byrne et al., 2013).The purpose of these experiments was to verify if the depth of the lower container boundaryimparted a significant effect on the overall model results. These experiments were set upsuch that they represented a thin elastic layer with a deep valley, and a thick elastic layerwith a deep valley (i.e. simulating the soft and hard rock scenarios, respectively). The hardrock experiment showed initial ridge compression and valley bulging, while the soft rockexperiments showed initial ridge extension. This suggests that the absolute thickness of thesilicone layer (Dt) does not impart a significant control on the overall nature of the results.The two characteristic responses to our experiments (i.e., initial ridge extension andinitial ridge contraction) are described in detail below.141Table 6.1: Experimental parameters and measured vertical and horizontal displacements in all analogue experiments.Parameter∗ DescriptionExperiment IDAW8 AW14 AW21 AW22 AW23 AW24 AW25 AW26 AW27 AW28 AW29 AW30 AW31 AW32 AW33Experimental SetupVw Valley width 120 180 140 220 120 80 120 150 100 220 100 200 150 100 150Vd Valley depth 60 40 60 60 40 40 45 40 35 50 60 60 60 60 60Vbw Valley base width 0 0 20 100 0 0 0 50 20 0 0 100 50 0 50Vsw Valley side width 60 90 60 60 60 40 60 50 40 110 60 60 60 60 60Rw Ridge width 155 95 135 55 155 195 155 125 175 55 175 75 125 175 125Rsw Ridge side width 77.5 47.5 67.5 27.5 77.5 97.5 77.5 62.5 87.5 27.5 87.5 37.5 62.5 87.5 62.5Θ Valley side angle (◦) 45.0 24.0 45.0 45.0 33.7 45.0 36.9 38.7 41.2 24.4 45.0 45.0 45.0 45.0 45.0Bt Brittle thickness 70 70 60 60 60 60 60 60 60 60 70 70 70 20 30Dt Ductile thickness 70 70 60 60 60 60 60 60 60 60 30 30 30 90 80Td Tank top depth 10 10 30 30 30 30 30 30 30 30 50 50 50 40 40ObservationsV1l Valley 1 end depth (left) 70 20 50 45 50 70 40 50 64 40 110 90 93 35 30V1m Valley 1 end depth (middle) 70 22 60 40 70 70 70 55 63 40 108 95 91 32 25V1r Valley 1 end depth (right) 70 20 60 45 50 70 80 58 65 40 110 90 93 35 30V2l Valley 2 end depth (left) 70 25 50 45 70 70 80 35 60 40 110 95 95 35 30V2m Valley 2 end depth (middle) 70 20 50 40 60 70 70 40 60 40 105 95 94 31 25V2r Valley 2 end depth (right) 70 20 50 45 45 70 50 40 60 40 110 95 92 35 30Vd(end) Valley depth (mean[V1,V2] - Td) 60.0 11.2 23.3 13.3 27.5 40.0 35.0 16.3 32.0 10.0 58.8 43.3 43.0 -6.2 -11.7t time (minutes) 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30Calculated Parameters⊓1 Vw / Vd 2.00 4.50 2.33 3.67 3.00 2.00 2.67 3.75 2.86 4.40 1.67 3.33 2.50 1.67 2.50⊓2 Vd/Vsw 1.00 0.44 1.00 1.00 0.67 1.00 0.75 0.80 0.88 0.45 1.00 1.00 1.00 1.00 1.00⊓3 Bt / Dt 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 2.33 2.33 2.33 0.22 0.38⊓4 Bt / Vd 1.17 1.75 1.00 1.00 1.50 1.50 1.33 1.50 1.71 1.20 1.17 1.17 1.17 0.33 0.50Vertical displacement (strain) (30 minutes) 0.00 0.72 0.61 0.78 0.31 0.00 0.22 0.59 0.09 0.80 0.02 0.28 0.28 1.10 1.19Horizontal displacement (strain) (1 minute) 8.1E-5 8.6E-5 1.3E-4 3.3E-4 8.9E-5 1.4E-5 3.4E-5 1.5E-4 2.1E-5 3.1E-5 9.2E-6 3.4E-5 1.8E-5 8.1E-4 3.7E-5Horizontal displacement (strain) (30 minutes) 1.2E-4 1.2E-4 1.5E-4 4.5E-4 1.4E-4 1.7E-5 6.1E-5 2.1E-4 2.9E-5 5.1E-5 1.5E-5 7.6E-5 7.3E-5 8.1E-4 5.4E-4Ridge extension (E) vs. contraction (C) C C C C C C C C C C C C C E E*All length-scale measurement units are in milimeters (mm). The dimensions of the container were 550 x 300 x 150 mm1426.3.1 Initial ridge contractionMost of the experiments showed initial ridge contraction, ridge lowering and upward val-ley bulging attending simulated deglaciation (e.g., Matheson and Thomson, 1973). Theseexperiments have ⊓3 ≥ 1 (Table 6.1; Figure 6.4A). The valley bulge had extensional struc-tures that were bound by twin basal thrusts (Figure 6.4A). The models show central ridgedepression and contraction as the viscous material beneath the ridges was displaced down-wards and flowed towards the valleys. Horizontal displacement vectors (determined usingPIV analysis) indicate that the magnitude of ridge contraction decreased towards the cen-ter of the ridge. Models with the highest deformation showed subtle ”fish-scale” surface-breaking thrusts, formed due to outward-verging convexities (Figure 6.4A; Byrne et al.,2013). The experiments displayed no visible extension around the ridge edges. The high-est contractional area change occurred at the center of the ridge. Where ⊓1 < ∼2.5 and⊓4 > ∼1.2, the experiments remained undeformed (Figure 6.1C, Figure 6.3A). Deforma-tion rate is strongly influenced by the topographic geometry, specifically, valley aspect ratio(⊓1) is the largest contributor to deformation rate, with the layer–thickness ratio (⊓3) andthe valley-depth-lithology ratio (⊓4) have second order effects (Figure 6.3A, Figure 6.3C).The largest deformations were from experiments with the deepest and widest valleys (Fig-ure 6.3A, Figure 6.3C). Valley bulging is a feature recognised in some deglaciating sedi-mentary landscapes (Matheson and Thomson, 1973) and rapidly eroding valleys cut in sedi-mentary rocks (Horswill et al., 1976; Parks, 1991). The features are commonly attributed tothe elastic rebound of a differentially-loaded landscape in response to deglaciation (Math-eson and Thomson, 1973; Parks, 1991). To the best of our knowledge, valley bulging inresponse to deglaciation has not been recognised in hard bedrocks, probably because theserock types deform at a rate that is obscured by erosion and tectonic uplift. It is possible,however, that future high-resolution GPS and InSAR measurements may record these de-formations.6.3.2 Initial ridge extensionIn two of the experiments, the central ridge showed initial extension instead of contrac-tion. These experiments had a thin brittle layer and a thick ductile layer, and the valleydepth exceeded the brittle layer thickness (i.e., ⊓3 < 1 and ⊓4 < 1). Large extensionalcracks developed in center of the ridges and the valley walls flowed inwards at a rate thatexceeded ridge subsidence and upward valley bulge (Figure 6.4B). Some upward bulgingof the valleys did occur in the initial stages of these models, however the valley bulgeswere rapidly overlain by inward-flowing viscous material. These experiments showed thehighest displacements overall (Figure 6.2). The experiment with the most viscous mate-1430.811.21.41.61.821.5 2 2.5 3 3.5 4 4.500.511.522.5300.20.40.60.811.2Vertical strain (measured from experiments)Average valley depth (Vd) strain0.811.21.41.61.821.5 2 2.5 3 3.5 4 4.500.511.522.5301234510 -4Horizontal strain (measured from PIV)∏ 4 (Bt / Vd)∏1 (Vw / Vd)∏ 3 (Bt / Dt)∏ 4 (Bt / Vd)∏ 3 (Bt / Dt)Average ridge strain∏1 (Vw / Vd)Ridge compressionABCDRidge extensionNo DeformationΠ3 = 1 Π3 = 1Ridge compression Ridge compressionΠ3&  Π4< 1Π3&  Π4> 1 Ridge compressionRidge extensionΠ3&  Π4< 1Π3&  Π4> 1Figure 6.3: Dimensionless diagrams showing cumulative vertical and horizontalstrain for all experiments. The white dots indicate the dimensionless parame-ters used to define each experiment. The stippled sections indicate zones of nodeformation. Vertical strain was measured using the depth of the valley at thestart of the experiments and after 30 minutes. Horizontal strain was measuredusing particle imaging velocimetry (PIV) on average displacements on the ridgeportion of the model (area indicated in Figure 6.1A). The greyscale color gradi-ent corresponds to strain and was created using a smoothed linear interpolationin MATLAB®. Figure panels A and C show ⊓1 (valley aspect ratio; Vw/Vd) ver-sus ⊓4 (valley depth–lithology ratio; Bt /Vd); Panels B and D show ⊓1 (valleyaspect ratio; Vw/Vd) versus ⊓3 (lithology-thickness ratio; Bt /Dt) for vertical andhorizontal strains, respectively. The dashed white lines separate experiments thatshowed ridge extension and those that showed ridge compression.rial deformed the most, suggesting that the strength of the rocks making up the topographyis the primary controller in this scenario (Figure 6.3B, Figure 6.3D). Ridge extension andvalley-side-wall creep resulting in tensile failure structures (i.e., sackung) are widely recog-nised features associated with highly incompetent lithologies worldwide (e.g., Me`ge et al.,2013), and in some assymetric ridge forms developed in competent bedrock (e.g., Kinakinand Stead, 2005). These features are attributed to plastic and elasto-plastic, gravity-induceddeformation of rocks and ridge failure under self-weight (Savage and Varnes, 1987).6.4 Discussion: Topographic and lithologic volcano–icecouplingIdeas concerning climate as a trigger for volcanism are mostly considered at crustal scales(i.e., >100’s of km) (e.g., Glazner et al., 1999; Hardarson and Fitton, 1991; Jellinek et al.,2004; Jull and McKenzie, 1996; Nakada and Yokose, 1992; Watt et al., 2013). There is,however, a compelling body of literature to suggest that eruption frequency and vent distri-144Compressional structures Extensional structures6 mm 16 mm +5x10-4A B20 cm 20 cmBeforeAfterBeforeAfter-5x10-4+5x10-4-5x10-4Sand-plasterSilicone puttySand-plasterSilicone puttyBasal thrust faultsExtensional normal faults“Fish-scale” thrustsNormal faultsFigure 6.4: Representative experiments showing A) ridge compression, and, B) ridgeextension. The top row shows photographs of the experiments after 8 min-utes. The second row shows traced interpreted deformation structures devel-oped. Contractional features (thrusts) are shown with blue lines, and extensionalfeatures (normal faults) are shown with red lines. The third row shows accu-mulated horizontal displacement vectors (arrows) and divergence fields (red andblue colour gradient) calculated using particle imaging velocimetry (PIV). Redcorresponds to positive dilation (extension), and blue corresponds to negativedilation (contraction). The bottom two rows show schematic cross sections ofthe experiments before and after deformation. The arrows indicate direction ofmaterial displacement.145bution may also be influenced by small-scale perturbations in shallow crustal stress fields(e.g., Gonnermann and Taisne, 2015; Gudmundsson, 2012; Kavanagh, 2018; Kervyn et al.,2009; Mathieu et al., 2008; Muller et al., 2001; Murray, 1988; vanWyk de Vries andMatela,1998). The linkage between these shallow stresses developed in response to valley-scaleglacial unloading has not been previously recognised. While these influences may be lessimportant than those driven by crustal-scale processes (Watt et al., 2013), they may stillimpart a significant control on the size, frequency and location of volcanic eruptions.In this study, we have evaluated the effect that deglaciating topography may have on theorientation and likelihood of dike propagation and, in turn, the frequency and distribution oferuptions. We use analogue experiments to show that the glacial unloading of valleys couldinduce a significant stress/strain response that reflects the local topography and rock typeswhich could affect the likelihood and orientation of propagating dikes. Our preliminaryresults suggest that the rate of the topographic deformation is mostly controlled by the ge-ometry of the topography and the style of deformation is mostly controlled by the rheologyof the rocks. The experiments employ a simplified topography featuring regular, flat-toppedridges and broad valleys combined with a crust represented as two homogenous layers ofvariable thickness. In reality, both rock rheology and topography are much more complex(e.g., Kinakin and Stead, 2005). However, in simplifying we allow the main relationshipsto be observed over a continuum of responses. In this manner, the experiments define twocharacteristic end-member responses to deglaciation that may influence the behaviour ofvolcanic systems. For topography composed mostly of stiff, hard rocks, valley-glacier re-moval causes downward ridge compression (subsidence) and corresponding upward valleybulging. For topography composed mostly of soft rocks, the opposite occurs, and the ridgeis extended outwards as the valley-sides flow inwards to fill the valleys. In all experimentsthe rate of deformation was greatest immediately following the valley excavation (i.e., thesimulated deglaciation event; Figure 6.2A). It is therefore reasonable to suggest that thestrain induced by deglaciation in natural systems will also be greatest immediately follow-ing deglaciation.In order to scale stress developed in the brittle portions of the models to natural values,we use the PIV-measured accumulated horizontal strains for the first minute of modelsand a Young’s Modulus scaling ratio of 1.33× 104 (Figure 6.2B, Table 6.1) (see scalingsection, above). Compressive stress is taken to be positive and tensile stress as negative.In the first minute of the model, following valley excavation (i.e., when deformation ratewas the greatest and the stresses are at their maximum), the experiments that underwentridge compression showed between 9.17× 10−6 and 3.26× 10−4 of strain after 1 minute.This translates to between 46 and 1630 Pa of compressive deviatoric stress in the models,and scaled values of 0.7 to 25 MPa in nature. In contrast, the models that underwent ridge146extension showed ridge strain of−3.66×10−5 and−8.05×10−4. Indicating between -183and -4026 Pa of tensile deviatoric stress and scaled values to -3 to -60 MPa. In practice,the calculated tensile stresses mostly exceed the tensile strength of the rocks (probably<10 MPa; Turcotte and Schubert, 2002) and would actually be limited by this strength.However, the scaled stresses do serve to indicate that in the extensional case, the rockmasses will be faulted, as seen by the structures observed in the models (e.g. Figure 6.4).Due to the wide range of lithologic and geometric scenarios we have tested, our resultsare difficult to compare, directly, with previous estimates of stress induced by topographicdeformation (e.g., Kinakin and Stead, 2005; Me`ge et al., 2013). The results, however, arein line with numerical studies of tensile elastic stress developed in asymmetric topographycomposed of moderately-strong rocks (typically <10 Mpa; Kinakin and Stead, 2005). Insummary, topography controls the distribution of glaciers on the landscape. Our stress-strain information shows that deglaciation of glaciated topography will impart the greatesteffect on topography composed of soft rocks versus harder rocks.Estimates of driving pressure in dikes range from a few MPa to several tens of MPa,depending on the magma composition, dike over pressure, and reservoir pressure conditionsGeshi et al. (2010); Gonnermann and Taisne (2015); Kusumoto et al. (2013); Pollard andMuller (2008). At these pressures, transient deviatoric stess induced by topographic andlithologic glacier coupling may impart a significant control on the orientation and thereforesurface location (e.g., Gaffney and Damjanac, 2006; Gudmundsson, 2012; Kervyn et al.,2009) and likelihood of dike initiation (e.g., Glazner et al., 1999; Jellinek and DePaolo,2003; Jellinek et al., 2004; Rubin, 1995). By suppressing or enhancing shallow magmachamber rupture, glacially-imposed deviatoric stress may also influence the timing, locationand depth of magma storage (Glazner et al., 1999; Jellinek et al., 2004; Nakada and Yokose,1992).Figure 6.5 shows, conceptually, how loading and unloading of valley-filling glaciers ontopography could influence volcanism by affecting the transportation and storage of near-surface, magma-filled dikes. Although ice has less than half of the density of most volcanicrocks, periods of ice loading, where glaciers fill-in the valleys, will act to partially equi-librate the uneven load imposed by the topography (Liu and Zoback, 1992; Savage andSwolfs, 1986). Glacial periods should, therefore act to slow or inhibit viscous deforma-tion of the rocks. Under these conditions, ascending dikes will generally propagate in thedirection of the most compressive stress, in this case, vertical across all parts of the topogra-phy (Gonnermann and Taisne, 2015). We interpret that periods of glaciation should favourundifferentiated, sub-vertical magma rise under all parts of the topography (Figure 6.5A).Deglaciation, however, will cause a significant stress/strain response in topography, thestyle of which will depend on the rheology of the rocks. Ridge compression and upward val-147Valley-filling glacierZone of relative compressionZone of relative extensionEruption LocationSoft rocksValleyRidge/mountainGlaciationDeglaciationHard RocksDeglaciationSoft RocksABCGlacierHard rocksValley bulgeRidge compressionMagma intrusion?Ridge  extensionArrested dikesbeneath valleyEruptions deflectedto valley wallsArrested dikesbeneath valleyAscending dikeDisplacement directionOriginal topographyFigure 6.5: Generalized conceptual model for climate-induced glacier loading/un-loading affecting volcanic eruptions. A) a glacial period, B) after deglacia-tion with near-surface crustal rocks that are hard and competent, and, C) afterdeglaciation with mostly soft, weak rocks. Three ascending dikes are shown ineach figure panel (red arrows) showing the interpreted consequence for dikesascending beneath valleys, ridges and under valley walls. The yellow stars in-dicate interpreted likely locations for eruptions. The black arrows indicate thedirection of ductile deformation of the rocks. The red and blue colour gradientsqualitatively represent zones of relative extension and compression, respectively.148ley bulging occurs if the rocks are mostly competent and hard and respond predominantlyelastically to the stresses induced during glacial unloading. In this scenario, the surfaces ofthe ridges and mountains are compressed, however, the cores of the mountains are extendedto accommodate the strain associated with valley bulging (Figure 6.5B). Dimensional scal-ing suggests that compressive deviatoric elastic stresses at the surface may reach up to 25MPa over 1.9 years to 1.9 million years of deformation (for rock viscosities of 1016 and1022 Pa·s, respectively). At these magnitudes, the stresses may cause failure (faulting) ofthe rocks and may also influence the likelihood and orientation of dike propagation beneaththe landscape (Geshi et al., 2010; Gudmundsson et al., 2012; Kusumoto et al., 2013; Pol-lard and Muller, 2008). Extension within the cores of the mountains (e.g., Figure 6.5B)may also favour magma intrusions as these areas are placed into extension. We speculatethat deglaciation may encourage shallow-crustal magma storage and fractionational crys-tallisation and crustal assimilation of the magma. In this manner, while deglaciation maycause an overall reduction in volcanism (i.e., due to the widespread development of com-pressional surface stress), the eruptions that occur during this time may be larger and moreexplosive. This is due to a tendency for magmas to differentiate and create more hydrousand silicic compositions while stored within the crust (these magma compositions typicallyproduce more explosive eruptions). Furthermore, dikes propagating beneath valley wallsmay be deflected away from the zones of ridge compression and towards the valleys (Fig-ure 6.5B). Dikes propagating beneath the valleys may be arrested at depth due to additionalcompression in this zone (Figure 6.5B). We found that the largest responses were fromexperiments with the highest relief (i.e., with deep and wide valleys). This indicates thatdeeply-dissected volcanic structures with large valleys should be more susceptible to theseeffects during deglaciation.If the topography mostly consists of weaker lithologies, however, deglaciation will pro-mote extension of the ridges and compression of the valleys (Figure 6.5C). Under theseconditions, the tensile deviatoric stresses in the ridges may reach up to 60 MPa, promptingtensile failure of the rocks and spreading (sackung) of the ridges. This stress scenario mayfavour sub-vertical dike propagation beneath the ridges and encourage magma flow to thesurface and eruptions in these areas (Figure 6.5C). Correspondingly, however, dikes propa-gating beneath the valleys may be arrested due to the compression associated with inwardflow of the valley walls (Figure 6.5C). Dikes propagating beneath valley walls may be de-flected towards the zones of extension in the ridge. We speculate that deglaciating valleyscomprising incompetent lithologies should promote frequent and less-violent volcanism.Our experiments make several simplifications concerning the topography, geometryand the rheology of the rocks involved in near-surface glacier-induced deformation. Ofnote, glaciated landscapes typically do not comprise sequences of flat-topped ridges and149U-shaped valleys, rather, they are made of steep mountains, areˆtes and cirques. Steep-sided ridges would serve to add mass to the ridge portion of the landscape (e.g., Kinakinand Stead, 2005; Savage and Swolfs, 1986). For the extensional and compressional ridgecases, respectively, we expect that this would: i) increase the gravity-induced stress driv-ing ridge lowering and thereby exacerbate ridge compression and associated valley bulging(and would presumably increase the compressive effect on propagating dikes), and, ii) in-crease the stress driving ridge spreading (and thereby encourage dikes). Furthermore, ouruse of two analogue materials to represent overall rock strength (i.e., silicone putty andsand-and-plaster mixture) is unrealistic. Near-surface lithology may be more accuratelyrepresented by using a single material to simulate landscape evolution (e.g., Kinakin andStead, 2005). This is difficult in practice, however, as there are no single materials rep-resenting a continuum from ductile to elasto-plastic rheology that are usable for analoguelaboratory experiments. Numerical methods (e.g., Kinakin and Stead, 2005) may be a usefulsupplement for future experiments, allowing testing realistic glaciated landscape topogra-phies and leading to better constraints on the critical boundary rheology and geometry fora deglaciating to impart compressional versus extensional deformation (i.e., Figure 6.5Bversus Figure 6.5C).6.5 SummaryOur analogue models illustrate linkages between changing climate (i.e., growth and decayof glaciers) and volcanism based on variations in the near surface stress fields imposedduring deglaciation of topography. For the most common topographic-rheologic configu-ration deglaciation and melting of ice sheets may be linked to suppressed dike transportand storage of magmas beneath ridges (and presumably less eruptions). However, for rocksequences composed predominantly of soft rocks that can flow, deglaciation may induceridge spreading and encouraged, sub-ridge dike transport. These effects may be importantfor the behaviour of volcanoes with shallow, sub-surface magma storage reservoirs, or theultimate fate of dikes that exist near to the ground surface.It is likely that the shallow-crustal topographic-lithologic-ice coupling effects we havedescribed exert a subordinate effect on volcanism when compared with large-scale influ-ences on magmatism (e.g., Hardarson and Fitton, 1991; Jull and McKenzie, 1996; Maclen-nan et al., 2002), or crustal-scale controls on magma transport and storage (e.g., Glazneret al., 1999; Jellinek et al., 2004; Nakada and Yokose, 1992; Watt et al., 2013). Shallow-crustal topography–ice coupling, however, may be important at the scale of individual ed-ifices that host (or have hosted previously) valley-filling glaciers. This makes our modelresults applicable to stratovolcanoes that are deeply glacially dissected and/or host retreat-150ing glaciers (e.g., Mount Meager, southwestern British Columbia, Canada; Roberti et al.,2018), and new and long-lived volcanic complexes that exist in high-relief topographic envi-ronments (e.g., the Garibaldi volcanic belt and Cordillera of southwestern British Columbia,Canada; Green et al., 1988; Kelman et al., 2002b; Russell et al., 2007). Although our mod-els are semi-quantitative and preliminary, they provide a first-order framework for isolatingthe geometric and lithologic parameters that may control gravity-induced deformation inglaciated/deglaciating volcanic structures and are an important first-step towards under-standing the full effect of climate change on volcanoes.151Chapter 7Glacial pumping of amagma-charged lithosphere7.1 IntroductionUnderstanding linkages between volcanic eruptions and climate-related growth and decayof glaciers is a critical and unresolved issue. These causal connections are important for im-proving eruption forecasting (e.g., Pagli and Sigmundsson, 2008; Pinel and Jaupart, 2005;Sigmundsson et al., 2010), managing hazards associated with deglaciating volcanic struc-tures (e.g., Major and Newhall, 1989; Roberti et al., 2018) and for using volcanoes as anunbiased record of ancient paleoenvironments (e.g., Russell et al., 2014; Smellie, 2018;Smellie et al., 1993, 2008; Smellie and Skilling, 1994). Ideas concerning climate-relatedforcing of volcanism are embedded in the literature (e.g., Aubry et al., 2016; Glazner et al.,1999; Grove, 1974; Hooper et al., 2011; Jellinek et al., 2004; Jull and McKenzie, 1996;Kutterolf et al., 2013; Mathews, 1958; Rampino et al., 1979; Rawson et al., 2016; Schmidtet al., 2013; Sigmundsson et al., 2010; Sternai et al., 2016), however, they remain to betested over a full range of tectonic environments and temporal scales (Edwards and Russell,2002; Smellie and Edwards, 2016; Tuffen and Betts, 2010; Watt et al., 2013). Specifically,the coupling between glaciers and volcanoes in continental arcs has received comparativelylittle attention when compared with studies examining linkages in thin-crusted regions suchas Iceland (Smellie and Edwards, 2016). For tectonic environments where the lithosphereis thin and the mantle is near to its melting point (e.g. oceanic ridges and continental riftzones), ice sheet unloading has been shown to increase magmatism by facilitating inwardflow and melting of hot asthenosphere (Jull and McKenzie, 1996; Maclennan et al., 2002;McGuire et al., 1997; Schmidt et al., 2013). The last deglaciation in Iceland, for exam-ple, was accompanied by a ∼20 to 30-fold increase in volcanic eruptions over the normal152background rate (Hardarson and Fitton, 1991; Sigvaldason et al., 1992). In addition, post-glacial and present-day volcanism in Iceland may also be influenced by modulated captureof magma within the crust as a consequence of evolving crustal stress fields associatedwith glacier melting (e.g., Hooper et al., 2011; Pagli and Sigmundsson, 2008; Sigmundssonet al., 2012, 2010).For volcanic arcs, however, where the lithosphere is substantially thicker (i.e., >30km), magmatism is driven by dehydration melting processes and magma fluxes vary ontectonic timescales (i.e., 10’s–100’s Ma) (e.g., Jellinek et al., 2004). Melting of the mantlelithosphere in these domains is mostly controlled by hydrous mineral dehydration processesthat are unlikely to be affected by the nominal isostatic influence (i.e., <∼ ∆20–30 MPa)of rapid (i.e., over 10’s–1000’s yr) glacier growth and decay (Cruden and Weinberg, 2018;Glazner et al., 1999; Grove et al., 2012; Jellinek et al., 2004; Nakada and Yokose, 1992;Peacock et al., 1994; Watt et al., 2013; Winter, 2013). This is evidenced by the wide rangeof crustal thickness and mantle solidus depths throughout continental volcanic arcs acrossthe globe (e.g., Arculus, 1994; Profeta et al., 2016; Tesauro et al., 2015), and the widerange in magma compositions that these arcs support (e.g., Ducea et al., 2015; Hildreth,2007; Peacock et al., 1994). Furthermore, the petrologic and geochemical signatures ofarc rocks often indicate long and interrupted crustal transit times that mostly exceed theduration of glacial cycles (Hawkesworth et al., 2002; Muller, 1997; O’Reilly and Griffin,2010; Paillard, 2001). If causal linkages between volcanoes and ice exist for volcanic arcs,they must relate to processes that modulate transport and storage of the magmas throughthe crust (Barnett and Gudmundsson, 2014; Edwards and Russell, 2002; Glazner et al.,1999; Jellinek et al., 2004; Kavanagh and Sparks, 2011; Kavanagh, 2018; Kavanagh et al.,2018; Kutterolf et al., 2013; McGuire et al., 1997; Nakada and Yokose, 1992; Nyland et al.,2013; Singer et al., 2008). In practice, glacial loading and unloading in arcs can affect thedynamics of dike propagation and the stability of crustal magma storage zones, therebymodulating volcanism (Gonnermann and Taisne, 2015; Hooper et al., 2011; Jellinek andDePaolo, 2003; Jellinek et al., 2004).We model the eruptive response of Quaternary volcanoes in the Garibaldi volcanic belt(GVB) of southwestern British Columbia (SWBC), Canada to changes in the thickness ofthe Cordilleran Ice Sheet (CIS) over the last glacial cycle (i.e., the past 40 ka). Our model isbased on the idea presented by Nakada and Yokose (1992), whereby bending of the elasticcrustal lithosphere, due to loading/unloading of seawater (in response to glaciation), inducescompressive and extensional deviatoric stress in the crust. These crustal stresses can reachvalues that facilitate (extensional) or suppress (compressive) the initiation and propagationof dikes (Glazner and Schubert, 1985; Hooper et al., 2011; Huybers and Langmuir, 2009;Jellinek et al., 2004; Rawson et al., 2016; Sigmundsson et al., 2010; Watt et al., 2013). We153use Monte Carlo simulations to model the potential effects of changing stress fields fromglacier loading and unloading of the crust on the rates of volcanic eruptions in a continentalarc setting.7.2 Causal linkages between arc volcanoes and glaciersA number of studies have suggested causal linkages between deglaciation and increasederuption rates at volcanoes that do not involve changes to magma source-regions. Tho-rarinsson (1953) suggested that removal of a 100 m-deep water column in the Grı´msvo¨tncaldera in Iceland may have triggered an eruption. Mathews (1958) speculated that thelarge-volume dacite effusions surrounding Mt. Garibaldi in British Columbia may have oc-cured in response to isostatic uplift associated with melting of the last CIS. Grove (1974)also suggested, without detailed analysis, that deglaciation and isostasy would alter crustalstresses and may have caused eruptions in northern British Columbia. Gudmundsson (1986)proposed that increased magmatism immediately following disappearance of ice from theReykjanes Peninsula, Iceland may have been caused, in part, by glacial-isostatic inducedextensional faulting. Later workers applied these ideas to sea level changes around volca-noes and indicated that water draw-down during glaciations can produce crustal stressesthat may generate eruptions (e.g., McGuire, 1992; Nakada and Yokose, 1992; Paterne andGuichard, 1993; Wallmann et al., 1988). The study by Nakada and Yokose (1992), in par-ticular, estimated the degree of crustal bending associated with water loading/unloadingaround island arcs. These authors speculate as to the effect that these transient stressesmay have on magma storage and transport through the crust and suggest that water unload-ing (during glaciation) may induce compressive deviatoric stress in the upper crust up to±13 MPa, enough to suppress dike transport and eruptions, while loading of sea water mayfacilitate dikes and increase eruptions. This is the opposite effect that was speculated byMcGuire (1992), who suggest that sea water unloading should be associated with increasedincidences of eruptions.The first detailed study of linkages at arc volcanoes, specifically, was by Glazner et al.(1999), who found an anti-correlation between loading of ∼300 m-thick glaciers in theSierra Nevada mountain ranges and volcanism in eastern California. These authors showthat loads imposed by glaciers produce horizontally-oriented compressive crustal stresses(up to 1 MPa) which may have suppressed dike formation and propagation. Jellinek et al.(2004) re-examined the same data and identified a 40 ka-cyclicity for volcanism and acorrelation between eruptive flux and the rate of change of glacier loading/unloading. Usinga model for evolving critical crustal magma chamber overpressure (Jellinek and DePaolo,2003), these authors calculate lag times (i.e., lagging behind the onset of deglaciation)154of 3.2 ± 4.2 ka and 11.2 ± 2.3 ka, for felsic and basaltic magmas respectively. Theseresults provided the first quantitative evidence that modulated volcanism in arc regions isnot caused by changing the rate of melt supply to the crust (c.f., Jull and McKenzie, 1996;Maclennan et al., 2002; Slater et al., 1998).A lack of precise geochronology measurements of volcanic arc rocks has largely prohib-ited detailed analysis of the timing of arc eruptions with respect to glacial cycles (Smellieand Edwards, 2016). Watt et al. (2013), however, examined Pleistocene eruptions in theKamchatka and Cascade volcanic arcs and found little variation in eruption rates duringthe early post glacial period. In the Andean, Southern Volcanic Zone, however, Watt et al.(2013) found that the early post-glacial eruption rate was found to be ∼2 times greaterthan the eruption rate later in the post-glacial period. These results are corroborated bythe analysis of Rawson et al. (2016), who examined Late Pleistocene eruptions at Mocho-Choshuenco. These deposits indicate an eruption response lagging behind the onset ofdeglaciation by ∼5 ka. In addition, Pleistocene ash records from the Pacific region suggestthat global eruption rates intensified following the last deglaciation and peak output laggedbehind the highest rate of increasing eustatic sea level (i.e. indicating melting glaciers) by4.0 ± 3.6 ka (Kutterolf et al., 2013).Several studies at Icelandic volcanoes have suggested that glacio-isostasy may play acritical role in crustal magma storage and transport (Albino et al., 2010; Kelemen et al.,1997; Sigmundsson et al., 2012, 2010). Using numerical models, these studies show thatmelting ice will relax crustal stresses, and thereby favour dike generation and eruptions. Fi-nally, several workers have also suggested that changes in ice loads may affect the buoyancypotential of magmas (Edwards and Russell, 2002; Gee et al., 1998; Wilson et al., 2013). Ed-wards and Russell (2002), in particular, showed through modelling assimilation-fractionalcrystallization of crustal-stored melts, that ice loading may influence whether a system islikely to produce a subglacial or supraglacial eruption by changing the static buoyancy po-tential of the magmas.There is general consensus in the literature that: i) glacier loading and unloading atcontinental arcs will not affect melt generation in the same way that it does in Iceland, ii)loading of continental crust by glaciers may suppress transport and eruption of magmasby increasing compressive deviatoric crustal stress and prohibiting dikes from transportingmagma, and iii) unloading of glaciers (deglaciation) may facilitate magma mobilizationthrough dikes and encourage more eruptions.1557.3 The Garibaldi volcanic belt, British Columbia, CanadaFor most of the last glacial cycle, the CIS in SWBC was restricted to the major mountainranges (Booth et al., 2003; Clague, 2011; Ryder et al., 1991) (Figure 7.1A). Near the coast,the CIS nucleated in the high Coast Mountains and maintained a central ice divide (i.e.,where the ice was thickest) over this area for most of the last glacial cycle (Clague, 2011;Clague and James, 2002; Jackson and Clague, 1991; James et al., 2000; Seguinot et al.,2016). At its maximum extent, ice overlying the Cordillera was up to 2 km thick anddisplaced the crust downward by ∼400 m (John Clague, pers. comm. 2019). Paleo-marinelimits show that glacio-isostatic depression was <250 m along the mainland coast (Clague,1983) and declined toward the west and southwest to<50 m on the west coast of VancouverIsland near the ice sheet margin (Clague, 1981; Mathews et al., 1970; Ryder et al., 1991).Assuming the greatest degree of bending to be situated beneath the axis of the Cordillera,these displacement measurements suggest that the deformed ground surface had an arcprofile with a length (l) of ∼400 km (Figure 7.2; Turcotte and Schubert, 2002).Growth of the last CIS began at ∼30 ka (Clague, 2011; Ryder et al., 1991). The icesheet nucleated in the high mountains and expanded gradually until it reached a peak at∼17 ka (Blaise et al., 1990; Clague, 2011; James et al., 2000) (Figure 7.1A). Deglaciationwas rapid and the present glacial configuration was achieved by∼11 ka (Clague and James,2002). The low viscosity of the mantle beneath the Canadain Cordillera facilitated rapidcrustal rebound which approximately matched the rate of ice melting (James et al., 2000).Glacio-isostatic rebound was also essentially complete by ∼11 ka (Clague, 2011; Clagueand James, 2002; Ryder et al., 1991).The scarcity of high-precision geochronology, typical of many arcs, has prohibitedrecovery of long-term volcano-ice correlations in the GVB (Hildreth, 2007; Sherrod andSmith, 1990; Watt et al., 2013). However, there is strong field evidence that the lastdeglaciation in southwestern British Columbia was attended by an increase in eruptive flux.The Rubble Creek, Culliton Creek and Ring Creek dacite lavas were all emplaced between9 and 13 ka (Figure 7.1) (Brooks and Friele, 1992; Bruno, 2011; Mathews, 1958; Mooreand Mathews, 1978). The age of Rubble and Culliton Creek lavas (∼13 ka) is inferred fromfield evidence (Friele and Clague, 2009; Mathews, 1958). The age of the Ring Creek lava isbracketed at between 9.3 and 10.7 ka using radiocarbon geochronology (Brooks and Friele,1992; Bruno, 2011). Combined, these post-glacial-aged dacite lavas have a surface areaof 5.0 x 104 m2, which is ∼15 % of the surface area of all Quaternary volcanic rocks inthe southern GVB (see chapter 2). We estimate, based on an average thickness of ∼50 m(Bruno, 2011; Mathews, 1958), that the rocks have a total volume of ∼2.5 x 109 m3. Giventhe timeframe for their emplacement, these deposits represent a volumetric eruption rate of156 50.0° N   52.5° N   120.0° W 130.0° W 47.5° N   122.5° W 125.0° W 127.5° W 4.4 4.6 4.8 5 5.25.55.555.65.65AB30 ka25 ka17 kaInterior PlateauCanadaUSAGVBPacificOceanGlaciovolcanicNon-glaciovolcanicRing creekRubblecreek Cullitoncreek CoastMountains400 km-long model line<-50 m-250 m-400 mx easting (m x 10 5)y northing (m x 10 6)Figure 7.1: A) Location and physiographic setting of the Garibaldi volcanic belt(GVB) in southern British Columbia, Canada and the last Cordilleran Ice Sheet(CIS). Dark blue contour lines correspond to the approximate distribution of thelast CIS at 30, 25 and 17 ka (peak) (after., Booth et al., 2003; Clague and Ward,2011). A representative 400 km-long model lines is shown in dark grey, withloci of isostatic surface depression measurements (yellow stars). B) Map show-ing distribution of subglacial (orange) and subaerial (blue) volcanic rocks in theGaribaldi volcanic belt (GVB).∼6.3 x 108 m3 kyr−1, indicating, at minimum, a ∼5-fold increase in volcanic flux over theaverage Pleistocene rate for the Northern Cascade volcanic arc (e.g., 130 km3 Ma−1, or, 1.3x 108 m3 kyr−1; Sherrod and Smith, 1990).7.4 Monte Carlo simulations of glacial loading and unloadingas a forcing function for magma stall or riseGlacier loading and unloading of large ice sheets causes downward displacement and bend-ing of the crust on short time scales of 100’s to 1000’s of years (Booth et al., 2003; Clagueand James, 2002; James et al., 2000; McGinnis, 1968; Mey et al., 2016). This bending mayinduce compressive and extensional deviatoric stress within the crust at a magnitude thatcould influence failure in tension of magma chamber walls and thereby facilitate or prohibit157hhℓ ℓ Δℓ or εxx ϕy y w ABRLength along plate (km) 100 150 200 250 300 350-30-25-20-15-10-5050Plate depth (km)Cordilleran ice sheetCordillera surface profile (Fig. 7.1A)<400 m Figure 7.2: Diagram (not to scale) showing cylindrical thin-plate bending method andvariables (after., Turcotte and Schubert, 2002). A) Plate section of 400 km-length(real-world extent and cross-section shown in Figure 7.1A, prior to bending. B)Bent plate after loading of the Cordilleran Ice Sheet. The blue and red colorgradients correspond to zones of extension and compression, respectively. Theplate boundary is fixed at the edges at the central mid-plane (black dots). Math-ematical notation is given in text.dike initiation and propagation (Gudmundsson et al., 2012; Pinel and Jaupart, 2005; Sig-mundsson et al., 2010; Turcotte and Schubert, 2002). Stress from glacio-isostatic bendingoperates over short timescales when compared with crustal stresses induced by regionaltectonic activity, and thus, may overprint and act independently of the background (i.e.,tectonic) state of stress (Jellinek et al., 2004; Stewart et al., 2000). Transient stresses causedby glacial loading and unloading may therefore control magma storage and eruptions onglacial time-scales (Albino et al., 2010; Glazner and Schubert, 1985; Gudmundsson, 1986;Jellinek et al., 2004; Kutterolf et al., 2013; Nakada and Yokose, 1992; Sigmundsson et al.,2010; Turcotte and Schubert, 2002).Here, we test the effect of loading/unloading of the crust by ice sheets and its resultanttransient stress fields to the ascent of magma to the point of eruption using a Monte Carlosimulation. Our model simplifies the magma transport through the crust into the threeprimary components, magma buoyancy, crustal rock strength in tension and changes indeviatoric stress due to glacier loading and unloading. Our model is based on the conceptdeveloped by Nakada and Yokose (1992), who suggested that downward bending of the158crust due to loading may induce stresses that trap magma within the crust, while unloadingwill release stored magmas to erupt (see above).7.4.1 Isostatic lithospheric plate flexureThe magnitude of deviatoric stress developed during glacio-isostatic plate bending dependson: i) the geometry of plate bending (i.e., the dimensions of the bent plate and the conditionsof boundary fixture at the plate edges; Turcotte and Schubert, 2002; Wickert, 2016), ii) theelastic lithospheric plate strength and thickness and viscosity of the underlying mantle (e.g.,James et al., 2000), and, iii) the magnitude and size of the surface load causing downwardsurface displacement (e.g., Mey et al., 2016). For thin tectonic plates, where the platethickness is much less than the plate length, the geometry of bending may be estimatedin two-dimensions using a cylindrical plate bending approximation (i.e., Kirchhoff–Lovetheory; Turcotte and Schubert, 2002). The key assumption of this theory is that the lateralmargins of the plate remain fixed at a central mid-plane position and there is a neutral planein the center of the plate that experiences no net extension or shortening (e.g., Figure 7.2).This approximation is widely used in geologic applications as the method is isovolumetricthroughout all stages of deformation (e.g., Mey et al., 2016; Quinlan, 1984; Turcotte andSchubert, 2002; Wickert, 2016; Zhou and Lin, 2018).We assess cylindrical plate curvature (bending) by using quantitative measurements ofsurface depression from growth and decay of the CIS through time (see above). We assumethe effective elastic thickness of the lithosphere beneath southwestern British Columbia tobe 34 km-thick (James et al., 2000). At the glacial maxima, the deformed ground surfacehad an cylindrical arc profile with a length (l) of ∼400 km (Figure 7.2; Turcotte and Schu-bert, 2002). The maximum downward displacement in the center of the arc was ∼400 m(John Clague, pers. comm. 2019). We are interested in the relative displacement of thecrust due to glacio-isostatic bending. As such, we do not account for curvature of the earthand assume the pre-bent (i.e., un-deformed) plate surface to be horizontal and flat.We construct a 40 ka-long model input curve that tracks the maximum displacementof the bent plate surface over a single glacial cycle (i.e., from 0 to 400 m of downwarddisplacement; Figure 7.3A). The model contains 10 ka at the beginning with no glacialloading to establish a modelled background rate of eruptions. We divide the model into 1ka-long time intervals and, calculate deviatoric stress developed within the bent plate usingthe method outlined by Turcotte and Schubert (2002).Relative strain throughout the plate from bending (εxx) is calculated as:εxx =yR, (7.1)1590510152025303540Time (ka)-400-300-200-1000Crustal depression (m)xx051015202530Depth (km)-12 -10 -8 -6 -4 -2 10 120 2 4 6 8Background(pre-glacial)Loading UnloadingLGM (17 ka)Background(post-glacial)No crustal bendingMaximum crustal bendingMid-plane of crustAB+ y- yFigure 7.3: Model loading curve of maximum crustal displacement and calculated de-viatoric stresses developed through the bent plate. A) 40 ka model loading curveshowing background (pre-glacial), loading (glaciation), unloading (deglaciation)and post-glacial periods. B) Maximum change in deviatoric stress (∆σxx) devel-oped during plate bending.where, y is the distance from the neutral center plane of the bent plate, and R is the radiusof the bent arc (Figure 7.2). The inverse of R is equal to:1R=φl, (7.2)where, φ is the small angle of the arc of the bent plate, and l is the length along the plate(Figure 7.2). φ is calculated as:φ = 2(−w0.5l), (7.3)where, w is the amount of downward deflection of the bent plate (Figure 7.2 and Figure 7.3).Deviatoric stress is therefore:160σxx =E(1− v2)εxx, (7.4)where, E, and v are Young’s modulus (equal to 70 GPa) and Poisson’s ratio (equal to∼0.28),respectively (Glazner and Schubert, 1985; Turcotte and Schubert, 2002). Positive values ofσxx are compressive, and negative values are extensional (Figure 7.3B).Loading causes downward bending and compressive deviatoric stress in the upper halfof the plate and tensile stress in the lower half (Figure 7.3B). The maximum deviatoric stressimposed by glacio-isostatic bending is ±12 MPa (Figure 7.3B). The relative magnitudes ofthese stress changes are strongly controlled by the dimensions of the plate and the degreeof bending. For example, increasing the length of the plate to 800 km reduces the overallbending and reduces maximum σxx to±2.9 MPa. Increasing the thickness of the plate to 50km increases the maximum σxx to ±16.8 MPa, while increasing the degree of bending to600 m increases the maximum σxx to ±17.3 MPa. In reality, the accumulated stress cannotexceed the tensile strength of rocks (i.e. ∼10 MPa, see below) as the rocks would fail intension and therefore σxx would be limited by this strength.7.4.2 Crustal dike migrationDikes are highly sensitive to the state of stress in the crust (Barnett and Gudmundsson, 2014;Gonnermann and Taisne, 2015; Gudmundsson, 2006; Gudmundsson et al., 2012; Kavanaghand Sparks, 2011; Kavanagh, 2018; Kavanagh et al., 2018; Kusumoto et al., 2013; Mulleret al., 2001). Upward dike migration is driven by buoyancy contrasts between the magmain the dike and the host rocks. In the simplest case, the threshold for dike initiation ismet when the driving pressure (i.e., the combined buoyancy potential and magma chamberoverpressure; Pd) reaches or exceeds the in situ tensile strength of the host rocks (τ), plus theminimum principal stress (i.e., the deviatoric stress acting perpendicular to the dike walls;Gonnermann and Taisne, 2015; Gudmundsson et al., 2012; Sigmundsson et al., 2010):Pd ≥ σxx+ τ. (7.5)Typical driving pressures for dikes range from a few MPa to a few tens of MPa (Geshiet al., 2010; Gonnermann and Taisne, 2015; Gudmundsson et al., 2012; Kusumoto et al.,2013; Pollard and Muller, 2008). A typical crustal rock strength (τ) is ∼10 MPa (Edwardsand Russell, 2002; Glazner and Schubert, 1985; Gudmundsson et al., 2012; Rawson et al.,2016). Tensile failure of the rocks and dike initiation may therefore be inhibited (i.e., thedikes will be stalled) under shortening (compressive) crustal conditions and encouraged(i.e., the dikes will propagate) under extension.161Migration of dikes through the crust is tested using a Monte Carlo simulation (Fig-ure 7.4). The dikes are treated as packets of magma with no practical length or height. Therelative positions of the dikes are tracked by comparing a randomly-assigned dike drivingpressure (Pd) (see below) to the tensile strength of the rocks (10 MPa), plus (or minus) thedeviatoric stress imposed during bending (σxx). If the conditions for dike propagation aremet, the dike is allowed to migrate upwards and eventually, erupt. All dikes are assumed tomove instantaneously over the 1 ka time intervals. Dike movement over a 1 ka is realistic asdikes of most volcanic-arc compositions show rise velocities that are sufficient to transit thecrust in well under 1 ka (Ho¨skuldsson and Sparks, 1997; Pinel et al., 2017; Rubin, 1995).The dikes begin at the base of the crust and are kept in the model for the remainder of the 40ka-simulation, unless they erupt (i.e., dikes are not able to freeze and crystallize within thecrust; see below for a discussion of this assumption). Our question addresses how loadingand unloading of glaciers may modulate storage and transport of dikes through the crust.As such, we assume that the flux of magma delivered to the base of the crust occurs at asteady rate over the timescale considered (e.g., Cruden and Weinberg, 2018; Grove et al.,2012; Peacock et al., 1994; Winter, 2013). We introduce d = 100 dikes to the base of thecrust for each time interval. In practice, this number is arbitrary. We tested simulationsadding 10, 50 and 500 dikes per time increment and found that it only affects the absolutenumber of eruptions, not the relative eruption rate, which is what we are interested in.The driving pressure (Pd) of each dike is treated as a random variable (Figure 7.5A).We construct a uniform distribution of Pd between 3 and 30 MPa, based on typical, ob-served driving pressures of dikes (Geshi et al., 2010; Kusumoto et al., 2013; Pollard andMuller, 2008). This range has the effect of simulating the full range of magma composi-tions (with different densities and viscosities) and reservoir sizes (affecting buoyancy po-tential) expected in arc environments without overly complicating the model (Geshi et al.,2010; Gonnermann and Taisne, 2015; Gudmundsson et al., 2012). The uniform Pd distri-bution was sampled in total d = 4× 106 times (i.e., for every trial (n = 1000), there were40 intervals that each added 100 dikes). The simulations (n) were repeated a sufficientnumber of times (n = 1000) so that the mean number of eruptions (x¯) per time incrementconverged and additional trials did not alter the mean (x¯) by more than 1 % (e.g., Byrne,2013) (Figure 7.5B).7.5 Model resultsPrior to bending (i.e., σxx = 0 throughout the plate), the modelled average backgrounderuption rate is ∼74 eruptions per 1 ka. (Figure 7.6A). All eruption rates are normalizedto this value (Figure 7.6B). Dikes that reach the surface during the pre-glacial period have16205101520253035Depth (km)Mantle LithosphereConstant supply of magma to the crust.Crustal LithosphereDikes with Pd < τ continue to accumulate in lower half of the bent plate.Dikes with Pd > τ are evacuated from upper half of the bent plate.05101520253035Depth (km)+Δσxx-ΔσxxEruptionsNeutral plane Neutral planeDeglaciationGlaciationIce sheet Melting ice+Δσxx Compression-Δσxx Extensionre-compressionde-compressionA BMantle LithosphereConstant supply of magma to the crust.Crustal LithosphereDikes with Pd < τ migrate upwards with progressive loading.Dikes with Pd > τ captured and accumulate in the upper half of bent plate. Figure 7.4: Cartoon illustrating key parameters controlling the Monte Carlo simula-tion. Dikes are represented as red ellipsoids. Dashed red arrows indicate direc-tion of dike movement. Black arrows indicate the direction of crustal movement.A) Glaciation induces downward crustal bending and a positive ∆σxx (compres-sion) in the upper half of the bent plate and a negative ∆σxx (extension) in thelower half of the bent plate. Dikes with Pd > τ ±σxx (Equation 7.5) are ablecross the neutral plane in the center of the plane and accumulate in the upperpart of the plate. B) Deglaciation induces de-compression in the upper half ofthe bent plate and re-compression in the lower half of the plate, evacuating thedikes stored in the upper part of the plate.Pd ≥ 10 MPa (i.e. equal to or exceeding the tensile strength (τ) of the host rocks). Thedikes with Pd < 10 MPa accumulate at base of the crust (Figure 7.4A). Bending-inducedstresses (σxx) caused by a superimposed load (Figure 7.2B) have the effect of encouragingdike propagation in the lower crust and discouraging dikes in the upper crust (Glazner andSchubert, 1985; Nakada and Yokose, 1992).Loading supports a suppression in eruptions by a factor of 0.4 ± 0.06. Specifically,during maximum loading the dikes that erupted have a Pd ≥ 22 MPa. Non-erupted dikesaccumulate throughout the bent plate. Dikes with Pd < 10 MPa accumulate in the lower163AB5 10 15 20 25 30024681012Frequency10410010110210305101520Pd (MPa)Trials (n)d = 4 x 106n = 1000Figure 7.5: Randomly-sampled uniform distribution of input dikes and change inmean number of eruptions. A) Histogram showing d = 4× 106 randomly sam-pled dike driving pressures (Pd) with a uniform distribution. B) Change in themean number of eruptions (∆x¯) at each time interval indicating statistical con-vergence of the mean after n= 1000 model trials. The coloured lines correspondto each of the 1 ka model time increments.crust, while dikes with Pd ≥ 10 MPa pass over the mid-plane and accumulate in the uppercrust (Figure 7.4B).Deglaciation is attended by a relaxation in compressive stress in the upper part of thecrust and a corresponding increase in eruptions (Figure 7.6B). This is caused by dikes thatwere accumulated in the upper half of the crust being released to erupt under the new stressfield (Figure 7.4C). The volcanic output rate curve is asymmetric and broadly the inverse ofthe input glacial loading curve. Peak volcanic output occurs at∼11 ka and is represented asan increase in eruptions by a factor of 2.7 ± 0.15. The peak output is synchronous with theend of crustal rebound and lags behind the LGM by ∼6 ka (Figure 7.6). The post-glacialeruption rate returns from the peak output rate to the background rate rapidly (i.e., within asingle 1 ka time increment).164050100150200250Number of eruptions0510152025303540Time (ka)00.511.522.53Eruptions (normalized to background)60 70 80Eruptions at 35 ka020406080100Frequency180 200 220 240Eruptions at 11 ka0204060801001201401  Standard Deviation    EruptionsBackgroundACBBackground(pre-glacial)Loading UnloadingLGM (17 ka)Background(post-glacial)DEruptions at 35 kaEruptions at 11 kaFigure 7.6: Model results over 40 ka loading/unloading cycle. A) n = 1000 modeltrials (coloured lines) showing the number of eruptions every 1 ka. The meannumber of eruptions (x¯) per 1 ka is shown with a thick black line. B) Eruptionsnormalized to the calculated background rate (red line and dots; established from40 to 30 ka). The yellow shaded region indicates 1σ standard deviation eitherside of the mean. The vertical blue dashed lines indicate eruption histogramdistributions at 35 (panel C) and 11 ka (panel D).1657.6 Discussion: Glacial pumping of a magma-chargedlithosphereOur Monte Carlo simulations suggest that eruptions at volcanic arcs may be strongly con-trolled by the effect of waxing and waning continental ice sheets (Figure 7.6). The linkageis formed through transient deviatoric stresses, developed during glacier-induced crustalbending and the effect of these stresses on the transport and storage of magma throughthe crust (Glazner et al., 1999; Glazner and Schubert, 1985; Jellinek et al., 2004; Nakadaand Yokose, 1992). This mechanism implies that loading and unloading of glaciers oncontinental arc magmatic systems may act as a “glacial pump” that modulates the flow ofmagma from the crust to the surface over glacial timescales (Figure 7.7). There are two keycomponents to this model: Firstly, glacial loading and downward crustal bending suppresseruptions (Figure 7.7). This is caused by compressive deviatoric stress developed in the up-per part of the plate, suppressing dike initiation and the transport of magma to the surface(Gonnermann and Taisne, 2015; Hooper et al., 2011). This modelled suppression is difficultto test against the geologic record, because, for the GVB, the number and precision of agedeterminations is inadequate (see chapter 2). However, the results are in good agreementwith sparse observations from other volcanic arcs and global eruption rates as a whole overthe last glacial period (e.g., Glazner et al., 1999; Jellinek et al., 2004; Jull and McKenzie,1996; Kutterolf et al., 2013, 2019; Rawson et al., 2016; Sigmundsson et al., 2010; Wattet al., 2013).Secondly, the model predicts an increase (by a factor of ∼3) in eruptions attendingdeglaciation (Figure 7.7). The peak output (at∼11 ka) lags behind the onset of deglaciationby ∼6 ka and is synchronous with the final stages of crustal rebound. This is consistentwith the pattern observed for volcanoes in the Southern Volcanic Zone in South America(e.g., Rawson et al., 2016; Watt et al., 2013), with observations of global volcanic activity inthe Pacific and California (e.g., Jellinek et al., 2004; Kutterolf et al., 2013), and in Iceland(Hardarson and Fitton, 1991; Sigvaldason et al., 1992). The modelled peak output lag times(∼6 ka) are in line with those suggested for both basaltic and silicic magmas in easternCalifornia (Jellinek and DePaolo, 2003; Jellinek et al., 2004) and align with those observedby Kutterolf et al. (2013) for the age of peak explosive eruptions in the surrounding thePacific. Moreover, our modelled peak output timings agree with the ages of the Rubble,Culliton and Ring Creek lavas, supporting the notion that the timing of these eruptions wasmodulated by glacial loading/unloading over the last glacial cycle.166ABCBefore GlaciationPeak GlaciationPost GlaciationCrustalLithosphereMantle Lithosphere~35 kmBackground eruptionsBackground flux of magma to the base of the crustCordilleran Ice SheetMagma accumulationCompressionExtensionStalled dikesWaning ice sheetPost-glacial reboundCrustalLithosphereMantle LithosphereCrustalLithosphereMantle LithosphereShallow stored magmas are released and eruptEruptionsMid-crust planeInitial stateGaribaldi volvcanic beltMOHOMOHOMOHOFigure 7.7: Conceptual model of glacial pumping of a magma-charged crustal litho-sphere. A) Prior to glaciation with the crust in a relaxed, initial state. A back-ground number of eruptions occur, reflecting the long term supply of magma tothe crust. B) Glacial build-up invokes compression in the upper crust and ex-tension in the lower crust. This encourages ponding of magma in the crust dueto arrested dike ascent and an overall reduction in eruptions. C) Immediatelyfollowing deglaciation, relaxation of accumulated stress in the crust encouragesevacuation of magmas stored in the upper crust and an overall increase in erup-tions.1677.6.1 Eruption implications of glacial pumpingThe primary glacial pumping mechanism affects volcanic eruptions by the capture and re-lease of dikes in the upper part of the crust during glacial loading/unloading (Figure 7.4and Figure 7.7). This mechanism requires some justification as magmas are unlikely to bestored as dikes for any significant length of time (i.e., the dikes would thermally freeze; Ru-bin, 1995). Our model assumes that all dikes that reach the upper part of the crust (i.e., thosewith Pd > τ) will eventually erupt over a full loading/unloading cycle. In practice, however,many of these dikes would never make it to the surface. We suggest that the dikes probablyfeed long-lived crustal magma-mush reservoirs (magma chambers) situated throughout thecrust (e.g., Jackson et al., 2018). Accumulation within these reservoirs would allow magmato be stored with minimal thermal degradation for lengths of time greater than the glacialcycles considered. Additional compressional stress, however, could also assist in exceedingthe critical threshold pressure for magma chamber failure and dike formation (e.g., Gonner-mann and Taisne, 2015; Gudmundsson et al., 2012). Although we have not considered thisin our model, this mechanism may assist in moving magma from the lower crust to theupper crust between loading and unloading cycles or promote emptying of shallow crustalreservoirs during the initial stages of glacial loading.Stored magmas, however, may undergo assimilation and fractional crystallisation pro-cesses, which increase buoyancy and thereby increase dike driving pressures (Edwards andRussell, 2002). These buoyancy reductions may encourage additional magma migrationwithin the crust, even under increased glacial load. Assimilation and fractional crystallisa-tion may contribute towards reduced magma residence times and promote eruptions closerto the onset of deglaciation (i.e., reducing the lag time between deglaciation onset and peakvolcanic output). As we do not consider evolving magma driving pressure (Pd) in our model,our simulations must be considered maximum scenarios.A key implication implicit in our model is that glacially-modulated magmas should re-side in the crust for longer periods of time than non-modulated equivalents. Accordingly,these magmas should display a geochemical/petrographic signature of extended crustal stor-age. In the GVB, this notion is supported by the evolved nature of the erupted Rubble, Cul-liton and Ring Creek lavas (i.e., they are dacitic and not basaltic) and that Green (1981b)calculated equilibration pressures for the magmas that indicate storage at 3.5 to 4 kbar(at ∼13 to 15 km-depth). As glacial pumping affects crustal magma residence time andmagma evolution, another consequence is that the eruptions following deglaciation may bemore violent and explosive, due to a tendency towards evolved, siliceous and hydrous meltcompositions (Edwards and Russell, 2002; Rawson et al., 2016; Watt et al., 2013).168Glacial pumping will affect magmas with different physical properties, differently. As-suming the driving pressures of basaltic dikes to be lower than rhyolitic ones (e.g., Gonner-mann and Taisne, 2015; Gudmundsson et al., 2012; Jellinek and DePaolo, 2003; Jellineket al., 2004; Kusumoto et al., 2013; Muller et al., 2001), basaltic dikes should be moresusceptible to capture within the crust during loading, and released more slowly during re-bound (i.e., small deviatoric stresses are developed quickly during loading, and the crustmust return to a near-undeformed state to evacuate magmas with low Pd). In accordancewith this: i) basaltic magmas will be more affected by glacial modulation than rhyolites,ii) most eruptions following deglaciation will tend towards basaltic compositions and, iii)basaltic eruptions should occur near to the completion of crustal rebound while rhyoliticeruptions will occur quickly during glacial retreat. This is in agreement with observationsof Californian volcanoes by Jellinek et al. (2004), who calculated lag times of 11.2 ± 2.3ka and 3.2 ± 4.2 ka for basaltic and rhyolitic magmas, respectively.7.6.2 Glacial asymmetry and the timing and magnitude of volcanic responseGlacial pumping connects the length, shape and intensity of glaciations with the magnitudeand timing of volcanic response. As such, the response from the different orbitally-forcedclimatic events (i.e., the 21 ka, 40 ka and 100 ka-long cycles; Lisiecki and Raymo, 2005;Shackleton, 1987), will vary. Firstly, peak volcanic output is always concordant with theend of crustal rebound (in our case, at 11 ka) and lags behind the onset of deglaciation bythe duration of the deglaciation event (in our case, ∼ 6 ka). This relationship is explainedby the interaction between the chosen minimum input dike driving pressure (3 MPa), thestrength of the crustal rocks in tension (10 MPa) and the maximum deviatoric stress inducedduring bending (up to±12 MPa). For example , If the minimum driving pressure were> 10MPa (i.e., > τ), the overall lag time between the onset of deglaciation and peak eruptionswould be increased (i.e,. more bending is required to modulate the dikes). Secondly, theintensity of volcanic response is directly influenced by the magnitude of glaciation and thelength of glacial events. More intense glacial periods (i.e., with more ice and greater crustaldeformation) will affect a larger portion of the dikes emplaced into the crust, and longer pe-riods of glacial build ups will facilitate correspondingly enhanced periods of crustal magmaaccumulation (charging). In this manner, we expect that the longer and more intense 100 kaorbital eccentricity-driven climate cycles will induce larger volcanic responses than shorterand less intense 41 ka orbital obliquity-driven events.Ultimately, it is the asymmetry and characteristic “sawtooth” pattern of glaciation thatshapes the magnitude and timing of the causal linkages between volcanism and ice. Longperiods of glacial build ups facilitate long periods of crustal magma-chamber charging,169while rapid deglaciations efficiently evacuate the crust of stored magma to produce a short,intense spike in volcanic activity. In this manner, volcanic output (i.e., both the numberof eruptions and output volume) is strongly correlated with the rate of change in ice loadrather than the absolute load exerted on the crust at any time (e.g., Jellinek and DePaolo,2003; Jellinek et al., 2004).Finally, in the simplified asymmetric loading case that we have modelled, the glacialbuild up and retreat is linear. In reality ice sheet growth and decay typically involves nu-merous, convoluted stages of advance and retreat (e.g., Clague and James, 2002; Clagueand Ward, 2011; Friele and Clague, 2002). If these interruptions are sufficient to affect thestate of stress in the crust, they will also influence the timing and magnitude of the volcanicresponse. As such, successive ice build-up/decay events that occur rapidly (i.e., faster thanthe crust can isostatically reset), may have compounding effects on the volcanic response,and feed longer-term, complex linkages operating over a wide range of geologic timescales.7.7 ConclusionWe have used stochastic simulations to track the migration of magmas through a thickcrustal lithosphere (i.e., a continental arc) undergoing bending in response to loading/un-loading of continental-scale ice sheets. Our results indicate a transient response from arcvolcanoes to deglaciation over glacial timescales: glacial loading will suppress volcanicactivity by inhibiting the propagation of dikes in the shallow crust while deglaciation willpromote dikes, and eruptions, causing an overall increase in eruption rate. We apply ourmodel to Quaternary volcanoes in the GVB of southwestern British Columbia, Canada andfind that loads induced by the last CIS, caused an overall reduction in eruptions during peakglaciation (by a factor of 0.4 ± 0.06), and an increase following deglaciation and crustalrebound (by a factor of 2.7 ± 0.15). These model results (i.e., a ∼3-fold increase in thenumber of eruptions) supports the field evidence from the GVB, which indicates that theeruptive flux was ∼5-times higher immediately following the last deglaciation (i.e., fromaround 13 to 9 ka). Although our model provides a simplistic view of the continental arcmagmatic system, the results speak to the existence of a significant causal linkage betweenerupting arc magmas and waxing and waning of glaciers: loading/unloading acts as a glacialpump, charging and evacuating the crust with magma by modulating transport and storageof dikes. Our results indicate: i) the asymmetry of climate cycling events affects the mag-nitude and timing of volcanic response; greater volcanic output should be expected fromlonger and more intense glacial periods that are followed by rapid deglaciations, ii) theresponse (timing and output) from magmas with different physical properties (i.e., compo-sition, viscosity and reservoir sizes) will be different, and, iii) glacial pumping will affect170crustal magma storage and magma evolution (e.g., assimilation and fractional crystalli-sation). These processes will influence the overall style of glacially-modulated volcaniceruptions (e.g., they should be more explosive and felsic), and may imprint a geochem-ical/petrographic signature of modulation as stored magmas tend towards more evolvedcompositions.171Chapter 8ConclusionThis thesis explores glaciovolcanism in the Garibaldi volcanic belt (GVB) of southwesternBritish Columbia (SWBC), Canada. I use field observations and modelling to: i) investigatethe paleoenvironmental implications of glaciovolcanism in SWBC and develop glaciovol-canoes as a powerful proxy for ancient and global paleoclimates, ii) analyse the physicalprocesses governing the emplacement and eruption dynamics of two glaciovolcanoes inSWBC, and iii) examine the causal linkages between volcanic eruptions and the growthand decay of terrestrial ice masses.8.1 Paleoenvironmental implications of glaciovolcanism in theGaribaldi volcanic beltGlaciovolcanoes provide a holistic, yet underutilized proxy for Earth’s ancient paleoenvi-ronment (Smellie, 2018). The deposits can preserve highly-specialized eruption environ-ment information, including: paleo-glacier surface elevation, paleo-glacier thickness andthe basal thermal regime of paleo-glaciers. Glaciovolcanoes are also robust (i.e., they sur-vive glacial and other erosion processes) and are readily susceptible to direct age determi-nations (e.g., using 40Ar/39Ar geochronology). Despite these advantages, glaciovolcanismas a proxy, is undeveloped and underutilized across the globe (Russell et al., 2014; Smellie,2013, 2018; Smellie and Edwards, 2016; Smellie et al., 2008).Chapter 2 investigates the glaciovolcanic deposits of SWBC in terms of their paleoen-vironmental importance. Using field observations, I provide a comprehensive summaryof all known occurrences of glaciovolcanism in the GVB. I discuss the diversity in thevolcanic deposits and attribute this diversity to the wide range of magma compositions as-sociated with arc volcanism, the extreme topography of SWBC and to the varied styles ofglaciers that the volcanoes have interacted with (i.e., ice sheets, local alpine glaciers etc.).172I develop new methods for extracting paleoenvironmental information from the deposits,some of which are unique to the GVB (e.g., lava-dominated tuyas). I also compile new andexisting geochronology from the GVB and discuss the importance of these rocks in the con-text of reconstructing the paleo-presence and dynamics of the Pleistocene Cordilleran IceSheet (CIS). The methods developed in this thesis (e.g., determining paleo-glacier surfaceheight from lava-dominated tuyas) will contribute towards recognizing and fully exploitingsimilar deposits elsewhere in SWBC (e.g., the Black Tusk) and globally (e.g., intermediate-composition glaciovolcanoes in the southern Cascade volcanic arc and the South AmericanAndes).In chapter 3, I further establish glaciovolcanism as a useable proxy for ancient andglobal paleoenvironments by developing a geometric model that utilises glaciovolcanoesto reconstruct the paleo-extent and thickness of ancient mountain glaciers and ice sheets.This model is used in conjunction with three well-constrained examples of glaciovolcan-ism to determine the extent and thickness of the CIS at three geologically important timesthroughout the Pleistocene. The modelling establishes the existence of a well-developedCIS in SWBC during late Marine Isotope Stage (MIS) 4 (at 49 ka), the absence of theCIS during the MIS 6–5e interglacial transition (at 141 ka), and the first physical recordof a well-developed CIS in SWBC during MIS 15 (at 598 ka). These reconstructions arethe first depictions of pre-Fraser (i.e., before the last glaciation) ice sheets in SWBC to beproposed.My geometric model addresses one of the fundamental limitations in using glaciovol-canoes for paleo-glacier reconstructions: glaciovolcanoes are isolated points in space andtime and are therefore not well-suited for reconstructing the absolute distribution of ancientterrestrial ice masses. My model, however, exploits the geometric relationship betweenmountain glacier surface slopes and empirical constraints on paleo-glacier surfaces derivedfrom glaciovolcanoes. There is potential for this model to be applied to some of the other,well dated glaciovolcanic (and non-glaciovolcanic) deposits in the GVB (e.g., The Table;see chapter 2), and also for reconstructing local and regional-scale glaciers and ice sheets inother areas where topography and volcanism coincide (e.g., the Transantarctic mountainsin Antarctica and the South American Andes). Because the model exploits the geometricrelationship between glacier surface slope and topography, it could also be constrained bynon-volcanic paleoenvironmental data (e.g., paleo-glacier presence derived from ancientglacial trim lines or ancient glacial moraine deposits). Together, these data could be used torefine and improve the spatial-temporal resolution of multi-cycle ice sheet reconstructionsin SWBC, and globally.However, there are still some limitations for using glaciovolcanoes as recorders ofEarth’s ancient climate. Among these, the most significant is the difficulty in obtaining173accurate and precise ages on the rocks. Currently, even the best 40Ar/39Ar geochronologydeterminations for volcanic rocks have uncertainties of at least several thousand to severaltens of thousands of years (e.g., Conway et al., 2016; Guillou et al., 2010; Smellie, 2018;Smellie and Edwards, 2016) and there are sometimes discordances between samples datedusing the older K-Ar method and the newer 40Ar/39Ar approach (e.g., Hart and Villeneuve,2011). This makes meaningful comparisons between glaciovolcanoes (i.e., a record ofpaleo-glaciers) and Milankovitch-length glacial cycles (i.e., typically ∼41 ka–duration) anongoing challenge. Typically, these errors are exacerbated when analyzing low-K sam-ples and extremely young (i.e., <50 ka) volcanic rocks that have limited radiogenic 40Arand a high atmospheric 36Ar component (Kelley, 2002). Glaciovolcanic rocks may alsobe more susceptible to radiogenic 40Ar loss than subaerial volcanic rocks due to their in-herent glassiness (caused by rapid quenching) and alteration (caused by interaction withwater) (Brian Jicha, pers. comm, 2019). Currently, there is a severe lack of geochronolog-ical information for GVB volcanoes (see chapter 2). Attempts to mitigate this lack of datain the GVB are in progress, however, the young age and composition of the GVB rockspresents an ongoing challenge. The GVB geochronology database will be enhanced by:i) continued detailed field mapping and sampling of the deposits, ii) age determinationsusing a wide variety of geochronometric techniques (e.g., paleomagnetism dating, surfaceexposure dating etc.), iii) repeated analysis of critical samples, and iv) future technologi-cal advancements in the precision and accuracy of the geochronological dating methods.Ultimately, studies of glaciovolcanism in SWBC will produce a high-resolution map (i.e.,perhaps comprising >100 volcanoes) of the Pleistocene CIS in space and time. This mapwill serve as an independent, terrestrial-based proxy that could be used to test and refine thecurrently-established global paleoclimate framework (e.g., the MIS record; Lisiecki andRaymo, 2005).8.2 Physical controls on eruption dynamics of glaciovolcanoesin the Garibaldi volcanic beltThe second part of my thesis (chapters 4 and 5) explores the eruption dynamics of twoeffusive glaciovolcanoes in the GVB. At the Lillooet Glacier basalts, I use field mapping todelineate the deposit lithofacies and elucidate the presence of an ephemeral englacial lakethat was >150 m deep and was sustained throughout the entire eruption (i.e., there are nodiscernible breaks or passage zones in the deposit sequence). The Lillooet Glacier basaltswere probably erupted at the peak of, or during the waning stages of, Fraser glaciation(17–13 ka). They were erupted beneath a >654 m-thick paleo-glacier that had a surfaceelevation of >1895 m above sea level. I use petrologic and thermodynamic modelling to174establish co-magmatism within the sample suite and show that all chemical variation isconsistent with sorting of the observed phenocryst assemblage: olivine + plagioclase. Themagma was stored in a shallow, crustal reservoir at <2 kbar and had a pre-eruptive H2Ocontent of 0.5–1 wt.%.The Table is an iconic, flat-topped volcano that is recognised globally as the type-example of a glaciovolcanic lava-dominated tuya. However, previous hypotheses concern-ing the emplacement of this edifice were vague and untested (Green, 1977; Green et al.,1988; Kelman et al., 2002a; Lescinsky and Fink, 2000; Mathews, 1951, 1958; Russell et al.,2014). I use field mapping, 3-D photogrammetry, lithofacies identification and 40Ar/39Argeochronology to show the Table to be a steep-sided, flat-topped mass of dense, porphyriticandesite preserving a near-original outer surface comprising hackly fracture sets. The Tablewas erupted early in the last glacial cycle at 100 ± 12 ka. I provide a new field-constrainedmodel for the emplacement of the edifice involving dike injection and endogenous inflationwithin an overlying ice mass that was ∼250–350 m thick. The physical eruption model as-sumes that the edifice expanded outwardly as an elliptical cylinder, endogenously inflatingand melting an equal volume of ice. The thermal exchange between the expanding mas-sif and the melting ice operates across a subvertical, well-drained, enveloping carapace ofquench breccia. For effusion rates of 0.5–2 m3 s−1, marginal heat fluxes are 43–186 Wm−2, implying low overall heat transfer coefficients (∼0.04 to ∼0.18 W m−2 K−1). Thetotal volume of melted ice serves as a calorimeter and constrains the average thickness(∼3 m) and properties (thermal conductivity; 0.1–0.4 W m−2 K−1) of the breccia carapace.These results suggest that the heat transfer efficiency of intermediate lava-dominated tuyasis significantly lower (∼2 %) than reported for other glaciovolcanic systems (commonly∼45–77%).The study of the Table makes several key contributions towards understanding intermediate-composition glaciovolcanoes. The Table is now the first recognized example of an andesiticvolcano that was injected directly into, and inflated within, a glacier (c.f., Mathews, 1951).This new model for emplacement is derived entirely on field analysis of the deposit litho-facies and the surface morphologies. The model may be applied directly to other lava-dominated tuyas that exist locally (e.g., the Black Tusk) and will assist recognizing theserare edifices around the world. Furthermore, my eruption model and heat-exchange calcula-tions help to understand ice melting rates and the ‘space problem’ attending glaciovolcaniceruptions (i.e., howmagma creates space for itself when intruding ice; Gudmundsson, 2003;Gudmundsson et al., 2004; Smellie et al., 2008; Tuffen et al., 2002). The extremely low effi-ciency of heat exchange (and low rate of ice melting) exhibited by the Table also elucidateshow glaciers may impound or redirect the flow of lava (e.g., Mathews, 1952a) and assistsin forecasting the volume of melt water (and lahar hazard) produced during eruptions em-175anating from high-elevation, ice-clad vents (e.g., Conway et al., 2015; Lescinsky and Fink,2000; Major and Newhall, 1989).My studies of the Table and the Lillooet Glacier basalts also highlight the critical impor-tance of an englacial lake in controlling the morphology and deposit lithofacies at effusiveglaciovolcanoes. If an englacial body of water is developed and maintained throughout theeruption, the accumulated meltwater may facilitate sustained heat-transfer from the lava tothe ice via convection versus conduction. Convective heat transfer is significantly more ef-ficient than conduction, and thus, more ice will be melted (and more meltwater produced).Both the Lillooet Glacier basalts and the Table were erupted effusively beneath broadlysimilar thicknesses of ice (∼645 versus ∼350 m-thick, respectively). At Lillooet Glacier,effusive, exogenous growth of the pillow pile probably involved a heat-exchange efficiencyof ∼45% (Gudmundsson, 2003). Here, this was high enough to sustain an englacial bodyof meltwater (and likely some convective heat transfer), even though the lake was ac-tively draining throughout the eruption (i.e., there are syn-eruptive, bedded hyaloclastitedeposits). The Lillooet Glacier deposit is an unconstrained, elongate pillow pile with pe-ripheral lava tubes and flows. At the Table, within-ice dike injection (involving andesiteversus basalt) and endogenous growth of the massif produced an extremely low (conduc-tive) heat-exchange between the lava and the ice. This exchange was not sufficient to meltthe ice at a rate that exceeded the rate of meltwater drainage, and, as a result, the Tablewas strongly controlled (i.e., physically confined) by the surrounding ice. The fundamentaldifferences between these deposits directly reflect the properties of the erupted lava (i.e.,low-viscosity, high-temperature basalt versus higher-viscosity, lower temperature andesite)and the initial eruption conditions (e.g., extrusive basalt versus intrusive andesite).8.3 Climate triggers for volcanism: Causal linkages betweenvolcanoes and glaciersIn the final two chapters of this thesis (chapters 6 and 7), I explore the causal linkagesbetween volcanic eruptions in arc environments and the growth and decay of terrestrialice masses. Although ideas concerning these linkages are embedded in the literature (e.g.,Grove, 1974; Hooper et al., 2011; Jull and McKenzie, 1996; Mathews, 1958; Rampinoet al., 1979) they remain to be tested over a wide range of tectonic environments and scales.Within the literature there is general agreement that if causal linkages between volcanoesand ice exist for thick-crusted volcanic arcs, they must relate to processes that modulatetransport and storage of the magmas through the crust (e.g., Glazner et al., 1999; Jellineket al., 2004; Nakada and Yokose, 1992; Rawson et al., 2016; Watt et al., 2013). In chapters6 and 7, I build on these ideas, and investigate how loading and unloading of ice masses176ranging in size from local valley glaciers to continental ice sheets, may impart transientdeviatoric stresses within the crust and thereby influence the transport and flow of magmato the surface. In this way, glacial loading/unloading may control the size, distribution,style and frequency of eruption events. Due to extreme bedrock topography, a wide range inmagma compositions and a proximity and temporal correlation of the GVB to the southernlimit of the waxing and waning Pleistocene CIS, the GVB is an excellent volcanic arc inwhich to investigate and test these linkages.In chapter 6, I develop a series of preliminary analogue experiments to determine theeffect of climate-induced valley-scale glacial loading and unloading of volcanic landscapeson magma transport, storage and eruption. I develop a conceptual model for the responseof volcanic systems to rapid, near-surface stress-field variations driven by glacial loadingand unloading. I find that the rate and style of landscape deformation is controlled by thetopographic geometry and the crustal rock rheology. When the near-surface crustal rocksare strong and resistant to ductile flow (e.g. coherent lavas), unloading of the valleys byglacier retreat creates compression on topographic high points (mountains and ridges) andan overall suppression of volcanism in these areas. However, when rock types that arecapable of ductile deformation dominate the near-surface crust (e.g. hydrothermally alteredpyroclastic rocks), the effect is reversed. These experiments show that the nature and extentof coupling between climate-induced glacial activity volcanic eruptions may be governedby near-surface lithology and topographic geometry.In chapter 7, I use stochastic modelling to investigate how loading and unloading ofcontinental-scale ice sheets may affect magma transport, storage, and eruptions. In theGVB, field evidence suggests that the last major deglaciation (from ∼17 to ∼11 ka) wasaccompanied by a∼5-fold increase in eruption flux over the normal Pleistocene backgroundrate (i.e., the Rubble, Culliton and Ring Creek lavas were all erupted within this time frame).Deglaciation of the CIS facilitated rapid isostatic rebound of the crust, which released storedelastic deviatoric stresses that were developed during glacial loading and downward bend-ing. I use a Monte Carlo simulation to show that these transient stresses may act as a pumpthat modulates volcanic eruptions; the crust is charged with magma during loading and thenevacuated of magma (i.e., the magmas erupt) during unloading. Results of my modellingindicate that the last glaciation in the GVB suppressed eruptions by a factor of 0.4 ± 0.06,while unloading (i.e., deglaciation) increased eruptions by a factor of 2.7± 0.15, consistentwith the observed output flux.These two chapters illustrate that loading and unloading of ice in thick-crusted tec-tonic environments can influence the size, distribution and frequency of volcanic eruptions,however, the effect may be different for small, valley-filling glaciers than it is for largecontinental ice sheets. These processes will alter the overall number, size and distribution177of glacially-modulated eruptions at several different magnitudes and scales. The long-termflux of magma to the surface is probably mostly influenced by crustal-scale processes suchas glacial pumping (e.g., chapter 7). The behaviour of individual edifices (e.g., a singleglaciated stratovolcano), however, may be controlled by local topography-glacier isostaticeffects (e.g., chapter 6).All volcano-glacier causal linkage hypotheses and models that involve modulated stor-age and transport of magmas through the crust imply that a petrographic/geochemical sig-nature of glacial modulation must exist. Presumably, glacially-modulated lavas should bemore evolved than their non-modulated counterparts. This signature may be difficult todemonstrate for arc lavas (due to their typical complex and prolonged crustal ascent histo-ries) and has not been recognized in the literature. It is plausible to suggest that eruptionsthat occur during deglaciation events should be more felsic, more explosive and larger, dueto a tendency for these magmas to differentiate and create more hydrous and silicic compo-sitions while stored (due to glacial modulation) within the crust.Finally, in Iceland, there is compelling evidence to suggest that the last major deglacia-tion was accompanied by a ∼20 to 30-fold increase in volcanic eruptions over the normalbackground rate (Hardarson and Fitton, 1991; Sigvaldason et al., 1992). This process hasbeen modelled geochemically; ice sheet unloading caused increased decompression mantlemelting due to enhanced inward flow of hot underlying asthenosphere (Jull and McKenzie,1996; Maclennan et al., 2002; McGuire et al., 1997; Schmidt et al., 2013). Although themodel applied to Iceland is not useful for volcanic arcs (and other thick-crusted tectonicregions), the concept of glacial pumping presented in this thesis may be applied to Iceland.Several studies have suggested that deglaciation did cause more eruptions in Iceland due toreduced crustal stress from the melting ice (c.f., Gudmundsson, 1986; Sigmundsson et al.,2010), however, the full effect of glacio-crustal bending and glacial pumping has not beenexplored. Iceland has a thin elastic crust that underwent demonstrable downward bendingand elastic isostatic rebound over the last glacial cycle (Ingo´lfson et al., 2008; Wickert,2016). It is reasonable to suggest that glacial pumping may also have contributed to thetiming, frequency, volume and composition of the erupted material. This presumption doesnot discredit the models currently used to explain the causal linkages between Icelandicvolcanism and glacier mass loss or gain (e.g., Jull and McKenzie, 1996; Maclennan et al.,2002; McGuire et al., 1997; Schmidt et al., 2013). Instead, it suggests that these linkages arecomplex and operate in a variety of ways and over several different scales (e.g., Kelemenet al., 1997; Sigmundsson et al., 2010; Watt et al., 2013). Exploring the effect of glacialpumping in Iceland is an intriguing prospect worthy of further investigation.178BibliographyAlbino, F.; Pinel, V.; and Sigmundsson, F. (2010). Influence of surface load variations oneruption likelihood: Application to two Icelandic subglacial volcanoes, Grı´msvo¨tn andKatla. Geophysical Journal International, 181(3):pages 1510–1524. ISSN 0956540X.doi:10.1111/j.1365-246X.2010.04603.x. → pages 134, 155, 158Allen, C.C. (1979). Volcano-ice interactions on Mars. Journal of Geophysical Research,84(B14):page 8048. ISSN 0148-0227. doi:10.1029/JB084iB14p08048. → page 2Allen, C.C. (1980). Icelandic subglacial volcanism: thermal and physical studies. TheJournal of Geology, 88(1):pages 108–117. ISSN 0022-1376. doi:10.1086/628478. →page 35Allen, C.C.; Jercinovic, M.J.; and Allen, J.S.B. (1982). Subglacial Volcanism inNorth-Central British Columbia and Iceland. The Journal of Geology, 90(6):pages699–715. ISSN 0022-1376. doi:10.1086/628725. → pages 36, 52Andrews, G.D.; Porritt, L.; and Russell, J. (2014a). Quaternary subglacial and explosivevolcanism in the Canadian Cascade arc (Sea-to-Sky Corridor), British Columbia. InTrials and Tribulations of Life on an Active Subduction Zone: Field Trips in and aroundVancouver, Canada, volume 38, pages 125–167. Geological Society of America.doi:10.1130/2014.0038(07). → pages 20, 34, 105Andrews, G.D.; Russell, J.; and Stewart, M.L. (2014b). The history and dynamics of awelded pyroclastic dam and its failure. Bulletin of Volcanology, 76(4):pages 1–16.ISSN 14320819. doi:10.1007/s00445-014-0811-0. → pages 4, 38, 101Arculus, R. (1994). Aspects of magma genesis in arcs. Lithos, 33(1-3):pages 189–208.ISSN 00244937. doi:10.1016/0024-4937(94)90060-4. → page 153Armstrong, J.E. (1981). Post-Vashon Wisconsin glaciation, Fraser Lowland, BritishColumbia. Geological Survey of Canada, Bulletin, 322(1):page 34. → page 43Armstrong, J.E.; Crandell, D.R.; Easterbrook, D.J.; and Noble, J.B. (1965). Latepleistocene stratigraphy and chronology in southwestern British Columbia andnorthwestern Washington. Bulletin of the Geological Society of America, 76(3):pages321–330. ISSN 00167606. doi:10.1130/0016-7606(1965)76[321:LPSACI]2.0.CO;2.→ pages 43, 64, 67179Aubry, T.J.; Jellinek, A.M.; Degruyter, W.; Bonadonna, C.; Radı´c, V.; Clyne, M.; andQuainoo, A. (2016). Impact of global warming on the rise of volcanic plumes andimplications for future volcanic aerosol forcing. Journal of Geophysical Research,121(22):pages 13,326–13,351. ISSN 21562202. doi:10.1002/2016JD025405. → pages3, 152Augustinus, P.C. (1995). Glacial valley cross-profile development: the influence of in siturock stress and rock mass strength, with examples from the Southern Alps, NewZealand. Geomorphology, 14(2):pages 87–97. ISSN 0169555X.doi:10.1016/0169-555X(95)00050-X. → page 135Bacon, C.R.; Bruggman, P.E.; Christiansen, R.L.; Clynne, M.A.; Donnelly-Nolan, J.M.;and Hildreth, W. (1997). Primitive magmas at five Cascade volcanic fields: Melts fromhot, heterogeneous sub-arc mantle. Canadian Mineralogist, 35(2):pages 397–423.ISSN 00084476. doi:scopus/2-s2.0-0031463904. → pages 13, 72, 101Bacon, C.R. and Lanphere, M.A. (2006). Eruptive history and geochronology of MountMazama and the Crater Lake region, Oregon. Geological Society of America Bulletin,118(11-12):pages 1331–1359. ISSN 0016-7606. doi:10.1130/B25906.1. → page 21Baldini, J.U.; Brown, R.J.; and McElwaine, J.N. (2015). Was millennial scale climatechange during the Last Glacial triggered by explosive volcanism? Scientific Reports,5(1):page 17,442. ISSN 2045-2322. doi:10.1038/srep17442. → page 3Barnett, Z.A. and Gudmundsson, A. (2014). Numerical modelling of dykes deflected intosills to form a magma chamber. Journal of Volcanology and Geothermal Research,281:pages 1–11. ISSN 03770273. doi:10.1016/j.jvolgeores.2014.05.018. → pages153, 161Benn, D.I. and Hulton, N.R. (2010). An ExcelTM spreadsheet program for reconstructingthe surface profile of former mountain glaciers and ice caps. Computers & Geosciences,36(5):pages 605–610. ISSN 00983004. doi:10.1016/j.cageo.2009.09.016. → page 59Benowitz, J.A.; Layer, P.W.; and Vanlaningham, S. (2014). Persistent long-term (c. 24 Ma)exhumation in the Eastern Alaska Range constrained by stacked thermochronology.Advances in 40Ar/39Ar Dating: from Archaeology to Planetary Sciences, 378(1):pages225–243. ISSN 0305-8719. doi:10.1144/SP378.12. → page 112Bjo¨rnsson, H. (2003). Subglacial lakes and jo¨kulhlaups in Iceland. Global and PlanetaryChange, 35(3-4):pages 255–271. ISSN 09218181.doi:10.1016/S0921-8181(02)00130-3. → pages 35, 42, 97, 100, 125Blaise, B.; Clague, J.J.; and Mathewes, R.W. (1990). Time of Maximum Late WisconsinGlaciation, West Coast of Canada. Quaternary Research, 34(03):pages 282–295. ISSN0033-5894. doi:10.1016/0033-5894(90)90041-I. → pages 1, 11, 156Bolch, T.; Menounos, B.; and Wheate, R. (2010). Landsat-based inventory of glaciers inwestern Canada, 1985-2005. Remote Sensing of Environment, 114(1):pages 127–137.ISSN 00344257. doi:10.1016/j.rse.2009.08.015. → page 59180Booth, D.B.; Troost, K.G.; Clague, J.J.; and Waitt, R.B. (2003). The Cordilleran Ice Sheet.In Developments in Quaternary Science, volume 1, pages 17–43. Elsevier. ISBN9780444514707. doi:10.1016/S1571-0866(03)01002-9. → pages1, 10, 11, 43, 51, 55, 156, 157Borgia, A.; Linneman, S.; Spencer, D.; Morales, L.D.; and Andre, J.B. (1983). Dynamicsof lava flow fronts, Arenal Volcano, Costa Rica. Journal of Volcanology andGeothermal Research, 19(3-4):pages 303–329. ISSN 03770273.doi:10.1016/0377-0273(83)90116-6. → pages 120, 121Borgia, A. and van Wyk de Vries, B. (2003). The volcano-tectonic evolution ofConcepcio´n, Nicaragua. Bulletin of Volcanology, 65(4):pages 248–266. ISSN0258-8900. doi:10.1007/s00445-002-0256-8. → page 134Bowen, D.; Richmond, G.; Fullerton, D.; Sˇibrava, V.; Fulton, R.; and Velichko, A. (1986).Correlation of Quaternary glaciations in the Northern Hemisphere. Quaternary ScienceReviews, 5(C):pages 509–510. ISSN 02773791. doi:10.1016/0277-3791(86)90218-0.→ pages 51, 69Brooks, G.R. and Friele, P.A. (1992). Bracketing ages for the formation of the Ring Creeklava flow, Mount Garibaldi volcanic field, southwestern British Columbia. CanadianJournal of Earth Sciences, 29(11):pages 2425–2428. ISSN 0008-4077.doi:10.1139/e92-190. → pages 19, 46, 156Bruno, S. (2011). Derivation of the Eruption History of the Prehistoric Ring Creek LavaFlow , Southern British Columbia. Ph.D. thesis, Western Washington University. →page 156Bustin, A.M.; Clowes, R.M.; Monger, J.W.; Journeay, J.M.; and Spence, G. (2013). Thesouthern Coast Mountains, British Columbia: New interpretations from geological,seismic reflection, and gravity data. Canadian Journal of Earth Sciences, 50(10):pages1033–1050. ISSN 0008-4077. doi:10.1139/cjes-2012-0122. → page 121Bye, A.; Edwards, B.; and Hickson, C.J. (2000). Preliminary field, petrographic, andgeochemical analysis of possible subglacial, dacitic volcanism at the Watts Pointvolcanic centre, southwestern British Columbia. Current Research, A20:page 9. →pages 19, 22, 23, 117Byrne, M.D. (2013). How Many Times should A Stochastic Model be Run? An ApproachBased on Confidence Intervals. Proceedings of the 12th International Conference onCognitive Modeling, pages 445–450. → page 162Byrne, P.K.; Holohan, E.P.; Kervyn, M.; Van Wyk de Vries, B.; Troll, V.R.; and Murray,J.B. (2013). A sagging-spreading continuum of large volcano structure: supplementarymaterial. Geology, 41(3):pages 339–342. ISSN 00917613. doi:10.1130/G33990.1. →pages 135, 137, 139, 141, 143181Carrivick, J.L.; Russell, A.J.; Lucy Rushmer, E.; Tweed, F.S.; Marren, P.M.; Deeming, H.;and Lowe, O.J. (2009). Geomorphological evidence towards a de-glacial control onvolcanism. Earth Surface Processes and Landforms, 34(8):pages 1164–1178. ISSN01979337. doi:10.1002/esp.1811. → page 20Cashman, K.; Pinkerton, H.; and Stephenson, J. (1998). Introduction to special section:long lava flows. Journal of Geophysical Research, 103(B11):pages 27,281–27,289.ISSN 21699356. doi:10.1029/98JB01820. → page 132Chapman, M.G. and Tanaka, K.L. (2001). Interior trough deposits on Mars: Subicevolcanoes? Journal of Geophysical Research: Planets, 106(E5):pages 10,087–10,100.ISSN 01480227. doi:10.1029/2000JE001303. → page 2Chapman, M.G. and Tanaka, K.L. (2002). Related magma-ice interactions: Possibleorigins of chasmata, chaos, and surface materials in Xanthe, Margaritifer, MeridianiTerrae, Mars. Icarus, 155(2):pages 324–339. ISSN 00191035.doi:10.1006/icar.2001.6735. → page 2Chayes, F. (1960). On correlation between variables of constant sum. Journal ofGeophysical Research, 65(12):pages 4185–4193. ISSN 01480227.doi:10.1029/JZ065i012p04185. → page 89Clague, J. and Evans, S.G. (2000). A review of catastrophic drainage of moraine-dammedlakes in British Columbia. Quaternary Science Reviews, 19(17-18):pages 1763–1783.ISSN 02773791. doi:10.1016/S0277-3791(00)00090-1. → page 97Clague, J.J. (1976). Quadra Sand and its Relation to the Late Wisconsin Glaciation ofSouthwest British-Columbia Canada. Canadian Journal of Earth Sciences, 13(6):pages803–815. ISSN 0008-4077. doi:10.1139/e76-083. → page 43Clague, J.J. (1980). Late Quaternary geology and geochronology of British Columbia:Part 1: Radiocarbon dates. Geological Survey of Canada Paper, 80-13:pages 1–28. →pages 11, 43Clague, J.J. (1981). Late Quaternary Geology and Geochronology of British Columbia.Part II: Summary and Discussion of Radiocarbon-Dated Quaernary History. GeologicalSurvey of Canada Paper, 80-35:page 47. ISSN 0036-8075.doi:10.1126/science.ns-6.149S.521-a. → pages 20, 35, 43, 156Clague, J.J. (1983). Glacio-isostatic effects of the Cordilleran ice sheet, British Columbia,Canada. In Shorelines and isostasy, volume 16, pages 321–343. Academic PressLondon. → page 156Clague, J.J. (2009). Cordilleran ice sheet. Encyclopedia of Paleoclimatology and AncientEnvironments, pages 206–211. ISSN 1402045514. → pages 74, 99Clague, J.J. (2011). The Quaternary stratigraphic record of British Columbia—evidencefor episodic sedimentation and erosion controlled by glaciation. Canadian Journal of182Earth Sciences, 23(6):pages 885–894. ISSN 0008-4077. doi:10.1139/e86-090. →pages 50, 51, 55, 56, 59, 65, 67, 69, 115, 156Clague, J.J.; Easterbrook, D.J.; Hughes, O.L.; and Matthews, J.V. (1992). TheSangamonian and early Wisconsinan stages in western Canada and northwestern UnitedStates. Geological Society of America Special Papers, 270:pages 253–268. ISSN00721077. doi:10.1130/SPE270-p253. → pages 48, 59, 67Clague, J.J. and James, T.S. (2002). History and isostatic effects of the last ice sheet insouthern British Columbia. Quaternary Science Reviews, 21(1-3):pages 71–87. ISSN02773791. doi:10.1016/S0277-3791(01)00070-1. → pages 156, 157, 170Clague, J.J. and Ward, B. (2011). Pleistocene Glaciation of British Columbia. InDevelopments in Quaternary Science, volume 15, pages 563–573. Elsevier Inc., 1edition. ISBN 9780444534477. doi:10.1016/B978-0-444-53447-7.00044-1. → pages1, 11, 38, 43, 47, 59, 63, 64, 67, 74, 99, 104, 157, 170Cole, R.P.; White, J.D.; Conway, C.E.; Leonard, G.S.; Townsend, D.B.; and Pure, L.R.(2018). The glaciovolcanic evolution of an andesitic edifice, South Crater, Tongarirovolcano, New Zealand. Journal of Volcanology and Geothermal Research, 352:pages55–77. ISSN 03770273. doi:10.1016/j.jvolgeores.2017.12.003. → pages 2, 3, 20Conrey, R.M.; Taylor, E.M.; Donnelly-Nolan, J.M.; and Sherrod, D.R. (2002).North-Central Oregon Cascades: Exploring Petrologic and Tectonic Intimacy in aPropagating Intra-Arc Rift. Field Guide to Geologic Processes in Cascadia: OregonDepartment of Geology and Mineral Industries, 36:pages 47–90. → page 21Conway, C.E.; Leonard, G.S.; Townsend, D.B.; Calvert, A.T.; Wilson, C.J.; Gamble, J.A.;and Eaves, S.R. (2016). A high-resolution 40Ar/39Ar lava chronology and edificeconstruction history for Ruapehu volcano, New Zealand. Journal of Volcanology andGeothermal Research, 327:pages 152–179. ISSN 03770273.doi:10.1016/j.jvolgeores.2016.07.006. → pages 2, 20, 70, 174Conway, C.E.; Townsend, D.B.; Leonard, G.S.; Wilson, C.J.N.; Calvert, A.T.; andGamble, J.A. (2015). Lava-ice interaction on a large composite volcano: a case studyfrom Ruapehu, New Zealand. Bulletin of Volcanology. ISSN 14320819.doi:10.1007/s00445-015-0906-2. → pages 20, 176Cosma, T.; Hendy, I.; and Chang, A. (2008). Chronological constraints on Cordilleran IceSheet glaciomarine sedimentation from core MD02-2496 off Vancouver Island (westernCanada). Quaternary Science Reviews, 27(9-10):pages 941–955. ISSN 02773791.doi:10.1016/j.quascirev.2008.01.013. → pages 64, 65Cruden, A. and Weinberg, R. (2018). Mechanisms of Magma Transport and Storage in theLower and Middle Crust—Magma Segregation, Ascent and Emplacement. In Volcanicand Igneous Plumbing Systems, pages 13–53. Elsevier.doi:10.1016/B978-0-12-809749-6.00002-9. → pages 153, 162183Cuffey, K.M. and Paterson, W.S..B. (2010). Physics of Glaciers, Fourth Edition.Academic Press. ISBN 0123694612. doi:10.1016. → pages 121, 126Dalla Santa, G.; Peron, F.; Galgaro, A.; Cultrera, M.; Bertermann, D.; Mueller, J.; andBernardi, A. (2017). Laboratory Measurements of Gravel Thermal Conductivity: AnUpdate Methodological Approach. Energy Procedia, 125:pages 671–677. ISSN18766102. doi:10.1016/j.egypro.2017.08.287. → page 129Dansgaard, W.; Clausen, H.B.; Gundestrup, N.; Hammer, C.U.; Johnsen, S.F.;Kristinsdottir, P.M.; and Reeh, N. (1982). A New Greenland Deep Ice Core. Science,218(4579):pages 1273–1277. ISSN 0036-8075. doi:10.1126/science.218.4579.1273. →pages 3, 50Davidson, J.P. and Frank J. Tepley (1997). Recharge in Volcanic Systems: Evidence fromIsotope Profiles of Phenocrysts. Science, 275(5301):pages 826–829. ISSN 00368075.doi:10.1126/science.275.5301.826. → page 81Davis, N. and Mathews, W. (1944). Four Phases of Glaciation with Illustrations fromSouthwestern British Columbia. The Journal of Geology, 52(6):pages 403–413. →pages 55, 56, 63de Boer, B.; Stocchi, P.; Whitehouse, P.L.; and van de Wal, R.S. (2017). Current state andfuture perspectives on coupled ice-sheet – sea-level modelling. Quaternary ScienceReviews, 169:pages 13–28. ISSN 02773791. doi:10.1016/j.quascirev.2017.05.013. →page 3Delcamp, A.; van Wyk de Vries, B.; James, M.R.; Gailler, L.S.; and Lebas, E. (2012).Relationships between volcano gravitational spreading and magma intrusion. Bulletin ofVolcanology, 74(3):pages 743–765. ISSN 0258-8900. doi:10.1007/s00445-011-0558-9.→ page 137Densmore, M.S.; Ehlers, T.A.; and Woodsworth, G.J. (2007). Effect of Alpine glaciationon thermochronometer age-elevation profiles. Geophysical Research Letters,34(2):pages 2–7. ISSN 00948276. doi:10.1029/2006GL028371. → page 59Driedger, C.L. and Kennard, P. (1986). Ice Volumes on Cascade Volcanoes: Mount Rainer,Mount Hood, Three Sisters, Mount Shasta. U.S. Geological Survey, 1365. ISSN10449612. → page 11Ducea, M.N.; Saleeby, J.B.; and Bergantz, G. (2015). The Architecture, Chemistry, andEvolution of Continental Magmatic Arcs. Annual Review of Earth and PlanetarySciences, 43(1):pages 299–331. ISSN 0084-6597.doi:10.1146/annurev-earth-060614-105049. → page 153Edwards, B.; Belousov, A.; Belousova, M.; and Melnikov, D. (2015a). Observations onlava, snowpack and their interactions during the 2012–13 Tolbachik eruption,Klyuchevskoy Group, Kamchatka, Russia. Journal of Volcanology and GeothermalResearch, 307:pages 107–119. ISSN 03770273. doi:10.1016/j.jvolgeores.2015.08.010.→ page 1184Edwards, B.; Gudmundsson, M.T.; and Russell, J. (2015b). Glaciovolcanism. In TheEncyclopedia of Volcanoes, June, chapter 20, pages 377–393. Elsevier. ISBN9780123859389. doi:10.1016/B978-0-12-385938-9.00020-1. → pages 100, 116Edwards, B.; Karson, J.; Wysocki, R.; Lev, E.; Bindeman, I.; and Kueppers, U. (2013).Insights on lava-ice/snow interactions from large-scale basaltic melt experiments.Geology. ISSN 00917613. doi:10.1130/G34305.1. → pages 122, 128Edwards, B.; Magnu´sson, E.; Thordarson, T.; Gumundsson, M.T.; Ho¨skuldsson, A.;Oddsson, B.; and Haklar, J. (2012). Interactions between lava and snow/ice during the2010 Fimmvo¨ ruha´ls eruption, South-Central Iceland. Journal of GeophysicalResearch: Solid Earth, 117(4):pages 1–21. ISSN 21699356.doi:10.1029/2011JB008985. → pages 1, 100Edwards, B. and Russell, J. (1999). Northern Cordilleran volcanic province: A northernBasin and Range? Geology, 27(3):page 243. ISSN 0091-7613.doi:10.1130/0091-7613(1999)027〈0243:NCVPAN〉2.3.CO;2. → pages 20, 34Edwards, B. and Russell, J. (2002). Glacial influences on morphology and eruptiveproducts of Hoodoo Mountain volcano, Canada. Geological Society, London, SpecialPublications, 202(1):pages 179–194. ISSN 0305-8719.doi:10.1144/GSL.SP.2002.202.01.09. → pages2, 3, 9, 20, 34, 74, 117, 152, 153, 155, 161, 168Edwards, B.; Russell, J.; and Anderson, R.G. (2002). Subglacial, phonolitic volcanism atHoodoo Mountain volcano, northern Canadian Cordillera. Bulletin of Volcanology,64(3-4):pages 254–272. ISSN 0258-8900. doi:10.1007/s00445-002-0202-9. → pages117, 121Edwards, B.; Russell, J.; Anderson, R.G.; and Harder, M. (2003). Overview of Neogene toRecent volcanism in the Atlin volcanic district, Northern Cordilleran volcanic province,northwestern British Columbia. Geological Survey of Canada, Current Research,A8:pages 1–6. doi:10.4095/214025. → page 20Edwards, B.; Russell, J.; and Simpson, K. (2011). Volcanology and petrology of MathewsTuya, northern British Columbia, Canada: glaciovolcanic constraints on interpretationsof the 0.730 Ma Cordilleran paleoclimate. Bulletin of Volcanology, 73(5):pages479–496. ISSN 0258-8900. doi:10.1007/s00445-010-0418-z. → pages1, 2, 9, 20, 42, 51, 52, 70Edwards, B.; Tuffen, H.; Skilling, I.P.; and Wilson, L. (2009). Introduction to special issueon volcano–ice interactions on Earth and Mars: The state of the science. Journal ofVolcanology and Geothermal Research, 185(4):pages 247–250. ISSN 03770273.doi:10.1016/j.jvolgeores.2009.06.003. → pages 20, 52, 74, 100Evans, D.J.; Clark, C.D.; and Mitchell, W.A. (2005). The last British Ice Sheet: A reviewof the evidence utilised in the compilation of the Glacial Map of Britain. Earth-Science185Reviews, 70(3-4):pages 253–312. ISSN 00128252. doi:10.1016/j.earscirev.2005.01.001.→ page 51Fillmore, J. and Coulson, I.M. (2013). Petrological and geochemical constraints on theorigin of adakites in the Garibaldi Volcanic Complex, southwestern British Columbia,Canada. Bulletin of Volcanology, 75(7):pages 1–23. ISSN 02588900.doi:10.1007/s00445-013-0730-5. → page 105Fink, J.H. and Griffiths, R.W. (1998). Morphology, eruption rates, and rheology of lavadomes: Insights from laboratory models. Journal of Geophysical Research: Solid Earth,103(B1):pages 527–545. ISSN 01480227. doi:10.1029/97JB02838. → pages 120, 121Forbes, A.; Blake, S.; McGarvie, D.; and Tuffen, H. (2012). Pseudopillow fracturesystems in lavas: Insights into cooling mechanisms and environments from lava flowfractures. Journal of Volcanology and Geothermal Research, 245-246:pages 68–80.ISSN 03770273. doi:10.1016/j.jvolgeores.2012.07.007. → pages 25, 109Forbes, A.E.S.; Blake, S.; and Tuffen, H. (2014). Entablature: Fracture types andmechanisms. Bulletin of Volcanology, 76(5). ISSN 14320819.doi:10.1007/s00445-014-0820-z. → page 109Friedman, E.; Polat, A.; Thorkelson, D.J.; and Frei, R. (2016). Lithospheric mantlexenoliths sampled by melts from upwelling asthenosphere: The Quaternary Tassealkaline basalts of southeastern British Columbia, Canada. Gondwana Research,33:pages 209–230. ISSN 1342937X. doi:10.1016/j.gr.2015.11.005. → page 72Friele, P.A. and Clague, J.J. (2002). Younger Dryas readvance in Squamish river valley,southern Coast mountains, British Columbia. Quaternary Science Reviews,21(18-19):pages 1925–1933. ISSN 02773791. doi:10.1016/S0277-3791(02)00081-1.→ pages 46, 170Friele, P.a. and Clague, J.J. (2009). Paraglacial geomorphology of Quaternary volcaniclandscapes in the southern Coast Mountains, British Columbia. Geological Society,London, Special Publications, 320(1):pages 219–233. ISSN 0305-8719.doi:10.1144/SP320.14. → pages 19, 24, 38, 46, 156Friele, P.a.; Ekes, C.; and Hickin, E.J. (1999). Evolution of Cheekye fan, Squamish,British Columbia: Holocene sedimentation and implications for hazard assessment.Canadian Journal of Earth Sciences, 36(12):pages 2023–2031. ISSN 0008-4077.doi:10.1139/e99-090. → page 46Fulton, R. (1989). Quaternary Geology of Canada and Greenland, volume K-1.Geological Society of America, North America. ISBN 0660131145.doi:10.1130/DNAG-GNA-K1. → page 50Fulton, R.J. (1991). A Conceptual Model for Growth and Decay of the Cordilleran IceSheet. Ge´ographie physique et Quaternaire, 45(3):page 281. ISSN 0705-7199.doi:10.7202/032875ar. → pages 11, 35, 38, 43, 55, 56, 63186Fulton, R.J.; Fenton, M.M.; and Rutter, N.W. (1986). Summary of quaternary stratigraphyand history, Western Canada. Quaternary Science Reviews, 5(C):pages 229–241. ISSN02773791. doi:10.1016/0277-3791(86)90189-7. → page 1Furnes, H.; Fridleifsson, I.; and Atkins, F. (1980). Subglacial volcanics — On theformation of acid hyaloclastites. Journal of Volcanology and Geothermal Research,8(1):pages 95–110. ISSN 03770273. doi:10.1016/0377-0273(80)90009-8. → page 9Gaffney, E.S. and Damjanac, B. (2006). Localization of volcanic activity: Topographiceffects on dike propagation, eruption and conduit formation. Geophysical ResearchLetters, 33(14):page L14,313. ISSN 0094-8276. doi:10.1029/2006GL026852. → page147Galland, O.; Burchardt, S.; Hallot, E.; Mourgues, R.; and Bulois, C. (2014). Dynamics ofdikes versus cone sheets in volcanic systems. Journal of Geophysical Research: SolidEarth, 119(8):pages 6178–6192. ISSN 21699313. doi:10.1002/2014JB011059. →pages 135, 137, 139Gascoyne, M.; Drake, J.J.; Ford, D.C.; Latham, A.G.; Harmon, R.S.; and Schwarcz, H.P.(2006). Estimates of the Age of the Existing Relief within the Southern RockyMountains of Canada. Arctic and Alpine Research, 13(1):page 1. ISSN 00040851.doi:10.2307/1550621. → page 59Gee, M.; Taylor, R.; Thirlwall, M.; and Murton, B. (1998). Glacioisostacy controlschemical and isotopic characteristics of tholeiites from the Reykjanes Peninsula, SWIceland. Earth and Planetary Science Letters, 164(1-2):pages 1–5. ISSN 0012821X.doi:10.1016/S0012-821X(98)00246-5. → page 155Geshi, N.; Kusumoto, S.; and Gudmundsson, A. (2010). Geometric difference betweennon-feeder and feeder dikes. Geology, 38(3):pages 195–198. ISSN 00917613.doi:10.1130/G30350.1. → pages 147, 149, 161, 162Ghiorso, M.S. and Gualda, G.A.R. (2015). An H2O–CO2 mixed fluid saturation modelcompatible with rhyolite-MELTS. Contributions to Mineralogy and Petrology,169(6):page 53. ISSN 0010-7999. doi:10.1007/s00410-015-1141-8. → pages 90, 94, 96Giordano, D.; Nichols, A.R.L.; and Dingwell, D.B. (2005). Glass transition temperaturesof natural hydrous melts: A relationship with shear viscosity and implications for thewelding process. Journal of Volcanology and Geothermal Research, 142(1-2 SPEC.ISS.):pages 105–118. ISSN 03770273. doi:10.1016/j.jvolgeores.2004.10.015. → page35Glazner, A.F.; Manley, C.R.; Marron, J.S.; and Rojstaczer, S. (1999). Fire or ice:anticorrelation of volcanism and glaciation in California over the past 800,000 years.Geophysical Research Letters, 26(12):pages 1759–1762. ISSN 00948276.doi:10.1029/1999GL900333. → pages 134, 144, 147, 150, 152, 153, 154, 166, 176187Glazner, A.F. and Schubert, G. (1985). Flexure of the North American lithosphere abovethe subducted Mendocino Fracture Zone and the formation of east-west faults in thetransverse ranges. Journal of Geophysical Research: Solid Earth, 90(B7):pages5405–5409. ISSN 01480227. doi:10.1029/JB090iB07p05405. → pages153, 158, 161, 163, 166Glen, J.W. (1952). Experiments on the Deformation of Ice. Journal of Glaciology,2(12):pages 111–114. ISSN 0022-1430. doi:10.1017/S0022143000034067. → page122Goehring, L. and Morris, S.W. (2008). Scaling of columnar joints in basalt. Journal ofGeophysical Research, 113(B10):page B10,203. ISSN 0148-0227.doi:10.1029/2007JB005018. → page 109Gonnermann, H. and Manga, M. (2013). Dynamics of magma ascent in the volcanicconduit. In S.A. Fagents; T.K.P. Gregg; and R.M. Lopes (editors),Modeling VolcanicProcesses, April, chapter 4, pages 55–84. Cambridge University Press, Cambridge.doi:10.1017/CBO9781139021562. → page 134Gonnermann, H. and Taisne, B. (2015). Magma Transport in Dikes. In The Encyclopediaof Volcanoes, pages 215–224. Elsevier. ISBN 9780123859389.doi:10.1016/B978-0-12-385938-9.00010-9. → pages146, 147, 153, 161, 162, 166, 168, 169Green, N. (1977). Multistage andesite genesis in the Garibaldi Lake area, southwesternBritish Columbia. Ph.D. thesis, The University of British Columbia. → pages13, 19, 22, 24, 25, 26, 27, 104, 175Green, N. (1981a). Geology and petrology of Quaternary volcanic rocks, Garibaldi Lakearea, southwestern British Columbia. Part II. Geological Society of America Bulletin,92:pages 1359–1470. ISSN 00167606.doi:10.1130/0016-7606(1981)92〈697:GAPOQV〉2.0.CO;2. → page 104Green, N. (1981b). Geology and petrology of Quaternary volcanic rocks, Garibaldi Lakearea, southwestern British Columbia: Summary. Geological Society of AmericaBulletin, 92(10):page 697. ISSN 0016-7606.doi:10.1130/0016-7606(1981)92〈697:GAPOQV〉2.0.CO;2. → pages13, 19, 27, 46, 53, 104, 168Green, N. (1990). Late Cenozoic Volcanism in the Mount Garibaldi and Garibaldi LakeVolcanic Fields, Garibaldi Volcanic Belt, Southwestern British Columbia. Journal of theGeological Association of Canada, 17(3):pages 171–175. doi:10.12789/gs.v17i3.3673.→ page 22Green, N. (1994). Mechanism for middle to upper crustal contamination: Evidence fromcontinental-margin magmas. Geology, 22(3):page 231. ISSN 0091-7613.doi:10.1130/0091-7613(1994)022〈0231:MFMTUC〉2.3.CO;2. → pages 19, 23188Green, N. (2006). Influence of slab thermal structure on basalt source regions and meltingconditions: REE and HFSE constraints from the Garibaldi volcanic belt, northernCascadia subduction system. Lithos, 87(1-2):pages 23–49. ISSN 00244937.doi:10.1016/j.lithos.2005.05.003. → pages 13, 72Green, N.; Armstrong, R.L.; Harakal, J.E.; Souther, J.G.; and Read, P.B. (1988). Eruptivehistory and K-Ar geochronology of the late Cenozoic Garibaldi volcanic belt,southwestern British Columbia. Geological Society of America Bulletin, 100(4):pages563–579. ISSN 00167606.doi:10.1130/0016-7606(1988)100〈0563:EHAKAG〉2.3.CO;2. → pages1, 4, 11, 13, 19, 20, 22, 23, 24, 25, 26, 29, 40, 52, 72, 101, 103, 151, 175Green, N. and Harry, D.L. (1999). On the relationship between subducted slab age and arcbasalt petrogenesis, Cascadia subduction system, North America. Earth and PlanetaryScience Letters, 171(3):pages 367–381. ISSN 0012821X.doi:10.1016/S0012-821X(99)00159-4. → page 101Green, N. and Sinha, A.K. (2005). Consequences of varied slab age and thermal structureon enrichment processes in the sub-arc mantle of the northern Cascadia subductionsystem. Journal of Volcanology and Geothermal Research, 140(1-3):pages 107–132.ISSN 03770273. doi:10.1016/j.jvolgeores.2004.07.017. → page 101Grossenbacher, K.A. and McDuffie, S.M. (1995). Conductive cooling of lava: columnarjoint diameter and stria width as functions of cooling rate and thermal gradient. Journalof Volcanology and Geothermal Research, 69(1-2):pages 95–103. ISSN 03770273.doi:10.1016/0377-0273(95)00032-1. → page 109Grove, E. (1974). Deglaciation - A Possible Triggering Mechanism for Recent Volcanism.Proceedings of the Symposium on Andean and Antarctic Volcanology Problems,Grove,E. W. (1974, September). Deglaciation—a possible triggering mechanism for recentvolcanism. In Proceedings of the Symposium on Andean and Antarctic VolcanologyProblems, Sant, pages 88–97. → pages 1, 3, 52, 152, 154, 176Grove, T.; Till, C.; and Krawczynski, M. (2012). The Role of H2O in Subduction ZoneMagmatism. Annual Review of Earth and Planetary Sciences, 40(1):pages 413–439.ISSN 0084-6597. doi:10.1146/annurev-earth-042711-105310. → pages 4, 153, 162Gualda, G.A.R.; Ghiorso, M.S.; Lemons, R.V.; and Carley, T.L. (2012). Rhyolite-MELTS:A modified calibration of MELTS optimized for silica-rich, fluid-bearing magmaticsystems. Journal of Petrology, 53(5):pages 875–890. ISSN 00223530.doi:10.1093/petrology/egr080. → pages 90, 94, 96Gudmundsson, A. (1983). Form and dimensions of dykes in eastern Iceland.Tectonophysics, 95(3-4):pages 295–307. ISSN 00401951.doi:10.1016/0040-1951(83)90074-4. → page 121Gudmundsson, A. (1984). Tectonic Aspects of Dykes in Northwestern Iceland. Jokull,34:pages 81–96. → pages 121, 134189Gudmundsson, A. (1986). Mechanical Aspects of Postglacial Volcanism and Tectonics ofthe Reykjanes Peninsula, Southwest Iceland. Journal of Geophysical Research,91:pages 12,711–12,721. ISSN 0148-0227. doi:10.1029/JB091iB12p12711. → pages154, 158, 178Gudmundsson, A. (2006). How local stresses control magma-chamber ruptures, dykeinjections, and eruptions in composite volcanoes. Earth-Science Reviews, 79(1-2):pages1–31. ISSN 00128252. doi:10.1016/j.earscirev.2006.06.006. → page 161Gudmundsson, A. (2012). Magma chambers: Formation, local stresses, excess pressures,and compartments. Journal of Volcanology and Geothermal Research, 237-238:pages19–41. ISSN 03770273. doi:10.1016/j.jvolgeores.2012.05.015. → pages 134, 146, 147Gudmundsson, M.T. (2003). Melting of ice by magma-ice-water interactions duringsubglacial eruptions as an indicator of heat transfer in subaqueous eruptions. InExplosive Subaqueous Volcanism, pages 61–72. American Geophysical Union. ISBN9781118668665. doi:10.1029/140GM04. → pages35, 99, 100, 120, 121, 125, 126, 127, 131, 132, 175, 176Gudmundsson, M.T.; Larsen, G.; Ho¨skuldsson, A´.; and Gylfason, A´.G. (2008). Volcanichazards in Iceland. Jokull, 58:pages 251–268. → pages 1, 2Gudmundsson, M.T.; Sigmundsson, F.; and Bjo¨rnsson, H. (1997). Ice-volcano interactionof the 1996 Gjalp subglacial eruption, Vatnajokull, Iceland. Nature, 389(6654):pages954–957. ISSN 00280836. doi:10.1038/40122. → pages 35, 42, 100, 120, 121, 125, 127Gudmundsson, M.T.; Sigmundsson, F.; Bjo¨rnsson, H.; and Ho¨gnado´ttir, T. (2004). The1996 eruption ay Gja´lp, Vatnajo¨kull ice cap, Iceland: Efficiency of heat transfer, icedeformation and subglacial water pressure. Bulletin of Volcanology, 66(1):pages 46–65.ISSN 02588900. doi:10.1007/s00445-003-0295-9. → pages35, 42, 100, 120, 125, 127, 131, 132, 175Gudmundsson, M.T.; Thordarson, T.; Ho¨skuldsson, A´.; Larsen, G.; Bjo¨rnsson, H.; Prata,F.J.; Oddsson, B.; Magnu´sson, E.; Ho¨gnado´ttir, T.; Petersen, G.N.; Hayward, C.L.;Stevenson, J.a.; and Jo´nsdo´ttir, I. (2012). Ash generation and distribution from theApril-May 2010 eruption of Eyjafjallajo¨kull, Iceland. Scientific Reports, 2(1):page 572.ISSN 2045-2322. doi:10.1038/srep00572. → pages 1, 2, 149, 158, 161, 162, 168, 169Guillou, H.; Van Vliet-Lanoe¨, B.; Guðmundsson, A.; and Nomade, S. (2010). Newunspiked K–Ar ages of Quaternary sub-glacial and sub-aerial volcanic activity inIceland. Quaternary Geochronology, 5(1):pages 10–19. ISSN 18711014.doi:10.1016/j.quageo.2009.08.007. → pages 9, 20, 42, 174Haeberli, W. and Hoelzle, M. (1995). Application of inventory data for estimatingcharacteristics of and regional climate-change effects on mountain glaciers: a pilotstudy with the European Alps. Annals of Glaciology, pages 206–212.doi:10.3198/1995AoG21-1-206-212. → page 55190Halleran, A.D. and Russell, J. (1990). Trace Elements Formulated As Pearce ElementRatios II: Recognition of Open System Behaviour. In Theory and Application of PearceElement Ratios to Geochemical Data Analysis, chapter 12, pages 243–270. GeologicalAssociation of Canada Special Papers. → page 89Hammond, P.E. (1980). Reconnaissance geologic map and cross sections of southernWashington Cascade Range: Portland, Oregon. Portland State University Departmentof Geology, 45, 122(scale 1:125,000):page 1. → page 21Hammond, P.E. (1987). Lone Butte and Crazy Hills: subglacial volcanic complexes,Cascade Range, Washington. Geological Society of America Centennial Field Guide -Cordilleran Section, 1:pages 339–344. → page 21Hansen, J.; Lacis, A.; Ruedy, R.; and Sato, M. (1992). Potential climate impact of MountPinatubo eruption. Geophysical Research Letters, 19(2):pages 215–218. ISSN00948276. doi:10.1029/91GL02788. → page 3Hardarson, B.S. and Fitton, J.G. (1991). Increased mantle melting beneath Snaefellsjo¨kullvolcano during Late Pleistocene deglaciation. Nature, 353(6339):pages 62–64. ISSN0028-0836. doi:10.1038/353062a0. → pages 134, 144, 150, 153, 166, 178Harder, M. and Russell, J. (2006). Basanite glaciovolcanism at Llangorse mountain,northern British Columbia, Canada. Bulletin of Volcanology, 69(3):pages 329–340.ISSN 0258-8900. doi:10.1007/s00445-006-0078-1. → page 25Harris, A.J.; Flynn, L.P.; Matı´as, O.; and Rose, W.I. (2002). The thermal stealth flows ofSantiaguito dome, Guatemala: Implications for the cooling and emplacement of daciticblock-lava flows. Bulletin of the Geological Society of America, 114(5):pages 533–546.ISSN 00167606. doi:10.1130/0016-7606(2002)114〈0533:TTSFOS〉2.0.CO;2. → page132Harris, A.J. and Rowland, S.K. (2009). Effusion rate controls on lava flow length and therole of heat loss: A review. Special Publications of IAVCEI, 2:pages 33–51. ISSN17508207. → page 132Harris, A.J.L.; Dehn, J.; and Calvari, S. (2007). Lava effusion rate definition andmeasurement: a review. Bulletin of Volcanology, 70(1):pages 1–22. ISSN 0258-8900.doi:10.1007/s00445-007-0120-y. → pages 120, 122Hart, C.J. and Villeneuve, M. (2011). Geochronology of Neogene alkaline volcanic rocks(Miles Canyon basalt), southern Yukon Territory, Canada: the relative effectiveness oflaser 40 Ar/ 39 Ar and K-Ar geochronology . Canadian Journal of Earth Sciences,36(9):pages 1495–1507. ISSN 0008-4077. doi:10.1139/e99-049. → page 174Hawkesworth, C.J.; Evans, P.J.; George, R.M.M.; Turner, S.P.; and Zellmer, G.F. (2002).Time-scales of magma formation, ascent and storage beneath subduction-zonevolcanoes. Philosophical Transactions of the Royal Society of London. Series A:Mathematical, Physical and Engineering Sciences, 358(1770):pages 1443–1464. ISSN1471-2962. doi:10.1098/rsta.2000.0598. → page 153191Head, J.W. and Wilson, L. (2002). Mars: a review and synthesis of general environmentsand geological settings of magma-H2O interactions. Geological Society, London,Special Publications, 202:pages 27–57. doi:10.1144/GSL.SP.1978.006.01.24. → page 2Heap, M.J.; Wadsworth, F.B.; Xu, T.; feng Chen, C.; and Tang, C. (2016). The strength ofheterogeneous volcanic rocks: A 2D approximation. Journal of Volcanology andGeothermal Research, 319:pages 1–11. ISSN 03770273.doi:10.1016/j.jvolgeores.2016.03.013. → page 136Hebda, R.J.; Lian, O.B.; Hicock, S.R.; and Fisher, T.G. (2016). Olympia Interstadial:vegetation, landscape history, and paleoclimatic implications of a mid-Wisconsinan(MIS3) nonglacial sequence from southwest British Columbia, Canada. CanadianJournal of Earth Sciences, 53(3):pages 304–320. ISSN 0008-4077.doi:10.1139/cjes-2015-0122. → pages 46, 64, 65Hickson, C.J. (1986). Quaternary Volcanism in the Wells Gray-Clearwater Area, EastCentral British Columbia. Ph.D. thesis, University of British Columbia. → pages 20, 34Hickson, C.J. (1994). Character of volcanism, volcanic hazards, and risk, northern end ofthe Cascade magmatic arc, British Columbia and Washington State. In J. Monger(editor), Geology and Geological Hazards of the Vancouver Region, SouthwesternBritish Columbia, pages 231–250. Geological Survey of Canada Bulletin 481.doi:10.4095/203253. → pages 2, 3, 4, 11, 13, 19, 20, 22, 24, 34, 36, 38, 72, 74Hickson, C.J. (2000). Physical controls and resulting morphological forms of Quaternaryice-contact volcanoes in western Canada. Geomorphology, 32(3-4):pages 239–261.ISSN 0169555X. doi:10.1016/S0169-555X(99)00099-9. → pages3, 20, 35, 36, 101, 105, 120Hickson, C.J.; Moore, J.G.; Calk, L.; and Metcalfe, P. (1995). Intraglacial volcanism in theWells Gray-Clearwater volcanic field, east-central British Columbia, Canada. CanadianJournal of Earth Sciences, 32(7):pages 838–851. ISSN 00084077 (ISSN).doi:10.1139/e95-070. → pages 52, 70Hickson, C.J.; Russell, J.; and Stasiuk, M.V. (1999). Volcanology of the 2350 B.P.Eruption of Mount Meager Volcanic Complex, British Columbia, Canada: implicationsfor Hazards from Eruptions in Topographically Complex Terrain. Bulletin ofVolcanology, 60(7):pages 489–507. ISSN 0258-8900. doi:10.1007/s004450050247. →pages 13, 19, 31Hicock, S.R. and Armstrong, J.E. (1983). 4 Pleistocene Formations in SouthwestBritish-Columbia Canada their Implications for Patterns of Sedimentation of PossibleSangamonian to Early Wisconsinan Age. Canadian Journal of Earth Sciences,20(8):pages 1232–1247. ISSN 0008-4077. doi:10.1139/e83-110. → pages43, 48, 59, 64, 65, 67Hildreth, W. (1996). Kulshan caldera: A Quaternary subglacial caldera in the NorthCascades, Washington. Geological Society of America Bulletin, 108(7):pages 786–793.192ISSN 00167606. doi:10.1130/0016-7606(1996)108〈0786:KCAQSC〉2.3.CO;2. → page21Hildreth, W. (2007). Quaternary magmatism in the Cascades; geologic perspectives. U.S.Geological Survey Professional Paper 1744, page 125. ISSN <null>.doi:http://pubs.usgs.gov/pp/pp1744/. → pages 3, 10, 13, 34, 52, 101, 153, 156Hooper, A.; O´feigsson, B.; Sigmundsson, F.; Lund, B.; Einarsson, P.; Geirsson, H.; andSturkell, E. (2011). Increased capture of magma in the crust promoted by ice-cap retreatin Iceland. Nature Geoscience, 4(11):pages 783–786. ISSN 1752-0894.doi:10.1038/ngeo1269. → pages 3, 152, 153, 166, 176Horswill, P.; Horton, A.; and Penman, A.D.M. (1976). Cambering and Valley Bulging inthe Gwash Valley at Empingham, Rutland. Philosophical Transactions of the RoyalSociety A: Mathematical, Physical and Engineering Sciences, 283(1315):pages427–462. ISSN 1364-503X. doi:10.1098/rsta.1976.0092. → page 143Ho¨skuldsson, A. and Sparks, R.S.J. (1997). Thermodynamics and fluid dynamics ofeffusive subglacial eruptions. Bulletin of Volcanology, 59(3):pages 219–230. ISSN0258-8900. doi:10.1007/s004450050187. → pages 35, 99, 100, 128, 131, 132, 162Hubbard, A.; Sugden, D.; Dugmore, A.; Norddahl, H.; and Pe´tursson, H.G. (2006). Amodelling insight into the Icelandic Last Glacial Maximum ice sheet. QuaternaryScience Reviews, 25(17-18):pages 2283–2296. ISSN 02773791.doi:10.1016/j.quascirev.2006.04.001. → page 101Hungerford, J.D.; Edwards, B.; Skilling, I.P.; and Cameron, B.I. (2014). Evolution of asubglacial basaltic lava flow field: Tennena volcanic center, Mount Edziza volcaniccomplex, British Columbia, Canada. Journal of Volcanology and Geothermal Research,272:pages 39–58. ISSN 03770273. doi:10.1016/j.jvolgeores.2013.09.012. → page 97Huscroft, C.A.; Ward, B.C.; Barendregt, R.W.; Jackson Jr., L.E.; and Opdyke, N.D.(2004). Pleistocene volcanic damming of Yukon River and the maximum age of theReid Glaciation, west-central Yukon. Canadian Journal of Earth Sciences, 41(2):pages151–164. ISSN 0008-4077. doi:10.1139/e03-098. → page 52Huybers, P. and Langmuir, C. (2009). Feedback between deglaciation, volcanism, andatmospheric CO2. Earth and Planetary Science Letters, 286(3-4):pages 479–491. ISSN0012821X. doi:10.1016/j.epsl.2009.07.014. → pages 3, 153Ingo´lfson, O´.; Norddahl, H.; and Haflidason, H. (2008). Rapid isostatic rebound insouthwestern Iceland at the end of the last glaciation. Boreas, 24(3):pages 245–259.ISSN 03009483. doi:10.1111/j.1502-3885.1995.tb00777.x. → page 178Jackson, L. and Clague, J.J. (1991). The Cordilleran Ice Sheet - 150 years of explorationand discovery. Geographie physique et Quaternaire, 45(3):pages 269–280. ISSN07057199. doi:http://www.erudit.org/revue/gpq/1991/v45/n3/032874ar.pdf. → pages1, 11, 38, 43, 50, 55, 56, 156193Jackson, L.; Nelson, F.; Huscroft, C.; Villeneuve, M.; Barendregt, R.; Storer, J.; and Ward,B. (2012). Pliocene and Pleistocene volcanic interaction with Cordilleran ice sheets,damming of the Yukon River and vertebrate Palaeontology, Fort Selkirk VolcanicGroup, west-central Yukon, Canada. Quaternary International, 260:pages 3–20. ISSN10406182. doi:10.1016/j.quaint.2011.08.033. → page 52Jackson, M.D.; Blundy, J.; and Sparks, R.S. (2018). Chemical differentiation, cold storageand remobilization of magma in the Earth’s crust. Nature, 564(7736):pages 405–409.ISSN 14764687. doi:10.1038/s41586-018-0746-2. → page 168Jakobsson, S.P. and Gudmundsson, M.T. (2008). Subglacial and intraglacial volcanicformations in Iceland. Jokull, 58:pages 179–196. ISSN 04490576. → pages3, 20, 22, 32, 34, 51, 74, 97James, T.S.; Clague, J.J.; Wang, K.; and Hutchinson, I. (2000). Postglacial rebound at thenorthern Cascadia subduction zone. Quaternary Science Reviews, 19(14-15):pages1527–1541. ISSN 02773791. doi:10.1016/S0277-3791(00)00076-7. → pages156, 157, 159Jellinek, A.M. and DePaolo, D.J. (2003). A model for the origin of large silicic magmachambers: Precursors of caldera-forming eruptions. Bulletin of Volcanology,65(5):pages 363–381. ISSN 02588900. doi:10.1007/s00445-003-0277-y. → pages147, 153, 154, 166, 169, 170Jellinek, A.M.; Manga, M.; and Saar, M.O. (2004). Did melting glaciers cause volcaniceruptions in eastern California? Probing the mechanics of dike formation. Journal ofGeophysical Research: Solid Earth, 109(B9):pages n/a–n/a. ISSN 01480227.doi:10.1029/2004JB002978. → pages4, 134, 144, 147, 150, 152, 153, 154, 158, 166, 169, 170, 176Jones, J. (1968). Intraglacial volcanoes of the Laugarvatn region, south-west Iceland–I.Quarterly Journal of the Geological Society, 124(1-4):pages 197–211. ISSN0370-291X. doi:10.1144/gsjgs.124.1.0197. → pages 3, 9, 36, 52, 74, 95, 103Jones, J.G. (1970). Intraglacial Volcanoes of the Laugarvatn Region, Southwest Iceland,II. Journal of Geology, 78(March):pages 127–140. → pages 3, 9, 36, 95, 103Jull, M. and McKenzie, D. (1996). The effect of deglaciation on mantle melting beneathIceland. Journal of Geophysical Research: Solid Earth, 101(B10):pages21,815–21,828. ISSN 01480227. doi:10.1029/96JB01308. → pages1, 4, 134, 144, 150, 152, 155, 166, 176, 178Kavanagh, J. and Sparks, R.J. (2011). Insights of dyke emplacement mechanics fromdetailed 3D dyke thickness datasets. Journal of the Geological Society, 168(4):pages965–978. ISSN 0016-7649. doi:10.1144/0016-76492010-137. → pages 153, 161Kavanagh, J.L. (2018). Mechanisms of Magma Transport in the Upper Crust—Dyking. InVolcanic and Igneous Plumbing Systems, pages 55–88. Elsevier. ISBN 9780128097496.doi:10.1016/B978-0-12-809749-6.00003-0. → pages 134, 146, 153, 161194Kavanagh, J.L.; Burns, A.J.; Hilmi Hazim, S.; Wood, E.P.; Martin, S.A.; Hignett, S.; andDennis, D.J. (2018). Challenging dyke ascent models using novel laboratoryexperiments: Implications for reinterpreting evidence of magma ascent and volcanism.Journal of Volcanology and Geothermal Research, 354:pages 87–101. ISSN 03770273.doi:10.1016/j.jvolgeores.2018.01.002. → pages 153, 161Kelemen, P.B.; Hirth, G.; Shimizu, N.; Spiegelman, M.; and Dick, H.J. (1997). A reviewof melt migration processes in the adiabatically upwelling mantle beneath oceanicspreading ridges. Philosophical Transactions of the Royal Society of London. Series A:Mathematical, Physical and Engineering Sciences, 355(1723):pages 283–318. ISSN1471-2962. doi:10.1098/rsta.1997.0010. → pages 155, 178Kelley, S. (2002). K-Ar and Ar-Ar Dating. Reviews in Mineralogy and Geochemistry,47(1):pages 785–818. ISSN 1529-6466. doi:10.2138/rmg.2002.47.17. → page 174Kelly, P.M. and Jones, P.D. (1996). The spatial response of the climate system to explosivevolcanic eruptions. International Journal of Climatology, 16(5):pages 537–550. ISSN0899-8418. doi:10.1002/(SICI)1097-0088(199605)16:5〈537::AID-JOC23〉3.0.CO;2-F.→ page 3Kelman, M.C. (2005). Glaciovolcanism at the Mount Cayley Volcanic Field, GaribaldiVolcanic Belt, Southwestern British Columbia. Ph.D. thesis, University of BritishColumbia. → pages 19, 22, 29, 30, 36, 38, 53, 104, 116Kelman, M.C.; Russell, J.; and Hickson, C.J. (2002a). Effusive intermediateglaciovolcanism in the Garibaldi Volcanic Belt, southwestern British Columbia, Canada.Geological Society, London, Special Publications,202(1):pages 195–211. ISSN 0305-8719. doi:10.1144/GSL.SP.2002.202.01.10.→ pages1, 2, 3, 4, 6, 9, 11, 19, 20, 22, 24, 29, 30, 34, 35, 36, 42, 53, 72, 74, 97, 100, 101, 103, 104, 114, 116, 117, 120, 122, 125,Kelman, M.C.; Russell, J.; and Hickson, C.J. (2002b). Glaciovolcanism at Ember Ridge,Mount Cayley volcanic field, southwestern British Columbia. Geological Survey ofCanada, Current Research, A15:page 7. doi:10.4095/213082. → pages22, 29, 74, 97, 99, 101, 103, 116, 117, 151Kerr, F.A. (1934). Glaciation in northern British Columbia. Proceedings and transactionsof the Royal Society of Canada, 28(4):pages 17–32. → page 55Kervyn, M.; Ernst, G.G.J.; van Wyk de Vries, B.; Mathieu, L.; and Jacobs, P. (2009).Volcano load control on dyke propagation and vent distribution: Insights from analoguemodeling. Journal of Geophysical Research, 114(B3):page B03,401. ISSN 0148-0227.doi:10.1029/2008JB005653. → pages 146, 147Kinakin, D. and Stead, D. (2005). Analysis of the distributions of stress in natural ridgeforms: Implications for the deformation mechanisms of rock slopes and the formationof sackung. Geomorphology, 65(1-2):pages 85–100. ISSN 0169555X.doi:10.1016/j.geomorph.2004.08.002. → pages 135, 144, 146, 147, 150195Kleman, J. (1994). Preservation of landforms under ice sheet and ice caps.Geomorphology, 9:pages 19–32. ISSN 0169555X.doi:10.1016/0169-555X(94)90028-0. → page 51Kleman, J.; Jansson, K.; De Angelis, H.; Stroeven, A.P.; Ha¨ttestrand, C.; Alm, G.; andGlasser, N. (2010). North American Ice Sheet build-up during the last glacial cycle,115-21kyr. Quaternary Science Reviews, 29(17-18):pages 2036–2051. ISSN 02773791.doi:10.1016/j.quascirev.2010.04.021. → page 51Kovanen, D.J. and Easterbrook, D.J. (2001). Late Pleistocene, post-Vashon, alpineglaciation of the Nooksack drainage, North Cascades, Washington. Bulletin of theGeological Society of America, 113(2):pages 274–288. ISSN 00167606.doi:10.1130/0016-7606(2001)113〈0274:LPPVAG〉2.0.CO;2. → page 11Kuiper, K.F.; Deino, A.; Hilgen, F.J.; Krijgsman, W.; Renne, P.R.; and Wijbrans, J.R.(2008). Synchronizing Rock Clocks of Earth History. Science, 320(5875):pages500–504. ISSN 0036-8075. doi:10.1126/science.1154339. → page 112Kusumoto, S.; Geshi, N.; and Gudmundsson, A. (2013). Aspect ratios and magmaoverpressures of non-feeder dikes observed in the Miyake-jima volcano (Japan), andfracture toughness of its upper part. Geophysical Research Letters, 40(6):pages1065–1068. ISSN 00948276. doi:10.1002/grl.50284. → pages121, 147, 149, 161, 162, 169Kutterolf, S.; Jegen, M.; Mitrovica, J.X.; Kwasnitschka, T.; Freundt, A.; and Huybers, P.J.(2013). A detection of Milankovitch frequencies in global volcanic activity. Geology,41(2):pages 227–230. ISSN 1943-2682. doi:10.1130/G33419.1. → pages4, 152, 153, 155, 158, 166Kutterolf, S.; Schindlbeck, J.; Jegen, M.; Freundt, A.; and Straub, S. (2019). Milankovitchfrequencies in tephra records at volcanic arcs: The relation of kyr-scale cyclic variationsin volcanism to global climate changes. Quaternary Science Reviews, 204:pages 1–16.ISSN 02773791. doi:10.1016/j.quascirev.2018.11.004. → page 166Lachowycz, S.M.; Pyle, D.M.; Gilbert, J.S.; Mather, T.A.; Mee, K.; Naranjo, J.A.; andHobbs, L.K. (2015). Glaciovolcanism at Volca´n Sollipulli, southern Chile: Lithofaciesanalysis and interpretation. Journal of Volcanology and Geothermal Research,303:pages 59–78. ISSN 03770273. doi:10.1016/j.jvolgeores.2015.06.021. → pages2, 42Lawrence, R.B.; Lee Armstrong, R.; and Berman, R.G. (1984). Garibaldi group volcanicrocks of the salal creek area, southwestern British Columbia: Alkaline lavas on thefringe of the predominantly calc-alkaline garibaldi (cascade) volcanic arc. Journal ofVolcanology and Geothermal Research, 21(3-4):pages 255–276. ISSN 03770273.doi:10.1016/0377-0273(84)90025-8. → pages13, 19, 22, 29, 31, 32, 40, 52, 53, 69, 72, 97196Layer, P.W.; Hall, C.M.; and York, D. (1987). The derivation of 40Ar/39Ar age spectra ofsingle grains of hornblende and biotite by laser step-heating. Geophysical ResearchLetters, 14(7):pages 757–760. ISSN 19448007. doi:10.1029/GL014i007p00757. →page 112Le Bas, M.; Le Maitre, R.; Streckeisen, a.; and Zanettin, B. (1986). A ChemicalClassificatron of Volcanic Rocks Based on the Total Alkali-Silica Diagram. Journal ofPetrology, 27(3):pages 745–750. ISSN 0022-3530. doi:10.1093/petrology/27.3.745. →pages 83, 86Le Corvec, N.; Muirhead, J.D.; and White, J.D. (2018). Shallow magma diversions duringexplosive diatreme-forming eruptions. Nature Communications, 9(1):pages 1–8. ISSN20411723. doi:10.1038/s41467-018-03865-x. → page 135Lescinsky, D.T. and Fink, J.H. (2000). Lava and ice interaction at stratovolcanoes: Use ofcharacteristic features to determine past glacial extents and future volcanic hazards.Journal of Geophysical Research: Solid Earth, 105(B10):pages 23,711–23,726. ISSN01480227. doi:10.1029/2000JB900214. → pages3, 11, 20, 21, 97, 104, 106, 109, 114, 115, 128, 175, 176Lescinsky, D.T. and Sisson, T.W. (1998). Ridge-forming, ice-bounded lava flows at MountRainier, Washington. Geology, 26(4):page 351. ISSN 0091-7613.doi:10.1130/0091-7613(1998)026〈0351:RFIBLF〉2.3.CO;2. → pages 20, 128Lisiecki, L.E. and Raymo, M.E. (2005). A Pliocene-Pleistocene stack of 57 globallydistributed benthic δ 18 O records. Paleoceanography, 20(1). ISSN 08838305.doi:10.1029/2004PA001071. → pages 3, 42, 44, 46, 50, 64, 68, 69, 169, 174Liu, L. and Zoback, M.D. (1992). The effect of topography on the state of stress in thecrust: Application to the site of the Cajon Pass Scientific Drilling Project. Journal ofGeophysical Research, 97(B4):page 5095. ISSN 0148-0227. doi:10.1029/91JB01355.→ page 147Livingstone, S.J.; Clark, C.D.; Woodward, J.; and Kingslake, J. (2013). Potentialsubglacial lake locations and meltwater drainage pathways beneath the Antarctic andGreenland ice sheets. The Cryosphere, 7(6):pages 1721–1740. ISSN 1994-0424.doi:10.5194/tc-7-1721-2013. → pages 43, 48Lodge, R.W. and Lescinsky, D.T. (2009). Fracture patterns at lava–ice contacts onKokostick Butte, OR, and Mazama Ridge, Mount Rainier, WA: Implications for flowemplacement and cooling histories. Journal of Volcanology and Geothermal Research,185(4):pages 298–310. ISSN 03770273. doi:10.1016/j.jvolgeores.2008.10.010. →pages 20, 24, 25, 97, 109, 114, 115Maccaferri, F.; Acocella, V.; and Rivalta, E. (2015). How the differential load induced bynormal fault scarps controls the distribution of monogenic volcanism. GeophysicalResearch Letters, 42:pages 7507–7512. doi:10.1002/2015GL065638. → page 134197MacDonald, G.A. (1968). Composition and origin of Hawaiian lavas. Studies involcanology: a memoir in honor of Howel Williams, 116:pages 477–522. → pages83, 86Maclennan, J.; Jull, M.; McKenzie, D.; Slater, L.; and Gro¨nvold, K. (2002). The linkbetween volcanism and deglaciation in Iceland. Geochemistry, Geophysics, Geosystems,3(11):pages 1–25. ISSN 15252027. doi:10.1029/2001GC000282. → pages1, 4, 134, 150, 152, 155, 178Major, J.J. and Newhall, C.G. (1989). Snow and ice perturbation during historical volcaniceruptions and the formation of lahars and floods - A global review. Bulletin ofVolcanology, 52(1):pages 1–27. ISSN 02588900. doi:10.1007/BF00641384. → pages2, 11, 100, 121, 127, 152, 176Makhnenko, R.Y. and Podladchikov, Y.Y. (2018). Experimental Poroviscoelasticity ofCommon Sedimentary Rocks. Journal of Geophysical Research: Solid Earth,123(9):pages 7586–7603. ISSN 21699356. doi:10.1029/2018JB015685. → page 139Martin, P.A.; Lea, D.W.; Rosenthal, Y.; Shackleton, N.J.; Sarnthein, M.; and Papenfuss, T.(2002). Quaternary deep sea temperature histories derived from benthic foraminiferalMg/Ca. Earth and Planetary Science Letters, 198(1-2):pages 193–209. ISSN0012821X. doi:10.1016/S0012-821X(02)00472-7. → page 43Martinson, D.G.; Pisias, N.G.; Hays, J.D.; Imbrie, J.; Moore, T.C.; and Shackleton, N.J.(1987). Age dating and the orbital theory of the ice ages: Development of a highresolution 0 to 300 000 year chronostratigr