UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effect of alternate stopbank alignments on the Waiho River, Westland, New Zealand : a microscale modelling… Beagley, Rosemary Patricia Jane 2017

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

Item Metadata

Download

Media
24-ubc_2018_february_beagley_rosemary.pdf [ 45.7MB ]
Metadata
JSON: 24-1.0358018.json
JSON-LD: 24-1.0358018-ld.json
RDF/XML (Pretty): 24-1.0358018-rdf.xml
RDF/JSON: 24-1.0358018-rdf.json
Turtle: 24-1.0358018-turtle.txt
N-Triples: 24-1.0358018-rdf-ntriples.txt
Original Record: 24-1.0358018-source.json
Full Text
24-1.0358018-fulltext.txt
Citation
24-1.0358018.ris

Full Text

Effect of alternate stopbankalignments on the Waiho River,Westland, New Zealand:A microscale modelling investigationbyRosemary Patricia Jane BeagleyB.Sc., The University of Melbourne, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Geography)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)November 2017c© Rosemary Patricia Jane Beagley 2017AbstractThe Waiho River in Westland, New Zealand has been rapidly aggrad-ing its bed as a result of lateral confinement by a stopbank network whichrestricts the river to 30% of its natural fan accommodation space. Theongoing aggradation has prompted the need to repeatedly raise the crestlevel of the stopbanks. This had led to the bed and stopbank elevationreaching unpreceded and dangerously high levels, putting the surroundingland, infrastructure and Franz Josef community at even greater risk thanbefore should the stopbanks fail. This thesis investigates an alternative so-lution to the current management practice. Using a microscale model ittests the response of an experimental Waiho River and fan to the removal ofthe Southern stopbank and replacement with two alternatives which allowthe river greater access to its Southern fan surface. In addition, the studyallowed for an exploration of several microscale modelling techniques.The results found that an experimental fan in a state of dynamic equilib-rium would not aggrade when confined as previously thought. Only whenthe fan was already aggrading did it continue to aggrade when confined. Inthis instance, when the confinement was removed it did not result in degra-dation to lower elevations. Aggradation continued, albeit at a reduced rate.This suggests that the Waiho was already in a state of aggradation prior tohuman interference, and that confinement exacerbated the rate. This resulthas implications for the future management of the Waiho. If the currentaggradation trend is to continue, then increasing stopbank crest height isnot a viable solution, however releasing the river to the South will reducethe rate of aggradation as well as the pressures on the Northern stopbankswhich protect the Franz Josef township. Effectively, this buys time for moredrastic action (i.e. relocation of the township) to be taken. In addition tothese results, the experiments found that measurement tools and model ma-terials used previously in other microscale models produced unreliable fanbehaviour and results. That they have failed in this study, motivates theneed for further investigation into the underlying principles of microscalemodelling and its practice.iiLay summarySince the 1980s the Waiho River in Westland, New Zealand has been rapidlyincreasing its bed elevation in response to confinement by stopbanks. Thesestopbanks are repeatedly built up to ensure the river does not overtop them.Both river bed and stopbank height have now reached unpreceded and dan-gerously high levels which puts the surrounding land, infrastructure andFranz Josef community at even greater risk than before should the stop-banks fail. This thesis investigates an alternate solution, using a microscalemodel to test the response of an experimental river to reduced confinement.The results indicate that reduced confinement does not prevent the bed el-evation from increasing, however it does slow the rate, as well as reducingthe pressures on the other remaining stopbanks. Effectively, this alternatesolution buys time for the planning and undertaking of more drastic actionsuch as town, highway, and farmland relocation, out of the Waiho’s way.iiiPrefaceThis thesis was completed by the author, under the guidance of a supervisorycommittee that included Brett Eaton (University of British Columbia) andTim Davies (University of Canterbury). The microscale model was designedbased off previous modelling experiments (Davies et al., 2003a, 2013), andconstructed with the support of Warwick Hill (Lincoln University) in theSoil and Water Laboratory at Lincoln University. All data collection andprocessing was completed by the author, with technical (ArcGIS) supportfrom Crile Doscher (Lincoln University). Both Brett Eaton and Tim Daviescontributed to edits in the thesis, however the author was solely responsiblefor its composition.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiLay summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . vAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Waiho River and Franz Josef township context . . . . . . . . 21.2 Method of research . . . . . . . . . . . . . . . . . . . . . . . . 51.2.1 Field observations and analysis . . . . . . . . . . . . . 71.2.2 Numerical modelling . . . . . . . . . . . . . . . . . . . 81.2.3 Physical modelling . . . . . . . . . . . . . . . . . . . . 81.2.4 Analogue models: microscale modelling . . . . . . . . 101.3 Thesis aims, objectives and structure . . . . . . . . . . . . . . 112 Alluvial fans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.1 Alluvial fan theory . . . . . . . . . . . . . . . . . . . . . . . . 142.1.1 Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.1.2 Processes . . . . . . . . . . . . . . . . . . . . . . . . . 182.1.3 Evolution . . . . . . . . . . . . . . . . . . . . . . . . . 202.2 The Waiho River and fan system . . . . . . . . . . . . . . . . 242.2.1 Geomorphic history . . . . . . . . . . . . . . . . . . . 292.2.2 Human interference . . . . . . . . . . . . . . . . . . . 303 Microscale modelling . . . . . . . . . . . . . . . . . . . . . . . . 38vTABLE OF CONTENTS3.1 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.1.1 Underlying principles . . . . . . . . . . . . . . . . . . 403.1.2 The justification for microscale modelling . . . . . . . 423.2 Practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.2.1 Model design . . . . . . . . . . . . . . . . . . . . . . . 463.2.2 Experimental procedure . . . . . . . . . . . . . . . . . 533.2.3 Measurement tools . . . . . . . . . . . . . . . . . . . . 543.3 The Waiho microscale model . . . . . . . . . . . . . . . . . . 563.3.1 Model construction . . . . . . . . . . . . . . . . . . . . 563.3.2 Experimental procedure . . . . . . . . . . . . . . . . . 593.3.3 Data collection and analysis . . . . . . . . . . . . . . . 614 Data analysis: steady input experiments . . . . . . . . . . . 694.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.2.1 S1: equilibrium fan . . . . . . . . . . . . . . . . . . . . 714.2.2 S2: equilibrium fan . . . . . . . . . . . . . . . . . . . . 744.2.3 S3: equilibrium fan . . . . . . . . . . . . . . . . . . . . 774.2.4 S4: aggrading fan . . . . . . . . . . . . . . . . . . . . . 814.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844.3.1 Importance of spatially dense data . . . . . . . . . . . 844.3.2 Effect of boundary surface roughness . . . . . . . . . . 865 Data analysis: variable input experiments . . . . . . . . . . 885.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 895.2.1 V1: small variability . . . . . . . . . . . . . . . . . . . 895.2.2 V2: large variability . . . . . . . . . . . . . . . . . . . 955.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 995.3.1 Fan behaviour . . . . . . . . . . . . . . . . . . . . . . 995.3.2 Fan surface state . . . . . . . . . . . . . . . . . . . . . 1006 Application to the Waiho River . . . . . . . . . . . . . . . . . 1026.1 Outcome of the alternate stopbank alignments . . . . . . . . 1026.2 Surface state of the Waiho fan . . . . . . . . . . . . . . . . . 1046.2.1 Was the Waiho in dynamic equilibrium? . . . . . . . . 1056.2.2 An aggrading Waiho scenario . . . . . . . . . . . . . . 1116.3 Implications for the Waiho . . . . . . . . . . . . . . . . . . . 1126.3.1 Recommendations . . . . . . . . . . . . . . . . . . . . 113viTABLE OF CONTENTS7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1157.1 Implications for microscale modelling . . . . . . . . . . . . . . 1167.2 Implications for the management of the Waiho River . . . . . 117Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127Appendix A: aerial imagery of the Waiho fan . . . . . . . . . 127Appendix B: digital vernier calliper elevation data . . . . . . 141viiList of Tables2.1 Current stopbank alignments . . . . . . . . . . . . . . . . . . 343.1 Experiment summary . . . . . . . . . . . . . . . . . . . . . . 603.2 Digital Vernier Calliper data . . . . . . . . . . . . . . . . . . 644.1 Steady input experiment summary . . . . . . . . . . . . . . . 704.2 Microscale model studies comparison . . . . . . . . . . . . . . 785.1 Variable input experiment summary . . . . . . . . . . . . . . 901 S1. Establishing dynamic equilibrium . . . . . . . . . . . . . 1422 S1. Current confinement elevation data . . . . . . . . . . . . 1433 S1. Current confinement elevation data continued . . . . . . . 1444 S1. Intermediate stopbank alignment . . . . . . . . . . . . . . 1455 S1. Intermediate stopbank alignment continued . . . . . . . . 1466 S1. Extreme stopbank alignment . . . . . . . . . . . . . . . . 1477 S1. Extreme stopbank alignment continued . . . . . . . . . . 148viiiList of Figures1.1 Google earth imagery of the Waiho River . . . . . . . . . . . 31.2 Topographic cross sectional profile of the upper Waiho Fan . 41.3 Gravel deposits from the 2016 Waiho flood . . . . . . . . . . 51.4 Stopbank alignments to be tested . . . . . . . . . . . . . . . . 122.1 Turkey flat alluvial fan and topographic map . . . . . . . . . 152.2 Fan components . . . . . . . . . . . . . . . . . . . . . . . . . 162.3 Fanhead trenching . . . . . . . . . . . . . . . . . . . . . . . . 202.4 Experimental fan evolution . . . . . . . . . . . . . . . . . . . 212.5 Landscape view of Waiho Fan . . . . . . . . . . . . . . . . . . 242.6 Waiho flood 2016 . . . . . . . . . . . . . . . . . . . . . . . . . 262.7 Franz Josef Glacier . . . . . . . . . . . . . . . . . . . . . . . . 272.8 Waiho River upper valley . . . . . . . . . . . . . . . . . . . . 282.9 1948 and 1997 aerial imagery . . . . . . . . . . . . . . . . . . 322.10 Failed 1980 stopbank . . . . . . . . . . . . . . . . . . . . . . . 332.11 True right stopbank downstream of SH6 bridge . . . . . . . . 352.12 True right lower terrace stopbank . . . . . . . . . . . . . . . . 352.13 True left stopbank upstream of SH6 bridge . . . . . . . . . . 362.14 True left spur groynes upstream of SH6 bridge . . . . . . . . 362.15 True left stopbank downstream of SH6 bridge . . . . . . . . 373.1 Labelled microscale model . . . . . . . . . . . . . . . . . . . . 483.2 Tinker sediment feeder . . . . . . . . . . . . . . . . . . . . . . 503.3 Labelled microscale model for this study . . . . . . . . . . . . 563.4 Labelled microscale model for this study . . . . . . . . . . . . 583.5 The two stopbank alignments used . . . . . . . . . . . . . . . 593.6 Digital Vernier Callipers . . . . . . . . . . . . . . . . . . . . . 623.7 Sections of model fan . . . . . . . . . . . . . . . . . . . . . . . 633.8 Aranz medical laser scanner . . . . . . . . . . . . . . . . . . . 653.9 ArcGIS data manipulation steps . . . . . . . . . . . . . . . . 663.10 Time lapse of change after stopbank removal . . . . . . . . . 68ixLIST OF FIGURES4.1 Time lapse of fan behaviour . . . . . . . . . . . . . . . . . . . 724.2 S1: net elevation change . . . . . . . . . . . . . . . . . . . . . 734.3 Increasing surface roughness . . . . . . . . . . . . . . . . . . . 754.4 S2. Establishing dynamic equilibrium . . . . . . . . . . . . . 764.5 S2. Establishing dynamic equilibrium . . . . . . . . . . . . . 774.6 S3. Net elevation change . . . . . . . . . . . . . . . . . . . . . 784.7 S3. DoD image showing aggradation . . . . . . . . . . . . . . 804.8 S3. Net elevation change (confined) . . . . . . . . . . . . . . . 814.9 S4. Net elevation change (confined) . . . . . . . . . . . . . . . 824.10 S4. Aggradation in confined vs. unconfined fan . . . . . . . . 835.1 V1a. net elevation change . . . . . . . . . . . . . . . . . . . . 915.2 Fanhead trenching . . . . . . . . . . . . . . . . . . . . . . . . 925.3 Fan catch up behaviour . . . . . . . . . . . . . . . . . . . . . 935.4 V1b. Net elevation change . . . . . . . . . . . . . . . . . . . . 945.5 V1b. Net elevation change . . . . . . . . . . . . . . . . . . . . 955.6 V2a. Attempting dynamic equilibrium . . . . . . . . . . . . . 965.7 V2a. 2 hour interval of change . . . . . . . . . . . . . . . . . 975.8 V2b. Net elevation change . . . . . . . . . . . . . . . . . . . . 986.1 Mean bed level data at Waiho River bridge . . . . . . . . . . 1036.2 Transport reach of the Fox River . . . . . . . . . . . . . . . . 1101 1948 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1282 1964 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1293 1973 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1304 1979 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1315 1980 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1326 1982 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1337 1985 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1348 1990 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1359 1994 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13610 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13711 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13812 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13913 2013 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140xAcknowledgementsFirst and foremost, a huge thank you to Brett Eaton and Tim Davies forgiving me the opportunity to work on a project I was so interested in, andproviding endless support and thoughts to my endeavour. Brett has en-couraged and pushed me to pursue my ideas throughout the last few yearswithout which I doubt I would have developed the knowledge and skills Ihave today. Tim’s knowledge of the Waiho, alluvial fans and microscalemodelling, and pursuit of a topic he’s obviously passionate about has beenmotivational. I feel like I always come away from our chats having learntsomething new, and with plenty to ponder. It has been amazing getting towork with both Brett and Tim as I’ve delved into the realms of microscalemodelling, alluvial fans, and the problem of the rapidly aggrading Waiho.To Crile and Warwick. Thank you so much for all the help you gave mewhilst I was working in the Soil and Water Laboratory at Lincoln Univer-sity. I don’t know what I would have done without Warwick’s expertise inmodel construction and problem solving (for the many hiccups and ideasthat turned up throughout the four months), it was an absolute pleasureworking in the lab with him. Also, I would have been lost in the GIS worldwithout Crile. He re-opened my eyes to working in ArcGIS.I have also felt so supported and encouraged this past year by both familyand friends. It would be hard to thank everyone individually, however Iwould like to make special mention to a particular few.To Tegan and Jessie, their bubbling enthusiasm both in Canada and NZhas been incredible. They’ve been there every step of the way, for outdooradventures when I could spare the time, delicious dinners or vegan treatsand coffee to give me a break from modelling or writing, and motivationaltexts or phone conversations when I needed them. I couldn’t have asked fortwo better friends. Thank you.xiAcknowledgementsCol. For for all the little things she has done unasked (and continues todo) to help make these past few years easier. It has all been so appreciated.And of course my parents, who not only helped proof read this thesis butwere (and are) always only a phone call or text message away when I’veneeded to talk or hear words of encouragement. (Sorry I don’t make it backto Australia as often as we’d all like).Finally, Linden. I’m so grateful to have had your suport from the oppositeside of the world (and island) for the last two and a half years. Thank youtoo for the reminders to put down the books, tuck away the thoughts andget outside for a break - it’s undoubtedly the best medicine to clear onesmind and refresh. I’m looking forward to all the adventures we can havenow that I’m finished and can come back home to the best West Coast.xiiDedicationThis thesis is dedicated to future microscale modellers, in the hope thatbeyond the practical reasons, they too may enjoy the “irresistible fascinationin watching a small, controlled landscape evolve, creating dynamic patternsthat seem to come from out of nowhere” (Paola et al., 2009).xiiiChapter 1IntroductionStopbanks, also known as flood control banks, levees, or longitudinal train-ing banks, are anthropogenically constructed walls which confine a river to arestricted width, and prevent it from utilizing the surrounding area (Daviesand McSaveney, 2006; Warburton, 1996). Not only do they act as a barrierbetween a river and surrounding land, they are also designed to increasethe mean flow depth, and therefore mean bed shear stress (Davies and Mc-Saveney, 2006; Warburton, 1996). This theoretically serves to increase thetransport capacity of the river and therefore prevent aggradation of the riverbed (Warburton, 1996; Acheson, 1968; Henderson, 1966).Stopbanks have been a common management practice for braided gravel-bed rivers in New Zealand (Grant, 1948; Acheson, 1968; Nevins, 1969; Daviesand McSaveney, 2006). This type of river is known for its highly dynamicnature which results in an everchanging network of intertwining channelsaround bars and islands (Me`tivier and Barrier, 2012; Bristow and Best,1993). The ability of these rivers to continually alter their channel mor-phology presents problems where humans and rivers interact (Piegay et al.,2006). Infrastructure and land use within and around the channel are at riskfrom any changes in the river pattern. This has led to the use of stopbanksas a means of restricting rivers to a set path, and protecting the surround-ing land and infrastructure from changes in channel location and pattern, aswell as from high flow events (Davies and Lee, 1988; Davies and McSaveney,2001; Fleming, 2002).However, there are a growing number of cases in New Zealand wherethe use of stopbanks has had the opposite effect to that intended (Daviesand McSaveney, 2006; Davies and Lee, 1988). Rather than reduce the riskposed by this type of river, the stopbanks have led to increased aggradationwithin the confined reach (Davies and McSaveney, 2006). Subsequently,stopbanks are then built up to restore the designed flood cross-sectionalarea and prevent overtopping, which would put the surrounding area atrisk from the flooding and/or avulsion (dramatic channel course change)11.1. Waiho River and Franz Josef township contexthazards which they had originally been installed to mitigate and control(Davies and Lee, 1988). In these cases, the original management strategyhas not eliminated the problem, and the ongoing addition to stopbank crestheight is only delaying its inevitable failure.A prominent example of this, and one which this study will be focussingon, is the Waiho River Franz Josef situation.1.1 Waiho River and Franz Josef townshipcontextThe Waiho River flows out of the steep mountains west of the main divideof the Southern Alps onto and across the foreland plains of the West Coastregion of the South Island to the Tasman Sea some 15km away (Davies et al.,2003a) (Figure 1.1). At the point of emergence from the mountains, an allu-vial fan has formed that extends as a valley train to the coast (Davies et al.,2003a). Alluvial fans are formed by the deposition of sediment by a riveras it exits a confined valley onto open plains (Whitehouse and McSaveney,1990; Thornbury, 1954). Sediment is distributed across the fan surface viaa network of channels that take on the characteristic of braided gravel-bedrivers (Rachocki, 1981). This is the case for the Waiho River and Fan.At present, the upper few km of the Waiho River is restricted to abouta third of its original fan surface (Davies and McSaveney, 2006). In theearly 1900s, stopbanks were installed along the Waiho River to protect thesurrounding land and infrastructure from flooding and minor aggradation(Davies et al., 2003a). However, over the last few decades, the upper fanwhere the Waiho River is confined has experienced rapid rates of aggradation(currently 300 mm a-1), which has prompted the need to repeatedly increasethe crest level of the stopbanks (Davies et al., 2003a; Davies and McSaveney,2006). This had led to the bed and stopbank elevation reaching unprecededand dangerously high levels, putting the surrounding land, infrastructureand Franz Josef community at even greater risk than before should thestopbanks fail (Figure 1.2) (Davies, 1997; Davies and McSaveney, 2001).21.1. Waiho River and Franz Josef township context Figure 1.1: Google Earth image of the Waiho River and fan from sourceto sea. Insert shows location in New Zealand.As the mean rate of sediment supply is believed to have changed littlein the last few thousand years (Davies et al., 2003a), and the position ofthe fan toe has been fixed in position at a constant base level as a resultof the stabilization of sea level 6000 years (Davies and McSaveney, 2001),it is believed that the basic cause of the aggradation is the confinement ofthe fan width (Davies and McSaveney, 2001). Using a 1:2000 microscalemodel Davies et al (2003a) replicated this situation; and with no changeto sediment supply, discharge or base level, the installation of the currentstopbank alignment resulted in rapid aggradation of the river bed in theconfined reach; with a spatial distribution of relative aggradation rates thatcorresponded to field data.Since the 1990s, corresponding to the increasing elevation of the riverbed, the severity of flood events has increased (Davies et al., 2003a). The31.1. Waiho River and Franz Josef township context Figure 1.2: A topographic cross sectional profile drawn across the Waihofan from the Lavender lobe in the SW to the edge of the Tartare fan, NE ofFranz Josef town using LiDAR data. The Waiho River bed is considerablyhigher than the land to either side of it. Figure taken from Langridge et al.,2016increasing bed elevation means that not only has the size of flood required toovertop the stopbanks been reduced, but the river bed elevation is now sig-nificantly higher than that of the surrounding land, as is the height of floodpeaks (Davies, 1997). Also, despite the increased height of the stopbanks,it would appear that during high flow events the Waiho River is still ableto overtop its banks and inundate the surrounding low lying land includingthe lower township area of Franz Josef (Davies, 1997). An example of thisis the flooding of the Heartland hotel The Mueller during a 2016 flood event(Aulakh and Mills, 2016) (Figure 1.3).In addition to this increasing river bed elevation and flood risk, recentriver behaviour during flood stages has begun to indicate an avulsion intothe Tatare River. Building upon the previous modelling done by Davies etal (2003a), a second microscale model was constructed to investigate if theincreasing rates of aggradation would result in such an avulsion, and theeffect on the township (Davies et al., 2013). Replicating the methodologyof Davies et al (2003a), ongoing aggradation resulted in an avulsion fromthe Waiho River into the Tatare River. This was followed by knickpointrecession of the junction between the two rivers upstream towards the FranzJosef township, placing it at greater risk (Davies et al., 2013).41.2. Method of research Figure 1.3: Gravel deposits around the Heartland hotel (the Mueller) leftbehind by the 2016 flood. Photo by Rob PieperThe consequences of an avulsion and the upstream knickpoint recession,the increasing flood risks, and the rising potential of stopbank failures haveprompted the need to find a solution to the rapidly aggrading river bed.However, this is not straightforward.1.2 Method of researchThe Waiho-Franz Josef situation presents a difficult scenario to study for anumber of reasons• Large spatial scaleThe area of confinement and aggradation is substantial. It extendsfrom the junction of the Callery and Waiho rivers down to an ancientterminal moraine (ca. 11,000 years B.P) known as the Waiho Loop(Davies and McSaveney, 2001). This area is approximately 4 km long,with a width of 1.3 km at its widest. The width of the entire unconfined51.2. Method of researchfan in this section is 4 km (Davies and McSaveney, 2001). This is alarge expanse of area to monitor regularly or model.• Long temporal scaleThe Waiho River is a braided gravel bed river (Davies and McSaveney,2001). It has a high sediment load made up by a significant fractionof bedload (Davies et al., 2003a). Bedload plays an important role ingravel-bed rivers. Comprising primarily the coarser particles, bedloadhas been found to account for a large fraction of the total mass trans-ported in these rivers (Malverti et al., 2008). The evolution of a riverbed through bedload transport operates at timescale from 10-1 to 104years (Malverti et al., 2008). Therefore, morphologic change cannotbe measured over a period of days, it can take months, years, or evenhundreds of years. This a considerable amount of time when consid-ering the human lifespan, and too long for people to directly perceiveand measure the river dynamics (Malverti et al., 2008).• Difficulties associated with monitoring a highly dynamic and braidedsystemBraided gravel bed rivers are highly dynamic systems, with a bed thatis constantly evolving, as the network of channels, bars, and islandschanges (Bertoldi et al., 2009; Schumm, 1985). These changes oftenoccur rapidly during periods of high flow (Davies and McSaveney,2006; Piegay et al., 2006). During these events, taking measurementsand observation of the processes and changes is near impossible, notto mention dangerous (Me`tivier and Barrier, 2012; Bristow and Best,1993). In addition to this, braiding makes measurement of sedimenttransport and water flow difficult, as there are too many channels (thatare regularly changing course) to measure (Ashmore, 1991; Davies,1987).• Lack of historical data availableDue to the difficulties discussed above there is a severe lack of historicaldata for bedload transport and water flow rates of the Waiho River(Davies and McSaveney, 2001). Additionally, aerial photography onlyreaches back to 1948 (Davies and McSaveney, 2001). The only recordof the river and fan behaviour prior to this are some sketches from the1850s and photos from 1870 (Davies and McSaveney, 2001).• Degree of urgency required for a solution61.2. Method of researchA solution is required urgently. The west coast of the South Island issubject to frequent periods of intense rainfall (Davies and McSaveney,2001). The head waters of the Waiho catchment receive an averageannual precipitation of 11,000 mm (McSaveney and Davies, 1998),with as high as 14,000 mm being recorded (Henderson and Thomp-son, 1999). 3-day rainfalls up to 750 mm have also been recorded.This means that the likelihood of large floods is high, as are the conse-quences of stopbank overtopping or failure for the neighbouring town-ship of Franz Josef Glacier (Davies, 1997).With these difficulties in mind, many of the common alternatives suchas field and analytical studies appeared impractical (Davies and Lee, 1988),and/or beyond the scope of the present Masters thesis. Without these stud-ies, real time (and historical) data on the sediment transport capacity, riverchannel pattern and location, and river hydrology are not available for thecalibration of many types of numerical and physical models.1.2.1 Field observations and analysisFieldwork has a long history in geomorphic research. It provides a directway to access, visualize and understand natural systems. In addition, his-torical data are invaluable for showcasing morphological change over time,as well as for the calibration of both numerical and physical model tech-niques (Popescu, 2014; Peakall et al., 1996). However, limitations exist interms of spatial and temporal scales that can be monitored and measured.These become apparent when considering field observation and analysis asa method of study for the Waiho River.• The temporal scale over which the Waiho River and fan has beenchanging and will continue to change, is too long to observe. There isjust not the time available to spend monitoring this system, especiallyunder the timeline of a Master of Science program and the degree ofurgency for a solution.• The spatial extent of the aggrading area. The stopbanked area andtotal fan accommodation space is too large to be monitored regularlyin detail. In addition to this, its highly dynamic nature and changesoccurring during high flows makes taking measurements and makingobservations difficult and dangerous.• There is lack of historical data. Current investigations would be start-ing from now without any detailed historical perspective.71.2. Method of research1.2.2 Numerical modellingNumerical modelling is the simulation of a system and/or component on acomputer (Popescu, 2014). It involves the quantification of a system andits properties through mathematical equations that can be either empiricalor derived from basic principles. These equations as well as the bound-ary conditions and any other input data necessary, are then translated intocomputer form using a range of numerical approaches (Popescu, 2014). Thistype of modelling requires a sound understanding of the system being mod-elled to ensure all the relevant physical processes and their mathematicalrepresentations, are included as well as real data from the prototype system(Popescu, 2014). Numerical modelling has proven to be an invaluable toolfor scientists, promoting significant advances in the understanding of theinterrelationships between sediment production, transport and depositionwithin dynamic fluvial environments (Peakall et al., 1996). Not limited tothe geomorphic discipline, its use extends from climate and weather predic-tions and forecasts (McKendry, 1992; Trenberth et al., 1998) to understand-ing tectonics (Hoink et al., 2013; Lenardi et al., 2003). However, it presentssome difficulties when considered for the Waiho-Franz Josef situation.• Numerical simulation models need to be calibrated to ensure that theyare representing what they’re supposed to be. The lack of historicaldata and the difficulties surrounding present day data collection, meanthat there is insufficient data to create an accurate numerical modelof the Waiho River.• In addition, the time, money and tools required to gather present-daydata are beyond the scope of this project.Whilst a numerically simulated model is a viable method for the Waiho-Franz Josef situation, without any data or means to collect data on waterflow rates or sediment transport and supply, it is not a suitable method forthis project.1.2.3 Physical modellingPhysical modelling is the physical replication of components of a system, orin some cases the entire system, in a laboratory (Mosley and Zimpfer, 1978).Physical models allow for sediment-flow interactions and bed morphology tobe reproduced and visualized under clear, safe and controlled conditions(Peakall et al., 1996). In addition to this they provide the option for dif-ferent temporal and spatial scales, depending on the type of model chosen81.2. Method of research(Peakall et al., 1996). There are three main types of models; one-to-one(unscaled reality; Chorley, 1967), scaled, and analogue (Peakall et al., 1996;Mosley and Zimpfer, 1978). All three have been used interchangeably bygeomorphologists to study form, process and evolution in the geomorphiclandscape (Peakall et al., 1996). In the case of the Waiho fan and river,where large time and spatial scales are required to represent the dynamics,behaviour and evolution of the system, and there is a lack of historical data,both one-to-one and scaled physical model types are impractical. However,analogue models provide a feasible option.• A one-to-one model has not been scaled down, it is the same sizeas its prototype. These models focus on a particular component ofa system, and therefore focus on the smallest scales of geomorphicinterest (Peakall et al., 1996). For example, hydraulic flumes havelong been used to study bedform generation in both sands and gravels(Peakall et al., 1996). Thus, there exists a limitation on the spatialextent of what a one-to-one model can replicate. Not only would itbe impossible to build a one to one model of the Waiho Franz Josefsituation in a laboratory, but there is a lack of sediment and waterflow data available to calibrate it with.• Scaled models are a scaled-down version of the prototype, with at-tempts made to replicate the flow and sediment conditions using di-mensionless values (Peakall et al., 1996; Malverti et al., 2008). In thecase of alluvial river systems dominated by bedload transport, thesevalues include the Froude, Reynolds and Shields numbers, relativedensity, and ratio of flow depth to grain size (Malverti et al., 2008).There are several types, the main two being Froude and distortedscaled models (Mosley and Zimpfer, 1978). Whilst this type of mod-elling requires less space and time than a one-to-one model, the sizeof model required for the Waiho-Franz Josef situation with accuratereplication of the flow and sediment conditions is too big for the scopeof this thesis due to laboratory space and finances. Warburton ( 1996) estimated the required size of a Froude scaled model to be of theorder of 50 m by 50 m. In addition, as with the other modelling types,there are no historical or current sediment or flow data to calibratethe model.91.2. Method of research1.2.4 Analogue models: microscale modellingAnalogue models are physical models, in which no attempt to replicate flowand sediment conditions has been made, and there is some degree of verticaldistortion (Mosley and Zimpfer, 1978; Malverti et al., 2008; Gaines andMaynord, 2001). They are often used to create scaled laboratory systemsthat are not based on a particular prototype (Peakall et al., 1996). Insteadthey are used to study form, process and evolution, as entities in themselves(Peakall et al., 1996).However, microscale models, a sub-branch of analogue models, can beused for case studies as long as the model reproduces some known responseor behaviour of the prototype (Gaines and Smith, 2002). This involves thereplication of geometric boundaries and forms, but relaxation of dynamicsimilarity criteria (Gaines and Maynord, 2001; Gaines and Smith, 2002;Maynord, 2006).• These models do not have a requirement for large of amounts of his-torical and present day data from the system they are replicating• They can handle large temporal and spatial scales• They are not affected by the difficulties associated with access andtaking measurements to collect data• They can provide a result within months.With these advantages in mind, microscale modelling appears to be aviable method of investigation for finding a solution to the Waiho-FranzJosef situation, and falls within the scope of this thesis. The model canbe built to include the entire fan area, without requiring large amounts ofdata, availability of which is limited. It will reduce the time required forfan evolution and response to a more manageable timeframe, which benefitsthis study, as a solution is required sooner rather than later; and it willprovide a controlled platform where fan behaviour can be directly observedand monitored.In addition, microscale models have been used previously to study theWaiho-Franz Josef situation. Davies et al. (2003a; 2013) constructed modelsof the area to investigate in the first instance, the cause of the aggradation,and in the second, the effect of ongoing aggradation and subsequent avulsioninto the Tatare River. These models all utilized the sandtray at the Soil and101.3. Thesis aims, objectives and structureWater Laboratory at Lincoln University, therefore the equipment requiredis already in place.1.3 Thesis aims, objectives and structureThe primary aim of this project is to investigate the effects of alternativestopbank alignments on the bed surface level of the Waiho River on theupper part of its fan using a microscale model (Figure 1.4). The effects ofthe two stopbank alignments investigated would then be used to determinethe future viability of the surrounding roads and infrastructure, and could beincorporated into the management strategy for the river. In addition to this,it would provide the opportunity to build upon the investigations by Davieset al. (2003a; 2013), who showed that stopbanks induced both increasedaggradation rates and an avulsion in their 1:2000 and 1:5000 microscalemodels of the Waiho River.To achieve this aim within the scope of the thesis the following objectiveswere undertaken:• Establish the long-term steady state of the fanhead bed level corre-sponding to alternative stopbank alignments.• Determine which alignment provides the better outcome for the sur-rounding land and infrastructure, and community of the Franz JosefTownship.A secondary aim was developed during the research phase of this thesis.This involved the exploration of the techniques used in microscale modelling.Microscale modelling is a relatively new type of physical modelling, and asa result the methodology and techniques in the literature have had littleexperimental assessment and/or discussion. As the experiments in this studyprogressed, problems with boundary roughness, input conditions, and thetools used to measure change occurred. Solving these problems developeda better understanding of microscale modelling and what is required for asuccessful model and experiment.111.3. Thesis aims, objectives and structure Figure 1.4: Google earth image of the Waiho River and fan with the 2stopbank alignments to be tested drawn in.This thesis is structured as follows.• Chapter two will review the literature available on alluvial fan form,process and evolution, before discussing the Waiho River and fan sys-tem.• Chapter three will then explore microscale modelling theory and prac-tice, followed by an outline of the Waiho microscale model design andconstruction, as well as experimental design.• Chapters four and five will discuss the experimental results and whatthey mean for microscale modelling. Chapter four will focus on thesteady input experiments, whilst five will focus on the variable inputexperiments.121.3. Thesis aims, objectives and structure• Chapter six will apply the results to the Waiho-Franz Josef situation.• Chapter seven is the conclusion. This will include implications andrecommendations for the Waiho River and fan system, as well as op-portunities for future research.13Chapter 2Alluvial fansFor the purpose of clarity and understanding it is important to outline theunderlying alluvial fan theory before discussing the study site itself. There-fore this chapter will first address what an alluvial fan is, the processes inplay and its evolution, before discussing the Waiho River and fan system.2.1 Alluvial fan theoryAlluvial fans are depositional landforms (Whitehouse and McSaveney, 1990)which are found in a diverse range of climatic and tectonic regimes (Clarke,2015; Harvey et al., 2005). They occur where there is a high sediment supply,and a lowland area upon which the sediment can be deposited (Schumm,1977) (Figure 2.1). Alluvial fans play an important role in fluvial systemsby acting as storage sites that moderate the transfer of sediment from theupper, sediment-producing part of the catchment to the lower, transitionaryand depositional region (Clarke, 2015).There are two types of alluvial fans and these are differentiated by theirdominant primary processes and resulting behaviour (Blair and McPherson,1994a). The first of these is the debris-flow fan, which forms by the forceof gravity acting directly on the sediment-water mixture in a stream andtransporting it. The second of these is the fluvially-formed fan, formed bythe transport of sediment by water (Blair and McPherson, 1994a,b). Thisstudy will focus on fluvial alluvial fans, and henceforth these will just becalled alluvial fans.142.1. Alluvial fan theory Figure 2.1: The Turkey flat alluvial fan formed by the Jordan Stream asit exits the Black mountain range into the Waimakariri River valley. Toetrimming by the Waimakariri River, braided channels, and depositional lobeson this beautiful fan specimen are clearly observable in the picture; whilstthe topographic map (right) provides a birds eye view of the symmetricalradial shape.Alluvial fans have been defined in different ways by geomorphologists us-ing emphasis on form, process, and evolution to differentiate them (Rachocki,1981). A simple definition may focus only on the form of the fan and sur-rounding landforms, describing it as:A stream deposit whose surface forms a segment of cone that radiatesdownslope from the point where the stream channel emerges from a moun-tainous area (Bull, 1964).In this case, the fan is described as a cone. In the literature, this termhas been used interchangeably with fan to describe alluvial fans (Rachocki,1981). Henceforth, this thesis will only use fan.In a more complex definition which includes a more process-based focus,Whitehouse and McSaveney (1990) define alluvial fans by the exit of flowfrom a steep confined channel to a more gently sloping surface where thechannel/s are no longer constrained to a fixed position and are free to spreadout. The decrease in stream gradient and increase in channel width which152.1. Alluvial fan theoryusually accompanies it, decrease the stream power per unit flow width avail-able to transport sediment and the excess sediment is deposited (Whitehouseand McSaveney, 1990).2.1.1 FormAn alluvial fan forms in a characteristic semicircular shape, which radiatesfrom a singular upstream point (Rachocki, 1981). It consists of three mainaspects:1. A fan apex, which is the highest point of the fan where the riveremerges from the mountains.2. A distributional channel network across the surface, and3. Fan lobes in the lower section which indicate the laterally migratingphases of deposition that result in toe progradation (figure 2.2; Blairand Mcpherson, 1994b) Apex Depositional channel network Lobes Figure 2.2: Sheet flow and channelized flows in an experimental fan (im-ages are from alluvial fan experiments run by Lucy Clarke at the Universityof Exeter Sediment Research Facility (Clarke, 2015))162.1. Alluvial fan theoryThe overall size, shape and slope of the fan reflect the inputs of water andsediment, determined by the upper drainage basin relief, size and geology,as well as the available accommodation space (Harvey et al., 2005; Schummet al., 1987; Kochel, 1990)Fan size, shape and slopeSeveral studies have indicated that quantitative relations exist betweenthe size, shape and slope of an alluvial fan and the relief, size and geologyof its upper drainage basin (Eckis, 1928; Bull, 1964; Denny, 1965; Ryder,1971).Fan size can be related to the size of its drainage basin (Denny, 1965; Bull,1964). The larger a drainage basin, the greater its supply of sediment in agiven region. Therefore, there is more sediment to be deposited, producinga larger fan. Bull’s (1964) studies of the alluvial fans in Fresno, Califor-nia, indicated a strong relationship between these two elements. He alsofound that the geology of the drainage basin played an important role infan size. Sediments from basins composed primarily of sandstone producedsmaller fans, than those of mudstone and shale (Schumm, 1977). Mudstoneand shale are highly erodible, and therefore produce more sediment thansandstone (Schumm, 1977).Fan shape, in terms of its profile concavity, can also be related to fansize. The profile is the topographic profile of a line drawn down the fansurface from the apex to the toe (Bull, 1964). Alluvial fans have a slightlyconcave long profiles (and convex cross-profiles; Clarke, 2015). Eckis (1928)observed that larger fans which have a correspondingly gentler slope have aless concave profile, compared to smaller fans with a steeper slope.Fan slope in a given region can be inversely related to the drainage basinsize and sequently fan volume (Bull, 1964). As basin area and therefore fanvolume increase, fan slope decreases. Thus, the larger a fan, the gentler itsoverall slope. Additionally Ryder (1971), in a study of fans located in BritishColumbia, found that steeper average relief of a drainage basin correspondedto a steeper fan gradient.172.1. Alluvial fan theory2.1.2 ProcessesAlluvial fans are depositional landforms (Clarke, 2015). They are deposi-tional because as the channel exits the mountainous terrain and flows outover open and often flat landscape, transport capacity declines, and sedi-ment is deposited (Clarke, 2015). This process occurs via a distributionalchannel network which radiates out from the apex of the fan, with individ-ual channels migrating laterally across the fan surface over time (Blair andMcPherson, 1994a).The distributional channel network has been found to share the charac-teristics of a braided river (Rachocki, 1981; Davies, 1997). Like a braidedriver, it is highly dynamic and characterised by a braided channel pattern(Bertoldi et al., 2009). This pattern is associated with wide and shallowchannels dissected by bars and/or islands, as well as low channel stabilitywhich results in lateral channel migration and flow diversion processes suchas bifurcation, crevasse splays and avulsion (Blair and McPherson, 1994a;Me`tivier and Barrier, 2012; Kleinhans et al., 2013).In addition, alluvial fans exhibit incisional, fan head trenching, and chan-nel backfilling behaviours (Blair and McPherson, 1994a), as well as sheetflows, which occur where flow becomes unchannelized (Hooke, 1967; Blairand McPherson, 1994b). Zarn and Davies (1994) noted a tendency for alter-nate build-up of different sides of the fan and toe progradation, accompaniedby channel migration which they termed the “windscreen-wiper effect”.At this point it is important to note that not all channels emerging frommountainous terrain will form alluvial fans (Blair and McPherson, 1994b).Bedload-dominated rivers may form instead. In these instances, the riveris capable of maintaining its transport capacity and channel banks withoutforming a fan (Blair and McPherson, 1994b).Flow diversion processesThe braided channel pattern on alluvial fans reflects three flow diversionprocesses: bifurcation, crevasse splays and avulsions (Bryant et al., 1995;Clarke et al., 2010). These three processes are the result of local deposition.Bifurcation refers to the points in the channel network where a single channelsplits into two downstream branches (Kleinhans et al., 2013); whilst crevassesplays occur when small channels break their natural boundaries depositingsediment in the surrounding area (Bryant et al., 1995). These two processes182.1. Alluvial fan theorycan also induce the gradual lateral migration of the channel network, whichallows for the distribution of sediment across the whole of the fan surfaceover time (Blair and McPherson, 1994a).A much larger shift in the network location can be induced by avulsion(Kleinhans et al., 2013). An avulsion is a major flow diversion process(Clarke et al., 2010). It involves the dramatic shift of the main channel to anew location (Clarke et al., 2010), and can be differentiated from the moreminor processes by carrying over 50% of the old channels flow, and leadingto the eventual abandonment of the previous channel (Bryant et al., 1995).Avulsions tend to occur in places where deposited sediment has aggraded thechannel bed level to the point where neighbouring surfaces are at lower ele-vation (Jerolmack and Mohrig, 2007). The now elevated channel then aban-dons its current location and shifts course (Reitz and Jerolmack, 2012). Theprocess of avulsion is typically associated with aggrading surfaces, such asa developing alluvial fan (Reitz and Jerolmack, 2012). However, it can alsooccur in systems in states of equilibrium and degradation, during episodicinputs of sediment or high periods of flow (Davies and McSaveney, 2008).Incisional processesFanhead trenching is a type of incision that occurs in the main channelat the fan apex, with the depth of the incision declining as it progressesdownslope (Schumm et al., 1987) (Figure 2.3). The trench tends to traversethe proximal region of the apex, and therefore can be described to be dis-secting the fan (Schumm et al., 1987). The entrenched channel transfersthe sediment away from the fanhead to the middle and lower reaches ofthe fan (Schumm et al., 1987). In the lower reaches, channel backfilling oc-curs, infilling the channel as the sediment is deposited from the toe upwards(Schumm et al., 1987).What induces channel entrenching, is a question that has produced manyresponses (Rachocki, 1981; Schumm et al., 1987). Reasoning tends to fallunder two categories relating to either changes in external influences or tointernal processes (Hooke, 1967; Wasson, 1977); a detailed discussion andlist of these can be found in Schumm et al (1987).192.1. Alluvial fan theory Figure 2.3: Fanhead trenching occurring in Lucy Clarke’s (Clarke, 2015)experimental fanHowever, in the last decade, the increasing use of physical modellingto study alluvial fans has brought about a different theory, related to thesurface state of the fan (Davies and Korup, 2007). Studies have found thatfan head trenching can also occur in experimental fan evolution, when itreaches a state of dynamic equilibrium (Davies and Korup, 2007; Clarkeet al., 2010).2.1.3 EvolutionAlluvial fans grow in size via the progradation of the fan toe across theiraccommodation space (Schumm et al., 1987). The accommodation space isthe extent of the lowland area available to the fan, and acts as a limitingfactor to fan growth (Davies and Korup, 2007).202.1. Alluvial fan theoryDuring progradation, the fans are constantly adjusting the number ofchannels, channel geometry, and overall morphology, thereby creating arange of spatial and temporal patterns (Clarke et al., 2010). In an exper-imental fan these adjustments and patterns can be categorized into threestages (Figure 2.4). These stages can be used to explain fan evolution, aswell as the response of the fan to changes in the supply of sediment andwater, and other extrinsic controls such as tectonics and human interferencesuch as river control works (Clarke et al., 2010). Figure 2.4: Diagram of the stages of experimental fan evolution; takenfrom Clarke et al (2010).Stage oneThis stage is deposition dominated; the fan grows rapidly, spreading outacross the accommodation space (Clarke et al., 2010). Often there are noclear channels, thus avulsions do not occur; instead water moves as sheet-flow, covering more than 50 % of the fan surface and spreading sedimentacross the fan surface (Bryant et al., 1995) (Figure 2.4 - 1a). However, thesheetflow does alternate with short lived and rapidly changing channelizedflows (Clarke et al., 2010) (Figure 2.4 -1b).212.1. Alluvial fan theoryStage twoFan growth continues; however, the rate of growth is slower. The fan toein some places has reached the end of the accommodation space, meaningthat not all the sediment is deposited onto the fan slopes but instead istransported out of the system (Clarke et al., 2010) (Figure 2.4 - 2).In this stage the water has now become channelized into one or two distinctchannels (Bryant et al., 1995). These channels are unstable and prone tobifurcation. There is a tendency for avulsion, which allows the channelnetwork to migrate laterally across the fan, depositing and/or reworkingsediment (Clarke et al., 2010). Short-lived crevasse splays that develop offthe main channel also contribute to the deposition (Bryant et al., 1995).An alluvial fan may alternate between stages one and two, switching be-tween sheet flow and channelized flow, depending on where the bulk of theflow is and the occurrence of high flow or sediment supply events (Clarkeet al., 2010; Davies and Korup, 2007).In addition, fan lobes form in the mid to lower reaches of the fan duringboth stages 1 and 2. These reflect periods of deposition, when the mainchannels are focussed on that particular area (Clarke et al., 2010).Stage threeIn stage three a majority of the sediment load is transported out of thesystem (Clarke et al., 2010), and continued growth depends on whether thefan is confined or unconfined (Davies and Korup, 2007).In confined fans, growth stabilizes once the fan toe has prograded to thecapacity of the accommodation space. This stage is characterised by a singlechannel (Bryant et al., 1995) which is entrenched in the upper part of thefan at the apex, but continues to avulse and migrate laterally over the fansurface in the mid and lower reaches (Clarke et al., 2010) (Figure 2.4 - 3).In this stage the fan is considered to have reached a dynamic equilibrium interms of its gradient and morphology. Therefore, the fans transport capacitymeets the amount of material supplied, such that sediment accumulation onthe fan surface is zero (Davies and Korup, 2007).222.1. Alluvial fan theoryIn comparison, an unconfined fan can continue to grow indefinitely (Daviesand Korup, 2007). It will episodically aggrade and prograde under constantsediment and water supplies; and whilst the rate of growth may become verysmall, the fan can never reach the dynamic equilibrium that a confined fancan (Davies and Korup, 2007).However, in both cases, the surface state of the fan experiences both shortand long term changes, which can result in the fan exhibiting behaviourcharacteristic of the previous two stages (Clarke et al., 2010). In the shortterm, minor alterations to the water and sediment supply i.e. floods whichintroduce higher flow, or landslides which increase the sediment load, andinduce continual alterations to the river profile and pattern (Harvey et al.,2005). In the long-term changes in the base level or climate, as well tectonicactivity can alter the dynamic equilibrium to the point that the fan canbecome aggradational or degradational (Harvey et al., 2005).232.2. The Waiho River and fan system2.2 The Waiho River and fan systemThe Waiho River and fan system is situated in and west of the Southern Alps,on the South Island of New Zealand (Davies, 1997; Davies and McSaveney,2001) (Figure 2.5). It lies just south of the Franz Josef Glacier (Waiau)township and is the location of the Franz Josef glacier, a major touristattraction for the town and region (Davies, 1997). Figure 2.5: Looking out over the upper Waiho Fan towards Canavan’sKnob from the river right stopbankThe Waiho River drains a total of 170 km2, a large portion of which isfrom the Callery River tributary, a complex system in its own right, whilstother smaller tributaries include the Tatare River and Dochertys Creek, aswell as several small streams (Davies and McSaveney, 2001). A proglacialriver, the Waiho flows west out of the Franz Josef glacier for 15 km, confinedby steep mountainous terrain (Davies, 1997; Davies and McSaveney, 2001).The Callery tributary is in fact 20% larger in area than the rest of theWaiho catchment, and hosts 3 small glaciers; it joins the Waiho about 1km upstream of Franz Josef Glacier township. The Waiho there departs themountains to enter a wide valley which it flows through for another 15 kmbefore meeting with the Tasman Sea (Davies and McSaveney, 2001). Fromwhere the Waiho River leaves the confines of the mountains, an elongatedalluvial fan has formed, the Waiho Fan.The Waiho Fan has formed in the 15 km long trough left behind by theonce coalesced Callery - Tatare - Waiho glacier, which at the last glacial max-imum extended some km beyond the present coastline (Davies and Scott,1997; McSaveney and Davies, 1998). Its width is limited to 4 km at its widest242.2. The Waiho River and fan systemby lateral moraines 100 m high that extend down to the coast (Davies andScott, 1997). At the coast the moraines form high cliffs which protect thefan from coastal erosion. Located where it is on the Western flanks of theSouthern Alps, the fan experiences high sediment supply, extreme rainfalland flood events, glacial interference, and earthquake movements from theAlpine and other faults. The combination of these factors have created anincredibly powerful and dynamic system.Water supplyThe predominant source of water for the Waiho River is precipitation inthe upper headwaters. This precipitation can fall as rain up to heights of3500 m, however in colder weather, this may fall as snow down to 1000m. Eighty one percent of the runoff generated by this rainfall interacts withglacial systems such as the Franz Josef Glacier, where it is transported underor through the glacier to the river network in the valley below; the remainingnineteen percent is delivered directly to the river (Davies and McSaveney,2001).The Waiho catchment is exposed to the ‘roaring forties’, a prevailing moistwesterly airflow (Davies and Scott, 1997). This means that average annualprecipitation is high. Whilst exact values are unknown, it has been estimatedfrom scattered measurements, that annual precipitation is around 11,000mm in the upper headwaters.Contributing to this high annual rainfall are intense storm events whichfrequently occur in the region (McSaveney and Davies, 1998). At the town-ship, rainfall from these storms can reach 200 mm over a 24 hour period atleast once a year, whilst storms with rainfall up to 600 mm over a 3 dayperiod can occur every few years. During these storms, maximum intensi-ties are likely to be about 2 mm per minute. However, rainfall is unevenlydistributed across the catchment, increasing inland, such that the upperheadwaters are likely to receive greater rainfall than the township.These storms typically produce large flood events. Catastrophic flooding,where the raging river has threatened and at times inundated both town andfarmland, punctuates the history of the region (Figure 2.6). The braidednature of the river, especially in the 15 km of alluvial fan, means that itsability to hold floods varies depending on the braided formation at the time252.2. The Waiho River and fan system(McSaveney and Davies, 1998). This makes predicting the response of theriver and fan to flood events extremely difficult. Figure 2.6: Waiho flood 2016. Photo RNZ/Conan Young.262.2. The Waiho River and fan systemSediment supplyThe Waiho River originates in very steep mountainous terrain, in someplaces 3000 m high. This terrain is a product of uplift resulting from thePacific plate in sliding collision with the Australian plate (Davies and Mc-Saveney, 2001). It is in an ongoing battle between the 1˜0 mm of uplift peryear, and rapid erosion caused by the high amounts of precipitation. Asa result, large amounts of sediment (metamorphosed schist/gneisses, andgreywacke) are constantly entering the Waiho catchment via subaerial ero-sion (i.e. rockfalls and landslides). Some of this debris falls onto the FranzJosef Glacier, a sediment source in its own right with high erosion rates atits base (Davies and McSaveney, 2001) (Figure 2.7). Figure 2.7: The retreating Franz Josef glacier. Photo by Linden BrownThe glacier transports sediment at a much faster rate to the Waiho Riveras the stormwater drainage flows through pressurised conduits (Davies andMcSaveney, 2001). Thus, the valley floor immediately downstream of the272.2. The Waiho River and fan systemglacier is constantly aggrading (Figure 2.8). Additionally, when a tempo-rary blockage or change in the subglacial drainage system occurs, brief floodscontaining large pulses of sediment and quantities of broken ice can be ex-pelled (Davies and McSaveney, 2001; Davies et al., 2003b). These generatetemporary sediment oversupply events to the river and fan below. Sedimentpulses from the Callery system also contribute to these and other oversupplyevents (Davies and McSaveney, 2001).The upper Waiho valley has a large storage capacity, therefore changesin sediment delivery from the glacier will take a while to become evidentat the confluence with the Callery, and the head of the Waiho fan (Daviesand McSaveney, 2001). However the Waiho River is still able to deliversediment to the fan head at its transport capacity (Davies and McSaveney,2001). This process is only limited by the velocity, depth, and slope ofthe river. By contrast, the Callery tributary runs in a 10 km long gorgeto its confluence with the Waiho, so any sediment input to this reach isimmediately transported to the confluence. Figure 2.8: the upper valley of the Waiho River, just below the glacierterminus. Photo by Linden Brown282.2. The Waiho River and fan system2.2.1 Geomorphic historyWaiho fan formationThe following information has been summarised from McSaveney andDavies (1998).Around 16,000 to 18,000 years ago, glacial activity dominated the southwestern Southern Alpine coastline, and stages of advance and retreat ofthe coalesced Callery — Tatare — Waiho glacier had carved out a broaddeep valley across the coastal lowland, now known as the Waiho Flats. Afew thousand years later (13,000 years ago) a huge rock avalanche onto thesignificantly retreated Callery — Waiho glacier covered the glacier to itsterminus. The glacier transported the rock avalanche debris to the terminusand there built the Waiho Loop terminal moraine (Alexander et al., 2014).It was only after the Callery — Waiho glacial lobe receded from the WaihoLoop, and the glacially-fed Tatare and Waiho Rivers began delivering sedi-ment to this upper lowland region, that the Tatare and Waiho alluvial fansbegan to form. At this time, the Waiho Loop would have extended muchfurther south than at present, and the isolated hillock known as Rata Knollwould have formed part of the continuous moraine deposit. River bed levelwould have been at least 100 m lower than the present, and sea level muchlower, thus the Loop would have acted as an obstacle to the rivers andrapidly developing fans. In addition, 10,000 years ago the Loop was 400 mfarther south-west relative to the Waiho Valley, due to tectonic movementon the Alpine fault.The Tatare River with probable assistance from the Waiho River rapidlyinfilled the area upstream of the Waiho Loop, until approximately about athousand years ago when its fan elevation became level with the lowest partof the Loop. The Tatare was then able to flow across the Loop, forming asubstantial waterfall on the other side. Downcutting saw the Tatare becomeincised into its fan, and prevented further aggradation of the upper fansurface.There is no evidence that the Waiho River ever flowed into the Tatareabove the Loop. Instead the Waiho is believed to have aggraded its fan to-wards the outer sea coast. When sea level stabilized about 6,000 years ago,292.2. The Waiho River and fan systemso did the position and elevation of the fan toe. Therefore long term aggra-dation would have ceased shortly after that, evident by shallowly buried oldsoils close to the South end of the Waiho Loop, either side of the WaihoRiver bed.Since its fan formation, it is believed that the Waiho River flowed southof Rata Knoll, its passage to the north being blocked by the higher Tatarefan. At the same time, farther upstream flow would have alternated eitherside of Canavan’s Knob, slowly burying and eroding the former westwardextent of the Waiho Loop moraine. However, in recent time, perhaps as aresult of either large sediment input or fault movement (or both), the upperWaiho fan experienced considerable and unusual aggradation which induceda shift of the main Waiho channel to the east of Canavan’s Knob on to theTatare fan, and then through a low point between the loop and Rata Knoll.This is its current configuration, which has been unchanged for at least acentury aside from an attempted break out to the south in 1982, which waspushed back by control measures.Fan surface stateIt is believed that by the time the Waiho river had shifted to occupythe land east of Canavan’s Knob and the low point between the Loop andRata Knoll, the fan surface had reached a state of long term dynamic equi-librium (Davies and McSaveney, 2001). Nevertheless, in the short termthis equilibrium would be affected by brief sediment oversupplies or deficits.Oversupply, a result of storm or earthquake triggered landslides, inducedaggradation and therefore the steepening of the fan surface profile (Daviesand McSaveney, 2001). Alternatively, deficits (or water surpluses) causedepisodes of degradation, resulting in lowering of the fan slope (Davies andMcSaveney, 2001).Overall, it is believed that the fan surface would have been operatingabout a well-defined mean profile, a result of steady long term sediment andwater inputs, a confined accommodation space, and a toe kept constant bythe stabilized sea level (Davies and McSaveney, 2001).2.2.2 Human interferenceThe Waiho region was settled around the 1890s. The town location itselfwas chosen as it was considered safe from flood events, and provided close302.2. The Waiho River and fan systemaccess to the reasonably safe and stable river crossing. However, over time itbecame evident that the Franz Josef township and farmlands were in dangerfrom flood events and the slowly aggrading river bed. As a result, stopbankswere gradually installed. These ad hoc works were designed to restrict theriver to the far north western side of the valley, preventing it from utilizing amajority of its valley as well as protecting the township and highway. Aerialphotography dating from 1948 to 2013, provides an excellent time lapse ofthe region since this time (Appendix A). It is interesting to note that sincethe initial stopbank installation in the 1930s the Waiho River has undergonedramatic and significant planform changes (Figure 2.9). The deeply incisedgorge just below the confluence between the Waiho and Callery rivers islong gone, lost under an unprecedented level of aggradation. These dramaticchanges have led to the belief that the stopbanks were the cause of the rapidrates of aggradation.Brief history of control worksThe first aerial photographs of the region were taken in 1948. Thus, theexact behaviour of the Waiho river and fan system prior to this time cannotbe accurately reconstructed. However, photographs and written accountsdo provide some insight into what was going on.• In the late 19th century the river bed adjacent to the site of the town-ship consisted of very large glacial lag boulders• The earliest evidence of river control works comes from a photo in1910 of the first car fording the Waiho River below a footbridge. Arock gabion had been put in place to control the location of the rivercrossing, and indicates that there was significant sediment movementand channel instability present.• In 1927, relocation of infrastructure indicates that significant sedimentmovement was resulting in aggradation of the river bed. The originalhotel had to be moved away from the lower terrace site because offlooding.• By the 1930s the river borne sediment was causing further problemsfor the lower terrace site, such as flooding of the former airstrip. At312.2. The Waiho River and fan system Figure 2.9: Aerial imagery of the upper Waiho fan from 1948 and 1997.Image from Davies and McSaveney (2001)this time the first permanent stopbank was installed to keep the riveraway from this lower terrace site.• The 1948 aerial photograph indicates several more stopbanks in place.Bank protection works are visible on the true right back opposite theHoliday Park, and there is a true left stopbank just upstream of theCanavan’s Knob.• From 1948 onwards, the aerial photographs provide an excellent recordof the changes taking place on the upper fan, and of the increasing andon occasion decreasing number of stopbanks (Appendix A). One of therare, but spectacular failures of the stopbank network, was of a longcentral bank installed in 1980 (figure 2.10). It was designed to restrict322.2. The Waiho River and fan systemthe river to a narrow bed against the south western bank (true left),downstream of the motor home. It lasted only four years, and wasbreached in a flood event; no evidence of it remains. Figure 2.10: Photo of the Waiho showing the 1980 stopbank. The (failed)stopbank runs down the centre of the fan.Present day stopbank alignmentThe present day stopbank system consists of nine main stopbanks (fiveon the left and four the right) which restrict the Waiho River to a thirdof its alluvial fan. These are described briefly below (Table 2.1). Otherstopbanks do exist in the lower reaches of the Waiho River, Tatare River,and Dochertys Creek, but are not discussed here, as the focus is on theupper and middle sections of the Waiho fan.332.2.TheWaihoRiverandfansystemTable 2.1: Description of the current stopbanks setup on the Waiho River (McSaveney and Davies, 1998)Type Location ReasonR1Interlocked heavy rock spurgroynesBetween the Callery Junction andthe SH6 bridgeTo deflect the river flow away from thebase of the high terrace and thereforeprevent further scour and loss of the terrace.Loss of this terrace edge would threaten theright bank (Northern) approach to the SH6bridge and the Waiho River frontage of theFranz Josef Glacier townshipR2280m embankment of interlockedheavy rockUpstream of the SH6 bridge to thedownstream edge of the terracewhere the Anglican church is sited.(Figure 2.11)Scour protection. The embankment is 5 mdeep to prevent future scouring, and provideongoing protection to the terrace, and church.R3 StopbankLower terrace where the former airstripwas located. It is now a heliport.This stopbank has been up to a kilometre longin an attempt to prevent the river overflowingonto the former airstrip.R4Upgraded R3 stopbank, as wellas fill and rock rip rap and spurgroynesLower terrace at the Franz Josef Glacierhotel and oxidation frontage. (This wasbreached in 2016; Figure 2.12)To protect the Franz Josef Glacier hotel andthe oxidation pond from the rapidly aggradingbed and flood events.L1 Raised road, and rock rip rapBetween the Callery confluence and theSH6 bridge (Figures 2.13 and 2.14)To protect the glacial access road fromaggradation and bank scourL2Stopbank armoured in part byheavy rock, and as well as stub groynesDownstream of the SH6 bridge along thefrontage of the Holiday Park (Figure 2.15)To protect the river frontage of the HolidayPark from continuing aggradation and scour.L3 Stopbank with stub groynesOverlapping with the L2 stopbank downto Canavan’s Knob.To protect the SH6 from aggradation andscour, and prevent the river from breakingout.L4 StopbankDownstream of Canavan’s Knob, actingas a continuation of a low terrace at itsupstream end.To protect the terrace from erosion bydeflecting flow away from the left bank(towards the unprotected right bank).L5 Stopbank faced with large rocks Immediately downstream of Rata KnollEatwells (formally Miltons) stopbankis designed to protect the farmland andWaiho Flats airstrip from flow breakouts, which could potentially result in ashift across the flats into Dochertys Creek.342.2. The Waiho River and fan system Figure 2.11: Looking downstream from the SH6 bridge at the true rightstopbank that protects the church Figure 2.12: The repaired lower terrace stopbank. The Heartland Hotelcan be seen in the top right hand corner. The Waiho breached this stopbankand flooded the hotel in 2016. Photo by Rob Pieper352.2. The Waiho River and fan system Figure 2.13: Looking upstream from the SH6 bridge at the true left stop-bank and spur groynes Figure 2.14: Looking downstream to the SH6 bridge at the spur groyneson the true left of the Waiho River362.2. The Waiho River and fan system Figure 2.15: Looking downstream from the SH6 bridge at the true leftstopbank. The river/fan bed is clearly much higher than the land on theother side of the stopbanks.37Chapter 3Microscale modellingMicroscale models are extremely small scale physical hydraulic models(Peakall et al., 1996; Clarke, 2015; Gaines and Maynord, 2001; Malvertiet al., 2008). Physical hydraulic models represent a selected fluvial geomor-phic feature, allowing it to be studied under closely monitored or controlledexperimental conditions (Mosley and Zimpfer, 1978). Microscale modelshave been successfully used to model alluvial fan behaviour and dynamics,and river engineering works (Davies et al., 2003a, 2013; Gaines and Maynord,2001; Clarke, 2015).The successful use of this type of modelling, and number of advantages asoutlined below, has shown them to be an invaluable tool to both engineersand geomorphologists alike (Peakall et al., 1996; Mosley and Zimpfer, 1978).• Due to their small size, usually tens of centimetres to a couple ofmeters (Malverti et al., 2008) these models are relatively easy andinexpensive to set up, and they take up very little space (Malvertiet al., 2008; Mosley and Zimpfer, 1978) For example, the Davies et al.(2013) microscale model of the Waiho alluvial fan and river systemwas constructed on a 2m by 3m wooden table, using readily availablepolystyrene for the boundaries (Campbell, 2012). Aside from the ini-tial assistance for the setup, the honours student was able to run theexperiments alone (Campbell, 2012).• The close control over the relevant variables that a small model al-lows, also means that they allow for precise measurement (Mosley andZimpfer, 1978), as was found in a study on growth dynamics on gravelbed river deltas (Wild, 2012). The relatively small scaled models of1:1500 and 1:2000, allowed the use of an instantaneous-profile laserscanner, providing bed elevation accuracy of +/- 0.5mm.• In addition, microscale models evolve at much shorter timescales thanother larger scaled models, which means that results can be achievedquickly (Malverti et al., 2008). This is particularly valuable when38Chapter 3. Microscale modellingconsidering that the timescale of river bed evolution through bedloadtransport is 10-1 to 104 years. Microscale models dramatically reducethis timescale. This has allowed for the study of landscape evolutionprocesses (Peakall et al., 1996). However, it must be borne in mindthat, unlike a properly scaled model, the timescale of operation ofmicroscale models is unknown.• Finally and perhaps most importantly when considering the study ofvery dynamic systems, they allow the observation of fluvial phenom-ena that are otherwise difficult to observe in the field (Warburton,1996; Davies and Lee, 1988). For example, braided gravel-bed riversexperience a majority of their planform change during flood conditions(Peakall et al., 1996). At this time, the highly turbulent and turbid na-ture of flow makes it virtually impossible to observe near-bed processesin the field (Peakall et al., 1996).However, microscale modelling is a relatively new form of physical hy-draulic modelling, and thus far has had only limited use, but received muchcriticism from the scientific community (Gaines and Maynord, 2001). Phys-ical hydraulic models are required to meet three similarity criteria, whichare geometric, kinematic and dynamic similarity (Schumm et al., 1987).Microscale models do not meet all of these criteria. They will maintain ge-ometric similarity, and some kinematic similarity, but place little emphasisupon dynamic similarity (Davinroy, 1994; Hong and Davies, 1979; Malvertiet al., 2008), instead appearing to adhere to a fourth criterion, Hooke’s sim-ilarity of process (Mosley and Zimpfer, 1978). This has led to a number ofconcerns in regards to its size and similitude (Gaines and Maynord, 2001).However, several studies have been conducted to address these concerns, andsubsequently justified the use of microscale modelling as a tool for scientificinvestigations.The following sections will explore microscale modelling theory and prac-tice, followed by an outline of the microscale model used in this study. Thetheory section will address underlying principles as well as the concerns overmodel size and similitude; whilst the practice section will discuss modelconstruction, experimental procedures and measurement tools, providing acomparison between the two main users of microscale models. The outline ofthe microscale model used in this study will also include model construction,experimental design, and the data collection and analysis tools used.393.1. Theory3.1 Theory3.1.1 Underlying principlesPhysical hydraulic models are often classified into three types: one-to-one,scaled, and analogue (Peakall et al., 1996). Of these types, microscale modelsfall under the analogue category. This reflects the nature of the modelling.Analogue implies not the exact scaled representation of a prototype butthat of the general population (Hooke, 1968; Peakall et al., 1996; Mosleyand Zimpfer, 1978). The models reproduce a specific aspect of the form andfunction of the chosen feature, but the forces, materials and processes maybe quite dissimilar to those in nature (Schumm et al., 1987). Thus, thesemodels must be considered as small systems in their own right (Peakallet al., 1996; Mosley and Zimpfer, 1978). Analogue modelling is based uponHooke’s similarity of process criterion.Hooke’s similarity of process criterionRoger Hooke’s (1968) similarity of process technique was developed as analternative to the formal scale-modelling procedures. He found the latterto be difficult to follow, restrictive and at times unsuitable to geomorphicstudies where the aim was to develop general theory rather than representa specific prototype (Schumm et al., 1987).The similarity of process technique ignores true dynamic similarity (Barr,1968), as Hooke (1968) viewed the quantitative extrapolation from a modelto a prototype as dubious. Instead it is based on the concept that as theunderlying principles are the same then the processes that occur in a labo-ratory system, and their morphologic effects, are similar to those in nature(i.e. in the field; Hooke, 1967).The similarity of process criterion requires that:1. gross scaling relationship be met2. the model reproduce some morphologic characteristic of the prototype3. the processes which produced this characteristic in the laboratory canlogically be assumed to have the same effect on the prototype.403.1. TheoryUnder these criteria, the model is treated as a system in its own right(Hooke, 1968). This type of model is considered the true analogue model.However, the treatment of these models as systems in their own right can beseen as limited to a specific mode of investigation. As there is no similarityto a prototype, analogues are better suited to geomorphic problems thatdo not involve an individual (prototype) in the population (Hooke, 1968).Instead they can be used to investigate general relationships applicable tothe entire population. For example, Hooke successfully conducted severalsimilarity of process experiments to study the processes and steady-staterelationships in arid region alluvial fans (1967; 1968). He did not focuson a particular fan, but on the behaviour of the group in general (Hooke,1967, 1968). Similarly, Schumm (1977; 1987) successfully used analoguemodels to reproduce the fundamental features and processes that one wouldexpect to see on natural alluvial fans (Clarke, 2015). These experiments wereinvaluable to the study of alluvial fans, and motivated a renewed interestinto experimental modelling of these landforms (Clarke, 2015).Whilst microscale models appear to adhere to the similarity of processcriterion, they differ from the traditional analogue models that Hooke andSchumm used in that they retain some aspect of geometric and kinematicsimilarity between model and prototype (Hong and Davies, 1979; Malvertiet al., 2008). This is because microscale models have and continue to be usedto investigate specific systems (prototypes; Davies et al., 2003a; 2013, Gainesand Maynord, 2001). They are able to do this by meeting the geometricsimilarity criterion required of physical hydraulic models (Hong and Davies,1979); and are calibrated to reproduce particular behaviours of the prototype(kinematic similarity).Geometric similarityGeometric similarity is one of the three similarity criteria required bya physical hydraulic model to precisely represent the prototype (Schummet al., 1987). In microscale models, geometric similarity is achieved by thescaling down of the boundaries and landscape features of the prototype tothe micro scale.Microscale models have been used at a model to prototype ratio of as smallas 1:20,000 (Gaines and Maynord, 2001). However, at such small scales itwould be impossible to model the required flow rates and depths. Therefore,413.1. Theoryin order to maintain adequate model flow depths, microscale models oftentake on a distortion effect (Gaines and Maynord, 2001; Peakall et al., 1996).This means that the vertical scale differs to the horizontal scale of the model,with vertical exaggeration (Gaines and Maynord, 2001; Peakall et al., 1996).3.1.2 The justification for microscale modellingDespite successful use (Gaines and Maynord, 2001; Davies et al., 2003a,2013), microscale modelling has and continues to be treated with some trep-idation and criticism by the scientific community (Gaines and Maynord,2001). This is because these models do not meet all the basic similarity cri-teria required for other forms of physical hydraulic modelling (Gaines andMaynord, 2001).Physical hydraulic models are required to meet three similarity criteria(Schumm et al., 1987):1. Geometric similarity (form): ratios of homologous dimensions areequal and equivalent angles are the same2. Kinematic similarity (motion): paths and patterns of motion are geo-metrically similar to those of homologous occurrence in the prototype3. Dynamic similarity (forces): the ratios of homologous masses andforces affecting matter are equal at all times.It is almost impossible to meet each of these similarity requirementsexactly in a scaled model, and many types of scaled models will relaxsome of the dynamic similarity (Malverti et al., 2008; Mosley and Zimpfer,1978). However, microscale models give little emphasis to these criteria.Whilst they conform to the (planform) geometric similarity criterion, andin part generate some kinematic similarity, there is no attempt made to-wards achieving dynamic similarity (Davinroy, 1994; Hong and Davies, 1979;Malverti et al., 2008).Therefore, in the same way that Hooke (1968) had doubts about the quan-titative extrapolation of results from model to prototype in scaled physicalmodels, there are many concerns in regards to the accuracy and reliabilityof microscale modelling (Gaines and Maynord, 2001; Malverti et al., 2008;Mosley and Zimpfer, 1978; Peakall et al., 1996). The short time span that423.1. Theorymicroscale models have been in existence, and the relatively few studies thathave used them, does not help their case, nor the present author’s job. So,criticism and trepidation are justified. Thus, the concerns in regards to theaccuracy and reliability of microscale modelling, as well as application to aspecific prototype, must be discussed.SizeUnderstandably concerns have focussed on the very small size of mi-croscale models (Gaines and Maynord, 2001). This is because there isincreasing evidence that some geomorphic processes are scale dependent(Schumm et al., 1987). However, considering that there is no direct quan-titative extrapolation from model to prototype, then the errors that mayoccur during the interpretation of gross scaling relationships cannot occur.One must reiterate, that the microscale models are treated as systems intheir own right (Hooke, 1968; Malverti et al., 2008).Other points to consider with respect to the small size of these modelsare the effects of surface tension on channel development and the physics ofchannelized flow, as well as the severe vertical distortion required to maintainacceptable flow depths and rates (Malverti et al., 2008).Surface tension is “the tensile force which results from the difference be-tween the internal molecular forces of a liquid and the forces between theliquid molecules and an adjacent [material]” (Malverti et al., 2008) for ex-ample, the interface between water and air. Based on this definition, itseems logical that at such small flow depths, surface tension would have aneffect on the model performance. However, studies by Peakall and Warbur-ton (1996) and Me`tivier and Meunier (2003) both found that the effect ofsurface tension was insignificant when the dimensionless Weber number waslarge i.e. between 10 and 100. Further, in a more recent study using laminarflow experiments, Malverti et al (2008) found that smaller Weber numbersranging between 0.1 and 2 also had negligible effects on surface tension.The very small size of microscale models also means that they employvery small horizontal scales (Gaines and Maynord, 2001). In order to main-tain the flow depths and rates necessary to allow sediment transport, andaccurate measurement with the available tools, a larger vertical scale must433.1. Theorybe applied (Malverti et al., 2008). This creates a vertical distortion. Con-cerns exist about the accuracy of flow distribution under this distortionaryeffect (Gaines and Maynord, 2001) as experiments conducted by the USArmy Corps of Engineers have shown that the vertical distortion meansthat riverbank effects extend over a greater portion of the model than theprototype (Gaines and Maynord, 2001).However, microscale models are not the first type of physical hydraulicmodel to use distortions (Gaines and Maynord, 2001). Indeed distorted mod-els themselves are considered a type of physical modelling (Peakall et al.,1996). Furthermore, scale distortion effects are irrelevant for analogue mod-els, as discussed by Mosley and Zimpfer (1978).SimilitudeMicroscale modelling appears to be almost flippant towards the similarityconsiderations that other physical hydraulic models are guided by (Gainesand Maynord, 2001). This is because there is no attempt made to meet thedynamic similarity criterion, meaning that dimensionless values such as theFroude, Reynolds and Shields numbers are ignored (Gaines and Maynord,2001; Paola et al., 2009). For example, flow in microscale models is laminar,when in field rivers it is always turbulent (Davies et al., 2003a).The use of laminar flow to model turbulent systems brings about severalconcerns about unrealistic friction coefficients, lack of suspended load trans-port, and as discussed earlier the effects of surface tension (Malverti et al.,2008). However, each of these can be defended.• A study by Lajeunesse et al (2010) showed in their own experimentsand in a discussion of other experimental studies (Armstrong, 2003)that the effect of unrealistic friction coefficients were negligible uponmodel performance (Lajeunesse et al., 2010; Malverti et al., 2008).• Whilst the lack of suspended sediment transport is a valid concern, itis not applicable to studies where the focus is on bedload transport(Malverti et al., 2008), as is the case for braided gravel-bed rivers, andin the Waiho River and fan system.443.1. TheoryIn addition, Hong and Davies (1979) have shown that whilst processes ofwater flow and sediment motion may differ between turbulent and laminarflows, this does not suppress similarity between channel pattern variabilityor Froude numbers (Davies et al., 2003a).• Hong and Davies (1979) were able to reproduce the braiding patternof several east coast rivers of the South Island of New Zealand in amicroscale model by adjusting flow rates, sediment supply, and slope,thus proving that microscale models can achieve kinematic similarity,at least in part.• Hong and Davies (1979) and Grant (1997) also found that Froude num-bers in microscale models are often similar to their prototype valuesregardless of the lack of attempt to achieve similarity.Another concern related to the similitude between model and prototypeis to do with the calibration of microscale models. The vertical scale rep-resented by flow depth, as well as flows rates, sediment supply, and slope,are not included in the design of microscale models (Gaines and Maynord,2001). In many cases, this is because field data are lacking, which is why themicroscale model has been utilized in the first place (Davies et al., 2013).Therefore there are insufficient data to generate a hydrograph or sedimentsupply rates (Gaines and Maynord, 2001). Thus, during the calibrationphase slope, flow depth and rate, and sediment supply are adjusted empiri-cally to ensure general bed movement as well as reproduce some morphologicand behavioural aspect of the natural system they are representing (Gainesand Maynord, 2001). However, to reiterate a previous point, the microscalemodel is not an exact scaled replica of the prototype. It may be similar inprocess and geometry, but it is still a system in its own right, and does notrequire the scaled inputs that other physical models do.Although there are concerns over the veracity of microscale modelling,these models have thus far been “unreasonably effective ” (Paola et al., 2009),having been used very successfully to study a number of problems (Clarke,2015). In both New Zealand and the United States of America, microscalemodels have been applied to investigate alluvial fan dynamics, and riverengineering works. Davies et al (2003a; 2013) successfully reproduced thebehaviour of the Waiho River on the upper part of its fan using 1:2000and 1:5000 scale micromodels (Clarke, 2015). The US Army Corps of engi-neers have and continue to use microscale models for a number of channel453.2. Practiceresponses studies, including evaluation of how navigational channel widthsand depths are affected by different channel designs on the Mississippi river(Gaines and Maynord, 2001). Given the success of the models in these stud-ies, it would seem that microscale models present an opportunity for furtheruse as well as exploration within the scientific community.3.2 PracticeIn the fluvial environment, microscale models have been successfully usedto study a range of evolutionary processes (Peakall et al., 1996). These in-clude sediment-flow interactions, bed morphology, channel morphology suchas meandering and braiding, channel evolution, and knickpoint migration(Malverti et al., 2008; Davies et al., 2003a, 2013; Davies and Korup, 2007;Gaines and Maynord, 2001; Guerit et al., 2014). Of note, they have provideda valuable insight into landscape evolution processes, such as that of alluvialfan and channel dynamics (Clarke et al., 2010; Peakall et al., 1996).In addition to developing general theory and understanding, they havebeen extensively applied to engineering and river management problems(Gaines and Maynord, 2001; Mosley and Zimpfer, 1978; Warburton, 1996).They have been used to investigate management upon specific reaches ofriver channel by studying the channels response to different control struc-tures such as weirs, dikes, groynes, and stopbanks (Gaines and Maynord,2001; Mosley and Zimpfer, 1978) as well as on the impact on fan behaviourwhen confined (Davies et al., 2003a, 2013); and to test engineering designequations (Warburton, 1996).The following three sections will explore microscale model design and ex-perimental procedure, as well as the tools used for measurements. This willdraw on and explore the limited number of microscale modelling experimentsassociated with alluvial fans and braided gravel-bed rivers.3.2.1 Model designThere appear to be two main centres of microscale modelling, one in theUnited States and the other in New Zealand.• In the United States, the St Louis District of the United States ArmyCorps of Engineers (USACE) has been using microscale modelling for463.2. Practicesome time now to investigate river training issues on the Mississippiand White rivers (Gaines and Maynord, 2001).• In New Zealand a number of microscale modelling experiments havebeen used to study fan systems on the South Island west coast and thedynamics of braided gravel-bed rivers on the east coast (Davies et al.,2003a, 2013; Davies and Korup, 2007).Comparison of the two microscale modelling styles reveals that whilst theydo differ in some respects, there is general trend in model design. Otherstudies which have adopted the microscale or analogue approach (Clarkeet al., 2010; Guerit et al., 2014; Me`tivier and Meunier, 2003; Paola et al.,2009) also share similar practices.The main components of microscale models include (Figure 3.1):• Hydraulic flume (structure, sediment/water feeders, and flow con-troller)• Geometric insert• Bed sediment• Materials to be used for control work structures (if required).473.2. Practice Sand tray Geometric inserts Sediment feeder Water system Butynol layer Figure 3.1: Campbell (2012) microscale model; components include thebutynol layer, geometric inserts, sediment feeder, and water system (headand footer tanks, connecting tubes and siphon).Hydraulic flumeHydraulic flumes (Gaines and Maynord, 2001), also called sand trays(Davies et al., 2013), are the self-contained structures on which the mi-croscale models are built.• Whilst the USACE structures are more flume-like, the tray used byDavies et al (2013) is simply a flat, wooden table top, with raisededges, and a waterproof butynol layer to prevent water loss (Figure3.1).• In both cases, they can be sized to meet a range of requirements.However the size tends to be limited by the area physically reachableby the modeller (Gaines and Maynord, 2001). The USACE models483.2. Practicetypically fall within 0.9 m wide by 1.90 m long dimensions, whilst theDavies et al. (2003a; 2013), Davies and Korup (2007) and Wild (2012)models shared a 2 m width, and a range of lengths between 2 and 5 m.These structures also have adjustable slopes in both the longitudinaland transverse direction, and planar tops (Gaines and Maynord, 2001;Wild, 2012).Microscale models receive sediment and water from a singular point at thehead of the model. In both centres of modelling, the water is recirculatedvia a pump, and header and footer tanks (Gaines and Maynord, 2001; Wild,2012; Davies et al., 2013).• The head tank is connected by a tube to the head of the outwashsection of the river, where flow enters the model via the flow con-troller. This control is electronic in the USACE experiments (Gainesand Maynord, 2001), whilst in the NZ set of microscale models, flowis manually controlled by adjustable clamps (Campbell, 2012).• Once the water has passed through the model, it enters a tray at thebottom. In the New Zealand models, a fine sieve mesh is secured atthe end of this tray, to ensure only the water drains down into thefooter tank below (Campbell, 2012). This tank is connected to theheader tank, and water recirculation is maintained between the twoby the pump and a pipe network.• In the USACE and Davies et al (2003a; 2013) models, water was fedinto the model at constant rates (Gaines and Maynord, 2001). How-ever, in the Davies and Korup (2007) model, water was fed into themodel via a series of repeated hydrographs between which flow wouldbriefly stop. The hydrographs were regulated by an automatic siphonand reserve tank which were fed by a steady inflow. The tank in whichthe siphon sat, would fill up to the point that the surface of the waterwas higher than its bend, at which point the siphon would be triggered.Sediment was also fed into the head of the models. In the USACE modelsthe sediment was recirculated (Gaines and Maynord, 2001), whilst in theNew Zealand set of models (Davies et al., 2003a, 2013; Davies and Korup,2007), it was not.• In the unrecirculating models, a Tinker sediment feeder was used tomove the sediment at a controlled rate into a funnel where it would be493.2. Practicejoined by the water. This allowed mixing of the two mediums beforeentering the actual model.• Tinker sediment feeders consist of open topped sediment holding con-tainer, and a revolving metal pipe connected to the bottom of thecontainer (Figure 3.2). Sediment is poured into the top of the feeder,and exits along the revolving pipe which is powered by the pump. Theangle of the pipe and container, and speed of the pump, dictate thesediment supply rate to the model. The steeper the angle, the greaterthe feed rate. Crucial to successful operation, the sediment must bedry, or it won’t move through the revolving pipe. (Campbell, 2012;Wild, 2012; Davies et al., 2013). Figure 3.2: the Tinker sediment feeder used in the author’s experiments.In this image, the double funnel and disc system used by Davies and Korup(2007) is present.• In addition, Davies and Korup (2007) had the feeder pipe deliver thesediment onto a disc above the funnel (Figure 3.2). A conical angle ofrepose sand pile would then form on the disc, and the sediment wouldavalanche naturally from it into the funnel, where it would mix with503.2. Practicethe water before entering the model. This generated irregular inputsof sediment in the short term.• Once the sediment left the model, the water was strained out, and thesediment dumped to dry for later use. Sediment could not be recir-culated as it is not possible to use wet sediment in a Tinker sedimentfeeder.Geometric insertThe geometric inserts represent the fixed boundaries of the river reach(Gaines and Maynord, 2001) or fan system (Davies et al., 2013) being stud-ied. These inserts have been cut out of high density expanded polystyrenethat is typically between 50 and 250 mm thick (Clarkson, 1999; Gaines andMaynord, 2001; Campbell, 2012; Wild, 2012). It is then attached via strong,water proof glue (ADOS Styrobond), and sealed with a silicone sealantSilaflex R©RTV (Wild, 2012).• The USACE designed the inserts from georeferenced aerial photographsof the prototype (Gaines and Maynord, 2001). The generated pho-tomap was scaled to the chosen horizontal scale of the model. It wasthen placed over the singular rectangular sheet of planar polystyrene,and the channels cut out vertically (Gaines and Maynord, 2001).• Wild (2012) designed her inserts from contours created in ArcGIS froma 25 m DEM of New Zealand which was produced by Land InformationNew Zealand using digital topography data. The inserts were chosenso that they represented the disparity of slope, where the mountainsrose steeply from the gentle valley slopes.• In the Davies et al. (2003a; 2013) experiments, inserts were cut fromstencils that had been generated from a scaled 1:50,000 topographicmap of the Waiho area (Clarkson, 1999; Campbell, 2012).• The polystyrene inserts also received layer/s of paint to make themwater tight (Wild, 2012; Clarkson, 1999; Campbell, 2012).• In addition to these inserts, the Davies et al. (2003a; 2013) models alsoused a free overfall. This meant a base level could be set, effectivelyconfining the fan systems they were working with. The overfall was a1cm thick strip of wood that was glued and sealed to the butynol layerat a point where field studies had shown that very little aggradation513.2. Practicehad occurred, suggesting that bed elevation downstream of that pointhad been fairly constant over the last few thousand years (Davies et al.,2003a).Bed sedimentThe bed sediment differed between the sets of microscale modelling.• The USACE used a “Urea Type II plastic ”, which has a typical specificgravity of 1.48. The mix used in the microscale models consisted offour gradations: 0.25 - 0.42 mm, 0.42 - 0.58 mm, 0.48 - 0.84 mm, and0.84 - 1.19 mm. The USACE had used a number of different types ofsediment, but has settled on the Urea type II plastic because it is lightenough to be transported, and retains the bed configuration after eachmodel test so that the bathymetric data can be collected (Gaines andMaynord, 2001).• The New Zealand set of models use a fine silica sand (Davies et al.,2003a, 2013; Davies and Korup, 2007; Wild, 2012), that has a mediandiameter of 0.19 - 0.22 mm.Control work structuresControl work structures include but are not limited to dikes, bendwayweirs, chevrons, closed structures, and stopbanks. In reality, these may bedesigned out of concrete, wood, a coarse mix of gravel, or large boulders.However, in the microscale models, they are not.• The USACE microscale models use pervious, steel mesh structures(Gaines and Maynord, 2001). The St Louis District of USACE testeda number of different materials using flume experiments and modelstudies. The pervious, steel mesh structures were found to producethe most realistic effect of flow and sediment dynamics. This was con-firmed by comparing localized scour and depositional trends betweenmodels and prototype.• Alternatively, the Davies et al. (2003a; 2013) models had sand rough-ened galvanized steel strips to represent the stopbanks.523.2. Practice3.2.2 Experimental procedureAfter initial construction, both sets of microscale modelling followed ageneral procedure to calibrate and prepare the models for experimentation.This procedure involved the running of the model with water and sedimentinputs until the respective river (Gaines and Maynord, 2001) or fan system(Davies et al., 2003a, 2013) had achieved a state of equilibrium. Equilibriumor rather dynamic equilibrium as termed by Davies and Korup (2007) refersto the bed slope. It is assumed that under specified channel alignment orboundaries, sediment size and supply rate, flow, and enough time, the bedslope will reach an equilibrium state. In this state the average sedimentinput is equal to the average sediment output, and as a result there is nochange in average sediment storage (Gaines and Maynord, 2001; Davies andKorup, 2007).• For the USACE models initial water discharge was one half to twothirds of the channel bankfull depth. This higher discharge was de-signed so as to form the bed configuration with respect to the channelalignment and instream structures, as well as flume tilt, amount of sed-iment in the model and the tailgate elevation (Gaines and Maynord,2001). A low water reference plane was also used in which the sur-face elevations of the micromodel corresponded to the water surfaceprofile that is exceeded 97% of the time in the prototype (Gaines andMaynord, 2001).• This procedure differed slightly for the Davies et al. (2003a; 2013),Davies and Korup (2007), and Wild (2012) models. In these modelsarbitrary values were initially used to develop a moderately slopedfan. After this, the values were gradually altered by trial and error togive a surface slope of about 7% (Davies et al., 2003a) and to providekinematic similarity to the prototype, meaning that the channel pat-terns and behaviour occurring in the model were similar to that of theprototype (Wild, 2012).As the sediment was not recirculated in these models, supply rate wasalso adjusted with the flow rate.Once the models were considered to be in “equilibrium ”, experimentscould begin.• For the USACE and Davies et al. (2003a; 2013) models this meant thatdifferent control work structures could be introduced and the resultsmonitored.533.2. Practice• The Davies and Korup (2007) experiments involved testing the effectof large inputs of sediment on different alluvial fan systems which hadformed via two types of steady input regimes.3.2.3 Measurement toolsA large component of analysis for microscale modelling is done via qualita-tive measures.• Analysis of channel response in the USACE models involved a qual-itative comparison between model and prototype bathymetry, with afocus on localized scour and depositional trends (Gaines and Maynord,2001).• Qualitative measurements in the New Zealand set were made via directobservation of processes at play in the microscale models, as well asthrough time lapse photography (Campbell, 2012).Campbell (2012) used a Nikon D40 digital SLR camera (6 MP) for thelong image sequences, capturing images every 30 seconds for analy-sis; whilst Wild (2012) used two 2-megapixel webcams with autofocus(Logitech QuickCam Pro 9000).The New Zealand microscale modelling also placed substantial focus onquantitative measurements. In this case, values were arbitrary and notextrapolated to the prototypes.• In the Davies et al. (2003a; 2013) models, bed surface elevation wasrecorded using a point gauge system. In the 2003 experiments, digitalVernier callipers collected the point measurements at 50 mm intervals,across 14 cross sections. The Bosch GLM150 laser range finder wasused for the 2013 experiments.The laser range finder was attached to a sliding rail system on whichit moved at 5 cm increments in both the longitudinal and transversedirections (Campbell, 2012), therefore generating a greater density ofpoints compared to the 2003 mode.Each point measurement was imported into GIS, where subtractionbetween new and old measurements provided arbitrary values whichrepresented the rate of change (Campbell, 2012).• Both the laser range finder and the callipers were also used to mea-sure knickpoint recession, fluvial incision, channel widening and otherchannel adjustments (Campbell, 2012).543.2. Practice• Davies and Korup (2007) generated longitudinal profiles using a hand-held laser scanner, which produced Digital Elevation Models of the fansurface at a vertical resolution of 1-2mm. This produced a spatial den-sity far greater than the point system utilized by Davies et al. (2003a;2013).• Wild (2012) also used a laser scanner. Her instantaneous-profile scan-ner (Darboux and Huang, 2003) was mounted on a rail system. Atvarious stages during the experiment it would take measurements atdifferent cross sections of the model surface. These allowed for the sur-face profile of the delta and river system, and volume of accumulatedsediment, to be quantified.553.3. The Waiho microscale model3.3 The Waiho microscale model3.3.1 Model constructionThe Waiho microscale model for this study was constructed in the Engi-neering Soil and Water Laboratory at Lincoln University, Canterbury, NewZealand, a purpose built facility containing a number of flumes as well as asand tray (Figure 3.3 ). The sand tray had previously been used for similarmodelling investigations on the Waiho River and fan system (Davies et al.,2003a, 2013), and was therefore considered suitable for this study. Figure 3.3: labelled microscale model used for this study; componentsinclude the butynol layer, bed sediment, geometric insert, sediment feeder,water feeder, header and footer tanksConstruction of the Waiho microscale model was based on the New Zealandset of microscale models discussed earlier in this chapter, and therefore con-sists of four main components: the hydraulic flume, geometric inserts, supplyof sediment and water, and stopbanks (Figure 3.4).• The hydraulic flume is the 2 by 3 m sand tray; a flat wooden struc-ture with an adjustable slope, covered by a waterproof butynol layer.563.3. The Waiho microscale model• Geometric inserts were built out of 100 mm thick painted polystyrene.Google Earth imagery and a scaled 1:50,000 Land Information NewZealand topographical map were used for the boundary dimensionsand shapes. For ease of scaling and to ensure all key boundaries andlandforms were included, a 6 km by 6 km area was chosen, creating a1:3000 scale model.Corresponding to the previous studies by Davies et al. (2003a; 2013),the free overfall was positioned at the scaled location where Davieset al. (2003a) reasoned there to have been fairly constant elevationover the last few decades as there is significant evidence of very littleaggradation. By shifting the base level inland, the length of the fanwas reduced. This meant that the length of experimental time wouldalso be reduced, as there would be less change of sediment volume forgiven change in circumstances.• The supply of sediment and water was delivered to the top (up-stream) end of the model via a funnel connected to two adjustablefeeders.An elevated constant head tank fed water through a rubber tube tothe funnel, where it mixed with the sediment entering from the Tinkersediment feeder. The header tank was connected to a tray at thebottom of the model which collected the exiting water. This was thenpumped back up to the header tank creating a looped system.The sediment (silica sand) was collected and removed from a sep-arate catcher to the one in the water circuit at the bottom of thesand tray, and was not reused. Instead new sediment, sifted througha 500 micron sieve, was supplied when needed into the feeder up top.• Unlike the Davies et al. (2003a; 2013) experiments, strips of mouldedfibre glass were used for the stopbanks. Coarse sediment was gluedto the strips to create a rough surface. Two stopbank alignments wereused in the experiments: the current confinement (Figure 3.5) and theintermediate option (Figure 3.5). The extreme stopbank option wassimply the fan in its natural unconfined state.To allow the alluvial fan to evolve naturally, sediment and water onlyentered the model area via the funnel system. This meant that the fanwas formed by aggradation as it would have done in the prototype. Ini-tial sediment supply and flow rates were taken from previous experiments573.3. The Waiho microscale modelby Davies et al. (2003a; 2013). Experimentation only began once the en-tire accommodation space had been filled with waterborne (and deposited)sediment. Bed sediment Geometric insert Sediment feeder Water feeder (header tank) Sand tray Butynol layer Footer tank Figure 3.4: Labelled microscale model used for this study; componentsinclude the butynol layer, bed sediment, geometric insert, sediment feeder,water feeder, header and footer tanks583.3. The Waiho microscale model A B Figure 3.5: The two stopbank alignments used in the experiments. A) isthe current stopbank alignment, whislt B) is the intermediate option. Nostopbanks were used for the extreme option as it is simply the fan in itsnatural state3.3.2 Experimental procedureIn total, eight experiments were conducted using the Waiho microscalemodel. These were based around four different types of conditions thatvaried about the type of input, and the surface state of the fan when thestopbanks were installed (Table 3.1). Steady input refers to a constant feedof sediment and water for the entirety of the experiment, whilst the variableinput involved having alternating 15 minute intervals of base and flood wa-ter flows. A state of aggradation was identified when it was visually (and593.3. The Waiho microscale modelquantitatively) obvious that the fan was growing in steepness and height.By comparison, in a state of dynamic equilibrium, change was little to none.Using the DEMs generated from the laser scanned point cloud, dynamicequilibrium was considered to have been achieved when:• the bed surface elevation had a net change with an absolute value lessthan 1 mm• the minimum and maximum (degradation and aggradation) values ofchange were equal• the histogram of net change showed a bell shaped curve centred at 0• the maximum elevation of the area scanned was no longer increasingIn addition, sediment output was measured during one experiment toascertain the surface state of the fan. Dynamic equilibrium was identifiedwhen the sediment output was within +/- 1 g min-1 of the input. This cor-responded well to the data collected using the laser scanner, and so furtheruse of sediment output data was considered unnecessary.Table 3.1: Brief summary of the 8 sets of experiments conductedNumber InputSurfacestateS1 SteadyDynamicequilibriumS2 SteadyDynamicequilibriumS3 SteadyDynamicequilibriumS4 Steady AggradingV1aSmallvariabilityDynamicequilibriumV1bSmallvariabilityAggradingV2aLargevariabilityDynamicequilibriumV2bLargevariabilityAggrading603.3. The Waiho microscale modelThe experiments were designed to replicate the prototype timeline ofevents. Therefore no stopbanks were in place for the initial hours of allexperiments. This was to replicate the natural state. The stopbanks werethen installed when the fan surface reached a certain state, either dynamicequilibrium or aggrading, replicating the anthropic interference with the sys-tem. Each experiment therefore consisted of a number of stages. However,not all stages were tested in each experiment.• Natural• Current stopbank alignment• Intermediate option• Natural/extreme optionThe experiments also differed based on water flow and sediment supplyrates. Initial experiments were found to have input rates too large for themodel. Subsequently these were gradually dropped down to be similar tothose of Davies et al. (2003a; 2013).Run durations varied between the experiments, and depended on the mod-els response to the rate and type of input as well as to the stopbanks. Theaggrading fan scenarios all fell under the 15 hour mark, whilst the equilib-rium fan scenarios extended up to 139 hours.3.3.3 Data collection and analysisBed surface elevationBed elevation data were extracted from the sand tray via two differentmethods.• Digital Vernier Callipers (DVC) were used for the first set of experi-ments (Figure 3.6)• The Aranz Medical Laser (AML) scanner was used for the remainingseven experimentsThe DVC measured cross sectional lines of points over the entire fansurface. The cross sections were chosen so that each of the upper, middleand lower sections of the fan were represented (Figure 3.7), whilst the points613.3. The Waiho microscale modelmeasured were evenly spaced out across each cross section. Earlier pointvalues were subtracted from the latest measurement to provide arbitraryrates of change (Table 3.2). Figure 3.6: Digital Vernier Callipers in action623.3. The Waiho microscale model Figure 3.7: The upper, middle and lower sections of the model fan633.3.TheWaihomicroscalemodelTable 3.2: Table of bed surface elevation points and calculated rates of change for the S1 experiment. Pink =aggrading, green = degrading, blue = no change. EC = elevation changeNrth East 945 - 1015am 1215 - 1230pm EC (2hrs) 130pm - 350pm EC (1hr) 450 - 5pm EC (1hr)30 20 35.49 35.46 0.03 31.78 3.68 30.16 1.6230 40 26.33 23 3.33 20.64 2.36 21.4 -0.7630 60 17.45 12.62 4.83 13.85 -1.23 12.06 1.7945 20 35.31 33.35 1.96 31.4 1.95 32.07 -0.6745 40 29.32 25.05 4.27 23.97 1.08 24.1 -0.1345 60 22.01 17.47 4.54 17.5 -0.03 16.11 1.3960 20 38.56 35.58 2.98 34.44 1.14 39.33 -4.8960 40 31.04 28.49 2.55 24.86 3.63 25.72 -0.8660 60 25 21.22 3.78 20.55 0.67 20.54 0.01105 90 51.32 49.57 1.75 49.37 0.2 48.94 0.43105 120 43.59 40.34 3.25 39.99 0.35 40.01 -0.02105 150 34.1 32.36 1.74 31.61 0.75 29.93 1.68165 65 51.79 51.86 -0.07 51.66 0.2 51.38 0.28165 90 54.12 52.98 1.14 52.97 0.01 53.9 -0.93165 115 47.02 46.41 0.61 45.32 1.09 46.33 -1.01643.3. The Waiho microscale modelThe AML scanner was used to create point clouds for Digital ElevationModels (DEMs) before, during and after each experiment. DEMs of Differ-ence (DoDs) could then be generated to show how the model was changingover time.The Aranz Medical Laser scanner consists of a receiver, transmitter andhandheld laser scanner. The receiver was placed below the sand tray andlevelled. It communicated with the moveable transmitter on top of themodel to increase the range of the laser scanner, and allow the full extentof the model to be measured. The handheld laser scanner was then wavedmethodically and slowly over the surface of the model to collect the elevationdata into a point cloud, assigning x, y, and z values to each point of elevation(Figure 3.8). The elevation data were then exported from the Fastscansoftware to ArcGIS programs. Figure 3.8: Aranz Medical Laser scanner in actionIn ArcMap, to create the DEM, each set of elevation data was interpo-lated using natural neighbour to an output cell size of 2.5 cm (2.5 by 2.5).Other interpolation tools were tested, with no major differences. The hill-shade tool was then applied to the DEM (Figure 3.9) as it simulates the653.3. The Waiho microscale modelfall of shadows on terrain, and therefore allows the surface topography tobe seen more clearly. The DEM was then imported into ArcScene and con-verted into a 3D map to allow for a thorough examination of the surface forany glitches. Figure 3.9: The five stages involved in generating the DEM and DEMof Difference in ArcGis. 1) Interpolated DEM using natural neighbour. 2)DEM with hillshade applied 3) the purple cookie cutter shapefile, used toextract the area of information required 4) the cut out area 5) the DEM ofdifference created by subtracting two cut out areas (4)DEMs of Difference (DoDs) were created by subtracting the newer DEMfrom the older, this involved several steps (Figure 3.9). First, a ‘cookiecutter’ was made. This consisted of a shapefile that covered the area ofinterest and excluded the boundaries. For each experiment natural, currentstopbank alignment, and the intermediate and extreme options, a different‘cookie cutter’ were created. The ‘clip (data management)’ tool used theshapefiles to cut out the same area and elevation data from each DEM. TheDEM cutouts could then be subtracted from each other using the ‘raster cal-culator’ tool to show where the model had changed. The range of values andtheir symbology were manually classified into three groups, aggrading (>1),degrading (<-1) and unchanged (-1< X < 1) to provide a clear picture of thechange. The range for the unchanged category was chosen corresponding tothe error of the scanner (±1mm).663.3. The Waiho microscale modelSediment outputSediment output rate from the base of the free overfall was comparedto input rate delivered via the sediment feeder to provide insight into theequilibrium state of the model. The output was measured at the start ofeach hour of run time for a period of five minutes before being dried andweighed.At the end of an hour, the sediment and water feeders were turned off andthe lower section of the sand tray (below the overfall) cleared of sediment.A 1 m-long tray with a 75 micron sieve sheet across the back to allow waterthrough was then placed squarely against the base of the overfall and helddown with weights. The sediment and water feeders were turned on again,and the five-minute period began when the flow of water entered the tray.After five minutes the tray was removed and the sediment washed througha coarse gridded sieve, into a 75 micron one. It was then dried before beingweighed. The bulk weight measurement was converted to grams per minute,and compared to the rate at which sediment is entered via the feeder at thetop of the model.Qualitative time sequencingPhotography was used to provide a qualitative measurement of change inthe model. For each experiment, different angles were chosen to provide theoptimal viewpoint for key areas of change. In the initial hour, photos weretaken every ten minutes. After this, change was considered to occur lessrapidly, and photos were taken on the hour. When the photos of the sameangle were arranged and viewed in order, it provided an accurate visualrecord of change over time (Figure 3.10).673.3. The Waiho microscale model T=0 T=10 T=20 T=30 T=40 T=50 T=60 Figure 3.10: Time lapse of change after the intermediate stopbank hasbeen removed. Images are spaced 10 minutes apart68Chapter 4Data analysis: steady inputexperiments4.1 IntroductionThis chapter reviews the results from the four steady input experimentsconducted on the Waiho River and fan microscale model. These experimentswere based on the methodology of the previous Waiho River and fan mi-croscale model experiments (Davies et al., 2003a, 2013). In the Davies et al.(2003a; 2013) experiments, steady inputs of water and sediment were usedto build microscale model fans to states of dynamic equilibrium. Installationof the current stopbank alignment then resulted in similar aggradational be-haviour to that of the prototype (Davies et al., 2003a). Ongoing confinementand therefore aggradation, induced an avulsion of the experimental WaihoRiver into the Tatare River (Davies et al., 2013).In the present study, the steady input experiments were initially designedto take the previous work a step further, so as to investigate how the mi-croscale model river and fan would respond to the removal of the currentstopbank alignment, and installation of two alternative alignments. How-ever, as the experiment sequence progressed, focus gradually switched to aninvestigation of the microscale modelling techniques. This was to addressinconsistencies that were occurring with data collection and the effect ofboundary surface roughness on fan behaviour; but above all to understandthe difference between the present experimental results and those of Davieset al. (2003a).4.2 ResultsThe following steady input experiments differ based on the surface state ofthe fan when the stopbanks were installed, water and sediment input rates,scenarios tested and measurement tools used (Table 4.1).694.2.ResultsTable 4.1: Summary of the four steady input experiments, including; surface state when stopbanks were installed,input conditions (sediment and water); scenarios tested (Nat = natural, CC = current confinement, Int = inter-mediate stopbank option, Nat2 = released from confinement (also the extreme stopbank option); measurementtools and outcomes.Input conditionsNumber Surface state Sediment(g/min)Water(ml/s)Scenarios Measurement tool/s OutcomeS1 Equilibrium 36 18Nat,CC, Int, Nat2CallipersTime lapse photographyFan aggraded when confined,and then continued to aggradewhen released.S2 Equilibrium 25 20Nat,CC, Nat2Laser scannerSediment outputTime lapse photographyNo aggradation when confined.S3 Equilibrium 16 8Nat,CCLaser scanner No aggradation when confined.S4 Aggrading 20 8Nat,CC, Nat2Laser scannerFan continued to aggrade whenconfined, and then when released.704.2. Results4.2.1 S1: equilibrium fanExperiment S1 was the first experiment to be run following the develop-ment to equilibrium of the fan on the sand tray. It was designed to addressthe primary aim of this thesis, which was investigating the response of thefan to removal of the current stopbank alignment and installation of lessrestrictive alternatives. Based on the previous studies (Davies et al., 2003a,2013) it was believed that after achieving a state of dynamic equilibrium, theinstallation of the current stopbank alignment would result in aggradation,followed by degradation upon removal.In this experiment, the fan bed surface elevation was measured at a num-ber of points using the digital Vernier Callipers. Observed equilibrium wasbased on the amount of change that occurred in these points over time.When a majority of the points showed no change (absolute values less than1 mm) the fan was considered to be in a state of dynamic equilibrium (Ap-pendix B). Once this had been established, the current stopbank alignmentwas installed and monitored, followed by the intermediate alignment option,and finally released back to its natural state, the extreme option (AppendixB).The experimental fan did not behave in the way it was expected to. Whenconfined by the current stopbank alignment, the fan did aggrade, as it hasdone in the prototype situation, and in the previous studies (Davies et al.,2003a, 2013). However, it did not degrade to the ‘dynamic equilibrium‘ bedsurface elevation when released from this confinement. Instead it infilledthe newly available areas of the fan, and once these were level with thebed surface elevation of what had been the confined fan, the whole surfacecontinued to aggrade (Figure 4.1).714.2. Results Hour 0 Hour 6 Hour 9 Figure 4.1: Photo time lapse of the fan behaviour when the current stop-bank alignment was removed and the intermediate alternative installed. Thethree photos show the initial, 6 hour and 9 hour stages. The fan is aggrading.724.2. ResultsThe aggradation also occurred at different rates depending on the degreeof confinement (Figure 4.2). The rate of aggradation was highest during thecurrent confinement at 1.77 cm per hour, and lowest with no confinement(extreme option) at 0.48 cm per hour. The intermediate option had anaverage aggradation rate of 1.12 cm per hour. Figure 4.2: plotted net elevation change during the current confinement(CC), intermediate option (int) and natural (extreme option - Ext) scenar-ios. Change in confinement is indicated by the black arrows. The rate ofaggradation decreased as confinement decreased.The outcome of this experiment did not correspond with what was ex-pected to happen according to geomorphic theory. That is, that there ex-ists one and only one fan surface configuration that uniquely correspondsto planform geometry under specified input rates of water and sediment.Therefore, following surface geometry disturbance the geometry will returnto the equilibrium state. The fan did not return to its defined state of dy-734.2. Resultsnamic equilibrium after the current stopbank alignment was removed. Thisprompted further experimentation involving the use of the Aranz MedicalLaser scanner. The Aranz Medical Laser scanner can be used to generatehigh density spatial data. Therefore, in these experiments it was able toestablish bed surface states with greater accuracy to ensure more reliableresults than the widely-spaced data from the vernier Callipers.In addition, the experimental focus widened to include investigation ofhow the fan responded to the installation and removal of the current stop-bank alignment when it was in an aggrading state. This was to provide acomparison to the steady state dynamic equilibrium experiments.4.2.2 S2: equilibrium fanThis experiment was designed to repeat the steps of the previous experi-ment (S1), with some modifications to supply rates, run time, boundarysurface roughness and measurement tools. As the river had been unable tobraid when confined, both the water flow and sediment supply rates werearbitrarily lowered. In addition, the time allowed for the fan to achieve‘equilibrium‘ was significantly greater (130 hours, compared to the 40 hoursof the S1 experiment).In addition, during the experiment it was identified that the smooth sur-face of the lateral boundaries resulted in the flow getting stuck against themand not migrating elsewhere on the fan. This was solved by the installationof coarse-sediment coated strips along the boundaries, after which the flowreturned to migrating across the fan surface (Figure 4.3).744.2. Results Figure 4.3: Coarse strips of sediment nailed to the boundaries to increasesurface roughnessIn order to obtain more detail on the behaviour of the model, the AranzMedical Laser scanner was used to measure bed surface elevation at hourlyintervals, and a sediment output collector tray was used to capture the sed-iment leaving the fan for five minutes every hour. Dynamic equilibrium wasconsidered to have been reached when both sets of data matched up; thatis, the scanner showed minimal total surface change and sediment outputrate equalled sediment input rate. For the laser scanner, this meant that theDigital Elevation Models (DEMs) of Difference (DoD) showed minimal netchange (i.e., absolute values less than 1 mm averaged over the entire sur-face)(Figure 4.4), whilst for the sediment output tray, the dried and weighedoutput sediment fell within ±1 g/min of the input amount (Figure 4.4).Once both measurement tools were indicating equilibrium, the stopbankswere installed, monitored, and then removed.754.2. Results 05101520253035-0.3-0.2-0.100.10.20.30.40.50.60.777 82 86 88 91 94 97 100 103 103 109 112 115Sediment (g/min)Net Change (cm)Time (hours)S2. Establishing Dynamic EquilibriumNet Elevation ChangeSediment output3 per. Mov. Avg. (Sediment output)Figure 4.4: The graph shows the final hours before dynamic equilibriumwas established in the S2 experiment. The columns show the net change(overall change in elevation) whilst the line shows the sediment output values(input was 25g/min)In contrast to the previous experiment, in its equilibrium state, the fandid not aggrade when confined. Instead in the first six hours of confinement,net change was negative (degradation), and in the following six hours whilstnet change was positive, it was minimal (<1 mm) (Figure 4.5). In addition,during the confined state it was noted that no braiding in the upper reachof the fan occurred, only sheet flow.764.2. Results  -1-0.8-0.6-0.4-0.200.20.40.60.891 94 97 100 103 103 109 112 115 118 121 124 127 130 133 136 139Change (cm)Time (hours)S2. Net Elevation ChangeUnconfinedConfinedFigure 4.5: Plot of the net change between hours 91 and 139. Hours 91to 127 were the lead up to the fan being considered to be in a state ofdynamic equilibrium, with hours 109 to 118 to be ignored as a batch offiner sediment unfortunately entered the model causing significant incisioninto the fan apex extending down to the middle of the fan and subsequentlyextra additions of sediment were added to infill the trench effecting the netchange. The installation of the current stopbanks configuration is markedby the black arrow between hours 127 and 130, and then from hours 127 to139 the net elevation change in the fan corresponding to the installation ofthe current stopbank alignment is shown.As the experimental river was unable to braid when it was confined,and the Waiho braids in its confined state, it was thought that the flowto sediment ratio was incorrect. This prompted the need to repeat the S2experiment with reduced supply rates and a smaller flow to sediment ratio.4.2.3 S3: equilibrium fanIn the S3 experiment sediment and water supply rates were adjusted to besimilar with those of the Davies et al. 2003 and 2013 models (Table 4.2).774.2. ResultsTable 4.2: Comparison of water flow and sediment supply rates vs modelscale between different studiesModel scale Water flow rate Sediment supply rateClarkson, 1999 1:2000 10 ml/s 20 g/minCampbell, 2012 1:5000 6 ml/s 18 g/minThis study, 2017 1:3000 8 ml/s 16 g/minThe fan was allowed to reach equilibrium and then the stopbanks wereinstalled. The model was then run for eight hours (Figure 4.6) before thestopbanks were removed, and the fan continued as before, in its naturalstate.  -0.6-0.4-0.200.20.40.60.811.21.41.60 2 4 6 8 10 12 18 24 30 36 40 42 44 46 48 50 52 54Change (cm)Time (hours)S3. Net Elevation ChangeUnconfinedConfinedFigure 4.6: Plotted net elevation change values over time. The naturalstate was monitored for its final 46 hours, at which point it was consideredto be in dynamic equilibrium, and the stopbanks were installed confining it(the black arrow indicates this change). The confinement was monitored foreight hours.In this experiment aggradation did occur when the stopbank was con-fined. However, this aggradation occurred in the middle reach of the fan,784.2. Resultsnot in the upper reach (Figure 4.7). The DEM generated from the pointcloud data collected by the laser scanner included both the upper and mid-dle reaches of the fan; and because the aggradation was significantly high inthe middle reach, the net change over both upper and middle was positive(Figure 4.8). However, when a second DEM was generated, focussing onlyon the upper reach, it was found that net change was negative, the fan sur-face was actually degrading in response to the installation of the stopbanks(Figure 4.8).The behaviour of the confined S3 experimental fan did not correspond tothe prototype behaviour or that of previous studies because no aggradationoccurred in the upper reach when the fan was confined. This promptedtwo further sets of experiments. The first one was to test the response ofan aggrading fan to confinement. The second involved the use of variablesediment and water inputs, as it was thought that perhaps some degree ofvariability was required to induce similar behaviour in the model to that ofthe prototype.794.2. ResultsFigure 4.7: DoD image showing aggradation (orange) in the middle reachand degradation (green) in the upper reach (Change between hours 0 and1)804.2. Results  Figure 4.8: Plotted mean net change values over time, upper and middlevs. upper only4.2.4 S4: aggrading fanThis experiment was designed to investigate the response of an aggradingfan to the current stopbank alignment. The response was then compared tothe S1 experiment results, to show that the ‘dynamic equilibrium‘ definedin the initial work of Davies et al. (2013) was unlikely to have been true.The sediment input was increased to 20 g/min inducing an aggradingfan, the stopbanks were installed and the model was run for 4 hours untilsubstantial (1 cm) aggradation occurred (Figure 4.9). The stopbanks werethen removed, and the model run for 6 hours in its natural state.814.2. Results  Figure 4.9: Plot of net elevation change for the four hours of confinement,and following (indicated by black arrow) 6 hours of unconfinement.In both the confined, and the following unconfined states the fan ag-graded (Figure 4.10a and 4.10b). This aggradation occurred in the upperreaches, similar to that of the prototype.The continued aggradation in both the confined and released states ofthe fan, whilst at greater rates, is similar to the behaviour of the fan toconfinement and release in the initial experiment (S1), in which the fanexperienced aggradation after the removal of the stopbanks (Figure 4.2).824.2.Results  A B Figure 4.10: Aggradation in both the (a) confined, and then (b) unconfined (after the stopbanks had beenremoved) fan. Orange = aggradation, blue = no change, green = degradation834.3. Discussion4.3 Discussion4.3.1 Importance of spatially dense dataA comparison between the results of the four steady input experiments inthis study indicated that a greater depth of detail is required to accuratelymeasure change and establish surface states in microscale fan systems. Inthe initial studies of the Waiho River and fan system, Digital Vernier Cal-lipers, time lapse photography and Structure from Motion (SfM) techniqueswere used to measure change in the bed surface elevation, and subsequentlyestablish dynamic equilibrium in the experimental fans (Campbell, 2012;Clarkson, 1999). In those dynamic equilibrium experiments, the fan ag-graded when confined. However in the present study, in experiments wheredynamic equilibrium was established using the Aranz Medical Laser scannerwhich provides high density spatial data, aggradation did not occur whenthe fan was confined. It was only in the test where the occurrence of dy-namic equilibrium fan was established using Digital Vernier Callipers thataggradation occurred. This suggests that the dynamic equilibrium surfacestates achieved in this and earlier studies may have been false.In this study, Digital Vernier Callipers were only used to measure changeand establish dynamic equilibrium for the first experimental fan (S1). Uponreaching the defined dynamic equilibrium, the S1 fan aggraded when con-fined and then continued to aggrade when released from its confinement.According to geomorphic theory in regards to the concept of dynamic equi-librium and the effects of perturbations, the fan should have degraded whenreleased from confinement. The fan should have degraded to its surfacestate prior to the confinement, if it really had been in state of dynamicequilibrium.Dynamic equilibrium refers to the oscillation about a system state orstable equilibria (Knighton, 1998). This oscillation occurs because of self-regulation, and relies on the mean values of the governing variables remain-ing relatively constant (Knighton, 1998). Self-regulation is characteristic ofnegative feedback mechanisms. These mechanisms moderate the effect ofperturbations, allowing the approximate return of the previous state pro-vided the perturbation is damped down (Knighton, 1998). They can alsotend to overcompensate which causes an ongoing adjustment process, char-acterised by oscillations. In the case of the S1 experiment, the perturbationwas the stopbank which confines the river and fan, and the governing vari-844.3. Discussionables were the geometric inserts and sediment and water inputs, all of whichwere kept constant. If the S1 experimental fan was operating about a dy-namic equilibrium prior to confinement, once the perturbation (stopbank)was removed, according to geomorphic theory, it should have been able todegrade back down to the dynamic equilibrium it had achieved earlier underthe given sediment and water inputs, constant fan toe location, and confinedspace. However, it did not.The next two experiments, which were repeats of the S1 experiment butwith slight changes, behaved differently. They did not aggrade when con-fined, behaviour continued as it had been prior to confinement in its dy-namic equilibrium state. In these experiments, dynamic equilibrium wasestablished using a high spatial density point cloud system achieved by theAranz Medical Laser Scanner and GIS.As a comparison to the dynamic equilibrium fan scenarios, a fourth ex-periment was conducted on an aggrading fan. It also used the laser scannerand GIS. In this experiment the fan behaved in the same way as the initialS1 experimental fan. Aggradation increased once confined, and continuedonce the confinement was released (albeit at a reduced rate).The similarity in fan response to stopbank installation and removal be-tween the S1 and S4 experiments, and the difference when compared tothe S2 and S3 experiments, suggests that the first experimental fan had notachieved dynamic equilibrium when the stopbanks were installed. The digitalVernier Callipers did not provide a spatial density of points high enough toaccurately capture change in the fan compared to that of the laser scannerin the other three experiments. This also then suggests that the microscalefans in the studies by Davies et al. (2003a; 2013) had also not reached a stateof dynamic equilibrium. In his thesis Campbell (2012) does mention thathe is unsure if his micro fan had reached equilibrium when the stopbankswere installed. However, in neither study was the response of the fans to theremoval of the stopbanks investigated, and as all three microscale modelsdiffer in scale and inputs, this makes comparisons unreliable. Nevertheless,it emphasises an important point for microscale modelling; the importanceof collecting spatially dense data to ensure the veracity of the experimentand results.854.3. DiscussionMicroscale models are very small replicas of nature, so change itself is alsoquite small. Therefore, measuring and monitoring such systems does requirea close attention to detail. Based upon the results from this study, and acomparison with the previous two studies on the Waiho system, the digitalVernier Callipers and qualitative methods such as time lapse photographyand Structure from Motion techniques cannot collect data at the spatialdensity required for accurate results. However, tools such as laser scannerswhich produce high density point cloud systems, can do this, as shown in thisstudy. This is not the first study to have utilized laser scanner technologyto generate high density point cloud systems. Both Wild (2012) and Daviesand Korup (2007) have utilized laser scanners successfully for studies on theDart and Rees rivers, and Westland fan systems respectively.However, further work could be done to investigate how accurate differentmeasuring techniques are for microscale modelling. Of interest would beto run an experiment using a range of measurement techniques includingthe Callipers, SfM, time lapse photography, and laser scanners to directlycompare the behaviour of the fan and the dynamic equilibrium established.4.3.2 Effect of boundary surface roughnessObservations from the steady input experiments, particularly the S2 ex-periment, indicate that the surface roughness of the boundary inserts hasan effect on model performance. Previously, it was believed that the effectsof surface tension and friction coefficients were negligible at the micro scale(Lajeunesse et al., 2010; Peakall and Warburton, 1996; Me`tivier and Meu-nier, 2003; Malverti et al., 2008). However, in this study, the polystyrenegeometric boundary inserts adversely affected model river and fan behaviour.In the S2 experiment it became obvious that the flow was getting ‘stuck‘against the true right boundary insert. At first it was thought that thepersistent flow down that side of the fan was indicating the fan surface hadachieved a state of dynamic equilibrium. But by observing the rest of thefan it was obvious that this dynamic equilibrium was false, as the rest of thefan surface was lower than the true right, with low spots untouched by theflow. What was actually happening was that the smooth wall of polystyrenewas creating a path of least resistance for the water, resulting in the flowremaining ‘stuck‘ to that side for up to ten hours. Tal and Paola (2010)also noticed this tendency in their flume studies. However by lining thewalls of their flume with coarse mesh, constructing a buffer zone of wet sand864.3. Discussionand installing wooden groins, they were able to create a rougher surfacewhich prevented the flow from sticking to the walls (Tal and Paola, 2010).Introducing surface roughness also worked for this study. Coarse strips ofsediment were added to all of the geometric boundary inserts, after whichflow returned to migrating back and forth across the fan.This seems like quite an obvious incident to observe in a microscale model,particularly as in both the New Zealand and the United States of Americastudies polystyrene is the material of choice for a majority of microscalemodels. However, perhaps due to the length run of times it may not havebeen noticeable in these experiments.In this study, the S2 experiment was run for 130 hours before it wasconsidered to be in dynamic equilibrium, with the coarse sediment stripsbeing introduced around the 70 hour mark. This was in addition to the73 hours of run time for the S1 experiment. By comparison, in the studyconducted by Clarkson (1999), experiments were monitored on average for 6hours; whilst in studies conducted by the USACE, hydrograph cycles rangedfrom 1.8 to 6 minutes (Gaines and Maynord, 2001; Maynord, 2006). In theUK, the maximum run time for a series of experiments conducted on alluvialfans by Clarke et al. (2010) was 12 hours. None of these times appear to belong enough to observe flow tendencies in the model. Furthermore, in hisstudy of the Waiho River and fan system, Campbell (2012) noted that at thetime of the stopbank installation it was unlikely that the fan had reacheda state of equilibrium, despite having been run for 81 hours. Perhaps thiswas due to unobserved interference from the smooth polystyrene boundarywalls.Microscale models have been used successfully to predict prototype fanbehaviour (Davies et al., 2003a, 2013), further understanding of alluvialfan processes (Davies and Korup, 2007), and facilitate the design of controlwork structure in rivers (Gaines and Maynord, 2001). However, they havenot been exposed to a thorough exploration of the underlying principlesor practice. The results of this study suggest an exploration is necessary,particularly in regards to the effects of surface roughness. Such a studycould investigate and compare how models fans respond to polystyrene andother materials.87Chapter 5Data analysis: variable inputexperiments5.1 IntroductionIn this chapter the results from the variable input experiments are de-scribed and reviewed. This set of experiments follow on from the previousset of steady input experiments (Chapter 4). The steady input experimentsdid not produce the fan behaviour that was expected, and it was hypothe-sized that by introducing input variability, this might be rectified. Ratherthan testing the fan response to the two alternative stopbank alignments,these experiments address the more fundamental topic of the response of thefan to variability, in terms of its behaviour and its ability to reach a stateof dynamic equilibrium.The variability introduced was based on experiments conducted by Daviesand Korup (2007). In their study of fan surface dynamics, a siphon was usedto simulate a fluctuating hydrograph of floods, base flows and no flow. Inaddition the Tinker sediment feeder, instead of delivering a constant rate ofsediment into the flow, deposited sediment onto a circular disc. The sedi-ment built up on this disc creating a conical angle of repose which generatedintermittent, variably-sized avalanches off the sides; these were collected bya funnel and delivered to the water inflow. This created irregular sedimentinput in the short term, with a long term measurable average.The siphon was initially used in this study, however it became unfeasibledue to the very low flows required. Instead, an electronic timer was usedto generate the fluctuating hydrograph. The hydrograph consists of twoperiods, a high flow (flood) and low flow (base).885.2. Results5.2 ResultsThe variable input experiments differed based on the surface state of thefan when the stopbanks were installed, and the hydrograph used (Table 5.1).Two different hydrographs were used, however they shared the same meanflow of 9.25 ml/s.5.2.1 V1: small variabilityThe V1 experiments were designed to be comparable to the S3 and S4steady input experiments in order to investigate the effect of variability onmicroscale model fan behaviour and its ability to reach dynamic equilibrium.The V1a and V1b experiments operated with a hydrograph of fifteen minutesat 8 ml/s followed by fifteen minutes at 10.5 ml/s, with a mean flow of 9.25ml/s. The response to confinement was tested for both an equilibrium fan(V1a) and an aggrading fan (V1b).In the V1a equilibrium experiment, the model was turned on and left torun for 54 hours. Hourly scans began at hour 20, and by the 54th hour,the fan was considered to be in dynamic equilibrium (Figure 5.1). The timetaken for the model fan seemed to be longer than that under the steady inputconditions. However as it is uncertain whether the experiments all startedfrom the same bed surface elevation, this cannot be categorically confirmed.Once dynamic equilibrium was established, the current stopbank alignmentwas installed and the model monitored for four hours (Figure 5.1).895.2.ResultsTable 5.1: Summary of the four variable input experiments, including surface state when stopbanks were installed,input conditions (sediment and water), scenarios tested (Nat = natural, CC = current confinement, and Nat2 =released from confinement), measurement tools and outcomes. (15min = 15 minutes)Input conditionsNumber Surface state Sediment(g/min)Water (ml/s)ScenariosMeasurementtool/sOutcomeV1a Equilibrium 18Mean 9.2515min (8)15min (10.5)NatCCLaser scanner No aggradaation when confined.V1b Aggrading 18Mean 9.2515min (8)15min(10.5)NatCCNat2Laser scannerFan continued to aggrade when confined,and also when releasedV2aAttemptedequilibrium18Mean 9.2515min (6.5)15min (12)NatCCLaser scanner Model DID NOT reach equilibriumV2b Aggrading 18Mean 9.2515min (6.5)15min (12)NatCCNat2Laser scannerFan continued to aggrade whenconfined, and also when released.905.2. Results -2-1.5-1-0.500.511.5220 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54/0 2 4Change (cm)Time (hours)V1a. Net Elevation ChangeUnconfinedConfinedFigure 5.1: Plotted net elevation change of the V1a experiment betweenhours 20 and 58. Hours 20 to 54 are of the fan in its natural state, whilsthours 54 to 58 are of the confined fan (beginning marked by the black arrow).In the lead up to achieving dynamic equilibrium the fan behaved quitedifferently to that of the steady input experimental fans. In particular, fanhead trenching became much more obvious. Towards the end of the 10.5ml/sfifteen minute interval, a relatively deep trench would form in the fan apex(Figure 5.2a and b), shifting sediment deposition to the middle and lowerreaches of the fan where it would then be transferred out of the model.Sediment accumulation below the free over fall was observably higher thanin the steady input experiments.915.2. Results  A B Figure 5.2: Photo (A) and DEM of Difference (B) of the unconfined fanshowing a definite trench in the apex at hour 24In addition, the fan exhibited catch up behaviour in response to the fanhead trenching. During the first half of the smaller flow interval, the fanhead trench would remain obvious (Figure 3a). However, as the flow tosediment ratio was smaller, sediment could accumulate in the trench untilin the second half of the interval the trench would be infilled and braidingwould reoccur (Figure 3b and c).925.2. Results   C B A Figure 5.3: Fan exhibiting catch up behaviour. A)first half of low flowinterval and the trench is still obvious B) second half of the low flow intervaland sediment has begun to accumulate C) end of the low flow interval andbraiding is obvious.As with the S2 and S3 equilibrium fan experiments, the fan did not ag-grade when the current stopbank alignment was installed. Consequently,the V1b aggrading experiment was conducted in order to confirm the fanwould respond in the same way the S4 aggrading fan had to confinement.935.2. ResultsIn the V1b experiments, the fan was monitored for four hours as it builtup in its natural state, and then whilst it was still aggrading the stopbankswere installed and it was monitored for a further four hours (Figure 5.4).The stopbanks were then removed and the fan monitored for a final fivehours (Figure 5.4). Figure 5.4: Plotted net elevation change of the V1b experiment. The graphshows three different states of the experiment, the initial natural state, theconfined state and then the return to natural state. These changes aremarked by the black arrows.As in the steady input aggrading fan experiment (S4), the fan aggradedwhen confined, and then when the stopbanks were removed it continued toaggrade, infilling the newly available space (Figure 5.5).945.2. Results   C B A Figure 5.5: Time lapse of change after the current stopbank alignmentis removed. A) 0 mins B) 20 minutes c) 40 minutes. Fan is continuing toaggrade regardless of stopbank removal.5.2.2 V2: large variabilityThe V2 experiments were designed as a comparison to the V1 experiments.They were run under a hydrograph with a much larger difference betweenthe two fifteen minute flows whilst still maintaining the mean of 9.25 ml/s.The two fifteen minute intervals comprised a 6.5 ml/s flow and a 12 ml/sflow. As with the V1 experiments, both an equilibrium fan (V2a) and anaggrading fan (V2b) were tested. In addition, the maximum elevation fromeach DEM was recorded, to show the change in fan head height over time.955.2. ResultsIn the V2a experiment the model was turned on and left to run for 62hours, with scans from hour 20 at hourly intervals. In addition to the meannet change data extracted from the DoD, maximum elevation data was alsorecorded (Figure 5.6). 208210212214216218220222224226228230-0.6-0.4-0.200.20.40.60.811.220 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62Maximum elevationNet Change (cm)Time (hours)V2a. Attempting Dynamic EquilibriumNet elevation changeMax bed elevation3 per. Mov. Avg. (Max bed elevation)Figure 5.6: Plotted net elevation change and maximum elevation valuesfor the V2a experiment. No stopbanks were installed, as this experimentwas ended before dynamic equilibrium was reached.During the experiment for a period of 2 hours (hours 55 and 56) scans weretaken every 15 minutes to see if much was being missed by only scanningat the end of the flood flow (Figure 5.7). The 2 hour period of 15 minuteinterval scans showed much the same fluctuating pattern about 0, just at asmaller scale because of the reduced time between scans.965.2. Results -0.5-0.4-0.3-0.2-0.100.10.20.30.455 55.25 55.5 55.75 56 56.25 56.5 56.75 57Change (cm)Time (hours)V2a. Net  Elevation Change15 min intervals Hourly intFigure 5.7: 2 hour period with scans completed every fifteen minutes inorder to capture the change between each stage in the hydrograph.The upper fan region did not reach dynamic equilibrium in this experiment(based on the definition of equilibrium for this study). Instead, the modelfan continued to aggrade slowly. At 62 hours its maximum elevation atthe head was only 227. At 54 hours the previous variability experiment(V1a) with the same water flow mean, had reached equilibrium with a headelevation of 240. The experiment was ended at 62 hours.In hindsight this experiment would have benefited from a longer run time.This would have allowed observation of the pattern that seems to begin athour 51. From this point onwards net change undergoes large fluctuationsabout zero. This could be evidence of a very dynamic equilibrium. The netchange data suggests that during this time the microscale fan experiencedepisodes of high aggradation followed by periods of slower aggradation, aswell as episodes of degradation. Overall, in the long term the fan was ag-grading.975.2. ResultsA second experiment (V2b) using the same hydrograph as V2a was thenrun to check the response of an aggrading fan. The V2b experiment pro-ceeded in a similar fashion to the V1b aggrading fan experiment. The modelwas run for 3 hours in its natural state, and then while it was still aggradingthe current stopbank alignment was installed and monitored for a further 3hours. The stopbanks were then removed and the fan monitored for another3 hours (Figure 5.8). 200202204206208210212214216218-0.4-0.200.20.40.60.810 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9Maximum Elevation Net Change (cm)Time (hours)V2b. Aggrading FanUnconfined ConfinedMax elevation 3 per. Mov. Avg. (Max elevation)Figure 5.8: Plotted net elevation change and maximum elevation values forthe V2b aggrading fan scenario. The experiment consisted of three stages,natural, confined and natural again, these are marked by the black arrows.There was an initial spike of high positive net change when the stopbankswere installed, however this did not last. A small fluctuating aggradingpattern developed after that first half hour (Figure 5.8). When the stop-banks were removed there was another initial spike, this time of negativenet change. Again, this did not last, the fluctuating aggrading behaviourresumed after that initial spike (Figure 5.8).985.3. Discussion5.3 Discussion5.3.1 Fan behaviourThe variable input experimental fans behaved differently to the steadyinput experimental fans. Many alluvial fan studies use constant input con-ditions (Davies et al., 2003a, 2013; Clarke et al., 2010; Whipple et al., 1998;Reitz and Jerolmack, 2012; Reitz et al., 2010). Under these steady inputconditions, the experimental fans follow a predicted set of developmentalsteps (Bryant et al., 1995), that involve a range of behaviours (Clarke et al.,2010). Of these behaviours, fan head trenching tends to occur in the laterstages of fan formation, when it has neared or reached maximum growth(Clarke et al., 2010; Whipple et al., 1998). This was not the case for thevariable input experimental fans in this study.In both the V1 and V2 sets of experimental fans, fan head trenchingoccurred every time the water discharge was increased. In the V1 exper-iments, this was a 31% increase whilst for the V2 experiments this was a85% increase. In both cases, after the water input was increased flow wouldbecome unchannelized forming sheet flow for several minutes before a deepwide trench in the apex of the fan formed (Figure 5.2). The trench alwaysformed in the apex, but it would switch to different alignments out into themiddle and lower reaches of the fan between each cycle of the hydrograph.During the higher flow intervals sediment accumulation below the free over-fall was also high. These behaviours appear to be a result of the increasedwater discharge increasing the ability of the fan to move sediment across itssurface. As there was no increase to sediment supply or lowering of the gra-dient when the water discharge was raised, the stream power was increasedand therefore the ability of the fan to transport sediment.By contrast, in the steady input experiments, fan head trenching oc-curred only when the fan began to approach a state of dynamic equilibrium.The response of the experimental fans to variability in this study is similarto another study. In their study of the Westland fan network, Davies andKorup (2007) constructed a microscale model of the Poerua river and fansystem. This model was fed by variable sediment and water inputs, and theresults showed that the irregular inputs of sediment had a direct effect uponfan head trenching behaviour that were not apparent in the steady inputexperiments.995.3. DiscussionWhen steady input conditions are used, the assumption must then bethat the fluctuations in sediment and water supply in nature can be av-eraged out over the scale of interest in a model i.e. long term evolution(Jerolmack, 2011). However, alluvial fans are incredibly dynamic systems,whose behaviour is often dictated by variability as has been shown by thisstudy and that of Davies and Korup (2007). The variability experimentshave produced behaviours that are not present in steady input experiments.This has implications for prototype-related investigations which use steadyinputs. A prototype fan will frequently experience and respond to some ran-dom higher than normal flood event or large sediment input. However, thisresponse will not be expected by engineers or hydrologists, if a steady inputexperiment has been used where there is no variability to produce such a re-sponse. For the sake of accurate and reliable investigations, it would appearthat modellers must find a way to incorporate variability into their modelsif they wish to reproduce those behaviours that occur in nature.5.3.2 Fan surface stateThe variable input conditions also affected the ability of the experimentalfan to reach a state of dynamic equilibrium. This effect differed dependingon the degree of variability. Small variability in the hydrograph between thebase and flood flow appeared to have less of an impact than larger variability.Similarly Davies and Korup (2007) showed that larger, irregular inputs ofsediment also had a greater impact on experimental fan behaviour.The V1a and V2a experiments were designed to allow the fans to reachdynamic equilibrium under variable conditions before the stopbanks wereinstalled. In the V1a experiment, variability was small. The base flow was8 ml/s, whilst the flood flow increased by 31% to 10.5 ml/s. Therefore therewas a 2.5 ml/s increase in flow between the two 15 minute intervals thatformed the hydrograph cycle. In this scenario, whilst the fan did reach adynamic equilibrium, it appeared to behave differently to the steady inputexperiments, with fluctuations between net change more gradual. How-ever, when dynamic equilibrium was reached and the stopbanks installed,it responded in the same way as the steady input experiments. No netaggradation occurred.By contrast the V2a experiment was ended before dynamic equilibriumwas reached, because of the extremely slow growth of the fan. In both theV1a and V2a experiments maximum elevation was measured at the head of1005.3. Discussionthe fans. In the V1a experiment, by the time the fan had reached a stateof dynamic equilibrium (fifty four hours), the head of the fan had aggradedto a relative elevation of 240. By 62 hours in the V2a experiment, the headof the fan was only at 228. Despite the same mean input values as the V1aexperiment, the greater variability in the V2a experiment with a base flowof 6.5 ml/s, and flood flow of 12 ml/s, clearly affected the progress of thefan towards dynamic equilibrium. In addition whilst the V1a fan was ableto develop a fluctuating net elevation change pattern early on (hour 21), theV2a did not. However the V2a fan does appear to achieve the beginningsof a fluctuating pattern at hour 51. Perhaps if the V2a experiment hadbeen left to run for much longer, it would have reached a higher maximumelevation, established its fluctuating pattern further and even achieved astate of dynamic equilibrium. Unfortunately, due to slow fan developmentand time constraints, it did not.These results raise the question of what effect irregular and large land-slides or flood events have upon a natural fan system and its evolution.For example, the Waiho river and fan system is in a location where sedi-ment input from landslides and glacial activity is high and irregular, andintense high rainfall storms frequently occur in the area causing high flowvariability. In addition the fan is located on the Alpine fault, which rup-tures several times per millennium, resulting in earthquakes which generateextreme shaking that causes very high sediment inputs due to coseismicland sliding (Robinson and Davies, 2013). The combination of these factorsmake for an extremely dynamic system. Can such active systems with somuch variability reach a state of dynamic equilibrium? And if they can,how long would this take or last? And how dynamic would this equilibriumbe? These types of questions need to be addressed and investigated, as theyhave implications for how we understand and manage such systems.Further studies on the impact of flow and sediment input variability onalluvial fan processes and evolution would be beneficial to our understand-ing of the dynamic nature of alluvial fans, and could perhaps improve ourmanagement of these systems. An avenue for research lies in further inves-tigation of the effects of irregular sediment and water inputs in laboratoryconditions and comparisons to both steady input experiments as well asobservations made in the field.101Chapter 6Application to the WaihoRiverThe purpose of this study was to investigate alternate stopbank align-ments for the Waiho River and fan system that would facilitate the returnof the bed to more manageable levels. This would reduce the severe floodrisk posed by the system to the neighbouring town, State Highway 6, andsurrounding farmland, and alleviate the need for ongoing and potentiallydangerous height additions to the current stopbank network. However, theresults of this study, discussed in depth in chapters four (steady input ex-periments) and five (variable input experiments) have produced some unex-pected conclusions and questions which indicate that a solution the Waiho-Franz Josef problem may not be able to be usefully based on micro-scalemodelling until better field information can be acquired.6.1 Outcome of the alternate stopbankalignmentsThe two stopbank alignments tested released the Waiho River from itspresent confined state to different degrees of confinement on the southernside of the fan. The intermediate option doubled the area that the rivercould utilize, whilst still protecting a large portion of the farmland for con-tinued human use. Alternatively, the extreme option freed the river com-pletely, allowing it full access to the farmland and the entirety of its naturalaccommodation space.The rationale and theory behind these options was based on the conceptsof dynamic equilibrium and the effects of channel confinement discussedearlier in chapter four. These assumptions have led to the belief that thecurrent stopbank alignment confined the river to an area of the fan thatwas too small for it to behave naturally. Traditionally, confinement of ariver has been designed to increase the depth of the flow, thereby increasing1026.1. Outcome of the alternate stopbank alignmentstransport capacity and subsequently promoting degradation. However, theWaiho River is still able braid at a range of flows within the confined reaches.This ability to braid suggests that confinement has not increased the depth,only limited the ability for it to laterally migrate. For this reason, theWaiho River has departed from its state of dynamic equilibrium, into oneof rapid aggradation (Figure 6.1). Therefore, realigning the stopbanks toreduce the degree of channel confinement should result in degradation, andthe return of the fan surface to its previous state of dynamic equilibrium.However, neither the extreme nor the intermediate realignment schemesinduced degradation in the micro-scale model. y = 0.0002x + 138.71R² = 0.9166138.000140.000142.000144.000146.000148.000150.000152.000Mean Bed LevelDateWaiho River Bridge Mean Bed Level - Long Term TrendMean Bed Level Linear (Mean Bed Level)Figure 6.1: Mean bed level data at the Waiho River bridge site plottedfrom 01/01/1907 to 11/09/2017 provides evidence for the aggrading riverbed over the last 111 years. Data was collected by NIWA and NZTA, andcompiled by Mark Healey at Opus Consulting Inc.In all of the dynamic equilibrium fan scenarios for the present modellingexercise, the fan did not aggrade when confined by the current stopbank1036.2. Surface state of the Waiho fanalignment, and additionally did not degrade or aggrade when the degree ofconfinement was reduced (Figured 4.5, 4.6 and 5.1). During the aggradingfan scenarios, channel confinement increased the local rate of aggradationby reducing the accommodation space available to the fan, which matchesobservations of the Waiho River in the field. However, when the degreeof confinement was reduced or removed altogether, the fan continued toaggrade albeit at a slower pace (Figures 4.9, 5.4 and 5.8). This suggest thatthere must be some aggradation present for the confinement to exacerbate.These results cast doubt on the assumptions that have been made withrespect to the behaviour and surface state of the Waiho fan prior to humaninterference.6.2 Surface state of the Waiho fanWhen people first arrived and settled across the Waiho fan surface, it isbelieved that the fan had reached a long term state of dynamic equilibriumfor a number of reasons based on:• Fan theory for dynamic equilibrium: constant fan toe position, limitedaccommodation space, river entrenched into the fan head• Assumed long-term steady inputs of water and sediment (based onassumed constant climate and tectonics)• Other fans in the Westland region appear to have (or are assumed tohave) reached a state of dynamic equilibrium• Young surface soil stratigraphyFollowing human settlement, stopbanks were gradually installed to protectthe growing town, highway and farmland from flood hazards. A recentconcept is that it was the confinement of the Waiho River by these stopbankswhich prompted the rivers shift from a state of dynamic equilibrium to oneof aggradation (e.g. Davies et al, 2003a).As outlined above, none of the dynamic equilibrium fan scenarios in thisstudy exhibited aggradation when confined by the current stopbank align-ment. It was only the aggrading fan scenarios in this study which exhibitedaggradation when confined. Similarly, Campbell (2012) suggested that his1046.2. Surface state of the Waiho fanWaiho experimental fan was also not in a state of dynamic equilibrium whenthe stopbanks were installed, but rather one of aggradation. In his modelthe fan aggraded when confined. Furthermore, in the various aggrading fanscenarios, confinement resulted in an increased rate of aggradation, whichis similar to what has been observed in the field over the past few decades.These discrepancies call the assumptions about the equilibrium nature ofthe Waiho fan into question.6.2.1 Was the Waiho in dynamic equilibrium?Alluvial fan theory and quantitative relationshipsAlluvial fan theory suggests that the Waiho fan surface had reached astate of dynamic equilibrium when people first arrived because the riverhad become entrenched into its fan head (Davies and McSaveney, 2001), sealevel had been constant for 6,000 years, and therefore so had the Waiho fantoe (Davies and McSaveney, 2001), and the fan has a limited accommoda-tion space, confined as it is by steep moraine walls on both sides. Therefore,it should have had enough time to fill its accommodation space to an equilib-rium state. However, a review of the quantitative relationships establishedbetween drainage basin and fan size and steepness suggest otherwise.During the mid-20th century, Bull (1964) found that the larger the basin,the larger the fan. In addition, drainage basins composed of mudstone orshale produced larger fans compared to those composed of sandstone dueto the higher erodibility of the former materials. Ryder (1971) also linkeddrainage basin relief to fan steepness. Steeper average basin relief resulted insteeper fans. More recently, experimental fan slopes have been shown to varyas a result of the sediment-discharge ratio, with steeper slopes associatedwith a higher ratio (Clarke et al., 2008; Zarn and Davies, 1994).When applying these relationships to the Waiho fan system, it would seemthat whilst the Waiho fan cannot increase its surface area it could still belarger in volume, and hence steeper in slope. Many of the mountains inthe Waiho drainage basin exceed 3000m and as a result of their location inan area of high tectonic uplift rate, the relief is on average very steep (Mc-Saveney and Davies, 1998). The geology of the region is that of greywackeand schist, and when combined with the high rainfall of the region, the re-sult is a high erosion rate and therefore high sediment supply (McSaveneyand Davies, 1998). Finally, the 170 km2 area could be considered quite large1056.2. Surface state of the Waiho fancompared to the 64 km2 fan. Thus, it would seem reasonable that the Waihofan could still have been growing when people first arrived. Despite its fixedtoe it could still have been aggrading, increasing in elevation and thereforein steepness.Furthermore, a constant fan toe does not necessarily imply equilibrium,as shown by the experiments in this study. Prior to any experimentation,the fan was allowed to build up without interference to where its toe hadreached the free overfall. This overfall acted as a constant fan toe. In allof the experiments aggradation continued despite the fan toe being reached,and sediment being expelled from the system.Long term steady inputs and a severe lack of dataDavies and McSaveney (2001) suggest that the Waiho and the other West-land fans would have been experiencing long term steady water and sedimentinputs, and as this has been for a substantial amount of time, the fans shouldtherefore be in a state of dynamic equilibrium. However, this is based uponthe average uplift of the mountains, and average rainfall data from an as-sumed climate. These are not recorded data. There are no such data forthe Waiho’s past, only an inferred history.The assumed history for the Waiho, whilst plausible, is not consistent towhat is understood about the impact of glacial behaviour or tectonic activ-ity upon sediment production rates. Nor is this history definite enough todetermine what the surface state of the fan or its behaviour was prior tohuman interference. Notes and sketches made by the early explorers and in-habitants do not extend further back than 150 years, and these accounts arenot accurate or reliable enough to make such a decision. Aerial photographydoes not go back to when the fan was unconfined by stopbanks.In addition, it is known that climate has changed dramatically over thelast 200 years (recovery from the Little Ice Age) and that the Franz JosefGlacier has undergone several advance and retreat phases during this time.The last Alpine fault earthquake was 1717, and would have had a majorsediment input; Davies et al. (2005) found evidence for several metres ofaggradation on the Waiho fanhead since before that event, again castingsuspicion on the steady-state assumption. These incidences could causeshort term deviation away from a long term equilibrium. By inferring long1066.2. Surface state of the Waiho fanterm inputs, the perturbations and resulting aggradational or degradationalepisodes are ignored.Long term sediment and water inputs can be inferred, but it is only aninference. Variability which exists everywhere in nature, and which may be acritical component of behaviour, is lost when something becomes consideredas steady in the long term. By doing this, the effect of large and small flowevents, sediment deficits and surpluses, climate change, glacial recession andadvance, and Alpine fault movements on fan behaviour and developmentare not taken into account and therefore not understood. People were notaround to measure and monitor such events, so inferences are made which actto simplify fundamentally dynamic and complex systems. Ultimately thisleaves us without the knowledge of how these events may have impacted andcontinue to impact fan behaviour and development.Even now, regular and ongoing collection of data sets of this system arenon-existent. Flow rates are not measured, bedload transport not moni-tored. There is little knowledge of sediment and water inputs and transportrates within the Franz Josef glacier, or how the glacier has and continuesto affect the river and fan below. If our knowledge of this system is limitednow, how can we accurately infer its past behaviour and surface state, orpredict future changes for the Waiho Fan?Soil stratigraphySoil stratigraphy shows that the soils on the inactive fan surface are allrecent, with few more than about 500 years old due to constant rework-ing of the face surface by the shifting river (Davies and McSaveney, 2001;Davies et al., 2003a). This indicates rapid process rates which suggests thatthere should have been enough time for the longitudinal profile of the fan tohave been in long term dynamic equilibrium (Davies and McSaveney, 2001).However, rapid process rates do not necessarily mean that there has beenenough time for the Waiho fan to have reached a state of dynamic equilib-rium 150 years ago. A review of how fans develop suggests that these youngsoil layers are also indicative of a fan still in development.Fans develop via a migrating channel network which shifts back and forthacross the entire fan surface, reworking, transporting and depositing sedi-ment (Blair and McPherson, 1994b). This migrating behaviour has been1076.2. Surface state of the Waiho fandescribed as the “windescreen-wiper effect ” (Zarn and Davies, 1994). Itoccurs as a result of local sediment deposition. The accumulating sedimentraises the channel bed until it is higher than that of the surrounding land,resulting in the river breaching its banks and forming a new channel (Reitzand Jerolmack, 2012). This process has been observed to gradually shift lat-erally across the fan from one side to the other and back (Zarn and Davies,1994).How many times would the Waiho channel network need to migrate backand forth across the fan surface to reach dynamic equilibrium? What wasthe past avulsion rate? Was it slow? If so, inactive areas would remain inac-tive for some time before the channel network returned. More importantlyhow did having a large feature like the Waiho Loop, which would originallyhave extended across the entire fan width, have affected fan development?The exact behaviour of the Waiho River beyond 150 years ago is unknownbecause humans were not a round to record it. Therefore behaviour is onlyinferred. The age of the soils in the inactive part of the fan could indicatedynamic equilibrium, but they could also indicate growth.In this study, of the two dynamic equilibrium fan scenarios under vari-able conditions (Chapter 5), only one reached dynamic equilibrium after 54hours. At the 62 hour mark the second experiment was ended, and thatfan never reached a state of dynamic equilibrium. Both experiments sharedthe same sediment input and mean water input. In the steady input ex-periments (Chapter 4), all dynamic equilibrium fan scenarios took similarlylong time periods to reach a state of dynamic equilibrium. Given also, thatthe records indicate that the stopbanks were originally installed to protectthe surrounding land from a slowly aggrading river bed, then it would seemreasonable to infer that the young age of the soils on the inactive part ofthe fan are from a phase of development for the migrating channel system.Therefore, the soil stratigraphy could suggest that the Waiho fan was stillincreasing its elevation, and had not reached a state of dynamic equilibriumwhen people first arrived.Westland region fans and the complexity of the WaihoDavies and Korup (2007) argue that many of the fans along the WestCoast have reached dynamic equilibrium, therefore inferring that the Waiho1086.2. Surface state of the Waiho fanhas also had enough time to establish dynamic equilibrium. However, de-spite being subject to identical environmental forces during their evolution,landforms in the same region can be in different stages of geomorphic devel-opment (Schumm et al., 1987). For instance, alluvial fans along the sameregion may be experiencing active growth (Beaty, 1970), dissection (Huntand Mabey, 1966) or steady-state equilibrium (Denny, 1967). This seemsa likely diagnosis for the Waiho, given the difference between its catchmentand that of its neighbours.Of the West Coast river and fan networks, only two have glaciers thatpersist down to near sea level, the Waiho and the Fox. Yet, the two networksdiffer significantly in terms of the shapes of the valleys, tributaries, andglacial deposits. For these reasons, it seems that the two may not necessarilybe in the same stage of development (Carrivick and Rushmer, 2009).There has been little study of the Fox glacier, river, and valley, therefore,the following information has been gleaned from topographical maps of theregion (NZTopo50-BX15). The proglacial Fox river emerges from the FoxGlacier, flowing through several kilometres of confined valley before exitinginto more open terrain (Figure 6.2). This point of exit is symmetricallyplaced, and feeds down into an elongated valley, where the Fox river eventu-ally joins the Cook River. The Cook, a larger river, emerges on the Southernpoint of the valley and has formed a gently sloped fan. Glacial activity inthe open section of valley has been dominated by the steady retreat of theFox glacier, from which no distinct deposits remain. However, there hasbeen a very large volume of ice buried beneath the upper part of the FoxRiver since the 1960s. This has been slowly melting since then, providingaccommodation space for Fox sediment and reducing sediment input to theFox system (Bull, 2004). There has been no such buried ice in the Waihoduring this time (Bull, 2004).1096.2. Surface state of the Waiho fan Figure 6.2: Looking upstream through the transport reach of the FoxRiver towards the Fox Glacier terminus. The channel is dominated by largeboulders, very different to the Waiho where such boulders have been buriedbeneath the aggrading bed. Photo by Greg HewgillSimilarly in the Waiho catchment, the proglacial Waiho River emergesfrom the Franz Josef Glacier and also flows through several kilometres ofconfined valley before exiting into open terrain. However this is where somedifferences occur.• Prior to leaving the confined valley, the Waiho is joined by the CalleryRiver. The Callery contributes a greater percentage of catchment tothe Waiho system (Davies and McSaveney, 2001). However, apartfrom inferences made about its sediment and water contributions tothe Waiho River very little is known about the Callery. There are nosediment or water supply data sets (Davies and Scott, 1997). Therehave been no studies of how sediment slugs are moved through thesystem, or the effects of the glaciers upon sediment and water inputs.The narrow, steep gorge, surrounded by rugged terrain and dense veg-etation make study difficult (Davies and Scott, 1997).• To the true right side of the Waiho River after it has emerged frommountains the substantial Tatare fan has formed and prevented the1106.2. Surface state of the Waiho fanWaiho from accessing this Northern area of the valley. The TatareRiver is a much smaller river than the Cook River; where the Fox is atributary to the Cook, the Tatare is a tributary to the Waiho.• Carrivick and Rushmer (2009) found that despite sharing the sameclimatic conditions, and neighbouring catchments, the Fox and FranzJosef glaciers and their proglacial systems differ considerably. Thesedifferences relate to glacier area and length, lower glacier debris cov-erage, valley orientation, and glacial behaviour, with subsequent ef-fect on the sandur (outwash plane) below (Carrivick and Rushmer,2009). The result is that the Franz Josef glacier sandur features mas-sive, poorly sorted sediments with outsized clasts, whereas the Foxglacier sandur is of much finer-grained material (Carrivick and Rush-mer, 2009).• Finally, the glacial history of the Waiho valley may differ to that ofthe Fox. Two very obvious features are observed on the Waiho fan andflats, Canavan’s Knob, and the Waiho Loop. These are ancient relicsof past glacial behaviour. The Waiho Loop is in fact only a portion ofa terminal moraine deposit that reached around to the southern sideof the valley (Davies and McSaveney, 2001). The missing portion waseroded by the Waiho river where it was not protected by the higherTatare fan (Alexander et al., 2014). How would this moraine wall, andthe substantial infilling which eventually buried it, have affected thebehaviour of the Waiho as it has developed? There is no evidence ofa feature like the Waiho Loop in the Fox valley, however having noequivalent to the Tatare fan to keep the Fox and Cook Rivers awayfrom such a feature, it may have been eroded. Aside from Rachocki’s(1981) studies of the Polish fans in post glacial environments, therehave been no experiments to look at how fan behaviour and develop-ment is affected by glacial remains.Considering these differences, it seems likely that the Fox and WaihoRivers could behave differently, and that their fans could also behave andhave developed differently.6.2.2 An aggrading Waiho scenarioIf, in fact, the Waiho fan was still developing when humans first settledthere, then removal of the current stopbank alignment and installation of1116.3. Implications for the Waihoone of the alternatives might not solve the problem of the aggrading riverbed in the long-term.The results from the aggrading fan scenarios (S4, V1b, V2b) in this studyindicate that removal of the current stopbank alignment will result in someinitial degradation, but the fan will eventually fill up the newly availablespace, continuing its aggradation trend. This does buy some time, but astime cannot be quantitatively scaled up from the microscale model to itsprototype, it is unknown how much time could be bought.However, continuing to increase the current stopbank crest height andadd more stopbanks to the current setup is certainly not going to solve theproblem of the aggrading river bed, even in the short-term. It serves onlyto exacerbate an already dangerous situation. Steps need to be taken nowto alleviate the pressures on the township of Franz Josef Glacier, SH6 andflats farmland.In this study, the aggrading fan scenarios showed that the two alternativestopbank alignments resulted in a reduced rate of aggradation. Therefore, acontrolled shift of the Waiho River to the South of the valley could be put inplace using the two alternatives from this study or similar. This would allowtime for the gradual relocation of farmland owners and the township, andshifting of the SH6 to the 160 m contour along the bush line where it is outof harms way. Whilst this solution does not solve the aggrading river bed, itshould reduce the rate of aggradation, pressures on the northern stopbanks,and buy some time for relocations.Ultimately, if the river was aggrading when people first arrived, and ifits basic tendency is to aggrade, then aggradation will go on regardless ofhow many stopbanks are installed and however high they are built, withinevitable impact upon the surrounding land use.6.3 Implications for the WaihoThe results of the microscale modelling experiments from this study haveimportant implications for the management of the Waiho River and Fan.1126.3. Implications for the WaihoPreviously it has been believed that it was the combination of the fanbeing in a state of dynamic equilibrium and then severe confinement bythe stopbanks which resulted in the rapid aggradational trend in the Waihosystem. However, in none of the dynamic equilibrium fan scenarios in thisstudy did the current stopbank alignment induce aggradation. It was only inthe aggrading fan scenarios that an aggradational response occurred. Thisoutcome suggests that the Waiho fan was not in a state of dynamic equilib-rium when humans first settled there. If so, then the alternative stopbankalignments tested in this study will not solve the problem of a rapidly ag-grading river bed, but will reduce the rate of the aggradation and relieve thecurrent pressures on the northern stopbanks. This will provide the FranzJosef Glacier township with time to develop a long-term flood risk manage-ment strategy. How much time is unknown, as time in microscale modelscannot be quantitatively scaled up to the prototype.6.3.1 RecommendationsThe rate at which the Waiho river bed is currently aggrading, and thehazards posed by this aggradation, are an immediate concern to the neigh-bouring Franz Josef Glacier township, State Highway 6, and landholders ofthe Waiho flats farmland. However, as the results of this study have indi-cated that a solution is not straightforward, the following recommendation isonly a suggestion for the future which could provide some time, in the hopeof finding a more favourable outcome for the town, highway and farmland.At present, ongoing height additions to the current set of stopbanks keepsthe rapidly aggrading river at bay, but for how much longer? Increasing thestopbank height is not going to solve the problem. Whilst maintenance andheight additions do protect the town, road and farmland, they are only atemporary solution. Ultimately the additions add to the already dangeroussituation. Each time the stopbanks are made higher, the river bed is alsoable to aggrade higher within its confines so that it sits considerably higherthan the surrounding land and town (Figure 1.2). This is an extremelyhazardous situation.Alternatively, one of the stopbank alignments tested in this study couldbe installed to replace the existing one, or even both which would allow agradual shift of the river to the south (Figure 1.2; Langridge et al., 2016).This would involve considerable effort, time and money, to move both high-way and stopbank, and as shown by the results of this study, is unlikely to1136.3. Implications for the Waihostop the river bed aggrading. However, it would allow the river more space,reduce the current pressures on the northern stopbank alignment and FranzJosef township (Langridge et al., 2016), potentially slow down the rapid rateof aggradation and provide time before other action such as town relocationcan be taken.114Chapter 7ConclusionsThis study was designed to investigate the response of an experimental fan toconfinement and release, in order to provide information relevant to the man-agement of the Waiho River on the South Island of New Zealand. Through aseries of experiments it was demonstrated that fan behaviour in response toconfinement by stopbanks, and its ability to reach equilibrium, were affectedby the surface state of the fan prior to its confinement and by introducedvariability in the water and sediment inputs.The surface state of the experimental fan prior to confinement affectedhow the fan responded to the installation of the stopbanks. When the fanhad reached a state of dynamic equilibrium before the stopbanks were in-stalled, in both the steady and the variable input experiments, no aggrada-tion took place as a consequence of confinement. Alternatively, when the fanwas in a state of aggradation and the stopbanks were installed, in both thesteady and the variable input experiments, aggradation continued. However,while the aggradation rate was increased by confinement under the steadyinput conditions, this was not the case for the variable input experiments.Introduced variability in the sediment and water inputs of the microscalemodel had a noticeable effect on the behaviour of the experimental fan.The achievement of equilibrium on the fan with strongly varying water andsediment inputs took much longer than when the degree of variation wasless or zero. In addition, when the degree of variation was greater thanzero, particular fan behaviours such as fan head trenching occurred morefrequently and under different circumstances than when variation was zero.These findings have implications for both microscale modelling and forthe relevance of this technique to management of the Waiho River.1157.1. Implications for microscale modelling7.1 Implications for microscale modellingMicroscale modelling has proven to be a valuable alternative to other formsof modelling as it does not require a large amount of input data. However,as it is a relatively new form of modelling it has not received the attentionthat more traditional forms have. Therefore, the underlying theory and thepractices used are still open to modification and improvement.The microscale modelling experiments of the Waiho River and Fan in thisstudy have demonstrated that:• It is extremely important to assess accurately whether or not the fanhas achieved equilibrium prior to modification, in order to draw correctconclusions about its behaviour in response to modification.• There are grounds to question whether a microscale model understeady water and sediment input conditions, thus lacking any vari-ability in the form of turbulence, flow and sediment input variation,and sediment grain-size, can be expected to realistically represent thebehaviour of a river that has all these variations.There are thus opportunities for future investigation into the underlyingtheory and practice of microscale modelling. Opportunities exist in:• The effect of surface roughness at such a small scale. In the presentstudy, moulded fibre glass was used to represent stopbanks, similar tothe non-porous stainless steel used by Clarkson (1999) and Campbell(2012). How do these materials affect the model behaviour comparedto the pervious, steel mesh structures utilized in the USACE models(Gaines and Maynord, 2001)?In addition, the smooth surface of the polystyrene used for the modelboundaries in this study and a majority of other microscale modelshad a definite effect on the model fan behaviour, and this could dowith further investigation.• The methods used to measure change in the model. The present studydemonstrated that point-gauging could not provide the detail requiredto accurately establish fan surface states, whilst laser technology could.Further work could be done in exploring how accurate and reliabledifferent measurement tools are to ensure reliable results.1167.2. Implications for the management of the Waiho River• The study of the effects of variations in water and sediment inputs onalluvial fans.In addition to these opportunities, there are no guidelines on microscalemodelling practice. Therefore, an exploration and reporting of the tech-niques would be beneficial for anyone wishing to employ this type of mod-elling.7.2 Implications for the management of theWaiho RiverThe findings of this study have important implications for relevance of theprevious microscale modelling work completed by Davies et al (2003a; 2013,as well as for the current and future management of the Waiho River.Davies et al. (2003a; 2013) were able to demonstrate a good match be-tween model and Waiho River relative aggradation rates. However, thepresent study has shown that this match was probably achieved by confin-ing a model fan that was a aggrading, and not in dynamic equilibrium aspreviously thought. The present study also indicated that the assumptionby Davies et al. (2003a; 2013) that the Waiho was in dynamic equilibriumin the late 20th century when the stopbanks were first installed, is ques-tionable. Therefore, the modelling of Davies et al (2003a; 2013) may still beappropriate, and so the implication from their work that reducing the degreeof confinement of the Waiho will lead to degradation at the fanhead seemslikely to be correct. However, as shown by this study, this degradation maynot be permanent and in the longer term aggradation is likely to resume,albeit at a slower rate than with the present degree of confinement.Ultimately, in order to be able to predict and understand the future be-haviour of the Waiho River, a better understanding of the system itself isrequired. As this study has shown, understanding is presently limited. Inaddition, there is a severe lack of data for the behaviour of the Waiho Riverand its fan, which makes it difficult to model and manage effectively. Ongo-ing and regular monitoring of a number of components in this system wouldprovide valuable information for how and where it is changing.At present Lidar data are being collected, and are proving to be invalu-able to understanding where the fan is aggrading and future aggradation1177.2. Implications for the management of the Waiho Riverpatterns. In addition, advances in Structure from Motion software may al-low for event-scale aggradation patterns, volumes and rates to be calculatedfrom aerial imagery.However, other data sets would be valuable. These include:• Bedload and suspended sediment load. At present bedload data ac-quisition is extremely difficult, and in reality unfeasible, due to thebraided nature of the Waiho River and its large-calibre bedload dur-ing floods. However, should a method of acquisition become available,then it its use on the Waiho would be invaluable.• River flow. Flow data are also difficult to collect due to the reasonsdescribed above. However, to accurately model a river system, thisinformation is vital.• Impact of glacial behaviour and fault rupture upon sediment supply• Rainfall• Braid formationsWith these types of data sets, other forms of physical models as well asnumerical models could be used to study the behaviour of the Waiho Riverand fan, assist with further microscale modelling studies, and in generalimprove our understanding of such a complicated, powerful and dynamicsystem so that we may better manage its impact on the human environment.118BibliographyAcheson, A. R. (1968). River control and drainage in New Zealand: andsome comparisons with overseas practices. Ministry of Works.Alexander, D., Davies, T. R. H., and Shulmeister, J. (2014). Formation ofthe Waiho Loop terminal moraine, New Zealand. Journal of QuaternaryScience.Armstrong, L. (2003). Bank erosion and sediment transport in a microscalestraight river. Doctoral thesis, Universite Paris, 7, Denis Diderot, France.Ashmore, P. (1991). Channel morphology and bed load pulses in braided,gravel-bed streams. Geografiska Annaler. Series A. Physical Geography,pages 37–52.Aulakh, B. and Mills, L. (2016). Franz Josef flooding: right in the firingline. NZ Herald.Barr, D. I. H. (1968). Discussion of scale model of urban runoff from stormrainfall. Journal of Hydraulics Division (American Society of Civil Engi-neers), 94(2):586–588.Beaty, C. B. (1970). Age and estimated rate of accumulation of an alluvialfan, White Mountains, California, U.S.A. American Journal of Science,268:50–70.Bertoldi, W., Zanoni, L., and Tubino, M. (2009). Planform dynamics ofbraided streams. Earth Surface Processes and Landforms, 34:547–557.Blair, T. C. and McPherson, J. G. (1994a). Alluvial fans and their nat-ural distinction from rivers based on morphology, hydraulic processes,sedimentary processes and facies assemblages. Journal of Sediment andResearch, A64:450489.Blair, T. C. and McPherson, J. G. (1994b). Geomorphology of Desert En-vironments, Alluvial fan processes and forms, pages 354–402. Chapmanand Hall, London.119BibliographyBristow, C. and Best, J. (1993). Braided rivers: perspectives and problems.Geological Society, London, Special Publications, 75(1):1–11.Bryant, M., Falk, P., and Paola, C. (1995). Experimental study of avulsionfrequency and rate of deposition. Geology, 23:365–368.Bull, M. P. (2004). New Zealand Tales and Tours: South Island Adventures,Tour four - Westland, page 268. Trafford Publishing.Bull, W. B. (1964). Geomorphology of segmented alluvial fans in westernFresno County, California. U.S. Geological Survey Professional Paper,352-E:89 – 128.Campbell, B. (2012). A microscale modelling experiment to investigate theeffects of an avulsion of the Waiho River into the Tatare River, southWestland, New Zealand. Honours dissertation, University of Canterbury,New Zealand.Carrivick, J. L. and Rushmer, E. L. (2009). Inter-and intra-catchment vari-ations in proglacial geomorphology: an example from Franz Josef glacierand Fox glacier, New Zealand. Arctic, Antarctic, and Alpine Research,41(1):18–36.Chorley, R. J. (1967). Models in geomorphology. Models in Geography, pages59–96.Clarke, L. E. (2015). Experimental alluvial fans: advances in understandingof fan dynamics and processes. Geomorphology, 244:135–145.Clarke, L. E., Quine, T. A., and Nicholas, A. P. (2008). Sediment Dynamicsin Changing Environments, An evaluation of the role of physical modelsin exploring form-procss feedbacks in alluvial fans, pages 175–183. Inter-national Association of Hydrological Sciences.Clarke, L. E., Quine, T. A., and Nicholas, A. P. (2010). An experimentalinvestigation of autogenic behaviour during alluvial fan evolution. Geo-morphology, 115:278–285.Clarkson, P. J. (1999). Small-scale hydraulic modelling of alluvial fans.Master of engineering, Lincoln University, New Zealand.Darboux, F. and Huang, C. (2003). An instantaneous profile laser scanner tomeasure soil surface microtopography. Soil Science Society of AmericanJournal, 67(1):92–99.120BibliographyDavies, T., Campbell, B., Hall, B., and Gomez, C. (2013). Recent behaviourand sustainable future management of the Waiho River, Westland, NewZealand. Journal of Hydrology (New Zealand), 52(1):27–42.Davies, T. and Lee, A. (1988). Physical hydraulic modelling of width reduc-tion and bed level change in braided rivers. Journal of Hydrology (NewZealand), pages 113–127.Davies, T. R. and McSaveney, M. (2006). Geomorphic constraints on themanagement of bedload-dominated rivers. Journal of Hydrology (NewZealand), 45(2):111.Davies, T. R. H. (1987). Sediment Transport in Gravel-Bed Rivers, Problemsof bedload transport in gravel-bed braided rivers, pages 793–828. J. Wileyand Sons.Davies, T. R. H. (1997). Long-term management of facilities on an active al-luvial fan Waiho River fan, Westland, New Zealand. Journal of Hydrology(New Zealand), 36(2):127–145.Davies, T. R. H. and Korup, O. (2007). Persistent alluvial fanhead trenchingresulting fromlarge, infrequent sediment inputs. Earth Surface Processesand Landforms, 32:725742.Davies, T. R. H. and McSaveney, M. J. (2001). Anthropogenic fanheadaggradation, Waiho River, Westland, New Zealand. In Mosley, M. P.,editor, Gravel-Bed Rivers V, pages 531–553, Wellington. New ZealandHydrological Society.Davies, T. R. H. and McSaveney, M. J. (2008). Principles of sustainabledevelopment on fans. Journal of Hydrology (New Zealand), 47(1):43 – 65.Davies, T. R. H., McSaveney, M. J., and Clarkson, P. J. (2003a). Anthropicaggradation of the Waiho River, Westland, New Zealand microscale mod-elling. Earth Surface Processes and Landforms, 28:209–218.Davies, T. R. H., McSaveney, M. J., and Doscher, C. (2005). Final reporton research project no. 03/499 monitoring and effects of landslide-inducedaggradation in the Poerua valley, Westland. Technical report, EarthquakeCommission, Wellington, New Zealand.Davies, T. R. H. and Scott, B. K. (1997). Dambreak flood hazard fromthe Callery River, Westland, New Zealand. Journal of Hydrology (NewZealand), 36(1):1 –13.121BibliographyDavies, T. R. H., Smart, C. C., and Turnbull, J. M. (2003b). Water and sed-iment outbursts from advanced Franz Josef glacier, New Zealand. EarthSurface Processes and Landforms, 28:1081–1096.Davinroy, R. D. (1994). Physical sediment modelling of the Mississippi Riveron a micro scale. Master of science, University of Missouri-Rolla, Rolla,Mo.Denny, C. S. (1965). Alluvial fans in the Death Valley region of Californiaand Nevada. U.S. Geological Survey Professional Paper, 466:62pp.Denny, C. S. (1967). Fans and pediments. American Journal of Science,265:81–105.Eckis, R. (1928). Alluvial fans of the Cucamonga district, southern Califor-nia. Journal of Geology, 36:225–247.Fleming, G. (2002). Flood risk management: learning to live with rivers.Thomas Telford.Gaines, R. and Smith, R. (2002). Micro-scale loose-bed physical models. InHydraulic Measurements and Experimental Methods, pages 1–12.Gaines, R. A. and Maynord, S. T. (2001). Microscale loose-bed hydraulicmodels. Journal of Hydraulic Engineering, 27:335 – 339.Grant, A. P. (1948). Channel improvements in alluvial streams. Proceedings,New Zealand Institute of Engineers, 34:231–279.Grant, G. E. (1997). Critical flow constrains flow hydraulics in mobile-bedstreams: a new hypothesis. Water Resources Research, 33:349–358.Guerit, L., Me`tivier, L. G., Devauchelle, O., Lajeunesse, E., and Barrier, L.(2014). Laboratory alluvial fans in one dimension. Physics Review, 90.Harvey, A. M., Mather, A. E., and Stokes, M., editors (2005). AlluvialFans: Geomorphology, Sedimentology, Dynamics, volume 251. GeologicalSociety Special Publications, London. PP 1- 7.Henderson, F. M. (1966). Open-Channel Flow. Macmillan.Henderson, R. and Thompson, S. M. (1999). Extreme rainfalls in the South-ern Alps of New Zealand. Journal of Hydrology (New Zealand), 38:309–330.122BibliographyHoink, T., Lenardic, A., and Jellinek, A. M. (2013). Earths thermal evolu-tion with multiple convection modes: A monte-carlo approach. Physicsof the Earth and Planetary Interiors, 221:22–26.Hong, L. B. and Davies, T. R. H. (1979). A study of stream braiding.Geological Society of American Bulletin, 90:10941095.Hooke, R. (1967). Processes on arid-region alluvial fans. Journal of Geology,75:438–460.Hooke, R. (1968). Model geology; prototype and laboratory streams; dis-cussion. Geological Society of American Bulletin, 79:391394.Hunt, C. B. and Mabey, D. R. (1966). Stratigraphy and structure,Death Valley, California. U.S. Geological Survey Professional Paper, 494-A:162pp.Jerolmack, D. J. (2011). Causes and effects of noise in landscape dynamics.EOS, transactions, American geophysical Union, 92(44):385–386.Jerolmack, D. J. and Mohrig, D. (2007). Conditions for branching of depo-sitional rivers. Geology, 35(5):463466.Kleinhans, M. G., Ferguson, R. I., Lane, S. N., and Hardy, R. J. (2013).Splitting rivers at their seams: bifurcations and avulsion. Earth SurfaceProcesses and Landforms, 38:47–61.Knighton, A. D. (1998). Fluvial Forms and Processes: A New Perspective,Chapter five. Edward Arnold, London.Kochel, R. C. (1990). Alluvial fans: a field approach, Humid alluvial fans ofthe Appalachian Mountains, pages 109 – 129. Wiley, Chichester.Lajeunesse, E., Malverti, L., Lancien, P., Armstrong, L., Me`tivier, F., Cole-man, S., Smith, C. E., Davies, T., Cantelli, A., and Parker, G. (2010).Fluvial and subaqueous morphodynamics of laminar and near-laminarflows: A synthesis. Sedimentology, 57(1):1–26.Langridge, R. M., Howard, J. D., Buxton, R., and Ries, W. F. (2016).A natural hazard assessment for the township of Franz Josef, Westlanddistrict. Consultancy Report 2016/33, GNS Science.Lenardi, A., Moresi, L.-N., and Mulhaus, H. (2003). Longevity and stabilityof cratonic lithosphere: insights from numerical simulations of coupled123Bibliographymantle convection and continental tectonics. Journal of Geophysical Re-search, 108(B6):2303.Malverti, L., Lajeunesse, E., and Me`tivier, F. (2008). Small is beautiful:upscaling from microscale laminar to natural turbulent rivers. Journal ofGeophysical Resarch of Earth Surfaces, 113.Maynord, S. T. (2006). Evaluation of the micromodel: an extremely small-scale movable bed model. Journal of Hydraulic Engineering, 132(4):343–353.McKendry, I. G. (1992). Numerical simulation of sea breeze interactionsover the Auckland region, New Zealand. New Zealand Journal of Geologyand Geophysics, 35(1):9–20.McSaveney, M. J. and Davies, T. R. H. (1998). Natural hazard assessmentfor the township of Franz Josef Glacier and its environs. Technical Re-port Client Report 43714B.10, Institue of Geological and Nuclear SciencesLimited.Me`tivier, F. and Barrier, L. (2012). Gravel-bed rivers: processes, tools, en-vironments, Alluvial landscape evolution: what do we know about meta-morphosis of gravel-bed meandering and braided streams?, pages 474–501.John Wiley and Sons, Ltd., Chichester.Me`tivier, F. and Meunier, P. (2003). Input and output mass flux correlationsin an experimental braided stream. implications on the dynamics of bedload transport. Journal of Hydrology, 271:2238.Mosley, M. P. and Zimpfer, G. L. (1978). Hardware models in geomorphol-ogy. Progress in Physical Geography, 2:438–461.Nevins, T. H. F. (1969). River training the single thread channel. NewZealand Engineering, 15:367–73.Paola, C., Straub, K., Mohrig, D., and Reinhardt, L. (2009). The “unrea-sonable effectivenes” of stratigraphic and geomorphic experiments. EarthScience Review, 97:143.Peakall, J., Ashworth, P., and Best, J. (1996). The Scientific Nature ofGeomorphology, Physical modelling in fluvial geomorphology: principles,applications, and unresolved issues, pages 221–253. John Wiley, Chich-ester, U. K.124BibliographyPeakall, J. and Warburton, J. (1996). Surface tension in small hydraulicriver models the significance of the weber number. Journal of Hydrology(New Zealand), 35:199212.Piegay, H., Grant, G., Nakamura, F., and Trustrum, N. (2006). Braidedrivers: process, deposits, ecology and management, Braided river manage-ment: from assessment of river behaviour to improved sustainable devel-opment. International Association of Sedimentologists.Popescu, I. (2014). Computational Hydraulics, Chapter 1: modelling theory,pages 1 – 10. IWA Publishing, London.Rachocki, A. (1981). Alluvial Fans. John Wiley and Sons Ltd., Chichester,London.Reitz, M. D. and Jerolmack, D. J. (2012). Experimental alluvial fan evolu-tion: channel dynamics, slope controls, and shoreline growth. Journal ofGeophysical Research, 117.Reitz, M. D., Jerolmack, D. J., and Swenson, J. B. (2010). Flooding and flowpath selection on alluvial fans and deltas. Geophysical Research Letters,37.Robinson, T. R. and Davies, T. R. H. (2013). Review article: potential ge-omorphic consequences of a future great (mw = 8.0+) alpine fault earth-quake, South Island, New Zealand. Natural Hazards and Earth SystemScience, 13:22792299.Ryder, J. M. (1971). Some aspects of the morphometry of paraglacial allu-vial fans in south-central British Columbia. Canadian Journal of EarthScience, 8:1252–1264.Schumm, S., Mosley, M., and Weaver, W. (1987). Experimental FluvialGeomorphology. John Wiley, New York.Schumm, S. A. (1977). The Fluvial System. Wiley, New York.Schumm, S. A. (1985). Patterns of alluvial rivers. Annual Review of EarthPlanetary Science, 13:5–27.Tal, M. and Paola, C. (2010). Effects of vegetation on channel morphody-namics: results and insights from laboratory experiments. Earth SurfaceProcesses and Landforms, 35:1014–1028.125Thornbury, W. B. (1954). Principles of Geomorphology. John Wiley, NewYork.Trenberth, K. E., Branstator, G. W., Karoly, D., Kumar, A., Lau, N. G.,and Ropelewski, C. (1998). Progress during toga in understanding andmodelling global teleconnections associated with tropical sea surface tem-peratures. Journal of Geophysical Research, 103(C7):14291–14324.Warburton, J. (1996). A brief review of hydraulic modelling of braidedgravel-bed rivers in New Zealand. Journal of Hydrology (New Zealand),35(2):157–173.Wasson, R. J. (1977). Last-glacial alluvial fan sedimentation in the lowerDerwent valley, Tasmania. Sedimentology, 24:781–799.Whipple, K. X., Parker, G., Paola, C., and Mohrig, D. (1998). Channeldynamics, sediment transport, and the slope of alluvial fans: experimentalstudy. Journal of Geology, 106:677 –693.Whitehouse, I. E. and McSaveney, M. J. (1990). Alluvial fans: a field ap-proach, Geomorphic appraisals for development on two steep, active allu-vial fans, Mt Cook, New Zealand, pages 369–384. Wiley, Chichester.Wild, M. A. (2012). Growth dynamics of braided gravel-bed river deltas inNew Zealand. phdthesis, University of Canterbury, New Zealand. 215p.Zarn, B. and Davies, T. R. H. (1994). The significance of processes on allu-vial fans to hazard assessment. Zeitschrift fr Geomorphologie, 38(4):487–500.126AppendicesAppendicesAppendix A: aerial imagery of the Waiho fanAerial imagery of the Upper Waiho Fan taken from 1948 to 2013.127Appendices Figure 1: 1948 aerial imagery128Appendices Figure 2: 1964 aerial imagery129Appendices3/4/1973Figure 3: 1973 aerial imagery130Appendices18/5/1979Figure 4: 1979 aerial imagery131Appendices4/1/1980Figure 5: 1980 aerial imagery132Appendices23/2/1982Figure 6: 1982 aerial imagery133Appendices12/2/1985Figure 7: 1985 aerial imagery134Appendices22/2/1990Figure 8: 1990 aerial imagery135Appendices13/7/1994Figure 9: 1994 aerial imagery136Appendices15/7/2002Figure 10: 2002 aerial imagery137Appendices23/7/2004Figure 11: 2004 aerial imagery138Appendices17/2/2011Figure 12: 2011 aerial imagery139Appendices Figure 13: 2013 aerial imagery140AppendicesAppendix B: digital vernier calliper elevation dataDigital vernier calliper elevation data for the intial steady input experiment(S1).141AppendicesTable 1: Point measurements from the INITIAL NATURAL SCENARIO. Points were taken in the upper, middleand lower sections of the fan at set intervals. Pink represents aggradation, green degradation, and blue no netchange (values +/-1 around 0mm). EC stands for elevation change, and was calculated by substracting theprevious elevation from the most recent elevation.Point Nrth East Initial (Hr33) Hr35 EC (2hrs) Hr36 EC (1hr) Hr37 EC (1hr) Hr38 EC (1hr) Hr39 EC (1hr) Hr40 EC (1hr)1 30 20 35.49 35.46 0.03 31.78 3.68 30.16 1.62 30.18 -0.02 30.13 0.05 29.88 0.252 30 40 26.33 23 3.33 20.64 2.36 21.4 -0.76 20.04 1.36 18.47 1.57 15.84 2.633 30 60 17.45 12.62 4.83 13.85 -1.23 12.06 1.79 9.74 2.32 9.08 0.66 7.7 1.384 45 20 35.31 33.35 1.96 31.4 1.95 32.07 -0.67 31.52 0.55 30.34 1.18 29.67 0.675 45 40 29.32 25.05 4.27 23.97 1.08 24.1 -0.13 24.52 -0.42 22.22 2.3 20.54 1.686 45 60 22.01 17.47 4.54 17.5 -0.03 16.11 1.39 15.04 1.07 13.63 1.41 13.83 -0.27 60 20 38.56 35.58 2.98 34.44 1.14 39.33 -4.89 34.47 4.86 32.32 2.15 31.77 0.558 60 40 31.04 28.49 2.55 24.86 3.63 25.72 -0.86 25.23 0.49 25.37 -0.14 25.21 0.169 60 60 25 21.22 3.78 20.55 0.67 20.54 0.01 19.64 0.9 18.7 0.94 17.05 1.6510 105 90 51.32 49.57 1.75 49.37 0.2 48.94 0.43 49.52 -0.58 49.11 0.41 48.63 0.4811 105 120 43.59 40.34 3.25 39.99 0.35 40.01 -0.02 38.11 1.9 38.7 -0.59 37.78 0.9212 105 150 34.1 32.36 1.74 31.61 0.75 29.93 1.68 28.84 1.09 28.07 0.77 28.21 -0.1413 165 65 51.79 51.86 -0.07 51.66 0.2 51.38 0.28 51.36 0.02 51.9 -0.54 51.73 0.1714 165 90 54.12 52.98 1.14 52.97 0.01 53.9 -0.93 52.68 1.22 52.01 0.67 51.43 0.5815 165 115 47.02 46.41 0.61 45.32 1.09 46.33 -1.01 46.13 0.2 48.95 -2.82 47.93 1.02142AppendicesTable 2: Digital Vernier Calliper elevation data for the current confinement alignment stagePoint Nrth East Hr40 Hr41 EC Hr42 EC Hr43 EC Hr44 EC Hr45 EC2 30 65 31.13 29.22 1.91 27.57 1.65 23.36 4.21 22.6 0.76 20.03 2.573 30 75 31.06 30.12 0.94 27.8 2.32 24.95 2.85 25 -0.05 22.41 2.594 40 55 41.06 35.92 5.14 32.35 3.57 30.62 1.73 28.48 2.14 23.91 4.575 40 65 39.26 35.61 3.65 31.79 3.82 29.52 2.27 28.78 0.74 24.68 4.16 40 75 33.13 32.87 0.26 29.69 3.18 26.06 3.63 24.94 1.12 23.17 1.777 60 55 45.47 39.97 5.5 39.79 0.18 36.71 3.08 32.95 3.76 32.64 0.318 60 65 43.17 41.43 1.74 38.33 3.1 34.5 3.83 34.92 -0.42 33.45 1.479 60 75 37.82 36.6 1.22 35.54 1.06 33.62 1.92 30.95 2.67 29.17 1.7810 80 55 49.99 47.6 2.39 45.39 2.21 43.34 2.05 42.24 1.1 38.99 3.2511 80 65 47.94 44.75 3.19 43.38 1.37 40.61 2.77 39.9 0.71 38.52 1.3812 80 75 43.29 41.55 1.74 41.33 0.22 40.59 0.74 39.04 1.55 36.23 2.8113 100 55 54.64 50.5 4.14 49.52 0.98 47.67 1.85 47.12 0.55 44.98 2.1414 100 65 48.96 48.38 0.58 46.52 1.86 45.71 0.81 45.96 -0.25 43.81 2.1515 100 75 42.23 43.26 -1.03 42.58 0.68 41.34 1.24 39.92 1.42 37.51 2.4116 120 55 53.56 52.96 0.6 52.85 0.11 53.64 -0.79 48.8 4.84 51.5 -2.717 120 65 46.76 47.99 -1.23 47.88 0.11 45.45 2.43 48.08 -2.63 43.24 4.8418 120 75 45.72 45.04 0.68 45.16 -0.12 43.6 1.56 43.32 0.28 41.45 1.87143AppendicesTable 3: Digital Vernier Calliper elevation data for the current confinement alignment stage continued...Point Nrth East Hr46 EC Hr47 EC Hr48 EC Hr49 EC Hr50 EC2 30 65 17.95 2.08 14.81 3.14 11.87 2.94 9.95 1.92 9.15 0.83 30 75 19.35 3.06 16.36 2.99 14.26 2.1 10.09 4.17 10.19 -0.14 40 55 24.26 -0.35 22.02 2.24 20.41 1.61 18.87 1.54 16.17 2.75 40 65 24.54 0.14 21.47 3.07 18.4 3.07 14.37 4.03 13.35 1.026 40 75 20.98 2.19 18.89 2.09 15.3 3.59 14.72 0.58 14.98 -0.267 60 55 27.48 5.16 26.67 0.81 26.13 0.54 21.43 4.7 19.73 1.78 60 65 30.39 3.06 27.72 2.67 25.71 2.01 22.8 2.91 19.43 3.379 60 75 29.24 -0.07 26.65 2.59 24.38 2.27 21.77 2.61 22.29 -0.5210 80 55 37.53 1.46 36.36 1.17 36.71 -0.35 33.51 3.2 30.79 2.7211 80 65 35.62 2.9 34.62 1 32.35 2.27 29.51 2.84 27.91 1.612 80 75 35.44 0.79 34.54 0.9 32.44 2.1 31 1.44 28.63 2.3713 100 55 42.77 2.21 42.83 -0.06 42.51 0.32 39.14 3.37 37.61 1.5314 100 65 39.64 4.17 39.59 0.05 38.72 0.87 36.51 2.21 35.43 1.0815 100 75 37.42 0.09 37.13 0.29 35.3 1.83 34.64 0.66 32.19 2.4516 120 55 48.16 3.34 47.64 0.52 49.13 -1.49 44.21 4.92 42.88 1.3317 120 65 44.35 -1.11 43.18 1.17 39.35 3.83 39.41 -0.06 38.49 0.9218 120 75 39.84 1.61 38.79 1.05 38.66 0.13 38.26 0.4 35.78 2.48144AppendicesTable 4: Digital Vernier Calliper elevation data for the intermediate stopbank alignmentPoint Nrth East Hr50 Hr51 EC Hr52 EC Hr53 EC Hr54 EC Hr55 EC1 20 45 40.68 32.31 8.37 29.52 2.79 28.21 1.31 23.88 4.33 22.16 1.722 20 55 29.38 25.25 4.13 22.5 2.75 20.16 2.34 17.29 2.87 15.84 1.453 20 65 6.39 19.85 -13.46 18.31 1.54 14.84 3.47 13.36 1.48 11.29 2.074 40 30 45.69 45.12 0.57 41.37 3.75 39.88 1.49 36.1 3.78 31.39 4.715 40 45 41.13 35.59 5.54 32.48 3.11 30.79 1.69 27.35 3.44 26.81 0.546 40 60 13.7 26.36 -12.66 25.16 1.2 23.34 1.82 20.61 2.73 20.03 0.587 40 75 11.91 13.06 -1.15 11.93 1.13 11.89 0.04 11.74 0.15 14.6 -2.868 60 15 57.37 57.7 -0.33 52.3 5.4 50.45 1.85 49.97 0.48 45.28 4.699 60 30 51.33 46.43 4.9 46.04 0.39 44.09 1.95 40.91 3.18 41.21 -0.310 60 45 45.15 37.17 7.98 36.81 0.36 36.36 0.45 33.41 2.95 32.2 1.2111 60 60 16.56 16.25 0.31 15.55 0.7 16.12 -0.57 16.7 -0.58 16.03 0.6712 60 75 18.31 18.16 0.15 18.7 -0.54 18.42 0.28 18.21 0.21 17.82 0.3913 80 20 60.41 54.16 6.25 51.82 2.34 51.92 -0.1 50.09 1.83 50.88 -0.7914 80 70 21.81 23.83 -2.02 24.1 -0.27 24.55 -0.45 24.46 0.09 24.06 0.415 100 20 61.25 59.2 2.05 57.72 1.48 55.28 2.44 55.17 0.11 53.73 1.4416 100 35 63.4 61.42 1.98 59.51 1.91 58.45 1.06 57.49 0.96 55.4 2.0917 100 60 32.8 32.93 -0.13 32.96 -0.03 32.73 0.23 33.06 -0.33 32.42 0.6418 100 70 31.28 31.44 -0.16 31.81 -0.37 31.44 0.37 30.83 0.61 30.74 0.0919 120 15 68.23 65.65 2.58 66.05 -0.4 63.49 2.56 63.69 -0.2 64.15 -0.4620 120 35 62.69 61.53 1.16 62.35 -0.82 64.77 -2.42 63.86 0.91 63.96 -0.121 120 55 40.51 40.21 0.3 39.53 0.68 40.41 -0.88 40.03 0.38 40.63 -0.622 120 75 33.48 33.91 -0.43 34.21 -0.3 34.24 -0.03 34.04 0.2 34.15 -0.11145AppendicesTable 5: Digital Vernier Calliper elevation data for the intermediate stopbank alignment continued...Point Nrth East Hr56 EC Hr57 EC Hr58 EC Hr59 EC Hr60 EC1 20 45 20.23 1.93 19.21 1.02 18.18 1.03 17.43 0.75 14.6 2.832 20 55 14.7 1.14 14.73 -0.03 12.49 2.24 10.23 2.26 8.7 1.533 20 65 10.43 0.86 8.9 1.53 7.46 1.44 6.24 1.22 3.69 2.554 40 30 30.31 1.08 28.36 1.95 28.37 -0.01 28.21 0.16 26.95 1.265 40 45 26.89 -0.08 25.87 1.02 24.46 1.41 22.69 1.77 20.98 1.716 40 60 20.26 -0.23 18.65 1.61 16.1 2.55 15.85 0.25 15.3 0.557 40 75 13.9 0.7 11.75 2.15 12.95 -1.2 12.79 0.16 9.18 3.618 60 15 41.84 3.44 41.43 0.41 41.49 -0.06 41.88 -0.39 41.13 0.759 60 30 37.65 3.56 37.89 -0.24 36.99 0.9 34.82 2.17 33.59 1.2310 60 45 33.28 -1.08 33.11 0.17 30.82 2.29 32.93 -2.11 28.83 4.111 60 60 20.47 -4.44 21.03 -0.56 20.7 0.33 20.29 0.41 20.18 0.1112 60 75 17.47 0.35 15.67 1.8 16.16 -0.49 16.06 0.1 15.05 1.0113 80 20 50.19 0.69 49.29 0.9 46.12 3.17 46.61 -0.49 45.82 0.7914 80 70 23.92 0.14 23.54 0.38 23.67 -0.13 24.12 -0.45 24.65 -0.5315 100 20 53.84 -0.11 52.02 1.82 51.5 0.52 50.01 1.49 47.3 2.7116 100 35 55.45 -0.05 55.38 0.07 55.95 -0.57 55.03 0.92 54.06 0.9717 100 60 32.84 -0.42 33.15 -0.31 33.49 -0.34 33.75 -0.26 33.5 0.2518 100 70 30.43 0.31 30.31 0.12 31.13 -0.82 30.32 0.81 31.03 -0.7119 120 15 59.47 4.68 59.07 0.4 57.91 1.16 54.87 3.04 56.4 -1.5320 120 35 64.04 -0.08 63.55 0.49 61.45 2.1 55.79 5.66 59.26 -3.4721 120 55 39.99 0.64 40.01 -0.02 40.75 -0.74 39.81 0.94 41.42 -1.6122 120 75 33.77 0.38 33.69 0.08 33.99 -0.3 33.41 0.58 33.6 -0.19146AppendicesTable 6: Digital Vernier Calliper elevation data for the extreme stopbank alignmentPoint Nrth East Hr60 Hr61 EC Hr62 EC Hr63 EC Hr64 EC Hr65 EC1 25 115 59.82 51.68 8.14 47.17 4.51 47.11 0.06 46.54 0.57 46.1 0.442 25 135 24.07 32.66 -8.59 28.85 3.81 28.41 0.44 26.67 1.74 25.74 0.933 25 150 13.74 20.89 -7.15 15.93 4.96 17.8 -1.87 17.92 -0.12 16.87 1.054 25 165 6.48 6.21 0.27 11.71 -5.5 12.08 -0.37 10.09 1.99 9.06 1.035 40 105 62.65 55.78 6.87 51.76 4.02 51.5 0.26 50.53 0.97 50.59 -0.066 40 130 27.47 27.22 0.25 26.9 0.32 30.68 -3.78 30.54 0.14 30.79 -0.257 40 150 15.06 17.98 -2.92 23.07 -5.09 21.49 1.58 19.09 2.4 18.4 0.698 40 170 9.54 9.91 -0.37 10.12 -0.21 9.8 0.32 11.23 -1.43 12.96 -1.739 55 85 73.72 66.27 7.45 64.82 1.45 65.21 -0.39 65.65 -0.44 64.39 1.2610 55 105 63.28 57.1 6.18 54.68 2.42 51.4 3.28 50.63 0.77 50.62 0.0111 55 125 34.5 34.41 0.09 33.83 0.58 33.3 0.53 34.2 -0.9 33.12 1.0812 55 145 25.83 25.77 0.06 30.25 -4.48 28.03 2.22 26.35 1.68 24.87 1.4813 55 165 18.18 18.08 0.1 18.07 0.01 17.66 0.41 18.98 -1.32 20.57 -1.5914 55 180 14.51 14.78 -0.27 14.45 0.33 14.5 -0.05 14.25 0.25 16.35 -2.115 75 75 72.24 74.08 -1.84 73.55 0.53 71.83 1.72 71.95 -0.12 71.63 0.3216 75 100 68.68 67.22 1.46 63.36 3.86 61.17 2.19 61.48 -0.31 61.23 0.2517 75 125 41.17 41.48 -0.31 42.62 -1.14 44 -1.38 43.9 0.1 39.49 4.4118 75 170 23.01 22.45 0.56 22.37 0.08 22.65 -0.28 22.79 -0.14 21.62 1.1719 100 75 76.09 77.49 -1.4 76.99 0.5 75.39 1.6 75.85 -0.46 74.56 1.2920 100 95 72.16 71.72 0.44 65.14 6.58 62.05 3.09 61.15 0.9 59.9 1.2521 100 120 49.58 48.27 1.31 48.64 -0.37 48.81 -0.17 47.54 1.27 47.68 -0.1422 100 150 40.23 40.55 -0.32 40.36 0.19 40.08 0.28 40.19 -0.11 38.73 1.4623 100 170 31.68 31.11 0.57 31.1 0.01 31.3 -0.2 30.11 1.19 30.89 -0.7824 125 70 80.1 78.02 2.08 77.44 0.58 77.47 -0.03 76.86 0.61 76.61 0.2525 125 95 74.11 73.66 0.45 71.27 2.39 66.66 4.61 64.02 2.64 62.89 1.1326 125 120 58.93 58.57 0.36 58.92 -0.35 59.04 -0.12 58.26 0.78 59.18 -0.9227 125 145 49.38 49.08 0.3 48.74 0.34 48.53 0.21 49.01 -0.48 48.56 0.4528 125 165 39.05 39.05 0 38.69 0.36 39.06 -0.37 38.91 0.15 37.48 1.4329 150 60 78.56 76.01 2.55 75.86 0.15 76.92 -1.06 76.33 0.59 78.04 -1.7130 150 80 77.17 76.46 0.71 75.66 0.8 74.36 1.3 72.92 1.44 72.79 0.1331 150 105 76.3 76.43 -0.13 72.39 4.04 69.96 2.43 69.4 0.56 68.7 0.732 150 125 66.69 69.48 -2.79 66.14 3.34 66.01 0.13 67.71 -1.7 63.05 4.66147AppendicesTable 7: Digital Vernier Calliper elevation data for the extreme stopbank alignment continued...Point Nrth East Hr66 EC Hr67 EC Hr68 EC Hr69 EC Hr70 EC1 25 115 45.91 0.19 44.04 1.87 45.04 -1 44.65 0.39 42.5 2.152 25 135 26.92 -1.18 22.8 4.12 23.24 -0.44 24.85 -1.61 22.26 2.593 25 150 15.03 1.84 14.04 0.99 14.88 -0.84 13.15 1.73 13.69 -0.544 25 165 8.05 1.01 8.09 -0.04 9.32 -1.23 6.97 2.35 6.12 0.855 40 105 49.7 0.89 48.89 0.81 50.08 -1.19 47.94 2.14 46.8 1.146 40 130 31.03 -0.24 29.28 1.75 30.23 -0.95 29.6 0.63 28.69 0.917 40 150 18.04 0.36 18.22 -0.18 17.57 0.65 16.9 0.67 16.14 0.768 40 170 12.99 -0.03 11.99 1 12.81 -0.82 12.02 0.79 11.15 0.879 55 85 64.68 -0.29 64.07 0.61 64.48 -0.41 64.07 0.41 62.9 1.1710 55 105 47.85 2.77 47.2 0.65 47.8 -0.6 47.57 0.23 48.05 -0.4811 55 125 34.03 -0.91 34.43 -0.4 35.89 -1.46 35.47 0.42 34.87 0.612 55 145 24.66 0.21 25.03 -0.37 22.92 2.11 24.79 -1.87 24.04 0.7513 55 165 19.61 0.96 20.08 -0.47 18.74 1.34 17.46 1.28 17.91 -0.4514 55 180 14.76 1.59 14.96 -0.2 15.04 -0.08 13.93 1.11 13.31 0.6215 75 75 72.16 -0.53 71.17 0.99 71.28 -0.11 71.42 -0.14 71.81 -0.3916 75 100 56.5 4.73 56.25 0.25 56.81 -0.56 55.12 1.69 54.08 1.0417 75 125 40.07 -0.58 40.05 0.02 37.7 2.35 39.51 -1.81 39.04 0.4718 75 170 21.88 -0.26 20.81 1.07 23.07 -2.26 22.3 0.77 20.29 2.0119 100 75 74.22 0.34 73.38 0.84 73.62 -0.24 73.77 -0.15 71.54 2.2320 100 95 60.38 -0.48 60.2 0.18 59.47 0.73 59.55 -0.08 58.87 0.6821 100 120 47.53 0.15 47 0.53 46.86 0.14 47.09 -0.23 46.62 0.4722 100 150 38.9 -0.17 37.86 1.04 37.06 0.8 37.76 -0.7 37.06 0.723 100 170 29.95 0.94 29.36 0.59 28.04 1.32 27.2 0.84 26.48 0.7224 125 70 77.19 -0.58 76.96 0.23 76.08 0.88 75.81 0.27 78.63 -2.8225 125 95 63.02 -0.13 62.78 0.24 63.05 -0.27 63.08 -0.03 63.38 -0.326 125 120 59.85 -0.67 60.87 -1.02 58.52 2.35 58.93 -0.41 59.38 -0.4527 125 145 47.92 0.64 46.92 1 45.69 1.23 46.66 -0.97 45.3 1.3628 125 165 37.4 0.08 35.49 1.91 34.6 0.89 32.66 1.94 32.94 -0.2829 150 60 77.92 0.12 77.2 0.72 78.55 -1.35 76.26 2.29 76.99 -0.7330 150 80 73.33 -0.54 72.72 0.61 73.59 -0.87 74.18 -0.59 73.66 0.5231 150 105 68.93 -0.23 69.36 -0.43 69.32 0.04 70.07 -0.75 66.76 3.3132 150 125 63.17 -0.12 62.03 1.14 60.68 1.35 59.41 1.27 58.91 0.5148

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0358018/manifest

Comment

Related Items