UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mitigating impacts of temperature-oxygen squeeze in a mesotrophic-eutrophic lake : Wood Lake, BC, Canada Young, Christopher J. 2016

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

Item Metadata

Download

Media
24-ubc_2017_february_young_christopher.pdf [ 21.15MB ]
Metadata
JSON: 24-1.0340487.json
JSON-LD: 24-1.0340487-ld.json
RDF/XML (Pretty): 24-1.0340487-rdf.xml
RDF/JSON: 24-1.0340487-rdf.json
Turtle: 24-1.0340487-turtle.txt
N-Triples: 24-1.0340487-rdf-ntriples.txt
Original Record: 24-1.0340487-source.json
Full Text
24-1.0340487-fulltext.txt
Citation
24-1.0340487.ris

Full Text

Mitigating Impacts of Temperature-Oxygen Squeeze In AMesotrophic-Eutrophic Lake: Wood Lake, BC, CanadabyChristopher J. YoungB.A.Sc., The University of British Columbia, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMaster of Applied ScienceinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Civil Engineering)The University of British Columbia(Vancouver)December 2016c© Christopher J. Young, 2016AbstractWarm surface waters and hypoxic hypolimnion during summer stratification in eu-trophic lakes, known as a temperature-oxygen squeeze, can limit available habitatfor aquatic species. This study of Wood Lake, a calcareous eutrophic lake situatedin the semi-arid Okanagan Valley of BC, Canada, was motivated by a decline inkokanee return spawners following a severe temperature-oxygen squeeze in 2011.Field data collected in 2015/2016, combined with DYRESM, a one-dimensionalphysical simulation model, was employed to investigate the factors influencing thethermal aspect of the temperature-oxygen squeeze. A relatively early and weakspring freshet in 2015 was followed by severely restricted inflows due to upstreamdiversions and coincident with relatively warm regional air temperatures from Maythrough July. These factors triggered a marling event in Wood Lake and also re-sulted in the early onset of high epilimnetic temperatures and hypoxic hypolimnion(>20◦C and <2 mg/L by mid June 2015). However, a cooler August and Septem-ber, compared to 2011, alleviated the temperature-oxygen squeeze in 2015 by al-lowing the surface layer to cool. Simulations of the lake’s thermal response to me-teorological perturbations on weekly and seasonal time scales consistently showedthe thermal component of the Wood Lake temperature-oxygen squeeze is sensi-tive to the degree and timing of changes in air temperature; elevated August andSeptember air temperatures, combined with high retention time due to limited in-flow, restricts kokanee fish habitat by increasing surface layer temperature duringSeptember when hypolimnetic oxygen is most diminished. While the marling eventreduced water clarity, it did not significantly influence the lake’s thermal structure.DYRESM was employed to evaluate the effect of several proposed managementscenarios involving the alteration of inflows and outflows with the goal of reducingiiSeptember surface layer temperature of Wood Lake. It was shown that return-ing Wood Lake’s volumetric fluxes to pre-industrial rates by supplementing inflowfrom one of two adjacent lakes has the potential to reduce the thermal componentof the seasonal temperature-oxygen squeeze.iiiPrefaceThis dissertation is an original unpublished work authored by myself, ChristopherYoung. I designed and executed the field work campaign (CTD transects, ther-mistor chains, and water sampling) involved for this study, and conducted all sub-sequent data analyses, as well as model calibration, validation, and evaluation ofmodelled scenarios. Gavin Young and Emily Young provided assistance duringthe field work campaign to operate the vessel and GPS. Raphael Nowak conducteda bathymetric survey of Wood Lake during 2015 solely for this study, which wasthen utilized by myself for modelling purposes. DYRESM v4 was provided by JörgImberger and Angus Stewart from the Centre for Water Research at the Universityof Western Australia. This thesis was written under the supervision and guidanceof Dr. Bernard Laval.BCMOE generously provided past data for Wood Lake that has been collectedsince the 1970’s as part of the Canada-Okanagan Basin Agreement and variousother water resource management studies in the Okanagan Valley, including reg-ular monthly monitoring of Wood Lake since 2012 during the stratified season.Larratt Aquatic Consulting provided all of their available water quality monitor-ing data for Wood Lake with formal permission from Pattie Meger (Water QualityTechnician, District of Lake Country) and Renee Clark (Water Quality Manager,Regional District of North Okanagan). BCMOE and Dr. Natasha Neumann gen-erously provided historical Middle Vernon Creek discharge (flow and temperature)data, which was collected as part of an ongoing project since 2012 (“The MiddleVernon Creek Action Plan”) for the Okanagan Nation Alliance Fisheries Depart-ment and the British Columbia Ministry of Forests, Lands, and Natural ResourceOperations (BCMFLNRO). Solar radiation data was provided by Coral Beach Farmsiv(Craig Dalgliesh) which has a Ranch Systems Weather Station located adjacent toWood Lake.Water chemistry analyses of (water) samples collected on 14 July 2015 and 20July 2015 were conducted by CARO Analytical Services in Kelowna, BC, Canada.A taxonomic analysis of water samples collected on 2 March 2016 was conductedby Jamie Self of Larratt Aquatic Consulting in West Kelowna, BC, Canada.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1 Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Motivation for Research . . . . . . . . . . . . . . . . . . . . . . 72.3 Current Research Objectives . . . . . . . . . . . . . . . . . . . . 162.4 Review of Historical Data . . . . . . . . . . . . . . . . . . . . . . 172.4.1 The Eutrophication of Wood Lake . . . . . . . . . . . . . 172.5 Water Movement Within Lakes and Transfer Between Basins . . . 182.6 Water Quality Parameters . . . . . . . . . . . . . . . . . . . . . . 20vi2.6.1 Water Clarity: Secchi Depth . . . . . . . . . . . . . . . . 212.6.2 Limiting Nutrients: Nitrogen and Phosphorus . . . . . . . 232.6.3 Middle Vernon Creek Hydrology and Spawning Habitat . 402.7 Review of Historically Proposed Management Strategies to Im-prove the Trophic Status in Wood Lake . . . . . . . . . . . . . . . 432.7.1 Alum Treatment (Or Alternative Sediment Cap) . . . . . . 442.8 Review of Current Remediation Strategies . . . . . . . . . . . . . 512.8.1 Proposed Management Options Not Investigated . . . . . 512.8.2 Proposed Management Options Investigated in Current Study 562.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663.1 Data: Previous & Ongoing Limnological Studies & Meteorology . 663.1.1 Complete Records . . . . . . . . . . . . . . . . . . . . . 673.1.2 Incomplete Records . . . . . . . . . . . . . . . . . . . . 693.2 Important Historic Key Studies . . . . . . . . . . . . . . . . . . . 733.3 Field Work: Thermistor Chains and CTD Profiles . . . . . . . . . 753.3.1 Central Thermistor Chain . . . . . . . . . . . . . . . . . 753.3.2 Seiche Thermistor Chains . . . . . . . . . . . . . . . . . 773.3.3 CTD and DO Transect Design . . . . . . . . . . . . . . . 823.3.4 CTD Profiles . . . . . . . . . . . . . . . . . . . . . . . . 833.3.5 Dissolved Oxygen Profiles . . . . . . . . . . . . . . . . . 853.4 Processing Field Data . . . . . . . . . . . . . . . . . . . . . . . . 863.4.1 Spatially Averaging Temperature Profiles for Each Field Day 863.4.2 Bathymetry and Establishing a Common Datum . . . . . . 893.4.3 Light Extinction Coefficient . . . . . . . . . . . . . . . . 933.5 Dynamic Reservoir Simulation Model (DYRESM) Construction . 973.5.1 Inflows and Outflows . . . . . . . . . . . . . . . . . . . . 993.5.2 Lake Surface Boundary Layer Fluxes and Exchanges . . . 1013.5.3 Surface Layer Mixing . . . . . . . . . . . . . . . . . . . 1033.5.4 Deep Water Column Mixing . . . . . . . . . . . . . . . . 1043.5.5 Modelling Adjustments and Specifications for Wood Lake 106vii4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1084.1 Water Clarity: Secchi Depth, PAR Profiles, and Light ExtinctionCoefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1094.2 Chlorophyll-a . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1124.3 Temperature and Dissolved Oxygen . . . . . . . . . . . . . . . . 1184.3.1 Spring and Summer 2015 . . . . . . . . . . . . . . . . . 1204.3.2 Early Fall 2015 . . . . . . . . . . . . . . . . . . . . . . . 1234.3.3 Late Fall 2015 . . . . . . . . . . . . . . . . . . . . . . . 1274.3.4 Winter 2015-2016 and Spring 2016 . . . . . . . . . . . . 1294.4 Progression of Temperature-Oxygen Squeeze in 2015 . . . . . . . 1354.5 Hypolimnetic Oxygen Deficit . . . . . . . . . . . . . . . . . . . . 1374.6 Middle Vernon Creek Inflow . . . . . . . . . . . . . . . . . . . . 1394.6.1 MVC 2016 Freshet Plume . . . . . . . . . . . . . . . . . 1524.7 Seiching in Wood Lake . . . . . . . . . . . . . . . . . . . . . . . 1584.8 Temperature Gradient Through Oyama Canal . . . . . . . . . . . 1634.9 Total Heat Content . . . . . . . . . . . . . . . . . . . . . . . . . 1664.10 Epilimnion and Hypolimnion Heating Rates . . . . . . . . . . . . 1714.11 Dynamic Reservoir Simulation Model (DYRESM) . . . . . . . . 1754.11.1 DYRESM Model Parameters . . . . . . . . . . . . . . . . 1764.11.2 2015 Modelled Results: Modelled and Observed Temper-ature Profiles . . . . . . . . . . . . . . . . . . . . . . . . 1794.11.3 2015 Model Error: Heat Content . . . . . . . . . . . . . . 1834.11.4 Model Validation (2013 and 2014) . . . . . . . . . . . . . 1864.11.5 Model Validation of Inflows . . . . . . . . . . . . . . . . 1934.12 Wood Lake: Application of DYRESM . . . . . . . . . . . . . . . 1994.12.1 Evaluating Proposed Management Scenarios . . . . . . . 1995 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2305.1 Physical Factors With A Key Influence on Wood Lake’s ThermalStructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2305.1.1 Tragedy of 2011: Why Similar Severe Limitations on koka-nee Habitat did not Occur in Recent Years . . . . . . . . . 232viii5.1.2 Investigation of Factors Responsible for Lake Heat Con-tent Anomalies in 2015 . . . . . . . . . . . . . . . . . . . 2375.1.3 Evaluating Scenarios with Increased Average Daily Air Tem-peratures from May until December . . . . . . . . . . . . 2425.2 Management Scenarios . . . . . . . . . . . . . . . . . . . . . . . 2535.2.1 Future Modelling Recommendations . . . . . . . . . . . . 2596 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265A DYRESM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286A.1 Model Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 286A.1.1 Time of Day for Output of Results . . . . . . . . . . . . . 286A.1.2 Layer Thickness Limits . . . . . . . . . . . . . . . . . . . 286A.1.3 Wind Multiplication Factor . . . . . . . . . . . . . . . . . 287A.1.4 Critical Wind Speed . . . . . . . . . . . . . . . . . . . . 287A.1.5 Vertical Mixing Coefficient . . . . . . . . . . . . . . . . . 288A.1.6 Wind Stirring Efficiency, Shear Production Efficiency, andPotential Energy Mixing Efficiency . . . . . . . . . . . . 288A.1.7 Emissivity of Water . . . . . . . . . . . . . . . . . . . . . 289A.1.8 Effective Surface Area Coefficient (Ac) . . . . . . . . . . 290A.1.9 Bottom Boundary Layer (BBL) Diffusivity . . . . . . . . 291A.1.10 Lake Surface Short Wave (SW) Radiation Albedo . . . . . 292A.1.11 Atmospheric (Air Column) Stability . . . . . . . . . . . . 295A.1.12 Bulk Aerodynamic Momentum Transport Coefficient, La-tent Heat and Sensible Heat Coefficients . . . . . . . . . . 296A.1.13 Entrainment Coefficients . . . . . . . . . . . . . . . . . . 297A.1.14 Computational Aquatic Ecosystem Dynamics Model (CAEDYM)Switch (on/off) . . . . . . . . . . . . . . . . . . . . . . . 298A.1.15 Extrapolation Required for Comparing Heat Content be-tween Field Data and DYRESM Outputs . . . . . . . . . 299A.2 1-D Approximation . . . . . . . . . . . . . . . . . . . . . . . . . 300A.2.1 The Lake Number . . . . . . . . . . . . . . . . . . . . . 301ixA.2.2 Wedderburn Number . . . . . . . . . . . . . . . . . . . . 303A.2.3 Inflow Lake Number . . . . . . . . . . . . . . . . . . . . 306A.2.4 Inflow Froude Number . . . . . . . . . . . . . . . . . . . 307A.2.5 Outflow Froude Number . . . . . . . . . . . . . . . . . . 308A.2.6 Burger Number . . . . . . . . . . . . . . . . . . . . . . . 309B Water Clarity, PAR, and Light Extinction Coefficient . . . . . . . . 311B.1 Factors Affecting Photosynthetic Active Radiation (PAR) Measure-ments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311B.2 Secchi Depth Measurements . . . . . . . . . . . . . . . . . . . . 314C Investigation into the 2015 Marling Event . . . . . . . . . . . . . . . 318C.1 Background Information on Marl Lakes . . . . . . . . . . . . . . 318C.2 2015 Marling Event in Wood Lake . . . . . . . . . . . . . . . . . 321xList of TablesTable 2.1 Physical Properties of Ellison Lake, Wood Lake, and KalamalkaLake. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Table 2.2 Kokanee salmon spawner enumeration, low eggs counts, andegg deposition in MVC. . . . . . . . . . . . . . . . . . . . . . 10Table 2.3 Date of onset of stratification in Wood Lake in 2013 - 2016. . . 15Table 2.4 Indices of trophic status and average values for Kalamalka, Wood,and Ellison Lakes from 1969-1999. . . . . . . . . . . . . . . . 21Table 2.5 Variation in water clarity in Wood Lake, as indicated by meansecchi depth (m) from 1969 - 2015. . . . . . . . . . . . . . . . 22Table 2.6 Annual Nutrient Loading into and out of Ellison Lake and intoWood Lake in 2013. . . . . . . . . . . . . . . . . . . . . . . . 26Table 2.7 Nutrient loading into Ellison & Wood Lake in 2013 comparedto 1969 to 1973 . . . . . . . . . . . . . . . . . . . . . . . . . 26Table 2.8 Trends in Nutrient Concentrations in Wood Lake from 1969 -1999. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30Table 2.9 Enumeration of stream-spawning kokanee from 2009 - 2015. . 43Table 3.1 BC MOE raw data record (1969 - 2014). . . . . . . . . . . . . 73Table 3.2 Central Thermistor Chain: Temperature logger heights fromlake bottom. . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Table 3.3 Duration, location, and depths of all temperature loggers in WoodLake. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80Table 3.4 Accuracy, stability, & resolution of SBE 19plus v2 SeaCATSensors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84xiTable 4.1 Taxonomic analysis of water samples (2 March 2016). . . . . . 117Table 4.2 Observations of ice cover on Wood Lake during the winter sea-son (2015). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130Table 4.3 Comparison of average/maximum/minimum daily air tempera-tures in winter 2013/2014,2014/2015, & 2015/2016. . . . . . . 131Table 4.4 Range of temperatures in water column in March and Aprilfrom 2013 to 2016. . . . . . . . . . . . . . . . . . . . . . . . 135Table 4.5 Hypolimnetic Oxygen Deficit in Wood Lake from 2011 - 2015. 139Table 4.6 Cumulative MVC inflow and water temperature during mod-elled periods. . . . . . . . . . . . . . . . . . . . . . . . . . . . 148Table 4.7 Background density, specific conductivity, turbidity, tempera-ture, & flow in MVC during 2015-2016. . . . . . . . . . . . . 151Table 4.8 Temperature, specific conductivity, and turbidity in MVC com-pared to Wood Lake on 25 April & 3 May 2016. . . . . . . . . 152Table 4.9 Total summer heat income and heating rates in Wood Lake (2013- 2015). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173Table 4.10 DYRESM Model Parameter Calibration . . . . . . . . . . . . 177Table 4.11 Comparison of modelled results & field observations for laketemperature & heat content during spring freshet in 2015 & 2016.197Table 4.12 Modelled outflow regimes for Wood Lake . . . . . . . . . . . 201Table 4.13 Historical measured/estimated outflows from Wood Lake andKalamalka Lake . . . . . . . . . . . . . . . . . . . . . . . . . 202Table 4.14 Ortho-P, TP, and TN in the hypolimnon and epilimnion of WoodLake in 2013. . . . . . . . . . . . . . . . . . . . . . . . . . . 207Table 4.15 Modelled Recirculation regimes for Wood Lake. . . . . . . . . 208Table 4.16 Modelled Okanagan Lake Inflow (no outflow) regimes for WoodLake. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214Table 4.17 Proposed Inflows from Okangan Lake into Wood Lake and per-mitted Outflows to Kalamalka Lake. . . . . . . . . . . . . . . 218Table 4.18 Comparison of secchi depth measurements in Wood Lake andKalamalka Lake in spring 2016. . . . . . . . . . . . . . . . . . 223Table 4.19 Proposed Inflows from Kalamalka Lake into Wood Lake andpermitted Outflows to Kalamalka Lake. . . . . . . . . . . . . . 226xiiTable 5.1 Average annual and seasonal historical air temperatures for WoodLake area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231Table 5.2 Average difference in daily air temperatures between 2011 andeach of 2013, 2014, & 2015. . . . . . . . . . . . . . . . . . . . 233Table 5.3 Comparison of heat content from 2015 with changes in meteo-rological conditions in July & August 2015, & reduction in lightextinction & surface albedo. . . . . . . . . . . . . . . . . . . . 241Table 5.4 Average annual & seasonal temperatures from 2015 & projectedvalues for 2025, 2055, and 2085. . . . . . . . . . . . . . . . . 243Table 5.5 Comparison of predicted total heat content in ATI-1 & ATI-2compared to 2015 observed and modelled conditions. . . . . . 246Table C.1 Key parameters from water samples taken on 14 July 2015 and20 July 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . 323Table C.2 Analyses of water samples from Wood Lake on 20 July 2015for IAP and Saturation Index . . . . . . . . . . . . . . . . . . 326xiiiList of FiguresFigure 2.1 The Vernon Creek catchment and area surrounding Wood Lake. 6Figure 2.2 Temperature and Dissolved Oxygen profiles in Wood Lake on13 September 2011. . . . . . . . . . . . . . . . . . . . . . . . 13Figure 2.3 Secchi Disk Readings in Wood Lake from 1970 to 2015. . . . 23Figure 2.4 Spring & fall TP in Wood Lake from 1970 - 2015. . . . . . . 32Figure 2.5 Spring & fall TN in Wood Lake from 1970 - 2015 in Wood Lake. 32Figure 2.6 TP in the epilimnion & hypolimnion of Wood Lake from Mar.- Oct. (2013 - 2015). . . . . . . . . . . . . . . . . . . . . . . 36Figure 2.7 Epilimnion chlorophyll-a in Wood Lake at “Central Site Sta-tion” (1975 - 1985). . . . . . . . . . . . . . . . . . . . . . . . 38Figure 2.8 Epilimnion chlorophyll-a in Wood Lake at “Mouth of MVCStation” (1975 - 2007). . . . . . . . . . . . . . . . . . . . . . 38Figure 2.9 Epilimnion chlorophyll-a in Wood Lake at “Deep Station” (1983- 2014). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Figure 2.10 Average chlorophyll-a in upper 10m of Wood Lake at “DeepStation” (2013 - 2015). . . . . . . . . . . . . . . . . . . . . . 40Figure 3.1 Locations for temperature loggers in Wood Lake, Oyama Canal,& Kalamalka Lake. . . . . . . . . . . . . . . . . . . . . . . . 79Figure 3.2 Wood Lake CTD and DO sample stations. . . . . . . . . . . . 83Figure 3.3 Temperature contours in Wood Lake on 23 September 2015.Distance along x-axis is measured from south to north alongthe long-axis of the lake (Figure 3.2). . . . . . . . . . . . . . 88Figure 3.4 Spatially averaged temperature profile on 23 Sept, 2015. . . . 89xivFigure 3.5 Bathymetric chart for Wood Lake. . . . . . . . . . . . . . . . 90Figure 3.6 Hypsographic curve for Wood Lake. . . . . . . . . . . . . . . 91Figure 3.7 Estimated maximum lake depth in Wood Lake during 2015 -2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93Figure 3.8 −ln(Iz/Io) vs. Depth (z) and linear regression of one PAR profile. 96Figure 4.1 Secchi depth in Wood Lake from 4 May 2015 to 25 April 2016. 110Figure 4.2 Secchi depth estimates of Kd compared to Kd(PAR). . . . . . 111Figure 4.3 Spatially averaged chlorophyll-a in Wood Lake from 4 May2015 to 25 April 2016. . . . . . . . . . . . . . . . . . . . . . 113Figure 4.4 Average chlorophyll-a from 1 January 2016 to 25 April 2016. 114Figure 4.5 Chlorophyll-a in Wood Lake on 16 March 2016 . . . . . . . . 115Figure 4.6 Taxonomic analysis of water samples (2 March 2016). . . . . 116Figure 4.7 Average chlorophyll-a from 4 May 2015 to 2 Dec. 2015 and 2Mar. 2015 to 25 Apr. 2016. . . . . . . . . . . . . . . . . . . . 118Figure 4.8 Temperature distribution in Wood Lake (4 May - 31 December2015) from central thermistor chain. . . . . . . . . . . . . . . 119Figure 4.9 Temperature and DO in Wood Lake from 4 May - 25 August2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122Figure 4.10 Thermal contours and DO contours on 9 September 2015. . . 123Figure 4.11 Temperature and DO in Wood Lake from 1 September to 28October 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . 125Figure 4.12 Temperature and DO profiles from 28 October 2015. . . . . . 126Figure 4.13 Temperature and DO profiles from Wood Lake on 5 November2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127Figure 4.14 Temperature and DO in Wood Lake from 5 November - 23December 2015. . . . . . . . . . . . . . . . . . . . . . . . . 128Figure 4.15 Thermal contours in Wood Lake on 2 December 2015. . . . . 129Figure 4.16 Temperature in Wood Lake from 1 January - 25 April 2016. . 132Figure 4.17 Absolute & average minimum & maximum temperatures inWood Lake (winter 2015-2016). . . . . . . . . . . . . . . . . 133xvFigure 4.18 Thermal contours in Wood Lake on 6 January 2016 (stronginverse stratification) & on 27 January 2016 (horizontal strati-fication). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134Figure 4.19 Comparison of temperatures in Wood Lake in March and Aprilfrom 2013-2016. . . . . . . . . . . . . . . . . . . . . . . . . 135Figure 4.20 Temperature-Oxygen squeeze during 2015. . . . . . . . . . . 136Figure 4.21 Average daily discharge in MVC in 2010 - 2015. . . . . . . . 141Figure 4.22 Cumulative volume discharged from MVC into Wood Lake in2010 - 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . 142Figure 4.23 Sandbar at mouth of MVC. . . . . . . . . . . . . . . . . . . . 143Figure 4.24 Water temperature in MVC (2010 - 2016). . . . . . . . . . . . 147Figure 4.25 Temperature distribution at mouth of MVC into Wood Lakealong “Middle Transect” on 26 May 2015. . . . . . . . . . . . 149Figure 4.26 Three CTD transects at mouth of MVC into Wood Lake. . . . 150Figure 4.27 Turbidity distribution from mouth of MVC into Wood Lakealong “West Transect” on 25 April 2016. . . . . . . . . . . . 153Figure 4.28 Turbidity distribution from mouth of MVC into Wood Lakealong “Middle Transect” on 25 April 2016. . . . . . . . . . . 153Figure 4.29 Turbidity distribution from mouth of MVC into Wood Lakealong “East Transect” on 25 April 2016. . . . . . . . . . . . . 154Figure 4.30 Background Turbidity in Wood Lake on 25 April 2016 . . . . 154Figure 4.31 Specific Conductivity distribution from mouth of MVC intoWood Lake along “West Transect” on 25 April 2016. . . . . . 155Figure 4.32 Specific Conductivity distribution from mouth of MVC intoWood Lake along “Middle Transect” on 25 April 2016. . . . . 156Figure 4.33 Specific Conductivity distribution from mouth of MVC intoWood Lake along “East Transect” on 25 April 2016. . . . . . 156Figure 4.34 Background Specific Conductivity in Wood Lake on 25 April2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157Figure 4.35 Overlaid thermistor chain data from north and south ends ofWood Lake (19-31 August 2015). . . . . . . . . . . . . . . . 160Figure 4.36 Overlaid thermistor chain data from the north and south endsof Wood Lake (1-30 September 2015). . . . . . . . . . . . . . 161xviFigure 4.37 Overlaid thermistor chain data from the north and south endsof Wood Lake (1-31 October 2015). . . . . . . . . . . . . . . 162Figure 4.38 Overlaid thermistor chain data from the north and south endsof Wood Lake (1-30 November 2015). . . . . . . . . . . . . . 163Figure 4.39 Average temperature range through Oyama Canal compared toaverage temperature range across Wood Lake (upper 2m). . . 165Figure 4.40 Heat content in Wood Lake from 4 May 2015 - 25 April 2016. 167Figure 4.41 Overplot of heat content in Wood Lake and Kd(PAR) from 4May 2015 to 25 April 2016. . . . . . . . . . . . . . . . . . . 168Figure 4.42 Short wave radiation flux and daily average air temperaturefrom 4 May - 31 October 2015. . . . . . . . . . . . . . . . . 169Figure 4.43 Temperature profiles from Wood Lake on 9 September 2015. . 171Figure 4.44 Comparison of DYRESM modelled results with spatial aver-age temperature profiles from 29 May - 23 December 2015. . 180Figure 4.45 RMSE comparing DYRESM output with field data in 2015compared to temperature range in Wood Lake on all field days. 182Figure 4.46 Temperature profiles on 23 September 2015 during seiching. . 183Figure 4.47 Comparison of heat content from DYRESM output to that basedon field work (2015). . . . . . . . . . . . . . . . . . . . . . . 185Figure 4.48 Error in heat content between simulated results compared tofield work in 2015. . . . . . . . . . . . . . . . . . . . . . . . 186Figure 4.49 S9-S13 CTD locations (2015) compared to geographic loca-tion of MOE Station 0500848. . . . . . . . . . . . . . . . . . 187Figure 4.50 Comparison of DYRESM output with temperature field pro-files in 2013 and 2014. . . . . . . . . . . . . . . . . . . . . . 189Figure 4.51 Comparison of heat content based on DYRESM output to thatbased on field profiles (2013 & 2014). . . . . . . . . . . . . . 191Figure 4.52 Hourly average wind speed and direction for 14-16 July2013and 9-11 Sept. 2013 . . . . . . . . . . . . . . . . . . . . . . 193Figure 4.53 MVC flow from 1 January to 1 September 2015 and 2016,showing portion of flow used in validating modelled inflows. . 195Figure 4.54 Inflowing MVC temperature & lake temperature profiles dur-ing modelled freshet period. . . . . . . . . . . . . . . . . . . 198xviiFigure 4.55 Actual flow in Lower VC in 2013 & 2015, compared to flowswith additional aliquots due to increasing release from Kala-malka Lake. . . . . . . . . . . . . . . . . . . . . . . . . . . . 203Figure 4.56 Change in height of 20◦C threshold due to Proposed Manage-ment Scenario #1. . . . . . . . . . . . . . . . . . . . . . . . . 204Figure 4.57 Predicted change in temperature in Wood Lake in August afteremploying Proposed Management Scenario #1. . . . . . . . . 205Figure 4.58 Predicted change in temperature in Wood Lake in Septemberafter employing Proposed Management Scenario #1. . . . . . 206Figure 4.59 Change in height of 20◦C threshold due to Proposed Manage-ment Scenario #2. . . . . . . . . . . . . . . . . . . . . . . . . 209Figure 4.60 Predicted change in temperature in Wood Lake in August afteremploying Proposed Management Scenario #2. . . . . . . . . 210Figure 4.61 Predicted change in temperature in Wood Lake in Septemberafter employing Proposed Management Scenario #2. . . . . . 211Figure 4.62 Flow in MVC (2012 - 2015) compared to flow with additionalaliquots from Okanagan Lake. . . . . . . . . . . . . . . . . . 213Figure 4.63 Change in height of 20◦C threshold due to Proposed Manage-ment Scenario #3(i). . . . . . . . . . . . . . . . . . . . . . . 215Figure 4.64 Predicted change in temperature in Wood Lake in August afteremploying Proposed Management Scenario #3(i). . . . . . . . 216Figure 4.65 Predicted change in temperature in Wood Lake in Septemberafter employing Proposed Management Scenario #3(i). . . . . 217Figure 4.66 Change in height of 20◦C threshold due to Proposed Manage-ment Scenario #3(ii). . . . . . . . . . . . . . . . . . . . . . . 219Figure 4.67 Predicted change in temperature in Wood Lake in August afteremploying Proposed Management Scenario #3(ii). . . . . . . 220Figure 4.68 Predicted change in temperature in Wood Lake in Septemberafter employing Proposed Management Scenario #3(ii). . . . . 221Figure 4.69 Average chlorophyll-a in Wood Lake (4 May 2015 to 25 April2016) compared to Kalamalka Lake. . . . . . . . . . . . . . . 223Figure 4.70 PAR from S10 in Wood Lake on 24 February 2016 comparedto one cast in Kalamalka Lake. . . . . . . . . . . . . . . . . . 224xviiiFigure 4.71 Change in height of 20◦C threshold due to Proposed Manage-ment Scenario #4. . . . . . . . . . . . . . . . . . . . . . . . . 227Figure 4.72 Predicted change in temperature in Wood Lake in August afteremploying Proposed Management Scenario #4. . . . . . . . . 228Figure 4.73 Predicted change in temperature in Wood Lake in Septemberafter employing Proposed Management Scenario #4. . . . . . 229Figure 5.1 Average annual historical air temperatures for Wood Lake area. 231Figure 5.2 Monthly average air temperatures for August & Septemberfrom 1930 - 2015 from ClimateBC. . . . . . . . . . . . . . . 234Figure 5.3 Temperature & DO profiles in September 2009, 2011, & 2013. 235Figure 5.4 Predicted change in temperature in Wood Lake in September2015 given a +2.62◦C change in daily average temperaturesfrom 20 Aug. - 30 Sept. 2015. . . . . . . . . . . . . . . . . . 237Figure 5.5 Average daily and weekly air temperatures from 4 May - 31October 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . 239Figure 5.6 Comparison of heat content from 2015 (field observations&simulated data) with simulations of changes in meteorologicalconditions and lake optical properties. . . . . . . . . . . . . . 242Figure 5.7 Historical (1930 - 2015) & predicted future (2025, 2055, 2085)seasonal average air temperatures. . . . . . . . . . . . . . . . 244Figure 5.8 Predicted heat content in Wood Lake for ATI-1 & ATI-2 com-pared to heat content in Wood Lake in 2015. . . . . . . . . . . 247Figure 5.9 Predicted change in temperature in Wood Lake in August dueto ATI-1 & ATI-2 scenarios. . . . . . . . . . . . . . . . . . . 248Figure 5.10 Predicted change in temperature in Wood Lake in Septemberdue to ATI-1 & ATI-2 scenarios. . . . . . . . . . . . . . . . . 249Figure 5.11 Height of 20◦C and 17◦C thresholds in response to modelledATI-1 scenario. . . . . . . . . . . . . . . . . . . . . . . . . . 250Figure 5.12 Height of 20◦C and 17◦C thresholds in response to modelledATI-2 scenario. . . . . . . . . . . . . . . . . . . . . . . . . . 251Figure 5.13 Temperature & DO profiles from 13 Sept. 2011 compared tosimulated results with change of -0.63◦C in upper 20-30m. . . 257xixFigure A.1 Lake Number and Wedderburn Number in Wood Lake duringthe simulation period. . . . . . . . . . . . . . . . . . . . . . . 303Figure A.2 Thermal contour plots for Wood Lake on 30 March 2015. . . . 305Figure A.3 Thermal contour plots for Wood Lake on 13 April 2015 show-ing re-establishment of stratification. . . . . . . . . . . . . . . 305Figure A.4 Thermal contour plots for Wood Lake on 25 April 2015. . . . 306xxGlossaryDYRESM Dynamic Reservoir Simulation ModelMVC Middle Vernon CreekUVC Upper Vernon CreekDLC District of Lake CountryBCMOE British Columbia Ministry of EnvironmentBCMFLNRO British Columbia Ministry of Forests, Lands, and Natural ResourceOperationsRDNO Regional District of the North OkanaganHWD Hiram Walker DistilleryDO dissolved oxygenTDP total dissolved phosphorousTSP total soluble phosphorousTKN total Kjeldahl nitrogenP phosphorousN nitrogenBC British ColumbiaxxiMVCAP Middle Vernon Creek Action PlanWUHW “Weighted Usable Habitat Width”MVCDSS Middle Vernon Creek Decision Support SystemCWR Centre for Water ResearchGUI graphical user interfaceCC cloud coverSML surface mixed layerKE kinetic energyBBL bottom boundary layerMAE mean absolute errorRMSE root mean squared errorxxiiAcknowledgmentsI must thank the Environmental Fluid Mechanics and Earth and Ocean ScienceDepartments at UBC for supplying me with the equipment and resources necessaryto complete this study. I would also like to extend a personal thank you to Dr.Bernard Laval, my supervisor, as well as Dr. Roger Pieters and Dr. Susan Allanfor your guidance, support and encouragement throughout my research. Thank youRaphael Nowak for your field work support.I extend my sincere thanks to BCMOE (in particular, Hillary Ward and MikeSokal), Paul Askey, Heather Larratt, Natasha Neumann, Craig Dalgliesh, and theOceala Fish and Game Club (OCFGC) for your support, encouragement, and gen-erosity in sharing equipment, data, knowledge and resources throughout this project.Thank you OCFGC for lending me your DO sensor for this project throughout the2015 season. Thank you Jörg Imberger and the Centre for Water Research forgenerously providing me with DYRESM.Thank you, Bernard Laval, for enabling me to complete this project, for yoursupport, teachings, and knowledge shared, as well as for your time dedicated toreviewing my final dissertation. Likewise, I would like to thank Greg Lawrence foryour time dedicated to reviewing my final dissertation.During the course of this project, I had the greatest pleasure of having myfather, Gavin Young, and my wife, Emily Young, assist me with my field work. Icould not thank the two of you more for all of your assistance and support in thefield, regardless of weather, throughout this project. Sharing this experience withboth of you was a very rare opportunity and something I will cherish for yearsto come. Finally, I would like to thank my mother, Lorna Young - for all of theyears of late nights spent editing, revising, and proof-reading my written works.xxiiiWords cannot express how grateful I am to each of you for your help and supportthroughout this study - none of this could have been possible without you.xxivDedicationTo: Mom, Dad, and Emily.Thank you for all of your support, love, and encouragement.xxvChapter 1IntroductionFish can become squeezed into a middle layer of preferred habitat in many eu-trophic lakes during summer stratification as a result of elevated temperatures insurface waters and diminished dissolved oxygen (DO) levels in the lower layers.This phenomenon, referred to here as a “temperature-oxygen squeeze”, has beenobserved to affect the behavior, distribution, and health of aquatic species in manylakes and other aquatic environments around the world, sometimes resulting intragic fish kills (Aku & Tonn, 1999; S. B. Brandt, Gerken, Hartman, & Demers,2009; C. C. Coutant, 1985, 1990; W. J. Matthews, Hill, & Schellhaass, 1985;B. Moore et al., 2014; Rowe & Chisnall, 1995; Zale, Wiechman, Lochmiller, &Burroughs, 1990). Global anthropogenic influences have intensified nutrient load-ing to aquatic ecosystems, resulting in eutrophication and the development of sea-sonal hypoxia (<2 mg/L DO) in the lower layers of many lakes (S. B. Brandt etal., 2009; Dent, Beutel, Gantzer, & Moore, 2014; Gächter & Wehrli, 1998; Horne& Goldman, 1994; Özkundakci, Hamilton, & Gibbs, 2011). Meanwhile, predic-tions of a globally warming climate (Field et al., 2014; IPCC, 2014), with im-plications for lake water temperatures and stratification patterns (Butcher, Nover,Johnson, & Clark, 2015; Field et al., 2014; Kraemer et al., 2015; D. M. Living-stone, 2003; Palmer, Yan, & Somers, 2014), and runoff timing and magnitude inresponse to changes in precipitation patterns (S. J. Cohen et al., 2006; Field et al.,2014; IPCC, 2014; Jiménez Cisneros et al., 2014; Merritt et al., 2006) threaten toexacerbate the already severely limited available habitat in many of these basins1(C. C. Coutant, 1990). Examples of lakes historically and currently affected bytemperature-oxygen squeezes include North and South Twin Lakes, WA, USA(B. Moore et al., 2014), Keystone Reservoir, OK, USA (Zale et al., 1990), LakeTexoma, OK-TX, USA (W. J. Matthews et al., 1985), Cherokee Reservoir, TN,USA (C. Coutant, 1987), the (Lower) Lake of Zurich, Switzerland (North, North,Livingstone, Köster, & Kipfer, 2014), Lake Rotoiti, New Zealand (Rowe & Chis-nall, 1995), Amisk Lake, AB, Canada (Aku & Tonn, 1999), Osoyoos Lake, BC,Canada (Hyatt et al., 2015; Paul Askey, Mike Sokal, Hillary Ward, pers. comm.,April 18, 2016), as well as reservoirs in Alabama, USA (Moss, 1985). Ultimatelythis can have lasting consequences for the lake’s natural biodiversity as well as thelocal fishery industry.Many different strategies to reduce trophic status and increase available fishhabitat in lakes have been applied around the world with varying results. One ap-proach is controlling diffuse and point sources of phosphorous (P) and nitrogen (N)(critical nutrients for primary production) via a variety of imposed watershed man-agement techniques in order to limit nutrient loading to streams and lakes (Boesch,Brinsfield, & Magnien, 2001; Brambilla, Lalumera, Terova, Crosa, & Saroglia,2007; G. D. Cooke, Welch, Peterson, & Nichols, 2005; Horne & Goldman, 1994;Prepas & Burke, 1997). Reduction of external nutrient input can have limited ef-fects when a major source of nutrients in the lake is internal loading from sedimentsthat occurs when the DO at the sediment-water interface falls below 1-2 mg/L (Aku& Tonn, 1999; Dent et al., 2014; Gächter & Wehrli, 1998; Gibbs & Özkundakci,2011; Mortimer, 1942, 1971; R. Nordin, 1980, 1987; North et al., 2014; Özkun-dakci, Hamilton, & Trolle, 2011; Özkundakci, Hamilton, & Gibbs, 2011; Prepas& Burke, 1997; Smolders, Lamers, Lucassen, Van Der Velde, & Roelofs, 2006;E. B. Welch & Cooke, 1999). In this case, in-situ methods may reduce internalnutrient loading from bottom sediments during periods of hypolimnetic anoxia tolevels that limit primary production thereby reducing the potential for algal blooms.Some examples include: biomanipulation (i.e. adding zooplankton to reduce algalmass) (G. D. Cooke et al., 2005; Smolders et al., 2006), biological manipulation(i.e. nitrate addition) (Özkundakci, Hamilton, & Gibbs, 2011; Smolders et al.,2006; Wauer, Gonsiorczyk, Kretschmer, Casper, & Koschel, 2005), application ofa “sediment cap” using a P-inactivation agent (i.e. alum Al2 (SO4)3 or modified2zeolite [Z2G1]) (Berg, Neumann, Donnert, Nüesch, & Stüben, 2004; G. D. Cookeet al., 2005; Gibbs & Özkundakci, 2011; Özkundakci, Hamilton, & Trolle, 2011;Özkundakci, Hamilton, & Gibbs, 2011; M. Robb, Greenop, Goss, Douglas, &Adeney, 2003; Smolders et al., 2006; E. Welch & Schrieve, 1994; E. B. Welch& Cooke, 1999; Yamada, Kayama, Saito, & Hara, 1987), lime treatment (i.e. co-precipitate nutrients and phytoplankton) (G. D. Cooke et al., 2005; Smolders et al.,2006), and sediment removal (G. D. Cooke et al., 2005; Smolders et al., 2006).One of the more widely used methods to reduce anoxia is hypolimnetic aeration oroxygenation (Ashley, 1983; G. D. Cooke et al., 2005; Dent et al., 2014; B. Mooreet al., 2014; Prepas & Burke, 1997; Smolders et al., 2006; Wauer et al., 2005; Xue,Gächter, & Sigg, 1997). Although even the overall effects of lake aeration on waterquality are not conclusive (Gächter & Wehrli, 1998; Prepas & Burke, 1997).Wood Lake, one of Canada’s premier kokanee fisheries (Askey, 2013), experi-ences the aforementioned temperature-oxygen squeeze on an annual basis duringlate summer and early fall (summer surface temperatures over 20◦C and bottomlayers with 0 mg/L of dissolved oxygen) as well as significant internal P-loading(Epp & Neumann, 2014; Field data received from British Columbia Ministry ofEnvironment (BCMOE), Mike Sokal, pers. comm., February 13, 2015 & March 15,2016). These conditions are believed to be the result of severely reduced inflowdue to upstream diversions and beaver dams (increased lake retention time), cou-pled with an increasingly warmer summer climate (S. Cohen et al., 2001, 2004;D.R., Bennett, Werner, Murdock, & Bronaugh, 2009; IPCC, 2007; Pacific Cli-mate Impacts Consortium [PCIC], 2013; BCMFLNRO, 2009; University of BritishColumbia [UBC], 2014; Hillary Ward, pers. comm., June 25, 2015; Paul Askey,pers. comm., 12 July 2015). In 2011, the habitat squeeze became severe enough toresult in the major die-off of thousands of kokanee despite many regional advancesin the past several decades that have reduced nutrient loading to the lake (Epp &Neumann, 2014; Webster, 2014). This tragic die-off in 2011, in addition to cli-mate warming predictions (S. Cohen et al., 2001, 2004; PCIC, 2013; UBC, 2014)that threaten to produce increased surface water temperatures and also increaseddemands on water from the lake’s primary tributary, have motivated the currentstudy.The primary goals of this study are to (1) provide a detailed account of the cur-3rent physical limnology of Wood Lake, (2) to apply a one-dimensional lake modelto assess a variety of feasible, affordable, and natural-design-approach proposedsolutions (i.e. management of the lake’s inflows and outflows) to mitigate futurefish kills by lowering surface water temperatures in the late summer and early fall;and (3) to utilize this model to discern the lake’s sensitivity to perturbations in dailyaverage air temperatures in terms of its thermal structure. Note that although thefield work investigations and analyses involve several key parameters of the lake’scurrent condition (i.e. temperature, DO, chlorophyll-a), only the thermal structureof the lake is modelled. The focus herein is to test the potential for various manage-ment scenarios to alleviate the thermal stress component of the temperature-oxygensqueeze for kokanee in Wood Lake during late summer and early fall.This document begins with a review of the relevant history and several keylimnological studies in Chapter 2, followed by a description of the fieldwork in-volved in the current study and the development of a one-dimensional DynamicReservoir Simulation Model (DYRESM) of Wood Lake in Chapter 3. Chapter 4presents an analysis of the key field work and modelled results for the 2015-2016field season (as well as model validation), concluding with an evaluation of severalmodelled management scenarios in terms of how they modify the thermal struc-ture of Wood Lake. Chapter 5 reveals a (modelled-approach) investigation intowhat factors made 2011 unique compared to subsequent years, studies the key fac-tors that influence Wood Lake’s thermal structure and heat content, and also dis-seminates Wood Lake’s response to seasonal/annual meteorological perturbationsand elevated air temperatures. Chapter 5 concludes with a discussion regardingthe management scenarios evaluated in Chapter 4 in terms of their potential forcooling the surface layer of Wood Lake during August and September when thetemperature-oxygen squeeze becomes critical.4Chapter 2Literature Review2.1 LocationWood Lake (391.4m amsl, 50.0292◦N, 119.4044◦W, 6.6km long, 1.7km wide, av-erage depth: 22m) is located in the Okanagan Valley, which is a semi-arid regionof the Southern Interior of British Columbia (BC). This region contains a largenumber of mountain and basin lakes that experience significant variations in runofffrom snowmelt and rainfall from year to year (Anonymous, 1974). Wood Lakereceives approximately 40 cm/year of precipitation, while surface evaporation isapproximately 76 cm/year. The regional average annual air temperature is 7◦C(BC Water Investigations Branch [BCWIB], 1974; Northcote et al., 1974).Wood Lake is a headwater for the Columbia River system. It receives inflowfrom Ellison Lake via Middle Vernon Creek (MVC) (limited by a temporary flowcontrol structure during the summer months), and discharges water to KalamalkaLake via a canal with no imposed control structure (Epp & Neumann, 2014).Ellison Lake receives its inflow via Upper Vernon Creek (UVC) from a pair ofheadwater lakes at higher elevation, Crooked Lake and Swalwell Lake (Jensen &Bryan, 2001). Wood Lake’s predominant inflow is from MVC (drainage area of104 km2) (Anonymous, 1974); however, it also receives minimal inflow (unmoni-tored) from Winfield Creek (south end), Ribbleworth Creek (east side), and Hay-ton Creek (north end) (Anonymous, 1974; R. Nordin, 1987). The total combineddrainage area for Wood Lake is 202 km2 (Anonymous, 1974). A canal, approx-5imately 180m long connects Kalamalka Lake and Wood Lake, which lie at thesame elevation (Jensen & Bryan, 2001). The canal was dredged in 1908, loweringthe water level of Wood Lake by about 0.6m (Walker, Reavie, Palmer, & Nordin,1993). Wood Lake generally flows north into Kalamalka Lake, which has an outletcontrol at the north end [Water Survey of Canada hydrometric station 08NM143]controlled by the City of Vernon and subsequently flows through Lower VernonCreek into Okanagan Lake (Ferguson et al., 1974). However, sometimes thereare wind-driven flow reversals between Kalamalka Lake and Wood Lake (BC Re-search, 1974; BCWIB, 1974; Larratt Aquatic Consulting [LAC], 2010). Figure 2.1below shows the area surrounding Wood Lake, as well as the relative size and ge-ographical relationship of all of the aforementioned lakes and primary tributaries.0 5 kmEllisonLk.UVCClarkCr.Swallwell Lk.Dee Lk.Island Lk.Deer Lk.Crooked Lk.WoodLk.KalamalkaLk.Oyama Lk.Oyama Cr.Coldwell Cr.Lower Vernon Cr.OkanaganLkMVCOyama CanalWinfieldVernonNFigure 2.1: The Vernon Creek catchment and area surrounding Wood Lake.Table 2.1 summarizes the physical properties of the three principal lakes rele-vant to this study. The retention time of these three lakes (range shown in Table 2.1)6has increased over recorded history since 1930 due to increased upstream diver-sions for irrigation and domestic water consumption (Anonymous, 1974; Jensen &Bryan, 2001; Walker et al., 1993).Table 2.1: Comparison of Selected Physical Properties of Ellison Lake, WoodLake, and Kalamalka Lake (Jensen & Bryan, 2001).Parameter Ellison Lake Wood Lake Kalamalka LakeSurface Area (ha.) 210 930 2590Volume (da m3) 5400 199,500 1520000Depth (Avg./Max) [m] 2.5/5 22/34 59/142Retention Time (yr) 0.3-2.1 14-30 37-65MVC 20%Wood LakeInflow UVC (& 3 minor streams)1 80% Coldstream Creek&groundwater2Thermal Structure Polymictic Dimictic Dimictic1 (Anonymous, 1974), 2 (LAC, 2010)Within the drainage basin for Wood Lake and Kalamalka Lake, approximately10,522 hectares of land is classified as having agricultural capability, with 8094hectares requiring irrigation due to insufficient precipitation relative to evapotran-spiration (BCWIB, 1974). In terms of Wood Lake’s catchment, the majority of thebenchlands and south-facing low elevation slopes on the east side of Wood Lakehave been developed into agricultural land (formerly semi-arid grassland) while thealluvial flat area south of Wood Lake that was formerly predominantly agriculturalland, is now primarily residential (Walker et al., 1993).2.2 Motivation for ResearchAlthough Wood Lake is one of Canada’s best kokanee fisheries (Askey, 2013),there has been a drastic decline in kokanee populations of all age classes in re-cent years. This decline has been attributed to unsuitable conditions that developin Wood Lake (warm surface waters and anoxic hypolimnion) during salmon mat-uration in the summer and early fall, as well as insufficient spawning habitat inthe lake’s major tributary, MVC (Webster, 2014). Anthropogenic influences arebelieved to be largely responsible for the deterioration in Wood Lake (and MVC)7kokanee habitat, which has then been exacerbated by a warming climate. For ex-ample, increased upstream diversions for agriculture and municipal use have re-duced inflow and thereby increased the retention time of the lake (reduced flush-ing) (Anonymous, 1974; Epp & Neumann, 2014; Walker et al., 1993), while theregional climate over the last 85-100 years persistently yields warmer summerswith less precipitation (S. Cohen et al., 2001, 2004; D.R. et al., 2009; Field et al.,2014; PCIC, 2013; BCMFLNRO, 2009; UBC, 2014). Meanwhile pollution and nu-trient loading from agriculture (fertilizers, manure), leaking septic tanks, and otherhuman activities in the watershed have accelerated eutrophication of Wood Lakesince the 1930’s (Anonymous, 1974; BCWIB, 1974; Walker et al., 1993). Severalmanagement solutions were first proposed in the 1970’s and 1980’s (Anonymous,1974; BCWIB, 1974; R. Nordin, 1980, 1987). However, other than improvingagricultural practices and transferring many homes from septic to sewage (Epp &Neumann, 2014), no affirmative action was ever employed, largely because suchoptions were expensive and had uncertain outcomes (BCWIB, 1974; R. Nordin,1980). This study utilizes a combined field and modelling approach to investigatethe impacts of warming seasonal/annual air temperatures as well as to study theinfluence of several management scenarios on the thermal structure of Wood Lake.The impact of European settlement is evident in past limnological and hydro-logical observations as well as core samples taken from Wood Lake by Walker et al.(1993). Agriculture was well established in this region prior to 1900 and has playedan important role in terms of nutrient loading into Wood Lake and water consump-tion from MVC. The runoff that enters UVC is regulated by multiple dams andreservoirs managed by the Winfield-Okanagan Centre Irrigation District. Since1930, the rapid increase in consumptive demands of freshwater from this reser-voir system for agricultural and municipal uses, in addition to increased pollution,and septic field leakage has caused a reduction in water quality (eutrophication)and quantity available for non-consumptive uses, such as recreation, aesthetics,and fisheries (Walker et al., 1993). Water losses from the system are estimated at:83% of water withdrawn for agricultural use (evapotranspiration) and 50% of waterwithdrawn for municipal use (BCWIB, 1974). From 1930 to 1970, it is estimatedthat diversions (reducing instream flows) resulted in an increase in the residencetime of Wood Lake from 12 to 30 years (Walker et al., 1993).8There have been many observations of poor reproductive health in kokaneesalmon populations in Wood Lake in the last eight years (Webster, 2014, 2015).Kokanee salmon populations of all age classes in Wood Lake remain currently ina depressed state and since 2011 the observed spawn returns have been the low-est recorded since 1994, when reliable recording began (Epp & Neumann, 2014).Return spawner counts, proportion of females sampled with low egg counts, andestimated egg deposition for MVC from 2009 to 2014 have been obtained fromkokanee stream spawner enumeration studies in the Okanagan (Table 2.9). 2009total run estimate was 4500 less than 2008 and the second lowest escapement es-timate since 1994 (Webster, 2010). No kokanee were observed in UVC in 2009or 2011 and only two kokanee were observed in 2012 due to low flows and beaverdam obstructions in MVC (Webster, 2013). Wood Lake is subject to significantfishing, thus escapement numbers do not reflect exact numbers of fish produced bythe lake (Webster, 2014). In 2008 - 2013, 18% (on average) of in-skein femalessampled had egg numbers well below the traditional trend (although the samplesizes are small) (Webster, 2014). Their skeins were well developed but smallerthan average. Note that some females sampled in previous years have had normalsized skeins with low egg counts due to having large eggs, which can result in lowfecundity (Webster, 2015). In 2011, 15 (38%) of the females sampled had low eggcounts (Webster, 2012). Underdeveloped ovaries with non-viable eggs as well asleftover secondary skeins with black eggs (uncommon) were noted in recent years(Webster, 2012). In 2011, the estimated egg deposition for MVC was 4,003,400(Webster, 2012); however, in 2012, MVC egg deposition estimates plummeted to1,072,638 (Webster, 2014).Warm surface waters and anoxic hypolimnion during late summer and earlyfall can generate a “temperature-oxygen squeeze” for aquatic wildlife that reducesaccess to food, generates thermal, respiratory, and confinement/crowding stresses,can limit fecundity and juvenile survival, and can lead to fish kills (C. C. Coutant,1985; B. Moore et al., 2014; Myrick & Cech Jr, 2000; Prepas & Burke, 1997; Rowe& Chisnall, 1995). In general, dissolved oxygen levels below 1 mg/L are consid-ered to be fatal to most fish species (Fry, 1971 as discussed in S. B. Brandt et al.(2009)); however, the lethal threshold for DO and the threshold for hypoxia im-pacts on aquatic life varies significantly for different species. The average thresh-9Table 2.2: Kokanee salmon spawner enumeration in tributary streams forWood Lake, number of females sampled with low eggs counts in MVC,and estimated egg deposition in MVC from 2009 to 2014.Year Spawner Counts Low Egg Counts Estimated Eggin MVC (Underdeveloped Ovaries) in Deposition in MVCMVC [n = sample size] & UVC20091 5250 2 (10%) [n = 20] 2,156,70020102 13508 1 (3%) [n = 36] 6,968,40220113 8301 15 (38%) [n = 39] 4,003,40020124 1566 2 (29%) [n = 7] 1,072,63820135 2897 3 (30%) [n = 10] 1,507,68020146 8879 0 (0%) [n = 10] 1,676,9251 (Webster, 2010) 2 (Webster, 2011) 3 (Webster, 2012) 4 (Webster, 2013)5 (Webster, 2014) 6 (Webster, 2015)old value reported in literature is 2.31 mg/L (Vaquer-Sunyer & Duarte, 2008). DOlevels between 2 - 4 mg/L have been observed to affect food consumption, growth,reproduction, distribution, and behaviour in fish (Aku & Tonn, 1999; Breitburg, Lo-her, Pacey, & Gerstein, 1997; Herbert & Steffensen, 2005; Kramer, 1987; T. Robb& Abrahams, 2003; Wannamaker & Rice, 2000). Temperature affects physiolog-ical and biochemical rates (Fry, 1971 as discussed in S. Brandt, Magnuson, andCrowder (1980)) and also has a strong influence on fish in terms of their reproduc-tion, metabolism, consumption and growth; hence, fish tend to migrate to layersin the lake with optimal temperatures (S. Brandt et al., 1980; Brett, 1971; Jobling,1981; Olla, Studholme, & Bejda, 1985). Fish are able to migrate to optimal envi-ronmental temperatures (thermoregulate behaviorally) (Aku & Tonn, 1999; Neill,Magnuson, & Chipman, 1972; Olla et al., 1985; Rozin & Mayer, 1961). Preferred,avoided, and lethal temperatures have been determined for many fish species (Fry,1947 as discussed in Olla et al. (1985))(Beitinger, Bennett, & McCauley, 2000;Black, 1953; Cherry, Dickson, Cairns, & Stauffer, 1977; C. Coutant, 1977); how-ever kokanee salmon were not included in many of these studies. During summerstratification, as a result of these elevated temperatures in surface waters and di-minished hypolimnetic DO levels, fish become squeezed into a middle layer ofpreferred suitable habitat in the lake.This temperature-oxygen squeeze phenomenon has been observed to develop10through the stratified season on an annual basis in Wood Lake to varying degrees ofseverity (BC MOE Data, Paul Askey, Mike Sokal, & Hillary Ward, pers. comm.,2014-2016). DO concentrations are elevated (>9 mg/L) above the hypolimnionin the spring due to photosynthetic activity and/or weather (surface mixing). Thewater column is close to 100% saturated (uniform) immediately following springturnover in March/April (BC Research, 1974), at which time the highest DO con-centrations are recorded (Epp & Neumann, 2014; (BCMOE Data)). By mid-July/August, the deepest levels in the lake are isolated from inputs of oxygen (strat-ification), becoming anoxic (DO < 0.5 mg/L) due to sedimentation and subsequentdecomposition of phytoplankton and other organic material in the sediments (Epp& Neumann, 2014). DO decreases with depth in the hypolimnion to 0 mg/L nearthe lake bottom at the beginning of August (clinograde DO profile is typical of eu-trophic lakes) (MacDougall, 1984). Autumn hypolimnion oxygen levels (August toOctober) serve as an indicator of the total biological production during the growingseason (Jensen & Bryan, 2001), thus anoxia suggests significant algal growth anddecomposition. The volume of water in the hypolimnion in Wood Lake is smallrelative to its surface area, which aids the depletion of hypolimnetic oxygen in thesummer (Jorgensen, Loffler, Rast, & Straskraba, 2005; LAC, 2010). Surface layertemperatures exceeding 20◦C (lethal limit for kokanee (Hillary Ward, Paul Askey,pers. comm., 2015)) can persist through September in some years (i.e. 2009, 2011,2013) (BCMOE data). Kokanee become squeezed into a middle layer during latesummer and early fall, limiting available habitat and access to food. DO continuesto be consumed in the lake without replenishment until turnover in late Novemberor early December, further raising the depth of this DO threshold and restrictingkokanee habitat. The anoxic zone typically reaches a maximum thickness in mid-October (Epp & Neumann, 2014), as observed in the current study (2015). In 2015,anoxic conditions in the hypolimnion persisted through November in 2015 and thelake did not fully turn over until 2 December 2015. No major DO changes havebeen reported during ice cover (BC Research, 1974).Severe temperature and oxygen stress conditions were observed in Wood Lakeon 13 September 2011 (BC MOE data), and anglers noted many fish mortalities (allage classes) that season. Despite limited empirical observations in Wood Lake from2011 (single profile at one location on 29 March & 13 September (Figure 2.2)),11kokanee appear unable to survive for extended periods in Wood Lake waters wheretemperatures rise over 20◦C or DO levels fall below 2mg/L. In fact, in other lakesin the interior (i.e. Osoyoos Lake), DFO recognizes 17◦C as the avoided temper-ature for sockeye (Hyatt et al., 2015; Paul Askey, pers. comm., April 18,2016).Observations in Wood Lake indicate that kokanee can survive in Wood Lake in17◦C - 20◦C waters and where DO is between 2 - 4 mg/L, although detrimentalhealth effects are considered imminent (Paul Askey, pers. comm., June 8, 2015,July 11, 2015 & April 18,2016). Black (1953) reported the upper limit at which50% of kokanee survive 24 hours is 22◦C when acclimatized to 11◦C, the lowestof 14 Okanagan species studied; these temperature limits (preferred/avoidance) aresimilar to those reported for other salmon species (C. Coutant, 1977). It is reason-able that the temperature threshold for extended exposure (>24 hours) for kokaneewould be less than 22◦C (Beitinger et al., 2000). Salmonidae have been reported tohave low temperature tolerances (compared to other fish families) in many studies(i.e. Beitinger et al. (2000); Black (1953); Cherry et al. (1977)). The recognizedDO thresholds in Wood Lake are consistent with those previously reported (Aku& Tonn, 1999; Breitburg et al., 1997; Herbert & Steffensen, 2005; Kramer, 1987;T. Robb & Abrahams, 2003; Vaquer-Sunyer & Duarte, 2008; Wannamaker & Rice,2000). In September 2011 (Figure 2.2) low DO was observed in Wood Lake be-low 10m depth, while water temperature exceeded 17◦C above this depth and 20◦Cabove 7m depth. There was no “ideal habitat” (temperature below 17◦C and DOabove 4mg/L) for kokanee in Wood Lake at this time. The “marginal habitat”(avoided region with 17◦C<Water Temperature<20◦C and 2mg/L<DO<4mg/L)may cause harmful effects to kokanee if exposed to it for extended periods, was3m thick at this time. The “available habitat” is habitat between the 20◦C watertemperature threshold and the 2 mg/L DO threshold; in this case all available habi-tat was “marginal habitat”. The remainder of the water column was inhabitable bykokanee on 13 September 2011. It is hypothesized that the extreme temperature-oxygen squeeze that occurred in 2011 was the result of two principal factors: aprolonged spring runoff, and warmer-than-average air temperatures in late summerand early fall (Epp & Neumann, 2016).12Figure 2.2: Temperature and Dissolved Oxygen profiles in Wood Lake on 13September 2011. MOE Field Data (EMS ID: 0500848) recorded at 2mintervals using a YSI multimeter probe. Data was interpolated into 0.5mdepth intervals. At this time, there was 3m of available kokanee fishhabitat between the 20◦C threshold and the 2 mg/L DO threshold andno ideal habitat available between the 17◦C threshold and the 4 mg/LDO threshold (Mike Sokal, pers. comm, 15 March 2016). The 20◦Cand 17◦C thresholds lay at 7m depth and 9.5m depth respectively, whilethe 2 mg/L and 4mg/L DO thresholds lay at 10m depth and 9.5m depthrespectively.From 2013 to 2015, the observed temperature-oxygen squeeze was less severethan 2011. In 2013 and 2014, the anoxic zone reached a maximum thickness inSeptember-October beginning at a depth of approximately 22m (down to lake bot-tom at 34m). In September 2013 the habitable layer reached a minimum thicknessof approximately 14m, and in Jul-Aug 2014, it reached a minimum of approx-imately 16m (BCMOE Data). In 2015, the anoxic zone reached its maximumthickness on 28 October 2015, extending from 18.6m depth down to the lake bot-tom. The minimum habitable layer thickness (13.6m) occurred on 25 August 2015(Section 4.3 and 4.4). Signs of improvement in the last two years have also been13aided by more rigorous kokanee harvest regulations (BCMFLNRO, 2013, 2015;Webster, 2015). In 2013, a 45-day kokanee harvest regulation (fishery open from15 April to 31 May only, with an imposed daily quota of 2 kokanee per angler)was instigated on Wood Lake to prevent overharvesting (Askey, 2013; Webster,2014). These limits remained in place for 2014 (BCMFLNRO, 2013). In 2015and 2016 (and proposed for 2017), the Wood Lake fishery was re-opened from 1April to 31 August, with a daily quota of two fish per angler (no fishing from 1September to 31 March) (BCMFLNRO, 2015). In 2014 the egg counts were con-sistent with traditional trends. Although 2014 spawner counts (via a fish trap/fencein the lower reach of MVC above Reimche Road) are notably higher than 2013,they are still only about 53% of 2010 parental stock. The mean kokanee length in2014 (341.46mm) from 2215 live-trapped fish was the lowest recorded since reli-able data recording was initiated in 2006 (Webster, 2015). Kokanee have made anotable recovery since 2011, although numbers remain below 2010 parental stockvalues (Table 2.9) (Webster, 2015). Wood Lake water quality has shown improve-ments since 2000 with the removal of septic tanks within the watershed and im-provements in farm practices (LAC, 2010).Profile data from recent years also suggests a climatological influence on thedevelopment of unsuitable conditions in Wood Lake. Epp and Neumann (2014) re-port that the waters in Wood Lake are well mixed, with temperatures relatively uni-form until about mid-May, and that a thermocline develops by June. In 1972/1973this was the condition reported by BC Research (1974) as well. In 2013 - 2015,the thermocline has become established progressively earlier than June (Table 2.3).The waters in Wood Lake are well mixed, with temperatures relatively uniform un-til about mid-March, as was observed in 2016 (Section 4.3). A thermocline beginsto develop in April-May. On 4 May 2015 the lake was already stratifying with av-erage surface (top 1m) and bottom temperatures in the middle 1.25km of the lakeof 12.5◦C and 5.2◦C respectively. Predictions of longer and warmer summers aremost alarming to future kokanee salmon populations, as this will also cause warmsurface temperatures to persist into September when the DO conditions are mostsevere and increase the severity of the temperature-oxygen squeeze (S. Cohen etal., 2001, 2004; Field et al., 2014; IPCC, 2014; PCIC, 2013; UBC, 2014).14Table 2.3: Date of onset of stratification in Wood Lake in 2013 - 2016.Year Date of First Observed Stratification2013 18 June 2013 (BCMOE Data)2014 21 May 2014 (BCMOE Data)2015 4-13 May 20152016 25 April 2016The effects of climate change on lakes have been realized in monitoring dataand simulations of future scenarios for many lakes around the world (Blumberg &Di Toro, 1990; Chang, Railsback, & Brown, 1992; Coats, Perez-Losada, Schladow,Richards, & Goldman, 2006; de Stasio, Hill, Kleinhans, Nibbelink, & Magnuson,1996; D.W, 2001; Fang & Stefan, 2009; Fee, Hecky, Kasian, & Cruikshank, 1996;Jiang & Fang, 2016; Palmer et al., 2014; Schindler et al., 1996, 1997; Schmid, Hun-ziker, & Wüest, 2014; Vincent, 2009). Common observed and expected changesinclude quality, quantity and availability (i.e. timing of stream flows) of freshwaterresources (S. Cohen et al., 2001, 2004; S. J. Cohen et al., 2006; D.W, 2001; IPCC,2014; Merritt et al., 2006; Schindler et al., 1996, 1997; Vincent, 2009), earlier on-set of stratification, increase in surface layer temperatures and thermal stability oflakes, warming autumn lake temperatures, changes in water clarity, changes in al-kalinity and concentrations of base cations, nitrogen, dissolved organic C, silica,and P, as well as delay of fall turnover (de Stasio et al., 1996; D.W, 2001; Fang &Stefan, 2009; Fee et al., 1996; Jiang & Fang, 2016; Palmer et al., 2014; Schindleret al., 1996, 1997; Schmid et al., 2014; Vincent, 2009). Some studies report de-creased mixing depths (due to increase in dissolved organic carbon concentrationsin lakes) (Palmer et al., 2014), whereas others indicate an increase in the depth ofthe surface mixed layer and lower concentrations of dissolved organic C (Fee etal., 1996; Schindler et al., 1996). Concurrently, studies suggest a decrease in hy-polimnetic DO leading to a lengthening and intensification of summer/fall anoxiain response to climate change (Blumberg & Di Toro, 1990; Fang & Stefan, 2009;Jiang & Fang, 2016; North et al., 2014; Stefan, Hondzo, & Fang, 1993). Globalwarming threatens increasingly warmer and progressively longer summer tempera-tures in the Okanagan Valley (S. Cohen et al., 2001, 2004; Field et al., 2014; IPCC,2014; BCMFLNRO, 2009; UBC, 2014), and the potential effects of these factors15on Wood Lake’s thermal structure and available habitat is of particular interest.There is considerable motivation from a variety of stakeholders (including BC-MOE) for researching Wood Lake’s past, current, and potential future conditions.The anthropogenic factors contributing to the conditions in Wood Lake and the re-sulting consequences on its chemical, biological, and physical properties are repre-sentative of the issues plighting many lakes around the world. Lake managers (i.e.BCMOE and Freshwater Fisheries Society of BC) are seeking to answer whether itis possible to mitigate fish kills from warm surface waters and anoxic hypolimnionin Wood Lake through water management inflowing from MVC and outflowing toKalamalka Lake, or through an alternative “natural-design” approach.2.3 Current Research ObjectivesThe research objectives of this study are to determine the principal factors that in-fluence Wood Lake’s annual and seasonal thermodynamic structure as well as topredictively model the potential thermal impacts of proposed management strate-gies for Wood Lake, and from these modelled scenarios infer impacts on availablefish habitat. This work includes reviewing existing limnological observations ofWood Lake, conducting an intensive one-year physical limnological field studyof Wood Lake to facilitate validating a one-dimensional thermodynamic model(DYRESM), and subsequently employing this model to better understand howWood Lake responds to thermodynamic forcing and to predictively test the ther-mal impacts of various management scenarios proposed for Wood Lake. Prior toanalyses and discussion of results from the current study (Chapter 4 and 5), thisinvestigation begins with a detailed summary of existing historical observationsand data from Wood Lake since the 1960’s in order to understand the developmentof present-day conditions. This review guided development of research methods(Chapter 3) for the current 2015-2016 study period and the development of ap-propriate and feasible modelled management scenarios. This chapter concludeswith a discussion of historical proposed management strategies (Section 2.7) anda detailed description of the six current proposed management strategies for WoodLake (four of which are investigated in detail in this current study) (Section 2.8).162.4 Review of Historical DataThis section provides a review of historical observations and data collected fromvarious limnological studies of Wood Lake during the 20th century. This analysisaids in understanding how Wood Lake and the surrounding watershed have evolvedover time and resulted in the development of current conditions in Wood Lake.2.4.1 The Eutrophication of Wood LakeLimited limnological data is available for Wood Lake from 1936 to 1970 and thuscore sediment samples help fill this observational gap (Walker et al., 1993). Theseanalyses provide details of the physical, chemical, biological, and geological pa-rameters of the lake (affected by upstream diversions of VC, and pollutant/nutri-ent loadings from urban development and agriculture) prior to 1970, after whichmore intensive field studies were performed on the regional lakes. Walker et al.(1993) revealed via a paleolimnological study (seven cores extracted in 1992) ofWood Lake sediments that the lake was oligo-mesotrophic (low-moderate nutrientconcentrations) prior to settlement in the area, and became increasingly eutrophicthereafter until Hiram Walker Distillery (HWD) began supplementing inflows intoWood Lake in the 1970’s. Regular algae blooms have been reported on Wood Lakesince 1935, but became of increasing concern to local residents in the early 1970’s(Walker et al., 1993).A comparison of the bottom fauna from 1935 to 1971 reveals a significantreduction in the diversity and density of benthic fauna. In 1935 there was an abun-dance of profundal bottom fauna, dominated by Chironomus, which are charac-teristic of eutrophic conditions (Walker et al., 1993). Bottom living invertebratesare often used as indicators of changes in trophic status because they are sedentaryorganisms that are sensitive to temporal and environmental changes (Anonymous,1974). Core samples analyzed by Walker et al. (1993) indicate that a significantdecrease in fauna density (i.e. Chironomus) occurred in 1940, concurrent with theincreased use of pesticides and chemical fertilizers. This was followed by a smallerdecrease in 1975.Diatom and Chrysophyte analyses revealed a reduction in planktonic speciesassociated with less eutrophic waters (Aulaocoseira and Cyclotella spp.) and an17increase in species associated with highly eutrophic waters (Stephanodiscus min-utulus) prior to 1940. Despite a decline in abundance from approximately 1940to 1992, S. minutulus was consistently dominant during this time, with a coupleof small exceptions where species associated with less eutrophic waters showedspikes in abundance (Walker et al., 1993). The decline after 1940 suggests achange in dominance from diatom to blue-green algae and increased eutrophic con-ditions. Diatoms are indicative of oligo-mesotrophic conditions (when abundant),and are replaced by blue-green alga as the most abundant species with increasingtrophic state (Anonymous, 1974). Chrysophyte abundance, another indicator ofoligotrophic conditions, also showed a decline towards surface sediments. Chi-ronomid taxa characteristic of profundal, oligotrophic environments with a coldwell-oxygenated hypolimnion (Parakiefferiella nigra and Heterotrissocladius) andmesotrophic environments (Sergentia) were nearly absent after 1940, once againindicating anoxic hypolimnetic conditions persisted during this time. The subse-quent decline in eutrophic diatom concentrations from 1970 to 1992 may be indica-tive of improvements in water quality due to HWD inputs. However, in 1992 theprofundal conditions were still adverse to establishment of benthic invertebrates(profundal chironomids), showing only a slight increase. Speculation exists overwhether the decline in profundal fauna is entirely explained by the depletion ofhypolimnetic oxygen, or if other factors are present (Walker et al., 1993).Oxygen depletion in the 1960’s in Wood Lake is also shown by an increase inmolybdenum [Mo] concentrations. Mo is typically depleted in anoxic water andis found in high concentrations in anoxic sediments. Elevated Mo concentrationswere found in Wood Lake sediments from 1940 to 1970. The peak concentra-tion was deposited around 1970, decreasing afterwards, during HWD operations(Walker et al., 1993).2.5 Water Movement Within Lakes and TransferBetween BasinsDye and drogue studies performed in the 1970’s revealed the general movement ofwater within Ellison, Wood, and Kalamalka Lake. UVC often short-circuits Elli-son Lake, travelling along the north shore, both during ice cover and similarly just18before/after ice cover when inflowing water temperatures exceed the lake temper-atures and/or when southerly winds dominate. Residence times as low as 1.2 hrsat an inflow rate of 2.3 m3/s were recorded. Short circuiting can occur during bothhigh and low inflows. Wind and water temperature have major roles in govern-ing the mixing of inflow water within Ellison Lake (BC Research, 1974). Studiessuggest that Ellison Lake is not an important source of nutrients for Wood Lake(Epp & Neumann, 2014). Due to Wood Lake stratification and inflowing MVCwater temperatures, MVC inflow mixes well with the layers above and within thethermocline during summer, but sinks down into the hypolimnion during fall (alsonoted in 1972/1973). Algal growth is somewhat limited to the epilimnion (higherturbidity and low transparency) and the bioavailability of nutrients is greatest dur-ing spring and summer. Wind on Wood Lake can redistribute nutrients and oxygenwithin the epilimnion, and north winds can cause incoming MVC water to becometrapped in the south end of the lake (BC Research, 1974).The pattern of water movement through the Oyama Canal was also realizedusing rhodamine WT dye and drogues (floats). Aerial photographs were takenduring these studies to qualitatively show the water movement patterns, whereingross water movement patterns were displayed using suspended solids as tracers.Warmer Wood Lake water tends to discharge along the surface of cooler KalamalkaLake waters near Oyama Canal and therefore, nutrients entering through the canalpass directly into the epilimnion where they can be assimilated immediately byphytoplankton. Winds during summer stratification tend to strongly influence and(north winds) impede the dispersion of nutrient-laden water from Wood Lake intothe epilimnion of Kalamalka Lake (BC Research, 1974; BCWIB, 1974). Althoughthere is a net northward flow of water from Wood Lake to Kalamalka Lake, windsand seiching often cause oscillations of flow through the canal (aided by the factthat these lakes lie at equal elevation). Flow from Wood Lake into Kalamalka Lakeis encouraged by south winds and spring freshet flows. North and/or south windstend to dominate in the summer, generating seiche events (Wiegand & Chamber-lain, 1987) in both basins that promote southerly movement and/or oscillations ofwater through the canal. Summer evaporation can also draw water in from Kala-malka Lake (BC Research, 1974; LAC, 2010). During late summer (dry years)there is a net southerly flow through the canal (BCWIB, 1974). Kalamalka Lake19experiences intense seiche events during warm autumns (main transport mecha-nism of surface contaminants in Kalamalka Lake). Seiche events are triggered bystorms and flows alternate 1-2 times/day (typical period is about 23.5 hrs) and lastfor several days with decreasing severity. North or southwest winds with gusts ex-ceeding 30km/hr can generate seiche events in this Kalamalka Lake. Each year, onaverage, seven to twelve major seiche events are detected (LAC, 2010).In 1971, HWD in Winfield began pumping water from Okanagan Lake to coolits stills and then discharged the water into UVC north of Ellison Lake at an av-erage rate of over 12,000 m3/day (estimates vary) from 1972-1990, reducing theresidence time of Wood Lake from 20-30 years to under 15 years (offsetting up-stream diversions) (Anonymous, 1974; BC Research, 1974; BCWIB, 1974; Jensen& Bryan, 2001; R. Nordin, Swain, Nijman, R., & Wetter, 1985). Ellison Lakeretention time decreased to around 0.3-1.4 years (from 0.6-2.1 years) and Kala-malka Lake residence time decreased from 46-65 years to 37-45 years (Anony-mous, 1974; BC Research, 1974; Jensen & Bryan, 2001; R. Nordin et al., 1985)During HWD discharge algal concentrations in Kalamalka Lake appeared to behigher in the epilimnion near the outlet of the Oyama Canal, inferring that thereis nutrient transfer from one basin to the other. Despite this, north winds tend todominate in the summer and reduce this influx of water and nutrients (BC Re-search, 1974). In 1992, the distillery ended operations (reducing flows to < 380m3/day) and ceased flows in 1995, increasing the residence time of Wood Lakeand Kalamalka Lake once again (Jensen & Bryan, 2001).2.6 Water Quality ParametersComparison of the typical ranges for phytoplankton and nutrient parameters forlakes of different trophic statuses with the average (1969 - 1999) deep basin valuesfor all available spring/autumn data in these three lakes suggests that KalamalkaLake is oligotrophic, Wood Lake is between mesotrophic and eutrophic, and Elli-son Lake is eutrophic (Table 2.4, adapted from Jensen and Bryan (2001)).20Table 2.4: Comparison of indices of trophic status to average values for Kala-malka, Wood, and Ellison Lakes from 1969 to 1999. [n] indicates samplesize. (Adapted from Jensen and Bryan (2001)).Trophic Status Chlorophyll-a Phytoplankton Total P Total N Secchi Depth Primary(µg/L) Abundance (µg/L) (µg/L) (m) Production1Growing (#/mL) Spring Spring Growing (C/m2/day)Season Avg. Growing Overturn Overturn Season Avg.Season Avg.Oligotrophic 0-2 <1000 1-10 <100 >6 50-300Mesotrophic 2-5 1000-5000 10-20 100-500 3-6 250-1000Eutrophic >5 >5000 >20 500-1000 <3 >>1000Kalamalka Lk. 1.7 [n=90] 3450 [n=29] 8 [n=52] 265 [n=38] 8 [n=64] UnavailableWood Lk. 13 [n=102] 2990 [n=13] 55 [n=58] 503 [n=47] 5.4 [n=63] UnavailableEllison Lk. 17 [n=82] 18270 [n=8] 42 [n=48] 440 [n=45] 1 [n=75] Unavailable1 (LAC, 2010)2.6.1 Water Clarity: Secchi DepthSecchi depth (the depth to which a secchi disk is visible from the surface whenlowered through the water column) provides a measure of water clarity and hasbeen measured typically at least twice/year in Wood Lake fairly consistently since1970 (Table 2.5; Figure 4.1). It is often correlated with chlorophyll-a and is utilizedas an indicator of algae biomass and free floating algae (Jensen & Bryan, 2001;R. Nordin, 1980). Note that secchi depth increased from 2m at the beginning ofHWD operations to 7.7m in 2007 during a low productivity summer (LAC, 2010).Investigation of measured secchi depths in Wood Lake from 1969 to 2015 suggeststhat water clarity in Wood Lake has increased on average since 1975. KalamalkaLake shows a similar trend with significantly more variability (R. Nordin, 1980).Prior to 1978, annual average secchi disk readings were between eutrophic andmesotrophic status; from 1978 to 2015, annual average secchi disk readings havebeen between mesotrophic-oligotrophic status (Table 2.4).21Table 2.5: Variation in water clarity in Wood Lake, as indicated by mean sec-chi depth (m) for selected years. This table outlines the coalition of secchidisk readings from a number of studies from 1970 to 2015, and indicateswhen the measurements were made that comprise the average.Year Secchi Depth (m) Year Secchi Depth (m)1939 (August)1 2 - 2.5 19923 6.75 (spring/fall only)1969-19792 2 19933 5.30 (spring/fall only)19703 3.65 (spring/fall only) 19943 4.30 (fall only)19714 2.5 (monthly Apr-Oct; 19953 7.15 (spring/fall only)max: 5.24m [May5])19725 2.84 (monthly May-Oct; 19963 7.20 (spring/fall only)max: 6.1m [May])19733,5 3.74 (monthly Apr-Aug) 19973 5.15 (spring/fall only)19743 3.06 (spring/fall only) 19983 7.35 (spring/fall only)19753 1.21 (spring/fall only) 19993 5.30 (spring/fall only)19763 1.75 (spring/fall only) 20003 5.25 (spring/fall only)19773 2.60 (spring/fall only) 20013 3.92 (spring/fall only)19783 3.70 (spring/fall only) 20023 5.05 (spring/fall only)19793 5.18 (spring/fall only) 20033 6.25 (spring/fall only)19803 5.88 (spring/fall only) 20043 5.85 (spring/fall only)19813 3.90 (spring/fall only) 20053 5.85 (spring/fall only)19823 6.40 (spring/fall only) 20063 5.85 (spring/fall only)19833 9.85 (spring/fall only) 2007 7.7 (monthly Mar-Oct))(low productivity summer)119846 5.90 (weekly/bi-weekly May-Oct; 2008 4.5 (monthly Mar-Oct)max: 10.1m [May14]) (high productivity summer)119853 4.30 (spring/fall only) 2009 4.9 (monthly Mar-Oct)(high productivity summer)119863 5.90 (fall only) 2010 4.5 (monthly Mar-Oct)(moderate productivity &marling event)119873 5.18 (spring/fall only) 20113 5.0 (spring/fall only)19883 6.11 (spring/fall only) 20123 7.1 (spring/fall only)19893 5.70 (spring/fall only) 20133 5.06 (monthly Mar-Oct)19903 6.50 (spring/fall only) 20143 5.14 (monthly Mar-Oct)19913 7.05 (spring/fall only) 20157 5.18 (monthly Mar-Oct)1(LAC, 2010), 2(R. Nordin, 1980), 3(BCMOE data acquired from Mike Sokal, pers. comm., Feb. 13, 2015)4(Anonymous, 1974), 5(BC Research, 1974), 6(MacDougall, 1984),7(BCMOE data acquired from Mike Sokal, pers. comm., April 13, 2016)221970 1975 1980 1985 1990 1995 2000 2005 2010 2015Year23456789Average Secchi Depth [m]Monthly Average During Field Season [m]Average of Spring/Fall Measurements [m]Fall Measurement Only [m]Figure 2.3: Secchi Disk Readings in Wood Lake from 1970 to 2015.2.6.2 Limiting Nutrients: Nitrogen and PhosphorusWood Lake appears to be co-limited by N and P, and thus addition of either nutrientcan stimulate productivity. Nutrient balance (TN:TP) definitions for microflora areas follows: P-limitation (>10:1 or 15:1), co-limitation of N and P (between 10-15:1and 5:1), and N-limitation (< 5:1) (LAC, 2010; R. Nordin et al., 1985).The dominant sources for nutrients to Wood Lake prior to HWD were reportedas subsurface flows from septic tanks and agriculture (fertilizer/irrigation/animalwaste), with a lesser percentage being derived from inflow through MVC (Anony-mous, 1974; BC Research, 1974; BCWIB, 1974; Northcote et al., 1974). In 1980,it was estimated that the non-point sources of P into Wood Lake were dominatedby septic tanks (42%), animal waste (27%), and logging (24%) (Jensen & Bryan,2001). N fertilizers began being used in the early 1900’s and were commonly usedfrom 1920 to 1960, but their use has since declined. P fertilizers on the other hand,have not been used extensively in the Okanagan due to sufficient natural levels in23soils (Walker et al., 1993). High spring freshet flows can contribute significant TP(from particulate P) and N (Horne & Goldman, 1994). Once these nutrients arereleased into the lakes, other factors including spring algae consumption, marl pre-cipitation, nutrient recycling, and exchange with sediments within the lake are im-portant processes in determining actual nutrient concentrations (Berg et al., 2004;Dittrich & Koschel, 2002; LAC, 2010; Robertson, Garn, & Rose, 2007). SedimentP concentrations have actually declined since settlement, particularly in the 1940’s(Walker et al., 1993). This may be explained by a number of factors: migrationfrom anoxic sediments to the sediment surface and release into anoxic hypolim-netic waters (Gibbs & Özkundakci, 2011; Özkundakci, Hamilton, & Gibbs, 2011;Prepas & Burke, 1997; Smolders et al., 2006; Walker et al., 1993), varying sourcesof sediments and erosion rates, dilution with accumulated CaCO3, and/or othercyclic biogeochemical reactions (Dent et al., 2014; Golterman, 2001; Mortimer,1942, 1971; R. Nordin, 1987; Smolders et al., 2006; Walker et al., 1993).Anonymous (1974) calculate the loading criteria (2000 - 3000 lbs/year) toachieve desirable concentrations of TP based on assimilative capacity of the lake.Every lake responds differently to nutrient loading. The assimilative capacity ofthe lake is the ratio of nutrients required for plant growth, respiration and produc-tion to the total nutrient loading. Anonymous (1974) note that salmonid growth isproportional to nutrient availability (measured as P-loading) until a threshold point(approximately equal to 77 µg/L), over which additional P results in a decrease ingrowth. Reproduction is also a factor of fish size, and thus fecundity is also de-pendent on P-loading. The average P in Wood Lake in 1971 was 219 µg/L. Theability of Wood Lake to support kokanee spawners based on productivity in 1970-1972 was assessed at being approximately 802,000 kokanee, whereas the numberof spawners in 1971 was estimated to be 3300 (0.4% of carrying capacity) (Anony-mous, 1974). The numbers of spawners in the past four years (Table 2.9) are stillwell below the estimated aforementioned carrying capacity of Wood Lake basedon productivity in 1970-1972.From 1969 to 1973 low nutrient concentrations were observed in the inflow andoutflow of Ellison Lake, despite HWD operations. In 1969 - 1971 it was estimatedthat approximately 50% P and 30% N was derived from subsurface flow from sep-tic tanks, approximately 20% P and 18% N was derived from agricultural sources,24and about 25%P and 37%N derived from inflow through MVC (Northcote et al.,1974). During HWD operations, an increase in nutrients (soluble P and nitrate) be-tween Ellison and Wood Lake in VC was reported in some studies. In 1972/1973,it was estimated that, in terms of total loading, Wood Lake received 51% of the Nand 87% of the P from MVC and also a large portion of nutrients from leaking sep-tic tanks. It was believed at this time that the major sources of P into Wood Lakewere surface flow from MVC, as well as diffuse loading from groundwater,septictanks, agricultural runoff, industrial discharge, and sewer overflows (Anonymous,1974; BC Research, 1974; BCWIB, 1974); however, this excluded internal load-ing, which has since been shown to be the most important source of P in WoodLake (Epp & Neumann, 2014; R. Nordin, 1980, 1987).As a part of the Middle Vernon Creek Action Plan [MVCAP] project (Sec-tion 2.6.3 and 3.1.2), water samples have been collected bi-weekly between Apriland October in 2012 - 2016 at three sites in VC [UVC (1) and MVC (2)] to assesstotal organic N, total Kjeldahl nitrogen (TKN), TN, ammonia (NH3), nitrite+nitrate(NO2+NO3), ortho-P (-PO4-3), total dissolved phosphorous (TDP), and TP. Datahas been provided by BC MOE (Mike Sokal, pers. corr., 13 February 2015 &Hillary Ward, pers. corr., 20 April 2016). Water chemistry data is combined withdischarge measurements to determine the mass flux of nutrients in VC. These mea-surements indicate that Ellison Lake primarily serves as a nutrient sink and re-moves some nutrients (consumption by algal and for macrophyte growth, and set-tling) before they enter MVC (Table 2.6 adapted from Epp and Neumann (2014)).Northcote et al. (1974) reported that 90% of the external P loading to Wood Lakeis retained in the lake. The total nutrient loading into and out of Ellison Lakeand into Wood Lake in recent years (i.e. 2013) has been lower than in 1969 to1973 (Table 2.7 adapted from Epp and Neumann (2014) with additional data fromBCWIB (1974); Northcote et al. (1974)). Note that 1972 was marked by signif-icant algae blooms, poor water transparency, and high nutrient inputs. However,during 1973, high transparency conditions prevailed, and few algae blooms oc-curred (green algae in thermocline) (BC Research, 1974). Annual nutrient loadings(particularly readily bioavailable dissolved and ortho-P fractions) from MVC arecomparatively low relative to amount of nutrients in Wood Lake (Epp & Neumann,2014). Wood Lake also serves as a nutrient sink to some extent, although results25are quite variable (BC Research, 1974; BCWIB, 1974). Anonymous (1974); BCResearch (1974); BCWIB (1974) note that groundwater flows and subsequent nu-trient loadings to surface waters are inherently difficult to estimate and thus thesesources incorporate the largest unknown source of error.Table 2.6: Annual Nutrient Loading into and out of Ellison Lake and intoWood Lake in 2013. (Adapted from Epp and Neumann (2014))Nutrients Into Ellison Lake Out of Ellison Lake Into Wood Lake(Total kg/yr) (Total kg/yr) (Total kg/yr)TKN 8023 7648 8405TON 7327 7180 7784TN 8392 7669 16004Ammonia 696 468 624NO2+NO3 366 41 7601Ortho-P 202 75 215TDP 460 256 431TP 624 546 1033Table 2.7: Comparison of TP and TN mass [kg/yr] loading into Ellison Lakeand Wood Lake in 2013 (surface inflow only) with 1969 to 1973 (totalloading). (Adapted from Epp and Neumann (2014)).Year TP [kg/yr] TN [kg/yr]19721 7053 17854Into Ellison Lake 19731 476 607420132 624 83921969-19713 1500-1805 2480019721 4069 26921Into Wood Lake 19731 1225 1959520132 1033 160041(BCWIB, 1974), 2(Epp & Neumann, 2014), 3(Northcote et al., 1974)26Controversy: Effects on Water Quality from Hiram Walker DistilleryOperationsWater quality and quantity investigations during and following HWD operationsoffer valuable insight into the limnological effects of supplementing flows in VCupstream of Wood Lake and offsetting upstream diversions; however, controversyremains regarding the ultimate positive and negative results for the water quality ofthese lakes (i.e. BC Research (1974); BCWIB (1974); Jensen and Bryan (2001);LAC (2010); R. Nordin (1987); R. Nordin et al. (1985); Walker et al. (1993)).Many studies indicate that the positive impacts on the water quality of WoodLake resulting from the additional inflow into VC outweighed any negative im-pacts. HWD obtained cooling water from the hypolimnion of Okanagan Lakewhere water quality is relatively consistent. The water quality of the dischargewas regularly monitored (nitrate, nitrite, TN, total soluble P, Ortho-P, TP, partic-ulate P, organic C, total Ca and Mg, and pH). 24 samples were collected duringthe study period of March to November, 1972 (weekly intervals) (BCWIB, 1974).Compared to background VC flows, the water after discharge had significantlylower TN, TKN, TP, and nitrate concentrations (BC Research, 1974) (although ni-trate leached from the exfiltration beds (MacDougall, 1984)), while calcium andmagnesium concentrations were higher (BC Research, 1974). The mean pH ofthis water was 8.2 (BC Research, 1974), similar to the mean pH (8.22) of WoodLake (1970 to 1988) (LAC, 2010). During the time of HWD operations, from theearly 1970’s to the early 1990’s, TP, TN, TDP, spring nitrate N, and planktonic lev-els (chlorophyll-a) decreased, while water clarity (secchi depth) increased in theselakes (Jensen & Bryan, 2001; R. Nordin, 1980; R. Nordin et al., 1985). Jensenand Bryan (2001); Walker et al. (1993) attribute the increase in water quality anddecline in eutrophic algal concentrations in Wood Lake from 1972 to 1995 to theincreased flushing rate from HWD’s discharge. Other studies indicate that nitratesin the HWD discharge may have stimulated spring P-uptake by phytoplankton,thereby preventing toxic summer/autumn blue-green algal blooms (Jasper & Gray(1982) as discussed in Jensen and Bryan (2001) and Jasper & Gray (1980) as dis-cussed by MacDougall (1984)). Jensen and Bryan (2001); R. Nordin et al. (1985);Walker et al. (1993) suggest that the reduction in residence time in Wood Lakehelped to improve its water quality at this time(Table 2.5). R. Nordin (1980) noted27an increase in water clarity (secchi depth) from 1975 to 1980 (1.21m to 5.88m).Furthermore, average fall chlorophyll-a declined from 18 to 2 µg/L from 1975 to1979, stabilizing at 4 µg/L by 1990 (Bryan (1990) as discussed by Walker et al.(1993)).Despite concern regarding the increased flushing of nutrients from Wood Lakeinto Kalamalka Lake with additional inflows (BC Research, 1974; BCWIB, 1974;R. Nordin et al., 1985), Kalamalka Lake data from 1975 to 2010 indicates that TPand TN levels in the south end of Kalamalka Lake decreased from 1980 to theearly 1990’s. There does not appear to be an increase in nutrients at the south endof Kalamalka Lake from 1970 - 1995 when HWD was operating (LAC, 2010). Al-gal consumption, followed by co-precipitation with calcite during marling eventsin the summer in Kalamalka Lake cause autumn N and P concentrations to declinefrom spring levels each year. Kalamalka Lake is typically P-limiting (R. Nordin etal., 1985), but is N-limiting in some years (Jensen & Bryan, 2001). The fact thatthe nutrient concentrations in the north and south ends of Kalamalka Lake rise andfall together indicates that whole-lake influences (i.e. freshet P-loading or ground-water N-loading) are more important than localized inputs. High-runoff duringfreshet tends to introduce more TP into the main-basin lakes due to particulate Pinputs. N is also not strongly bound to soils in the valley and thus nitrates tend totravel with groundwater (especially noted during high runoff years with increasedgroundwater flows) (Dill (1972) as discussed by LAC (2010)). Large peaks in TNand TP in the north and south ends of Kalamalka Lake in 1999 and 2000 occurredas a result of large freshet flows (LAC, 2010). About 80% of the inflow to Kala-malka lake is actually from Coldstream Creek at the north end of the lake, whichhas shown turbidity spikes around 500 NTU in 2008 (LAC, 2010) and is also themajor source of N and P for Kalamalka Lake (Anonymous, 1974; LAC, 2010).During 1972 and 1973 (when HWD was in operation), 56% of the N and 52% ofthe P loading into Kalamalka Lake was estimated to be from Coldstream Creek(BC Research, 1974; BCWIB, 1974).Although recent studies indicate that inflows to Ellison Lake short-circuit thelake around the north shore and that the lake serves as a nutrient sink, particularlyduring high inflows (BC Research, 1974; BCWIB, 1974; Epp & Neumann, 2014),there are some within the scientific community who suggest that Wood Lake and28Kalamalka Lake are better off without the increased inflow upstream of EllisonLake (i.e. LAC (2010); R. Nordin et al. (1985)). It is argued that although a re-duced flushing time is typically associated with decreased nutrient concentrations,nutrient-rich waters from Ellison Lake stimulated algal growth in Wood Lake andcaused a secondary effect in Kalamalka Lake (LAC, 2010). The BC MOE esti-mated that Wood Lake water was estimated to provide 20% of the N and 29% ofthe P into Kalamalka Lake during HWD operations and that nutrient concentrationsdeclined following the decommissioning of HWD (BC Research, 1974; BCWIB,1974; LAC, 2010). Estimates in the reduction of nutrient loading after removingthis discharge were: 15%N and 23%P to Wood lake, and 5%N and 6%P to Kala-malka Lake. Considering only flow through the Oyama Canal, it was estimatedthat this would reduce nutrient transfer from Wood Lake to Kalamalka Lake by31% for N and 32% for P (BC Research, 1974; BCWIB, 1974). It is uncertain howreadily Kalamalka Lake responds to nutrient inputs from Wood Lake. The con-cern regarding the increased inflow is that it may push nutrient-laden water fromEllison Lake into Wood Lake (encouraging algal blooms) and subsequently intoKalamalka Lake. Kalmalka Lake may respond to the increased epilimnetic nu-trients with increased algal production, or even the algal blooms themselves maytransfer into the south end of Kalamalka Lake (BC Research, 1974; BCWIB, 1974).Because of Kalamalka Lake’s long residence time (approximately 55yrs), theoreti-cally only 2% of the water volume in the lake is replaced in a given year; thereforeadverse nutrient, chemical, or biological impacts could potentially have lastingeffects (LAC, 2010). Although water quality investigations did not note any differ-ence in nutrient concentrations in Kalamalka Lake related to this influx of water,algal concentrations in surface waters at the south end of Kalamalka Lake wereelevated in 1972/1973 near the canal, and decreased with distance from the canal(BC Research, 1974). LAC (2010) and R. Nordin et al. (1985) suggest that Kala-malka Lake water quality is enhanced without this additional flushing of nutrientsinto the south end of the lake from Wood Lake.29Trends in Nutrient Concentrations in Wood Lake (1970 - 2015)Total Phosphorus [TP] and Total Nitrogen [TN] concentrations are often used asindicators of trophic status in lakes. Higher concentrations are associated withmore eutrophic waters (Table 2.4). TP concentrations in late 1800’s to early 1900’swere around 0.015 to 0.020 mg/L and increased to 0.025 to 0.040 mg/L betweenthe 1930’s and 1970’s (Epp & Neumann, 2014). TP in Wood Lake averaged 0.075mg/L in the 1980’s (in comparison to recommended 0.005 - 0.015 mg/L for aquaticlife) (LAC, 2010; R. Nordin et al., 1985). Trends in spring/autumn nutrient andphytoplankton concentrations (<10m depth) from 1969 to 1999 have been welldocumented (Table 2.8).Table 2.8: Trends in Nutrient Concentrations in Wood Lake from 1969 to1999 (Jensen & Bryan, 2001).Parameter Season Wood LakeTP Spring Decreasing trend from 1969 - 1993, increasing trend from 1993 - 1999.Autumn Typically less than spring. Less than Ellison Lake; similar to KalamalkaLake; decreasing trend from 1975 - 1999.TDP Spring Significant decrease from early 1980’s - 1990’s(dramatic drop from 1984 - 1989).TN Spring Dramatic decrease from 1970’s - 1990’s; increase from 1995 - 1999.Autumn Decrease 1975 - 1989; 0 net apparent change after(insufficient data prior to 1975).Nitrate N Spring Possible decrease from 1986 (0.2 mg/L) to 1999 (0.082 mg/L);considerable annual variation.TN:TP Ratio Spring Often < 14:1 (values < 5:1 reported).Phytoplankton (chl-a) Spring Moderate annual variation; possible upward trend from 1996 - 1999.Autumn No difference between sites. Decreasing 1975 - 1978; no trend after.Generally less than spring.Despite still receiving untreated stormwater, the water quality of Wood Lakehas continued to improve since 2000, and was actually considered mesotrophicby 2007 (LAC, 2010). This progress was aided in part by changing agriculturalpractices, removal of P from detergents, and closure of septic tanks. Since 2000,many of the septic tanks in proximity to Wood Lake were closed and householdswere connected to the municipal wastewater system (Epp & Neumann, 2014; LAC,2010). The recent relocation of Highway 97 (completed in August, 2013) awayfrom Wood Lake also permitted planting of native willow to serve as a natural30riparian filter along the lakeside (LAC, 2010). Spring and fall TN and TP concen-trations from 1970 to 2015 have been compiled from BCMOE data (Figure 2.4 and2.5). Winter and spring TP concentrations have averaged 0.045 mg/L over the last10 years (0.033 - 0.066 mg/L), with approximately 60% being dissolved and read-ily bioavailable as ortho-P (Epp & Neumann, 2014). Note this TP still exceeds therecommended 5-15 µg/L for aquatic life, but shows improvement from the 1980’s(LAC, 2010; R. Nordin et al., 1985). TN averaged 0.50 mg/L from 1969 to 1999(Jensen & Bryan, 2001). TN has not shown any recent observable trends, exceptperhaps some increased inter-annual variability between 2010 - 2015; values havecertainly decreased overall from those observed in the 1970’s and early 1980’s. TPconcentrations have also decreased from the high maximum values observed in the1970’s and early 1980’s. Spring TP (<10m depth and > 20m depth) and fall TP(<10m depth) have shown a decrease since the 1970’s until about 1990 after whichlevels appear to have stabilized. However, fall TP (>20m depth) have appeared toshow a gradual increasing trend since 1990. This corresponds to the thickening ofthe anoxic layer in the summer and fall and self-loading of P from the sedimentsthat has been noted in recent years (Epp & Neumann, 2014; LAC, 2010; MikeSokal, Paul Askey, Hillary Ward, & Heather Larratt, pers. comm., 2015-2016). InSeptember 2012, TN peaked at 0.93 mg/L in the hypolimnion, and subsequentlypeaked at 0.94 mg/L in in the epilimnion in March, 2013. TN decreased suddenlyin the epilimnion in spring (2013) and continued to decrease throughout summer,whereas it increased throughout the growing season in the hypolimnion (Epp &Neumann, 2014). TP peaked in the hypolimnion in September 2013 at 0.177 mg/L,which was the largest value recorded since 1982 (0.203 mg/L). The average TN in2013, 2014, and 2015 from March to October was 0.54 mg/L, 0.49 mg/L, and 0.49mg/L respectively. The average TP in 2013, 2014, and 2015 from March to Octo-ber was 0.054 mg/L, 0.052 mg/L, and 0.049 mg/L respectively. Note the differencebetween the hypolimnion (>20m depth) and epilimnion (<10m depth) for TP. Theaverage epilimnion:hypolimnion TP ratio in 2013, 2014, and 2015 was 0.020:0.086mg/L, 0.018:0.086 mg/L, and 0.019:0.080 mg/L respectively.311970 1975 1980 1985 1990 1995 2000 2005 2010 2015Year00.050.10.150.20.25Total Phosphorus [mg/L]Spring TP <10m [mg/L]Spring TP >20m [mg/L]Fall TP <10m [mg/L]Fall TP >20m [mg/L]Figure 2.4: Spring and fall TP from 1970 to 2015 in the hypolimnion andepilimnion of Wood Lake (BCMOE Data).1970 1975 1980 1985 1990 1995 2000 2005 2010 2015Year00.20.40.60.811.21.4Total Nitrogen [mg/L]Spring TN <10m [mg/L]Spring TN >20m [mg/L]Fall TN <10m [mg/L]Fall TN >20m [mg/L]Figure 2.5: Spring and fall TN in Wood Lake from 1970 to 2015 in the hy-polimnion and epilimnion of Wood Lake (BCMOE Data).32Algal blooms typically occur twice per year in Wood Lake in response tonutrient-rich freshet flows in the spring and during turnover in the fall (P fromthe hypolimnion replenishes nutrients near the surface). High spring freshet flows(i.e., 2006-2009 and 2011) from MVC contribute an excessive influx of nutrientsto Wood Lake, resulting in moderate algae blooms (Epp & Neumann, 2014). Phos-phate concentrations in excess of 0.020 mg/L (lower limit for eutrophic status forlakes (Table 2.4)) will promote plankton/benthic algal blooms (LAC, 2010). Min-imum spring TP >20m depth (Figure 2.4) was 0.019 mg/L in 1995, but has other-wise been greater than 0.020 mg/L since the mid 1970’s. Minimum fall TP >20mdepth was 0.073 (1994) and is always greater than 0.02 mg/L for all years on recordsince the mid 1970’s. Minimum spring TP in <10m depth is typically greater than0.02 mg/L except in 1993 (0.016 mg/L). Minimum fall TP <10m depth is typi-cally less than 0.02 mg/L since 1984, with an absolute minimum of 0.004 mg/Lin 1999. These fall measurements are recorded prior to turnover (generally in Au-gust to October) (BCMOE Data). Despite recent progress in reducing nutrientloading to Wood Lake, blue-green anacystis cyanea (cyanobacteria) bloomed insprings of 2006 - 2009, with a severe bloom in 2008 (high freshet year) in responseto warm water and high nutrient levels (LAC, 2010). Overall trends indicate areduction in excess algal production since the 1970’s and that nutrient concentra-tions have declined from their eutrophic levels in the 1980’s to levels classified asmesotrophic in 2007 to 2009. Evidence also suggests that the frequency of mar-ling events has increased since 2000 (LAC, 2010), which could help reduce nutri-ent and algal concentrations in Wood Lake (Brunskill, 1969; Dittrich & Koschel,2002; S. Effler, Perkins, Greer, & Johnson, 1987; Gilbert & Leask, 1981; Hamil-ton et al., 2009; LAC, 2010; Otsuki & Wetzel, 1972; Solim & Wanganeo, 2007;Wetzel, 2001; Williams, 1972).TP and TN in Wood Lake annually maintain a repeated seasonal cycle. Allforms of nutrients (except total organic N) decrease from March to October in theepilimnion (<10m depth), indicating consumption of nutrients by algae (Epp &Neumann, 2014). Most of Wood Lake’s N appears to be absorbed in algal pro-duction (LAC, 2010). Nitrate, nitrite (inorganic N) and all forms of P increase inthe hypolimnion (>20m depth) from March to October. Phytoplankton increasesin spring in response to increased light (indicated by chlorophyll-a), at which time33P is transferred to particulate form as algal cells. Concurrently, zooplankton typ-ically increases, consuming algae and decreasing algal biomass. Algal dies andsettles into the hypolimnion where it decomposes. The most important source ofP in Wood Lake remains self-loading, contributing significantly more to the nu-trient load in Wood Lake than MVC discharge. During the summer, P is trappedin the hypolimnion and increases during the growing season due to release fromsediments and from decaying organic matter, while epilimnetic P remains rela-tively stable. As hypolimnion TP levels increase during the growing season, theportion of TP that is ortho-P, one of the most bio-available forms of P, increases(i.e. up to 85-99% of TP from June to October compared to only 9-20% in theepilimnion in 2013) (Epp & Neumann, 2014) (Table 4.14 in Section 4.12.1). Pmay be associated with alumino-silicate minerals and incorporated into the matrixof silt, sand, and clay particles through ion substitution or bound to the surfaces ofthese particles at the lake bottom (Walker et al., 1993). Anoxic conditions abovethe sediments result in the release of P bound to sediments and exacerbates highconcentrations of P in the water column. When DO falls below 1-2 mg/L at thesediment water interface as a result of high sedimentation rates of biodegradablematter (i.e. when biological production is high and/or when when subject to or-ganic pollution) coupled with a relatively small hypolimnion volume, it results inthe reduction and mobilization of Manganese and Iron, and the subsequent releaseof phosphate, ammonia and silicate (Mortimer, 1971). P is often bound in sedi-ment as a FeOOH-phosphate complex, however when Fe(III) is reduced to Fe(II)and sulphate (SO4-2) is reduced concurrently, FeS may form (fewer sites for P toadsorb to than FeOOH); the ferrous phosphoric compound becomes soluble and Pis released (Berg et al., 2004; Gächter & Wehrli, 1998; Golterman, 2001; Mortimer,1942, 1971; Smolders et al., 2006). This can also occur if high concentrations ofSO4-2 exist at the sediment interface for an extended period, which can act as anelectron acceptor in sediments and enhance organic biodegradation and consump-tion of O2, compete with PO4-2 for anion adsorption sites, and can result in most ofthe iron being bound to reduced sulfur as FeSx (releases and mobilizes P formerlybound to Iron(III)) (Cole, Caraco, & Likens, 1989; Smolders et al., 2006). Alter-natively, the presence of H2S in anoxic conditions also encourages the formationof FeSx and release of P (Golterman, 2001). Recent studies show that this internal34loading of P to Wood Lake exceeds all other inputs and occurs for approximatelyfive months/year (Figure 2.6) (Epp & Neumann, 2014). There does not appear tobe evidence of notable internal nutrient loading during winter ice cover (BC Re-search, 1974). Maximum and mean nutrient mass [kg] values in the hypolimnionand epilimnion in Wood Lake (Table 4.14 in Section 4.12.1) were recorded in 2013by Epp and Neumann (2014). Investigation of TP in Wood Lake above 10m depthand below 20m depth throughout the field season (March to October) in 2013 -2015 (Figure 2.6) clearly shows the period of internal P loading (relative to surfacewater loading) into the hypolimnion from sediments that result from anoxic condi-tions in the hypolimnion throughout the summer and fall. On average in these threeyears, the hypolimnion TP concentration in October is 3.41 times greater (241%larger) than in March, whereas the epilimnion TP concentration in October is ac-tually 0.21 times (79% less) that in March. In terms of TP at spring overturn (i.e.March in Figure 2.6), Wood Lake is generally considered eutrophic (i.e. TP > 20µg/L)(Table 2.4). Since 2010, the average TP at spring overturn in the hypolimnionhas been 47.7 µg/L and in the epilimnion has been 45.8 µg/L. In terms of TN atspring overturn, Wood Lake is generally considered mesotrophic-eutrophic (TNbetween 100 - 1000 µg/L, Table 2.4). Since 2010, the average TN at spring over-turn in the hypolimnion has been 433.8 µg/L and in the epilimnion has been 517.7µg/L (BCMOE Data).In 2015, TDP measured from water samples taken on 20 July 2015 at onelocation near the centre of the lake, showed TDP was fairly constant up to about20m depth, but then increased about 4.5-fold by 30m depth at the lake bottom(Appendix C).35Mar Apr May Jun Jul Aug Sep OctMonth00.020.040.060.080.10.120.140.160.18Total Phosphorus [mg/L]2013: TP < 10m [mg/L]2013: TP > 20m [mg/L]2014: TP < 10m [mg/L]2014: TP > 20m [mg/L]2015: TP < 10m [mg/L]2015: TP > 20m [mg/L]Figure 2.6: TP in the epilimnion and hypolimnion of Wood Lake from Marchto October, 2013 - 2015 [BCMOE data]. Figure outlines TP in WoodLake above 10m depth and below 20m depth throughout the field season(March to October) in 2013 - 2015. This data was collected, processed,and analyzed by BCMOE and provided by Mike Sokal (pers. corr., 13February 2015 & 13 April 2016).Chlorophyll-aChlorophyll-a concentrations, secchi depth, or TP concentrations can be used asindependent estimates of algal biomass in a lake and are often referenced with re-gards to the trophic status of a lake (Wetzel, 2001). Chlorophyll-a measurementsare a favourable alternate for algal biomass, since the cost and time of measuringand analyzing chlorophyll-a is much less than that for algal biomass (via alternategravimetric methods). Thus, algal biomass is often characterized by chlorophyll-a concentrations (Hambrook Berkman & Canova, 2009; Lindenberg, Hoilman, &Wood, 2009). Chlorophyll-a has been measured rather inconsistently in Wood Lakefrom 1970 to 2015 at three different stations in Wood Lake: “Wood Lake Central36Station” (at 50.0875◦N, 119.3889◦W from 1975-1985 (Figure 2.7)), “Wood Lakeat Mouth of Middle Vernon Creek (West)” (at 50.057◦N, 119.4028◦W from 1975 -2007 (Figure 2.8)), and “Wood Lake ”Deep Station“” (at 50.0749◦N, 119.3917◦Wfrom 1983 - 2015 (Figure 2.9)). In some cases, single data point values for theepilimnion (i.e. at 0m depth, 4m depth, or 10m depth) were recorded, whereas, onother occasions, only pre-averaged values from the upper 10m of the water columnwere recorded. Amidst a few exceptions, the spring chlorophyll-a values in the up-per 10m of the water column always exceed the fall values in the same year. Springchlorophyll-a values show an overall increasing trend since 1983 at the “Deep Sta-tion” (500848), and are becoming increasingly greater than the fall values recordedat the same station. Chlorophyll-a in the Wood Lake “Deep Station” (500848) haveappeared to show no overall trend in fall chlorophyll-a values, other than perhapssome increased inter-annual variability from 2007 to 2011. Chlorophyll-a mea-sured at the mouth of MVC (until 2007) has shown considerable variability (par-ticularly in spring values), but has not shown an apparent upward or downwardtrend. Peak averaged spring chlorophyll-a values at the “Deep Station” occurredin 2001 (20.35 µg/L) and in 2010 (13.8 µg/L). BC MOE is continuing to recorddata at this latter station monthly. Peak averaged spring chlorophyll-a values at theMouth of Middle Vernon Creek Station occurred in 1990 (25.8 µg/L). Peak aver-aged fall chlorophyll-a values occurred in 1975 at the Central Station (18.5 µg/L)and at the Mouth of Middle Vernon Creek Station (26.85 µg/L). Data recording atthe Central Station ended in 1985 (BCMOE Data).371975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985Year24681012141618Chlorophyll-a [ug/L]Spring Chlorophyll-a [ug/L] Pre-Averaged Value <10m depthSpring Chlorophyll-a [ug/L] Single Data Point <10m depthSpring Average Chlorophyll-a [ug/L]Fall Chlorophyll-a [ug/L] Pre-Average Value <10m depthFall Chlorophyll-a [ug/L] Single Data Point <10m depthFall Average Chlorophyll-a [ug/L]Figure 2.7: Epilimnion chlorophyll-a in Wood Lake at the “Central Site Sta-tion” (1975 - 1985) (Data provided by BCMOE, Mike Sokal, pers.comm., 13 February 2015).1975 1980 1985 1990 1995 2000 2005Year5101520253035Chlorophyll-a [ug/L]Spring Chlorophyll-a [ug/L] Pre-Averaged Value <10m depthSpring Average Chlorophyll-a [ug/L]Fall Chlorophyll-a [ug/L] Pre-Average Value <10m depthFall Chlorophyll-a [ug/L] Single Data Point <10m depthFall Average Chlorophyll-a [ug/L]Figure 2.8: Epilimnion chlorophyll-a in Wood Lake at the “Mouth of MVCStation” (1975 - 2007) (Data provided by BCMOE, Mike Sokal, pers.corr., 13 February 2015).381985 1990 1995 2000 2005 2010Year051015202530Chlorophyll-a [ug/L]Spring Chlorophyll-a [ug/L] Pre-Averaged Value <10m depthSpring Chlorophyll-a [ug/L] Single Data Point <10m depthSpring Average Chlorophyll-a [ug/L]Fall Chlorophyll-a [ug/L] Pre-Average Value <10m depthFall Chlorophyll-a [ug/L] Single Data Point <10m depthFall Average Chlorophyll-a [ug/L]Figure 2.9: Epilimnion chlorophyll-a in Wood Lake at the “Deep Station”(1983 - 2014) (Data provided by BCMOE, Mike Sokal, pers. corr., 13February 2015).Monthly chlorophyll-a levels in the upper 10m of the water column in WoodLake have been fairly consistent in the past three years (Figure 2.10), with peakvalues occurring in March during the spring algal bloom, and then decreasingthroughout the year. The average chlorophyll-a values in the upper 10m of the wa-ter column from March to October in 2013 - 2015 were: 4.13 µg/L, 3.16 µg/L, and3.29 µg/L respectively. Therefore, in terms of chlorophyll-a, Wood Lake could beconsidered mesotrophic for the last three years (Table 2.4). Chlorophyll-a values inWood Lake are typically highest after spring overturn (i.e. in March, Figure 2.10).However, P that builds up in the hypolimnion during the summer and fall (Fig-ure 2.6) is mixed throughout the water column during fall turnover. Despite thecooler temperatures and shorter day lengths (less sunlight for photosynthesis), thistypically results in a blue-green algal bloom in November/December every year(LAC, 2010). At this time, we see a second spike in chlorophyll-a concentrations.Note that blue-green cyanobacteria have cyanochrome as their primary photosyn-thetic pigment (chlorophyll is their secondary pigment); diatoms on the other hand,contain chlorophyll as their primary photosynthetic pigment (LAC, 2010).39Mar Apr May Jun Jul Aug Sep OctMonth024681012141618Chlorophyll-a [ug/L]2013: chl-a [ug/L]2014: chl-a [ug/L]2015: chl-a [ug/L]Figure 2.10: Average chlorophyll-a in upper 10m of Water Column measuredat “Deep Station” (500848) in 2013 - 2015 (Data provided by BC-MOE)2.6.3 Middle Vernon Creek Hydrology and Spawning HabitatThe hydrology of VC from Swalwell Lake to Kalamalka Lake has been studied byBCMOE and Epp and Neumann as part of an ongoing project since 2012 (“TheMiddle Vernon Creek Action Plan” [MVCAP]) for the Okanagan Nation AllianceFisheries Department and the British Columbia Ministry of Forests, Lands, andNatural Resource Operations (BCMFLNRO) (Epp & Neumann, 2014, 2016). Un-derstanding VC hydrology is a key element of investigating the Wood Lake fisherysince low flows in MVC are believed to be one of the root causes of the declinein water quality (since 1930) and subsequent collapse in the kokanee fishery wit-nessed in recent years (Anonymous, 1974; Epp & Neumann, 2014; Walker et al.,1993; Webster, 2012, 2013, 2015). The primary purpose of the MVCAP is tosustain native kokanee populations in Wood Lake by ensuring a consistent watersupply within MVC while maintaining agricultural and municipal water supplydemands. Key elements of the water balance analyzed include: water licenses,groundwater seepage and infiltration, evaporation, runoff, and tributary inflows.40Swalwell Lake drains through a flow control structure managed by the DLC andthrough a spillway during high water levels. Swalwell Lake outflows to UVC (sup-plemented by Clark Creek) are compared to diversions from UVC by the DLC todetermine the new inflows into Ellison Lake. Ellison Lake discharges into MVC.Annual low flows in late summer and early fall have occurred in MVC since thecooling discharge from HWD ceased in 1995. For example, in 2013 Swalwell Lakereleases in August only exceeded District of Lake Country (DLC) demands by 0.07m3/s. This increased to 0.2 - 0.3 m3/s from September to November (Epp & Neu-mann, 2014). Low summer flows are exacerbated by the use of a temporary flowcontrol structure [TFCS] (sandbags) first installed in 2003 by the Oceola Fish andGame Club and the Okanagan Indian Band to store water in Ellison Lake and tocontrol the subsequent release into MVC during kokanee spawning in October (en-sure sufficient flows for egg deposition and alevin emergence). Although the TFCShelps provide required fall fish flows, it is not an ideal solution. Beaver dams onMVC also affect spawning flows. Discharge measurements downstream in MVCreveal that flows out of Ellison Lake into MVC are largely controlled by the waterlevel within the lake, by the TFCS, and by beaver dams (Epp & Neumann, 2014,2016; Webster, 2012, 2013, 2015). When Ellison Lake outflows are zero, MVCreceives minimal flow from a stormwater detention pond and Knopf Brook (smallgroundwater-fed creek) (Epp & Neumann, 2014). Previous studies conducted onVC have failed to formulate a formal water use plan and show a lack of sufficientknowledge about required minimum fish flows and how to best meet these throughoptimal water management practices (Epp & Neumann, 2014).Kokanee Salmon Spawning and Weighted Usable Habitat WidthsAvailable kokanee habitat in MVC is directly related to the creek’s flowrate andtemperature. Awareness of requisite flows for kokanee spawning is critical whenproposing solutions for Wood Lake that will affect (increase) discharge in MVC.Maximum visual kokanee counts in MVC tend to occur around early-to-mid Octo-ber. A “Physical Habitat Simulation technique” was applied to discharge data fromfour sites in MVC in 2012 and 2013 to determine the potential available habitat forkokanee spawning (i.e. “Weighted Usable Habitat Width” (WUHW)) as a function41of flow (Epp & Neumann, 2014). This process is detailed in Epp and Neumann(2014). It was determined that maximum WUHW is achieved at flows of 0.45 -0.60 m3/s, while 50% of maximum WUHW is available at 0.15 m3/s. Flow of0.050 m3/s was identified as the recommended minimum reference level, and 0.15m3/s is identified as the recommended kokanee spawning flow (20% of long termmean annual discharge of 0.775 m3/s). It is important to have stable flows duringmigration and spawning in order to ensure sufficient usable habitat for egg deposi-tion, eggs are not desiccated (by sudden flow decrease) and eggs are not scoured (bysudden flow increase) (Epp & Neumann, 2014). Beaver dams have been observedto hinder or prevent upstream migration of kokanee in MVC during a portion orall of the spawning period, and typically must be removed manually to enable pas-sage (i.e. 2009, 2010, 2011, 2012, 2014) (Webster, 2010, 2011, 2012, 2013, 2015).The WUHW in 2013 ranged between 0.5m on 15 September 2013 to 3.5m on 31October 2013 and the largest flux of fish counts occurred at the end of Septemberin response to increased flow (over 0.1 m3/s). In 2014, the largest spike in fishcounts occurred at low uniform flows of 0.02 m3/s at the end of September, and asmaller spike in mid-October at flows of 0.06 m3/s (increased flow due to removalof beaver dam upstream). However, in 2015, the WUHW remained below 0.5mthroughout the spawning season (15 September - 26 October 2015) (Epp & Neu-mann, 2016). Despite the inclination to formulate a relationship between increasedflow and fish counts (Epp & Neumann, 2014), the highest counts on record in theprevious four years surprisingly occurred in 2015 when flows in MVC remainedlower than the previous four years at 0.02 m3/s. Fish counts in MVC (Table 2.9)are conducted visually by walking the length of the stream every three days duringthe spawning run and counting the number of live and dead fish, and also by usinga counting fence (more reliable) in 2013-2015 (Epp & Neumann, 2016). It is ap-parent from 2011, 2013, and 2014 (Table 2.9) that the current factor (1.5) used toconvert peak visual count to total spawner count is too conservative and should beadjusted upward (i.e. should have been approximately 3.3 in 2011, 2.3 in 2013, and2.7 in 2014 in order to provide estimates on par with actual spawner abundance)(Webster, 2015).42Table 2.9: Enumeration of stream-spawning kokanee in MVC and the othertributary streams for Wood Lake (2009 - 2015).Year Fish Counting Fence (#) Visual Counts (#)2009 No Data 5250 (MVC)1525 (Winfield Creek)1 & 0 (UVC)15775 (all tributary streams)12010 No Data 13508 (MVC only)2801(MVC)2 & 2490 (UVC)216798 (all tributary streams)22011 8301(MVC only)3 82 (Winfield Creek)3 & 0 (UVC)38383 (all tributary streams)32012 No Data 1566 (MVC only)4136 (Winfield Creek)4 & 0 (UVC)41702 (all tributary streams)42013 2857 (MVC only)(22 Sept. - 1 Nov.)5 1256 (MVC only)62897 (MVC only)7 1884 (MVC only)73018 (all tributary streams)7 75 (Winfield Creek) 6 & 46 (UVC)72005 (All tributary streams)72014 8879 (MVC only) (22 Sept. - 1 Nov.)5 3230 (MVC only)89072 (all tributary streams)9 4845 (MVC only)9189 (Winfield Creek)9 & 4 (UVC)95038 (All tributary streams)92015 21330 (MVC only) (28 Sept. - 26 Oct.)5 9422 (MVC only)101(Webster, 2010), 2(Webster, 2011), 3(Webster, 2012), 4(Webster, 2013),5(Epp & Neumann, 2016)6(Askey (2014) as discussed in Epp and Neumann (2016)), 7(Webster, 2014)8(Ward (2015) as discussed in Epp and Neumann (2016)), 9(Webster, 2015)10(Ward (2016) as discussed in Epp and Neumann (2016))2.7 Review of Historically Proposed ManagementStrategies to Improve the Trophic Status in WoodLakeInternal P-loading is one of the most important considerations in selecting a strat-egy for controlling water quality in Wood Lake. Controlling external sources ofnutrient loading to Wood Lake may not be sufficient since the internal loading cy-43cle will not be interrupted and algal growth will continue to be stimulated by thisrelease of P on an annual basis (Aku & Tonn, 1999; Dent et al., 2014; Gächter& Wehrli, 1998; Gibbs & Özkundakci, 2011; Mortimer, 1942, 1971; R. Nordin,1987; Özkundakci, Hamilton, & Gibbs, 2011; Prepas & Burke, 1997; Smolders etal., 2006; E. B. Welch & Cooke, 1999). In the 1970’s - 1980’s several possible op-tions for water quality improvement in Wood Lake were suggested (BCWIB, 1974;R. Nordin, 1980, 1987; R. Nordin et al., 1985). Some of these options were pre-viously presented in the Kalamalka-Wood Lake Basin Study (1974) and evaluatedby the Okanagan Implementation Program (R. Nordin, 1980).2.7.1 Alum [Al2(SO4)3] Treatment (Or Alternative Sediment CapTreatment)Alum is a flocculating agent (aluminum hydroxide floc) that settles to the lake bot-tom relatively quickly, coagulating particulate matter along the way, and creatingan effective blanket over the sediment at the lake bottom (E. Welch & Schrieve,1994; E. B. Welch & Cooke, 1999). This can remove nutrients from the water col-umn by spreading it over the surface of the lake in spring by helicopter (when TDPis highest and before spring/summer plant growth). Lakes in Sweden and USAhave seen reductions in TP between 57-80%. The average TP in Wood Lake in1972/1973 was 87µg/L and 61µg/L respectively; thus alum treatment could reduceTP to 30-45µg/L assuming a 50% efficiency rate (BCWIB, 1974). Alum wouldalso lower the pH of the water (CO2 released from dissolved bicarbonates in water),which impedes the growth of blue-green algal (BCWIB, 1974). BCWIB (1974)suggested a dose of 1,100 tons over the lake surface at an estimated treatment costof $160,000. Pilot scale experiments in 1975 indicated dosage required was sig-nificantly greater than other treatments and lab experiments indicated (R. Nordin,1980). Treatment on six shallow lakes (mean depth 1.5m to 4.0m) in the 1980’sshowed an effective treatment rate between 50-80% with results lasting for at leastfive years (E. Welch & Schrieve, 1994).There are several other options also available to serve as sediments caps. Theapplication of a calcite (i..e limestone, industrially manufactured calcites, analyt-ical reagents, etc) can serve as a sediment cap to reduce the P-flux from the sed-iment to the water (Berg et al., 2004; Dittrich & Koschel, 2002). In addition to44providing a sediment cap, lime treatment (CaO, CaCO3, Ca(OH)2) would also sat-urate the lake with Ca+2, elevate pH, generate hydroxyl-apatite precipitates andco-precipitate phytoplankton, macrophytes and phosphorus (Berg et al., 2004; Dit-trich & Koschel, 2002; T. Murphy & Prepas, 1990; Prepas et al., 2001; Smolderset al., 2006). Although the addition of lime has shown promising results for bind-ing P, some research suggests that co-precipitation of P with calcite may providea weak mechanism for removing P from a lake because it is often resuspended atturnover and can redissolve in the hypolimnion during settling (Dittrich & Koschel,2002; Gilbert & Leask, 1981; Hamilton et al., 2009; R. Nordin, 1987; Wetzel, 2001;White & Wetzel, 1975; Williams, 1972). R. Nordin et al. (1985) suggested usingiron, alum, lime, manganese, or a pickling liquor (R. Nordin, 1987) to immobilizeP bound to sediment at the lake bottom. R. Nordin (1987) suggests that although Pcan be bound by several elements (i.e. calcium, aluminum, manganese, and iron),iron provides the strongest binding relationship. However, without the additionof iron (i.e. pickling liquor) the natural concentration of available iron in WoodLake sediments (in terms of ratio of iron to P) is too low to provide adequate sitesfor binding P (R. Nordin, 1987). Alternative options for application of a sedimentcap include using a modified zeolite mineral [Z2G1] (Gibbs & Özkundakci, 2011;Özkundakci, Hamilton, & Gibbs, 2011), modified clay minerals (T. Robb & Abra-hams, 2003), or iron slag (Yamada et al., 1987) (Fe+3 can also bind and immobilizePO4-3 (Smolders et al., 2006)).Mix the Contents of Kalamalka Lake and Wood Lake or Okanagan Lakeand Wood LakeMixing Kalamalka and Wood Lake water (i.e. to replace entire Wood Lake vol-ume) was considered an attractive option, but was expensive ($340,000 in 1970’s)and too likely to receive public disapproval due to unknown consequences for wa-ter quality in Kalamalka Lake and Wood Lake (R. Nordin, 1980, 1987). MixingOkanagan and Wood Lake water through a tunnel was another proposed option,but considered to be too expensive ($4.75M in 1974) because of the elevation dif-ference between the lakes and the pumping system required (HWD infrastructurenot available at the time) (R. Nordin, 1980, 1987). Revised variations of thesepropositions are discussed in more detail in Section 2.8 and 4.12.1.45Biological ManipulationOne suggested technique was to add nutrients (nitrate) to the hypolimnion of WoodLake to stimulate denitrification and oxidize the sediment at the bottom of thelake Gray&Jasper (1980) as discussed by R. Nordin (1987)(R. Nordin, 1980).This treatment can prevent the reduction of Fe(III) to Fe(II) (as well as reduc-tion of SO4-2) and thus inhibit the subsequent release of P bound to the solid ferrichydroxide and iron-phosphate complexes (Gächter & Wehrli, 1998; Özkundakci,Hamilton, & Gibbs, 2011; Smolders et al., 2006; Wauer et al., 2005). It was hy-pothesized that this treatment could stimulate spring diatom algal blooms, increas-ing spring uptake of P and thus limit its availability for summer blue-green algalblooms (MacDougall, 1984; R. Nordin et al., 1985). Diatoms do not reduce waterclarity as dramatically as blue-green algal (MacDougall, 1984) and are typicallyassociated with oligo-mesotrophic conditions (Anonymous, 1974). MacDougall(1984) indicated that nitrate was leaching from the exfiltration beds of HWD intothe groundwater system. This was suggested to possibly have a positive impacton the water quality and nutrient balance of Wood Lake during this period Mac-Dougall (1984), as indicated by a decline in blue-green (cyanophyte) algal bloomsthat were more regular before 1970 (Jasper &Gray (1980) as discussed by Mac-Dougall (1984)). However, this nitrate addition method has not been implementedin many cases and it was expected that the public would disapprove of the ideaof adding more nutrients to the lake (R. Nordin, 1980). R. Nordin et al. (1985)estimated the cost of this treatment at $25,000 each year, but also suggested that itwould not be able to significantly improve fish habitat or mitigate P-loading in thelong term.Hypolimnetic Aeration or OxygenationOne of the more widely used methods to reduce anoxia is hypolimnetic aerationor oxygenation (Aku & Tonn, 1999; Ashley, 1983; G. D. Cooke et al., 2005; Dentet al., 2014; B. Moore et al., 2014; Prepas & Burke, 1997; Smolders et al., 2006;Wauer et al., 2005; Xue et al., 1997). The desired objectives for aeration are tomaintain desired levels of hypolimnetic DO in order to provide oxygen for fishand mitigate the release of nutrients [P] from sediment that occurs under anaerobic46conditions (Ashley, 1983). This can be achieved by using a destratifcation sys-tem that adds air to the bottom and allows the lake to mix, or by aerating only thehypolimnion without influencing the epilimnion (BCWIB, 1974). Several differentsystems are available including passive flow pure oxygen injection systems (Prepas& Burke, 1997), bubbler systems with compressed air (Xue et al., 1997), “down-flow bubble contact” hypolimnetic aerators (Kosari, Mavinic, Fattah, & Ashley,2014), or line diffuser hypolimnetic oxygenation (Dent et al., 2014; B. Moore etal., 2014).In the 1970’s, BCWIB (1974) suggested that 10 aerators specially designed toaerate only the hypolimnion could be used at a cost of $10,000 each and an an-nual operating cost of $20,000. This was considered to be a long-term method interms of reducing nutrients in Wood Lake and it was suggested that it should beused in concert with an alternative method (i.e. alum treatment). Aeration wouldlikely be required for many years (BCWIB, 1974). The estimated cost associatedwith hypolimnetic aeration based on 15 lakes is $3000/ha/year, which is equal tothe operating cost in addition to the average installed cost over a ten year lifetime,when operating at160 days/yr. Other examples include projects in Lake Stevens(10 yr lifetime operating at 160 days/yr) at a cost of $1240/ha/yr and Lake Tegel(12 yr lifetime operating at 160 days/yr at a cost of $1052/ha/yr). Based on theseestimates, aeration for Wood Lake (930 ha) would cost $978,360/yr - $2,790,000(2002-USD) (G. D. Cooke et al., 2005). In 1980, the optimal strategy for WoodLake was considered aeration (a destratification-type system or a hypolimnetic aer-ator) (R. Nordin, 1980). In 1980, the Okanagan Implementation Program obtaineda quote from Atlas-Copco for $1,151,000 to aerate Wood Lake, which was con-sidered too expensive (R. Nordin, 1987). Additionally, it has been found in somecases that even aeration and reduction of external sources of P in conjunction arenot always successful at eliminating the anoxic conditions at the sediment-waterinterface as a result of high sedimentation of dissolved organic matter (G. Cooke,Welch, Peterson, & Newroth, 1986; G. D. Cooke et al., 2005; Gächter & Wehrli,1998). Sometimes parameters may rebound to pre-oxygenation conditions uponterminating aeration treatment (G. Cooke et al., 1986; G. D. Cooke et al., 2005;Dent et al., 2014), and/or the expected benefits for certain species of fish are notalways produced (Aku & Tonn, 1999; Skinner, Moore, & Swanson, 2014). Wood47Lake would require a massive system to provide the required hypolimnetic oxygendemand (6000-7000 kg/day) during summer stratification (R. Nordin, 1980). Dis-cussions with BCMOE and LAC suggest that there is not sufficient funding at thistime to implement this solution.Watershed Management Techniques (Setbacks, Green-belting Creeks, LandUse Zoning, Removal of Septic Tanks, and Implementation of SewageSystems)Watershed management techniques such as setbacks, green-belting creeks (bufferzone along tributary streams to minimize pollution into surface waters), and landuse zoning can help control nutrient loading to streams and lakes (Boesch et al.,2001; Brambilla et al., 2007; Horne & Goldman, 1994; Prepas & Burke, 1997).BCWIB (1974) discusses creating a buffer zone (i.e. 30m on either side) alongthe tributary streams to Wood Lake by way of land-zoning bylaws and avoidingdevelopments that have nutrient-loading potential within the buffer zone. This wasproposed to improve the stream fishery and prevent further erosion along the creekbank (Anonymous, 1974; BCWIB, 1974). Preliminary estimates suggested a pos-sible reduction in N loading of 35-40% and P-loading of 63-73% could be achievedwith green-belting and setbacks. Assuming a greenbelt of 30m on either side, theestimated cost of purchasing this land and fencing the boundary through Winfieldin 1974 was estimated at $100,000 in Winfield (or $330,000 if implemented from“Indian Reserve No.7” north of Ellison Lake to Wood Lake)(BCWIB, 1974).Anonymous (1974); BCWIB (1974) also discuss restricting the use of land(i.e. septic tanks, feed lots, etc) and/or removal of septic tanks (R. Nordin, 1987) insome areas. It was estimated that cultural sources accounted for 57% of N-loadingand 35% of P-loading (irrigation and septic tanks were the largest contributors).These restrictions would limit the development of residences with septic tanks toareas where the soil type, distance to the lake/creek, and groundwater levels wouldensure optimal nutrient removal efficacy (soil type and depth of groundwater tablemajorly affect septic tank nutrient removal efficiency). The number of feed lots inthe Wood Lake drainage basin is not significant (10% of N loading and 7% of Ploading); however, the largest factor in determining the effects of feed lots on nutri-ent loading via surface/sub-surface runoff is location. A minimum recommended48distance of feedlots from creeks/lakes was established at 150m (BCWIB, 1974).In 1974 the field data obtained from studies were inconclusive in terms of eval-uating the improvement of nutrient removal efficiency using sewage systems overseptic tanks, and there was concern that leaching from abandoned septic tankswould continually add nutrients to the ground for several years. The nutrient re-moval efficiency of septic tanks in Winfield was estimated at between 85-95% forN and P (BCWIB, 1974). However, septic tanks were estimated to still contributearound 44% of N and 62% of P loading from cultural sources from the Winfieldarea (Anonymous, 1974; BC Research, 1974; BCWIB, 1974). Implementing asewage system was estimated to be able to reduce cultural loading of nutrients toWood Lake by 2700lb for N and 550lb for P (BCWIB, 1974). The Lake Coun-try sewer was completed in 2000, and the majority of households were connectedto the municipal system. Closure of septic tanks around Wood Lake significantlyreduced nutrient loading into Wood Lake, although Wood Lake still receives un-treated stormwater effluent (Epp & Neumann, 2014; LAC, 2010).Other OptionsOne option that was subject to discussion in the 1980’s was using a tertiary treat-ment facility to treat Wood Lake water; however, this was considered impracticalbecause technology did not exist for handling low (75 µg/L) nutrient concentra-tions (R. Nordin, 1980). Using hypolimnetic water for irrigation was also consid-ered, but this option was not discussed in great detail in terms of its ability to helpimprove water quality in Wood Lake (R. Nordin, 1980). Several farms in closeproximity to Wood Lake do currently use Wood Lake water for irrigation.R. Nordin et al. (1985) also discuss two different flow control methods to limittransfer of nutrients between basins. The first of these involved placing a flowcontrol structure (i.e. stop-log) in the Oyama Canal during winter when peakP concentrations in Wood Lake occur and there is minimal boat traffic betweenlakes; however, this was considered unlikely to be favoured by local residents.The second involved diverting inflow away from Ellison Lake via a dyke whenP concentrations in UVC inflow were elevated and permitting UVC inflow to en-ter Ellison Lake when P concentrations subside. The cost of construction for this49latter option was estimated at $150,000 (in 1980’s) (R. Nordin et al., 1985). Fur-thermore, R. Nordin et al. (1985) who was not a proponent of increased inflowsinto VC from HWD suggested that diverting HWD discharge away from UVC intoKelowna Creek (flows into Okanagan Lake) would reduce P-loading to KalamalkaLake by approximately 0.69 tonnes/year. The cost of this diversion was estimatedat $1.3M in the 1980’s (R. Nordin et al., 1985).Other options for lake management and remediation that have been subjectto extensive review around the world include bio-manipulation (i.e. reduce algalmass by deliberately increasing zooplankton that feed on phytoplankton via reduc-ing zooplanktivorous and benthivorous fish stocks) (Smolders et al., 2006), andsediment removal via dredging (remove PO4-3-enriched or Fe-depleted sediments)(G. Cooke et al., 1986; G. D. Cooke et al., 2005; Jorgensen et al., 2005; Smolderset al., 2006). The cost of dredging is prohibitive and not practical for Wood Lakedue to its size; the cost for dredging has been estimated at between USD$0.23- $15.00/m2 in 1981 (not including transport and disposal) (Peterson (1981), asdiscussed by Jorgensen et al. (2005)).Additional ConsiderationsIn the development of a proposed method to remediate the conditions in WoodLake, it is critical to understand the relationship between Kalamalka Lake andWood Lake and any consequences that any implemented human interjection mayhave on both basins, as well as the trends in water quality over time and their rela-tionships with anthropogenic influence, climate change, and with kokanee salmonpopulations. Since Kalamalka Lake is much larger, it is only practical to considernutrient source control to limit nutrient loading into the lake. Available options in-clude converting residential areas to sewers, improving agricultural practices, greenbelting, and erosion control (R. Nordin, 1980). Since HWD closed, 20% of the flowinput to Kalamalka Lake is from Wood Lake and 80% is from Coldstream Creekand groundwater (LAC, 2010). Nutrient inputs can have long-term effects in Kala-malka Lake since its retention time is 55-65 years (LAC, 2010). There is remainingconcern regarding additional flushing of nutrients and algal cells from Wood Lakeinto Kalamalka Lake with management scenarios that increasing the flushing rate50of Wood Lake (BC Research, 1974; BCWIB, 1974; LAC, 2010; R. Nordin, 1987).However, returning Wood Lake inflows to their pre-industrialization rates (prior tosettlement) and thus improving the conditions in Wood Lake will also have a posi-tive impact on controlling the source of some nutrients into Kalamalka Lake in thelong term.2.8 Review of Current Remediation Strategies for WoodLake Under ConsiderationConversations with various stakeholders who have been involved in studying theOkanagan’s main valley lakes in the past have resulted in the development of sixpossible solutions for Wood Lake. The primary goals of all of these proposedsolutions are to influence either/both the thermal structure and/or the available hy-polimnetic DO in the late summer and early fall. The six possible solutions are:(1) bypassing Ellison Lake with a portion/all of UVC flows; (2) triggering annualmarling events in Wood Lake; (3) increasing outflows from Wood Lake; (4) recir-culating water within Wood Lake; (5) pumping water from Okanagan Lake intoWood Lake while controlling outflows; and (6) pumping water from KalamalkaLake into Wood Lake while controlling outflows. The first two options are de-scribed herein (Section 2.8.1), however, they are not investigated in further detailin the current study. The knowledge base required to understand the history andfeasibility of the latter four options is discussed in Section 2.8.2; these four options,(3) to (6), are evaluated in detail via DYRESM (Section 4.12.1).2.8.1 Proposed Management Options Not Subject to DetailedInvestigation in Current StudyTwo proposed options that are not evaluated with the model, but that warrantbrief discussion are: (1) bypassing Ellison Lake with a portion of UVC flows;and (2) utilizing a natural chemical process (calcium carbonate (CaCO3) precipita-tion, known as “marling” or “whitening”) to remove nutrients, algae, vitamins, andminerals from the water column in Wood Lake, which could prevent autumn algalblooms and reduce biological oxygen demand [BOD] in the hypolimnion (Berget al., 2004; Dittrich & Koschel, 2002; Gilbert & Leask, 1981; Hamilton et al.,512009; Koschel, 1990; Koschel, Benndorf, Proft, & Recknagel, 1983; LAC, 2010;T. P. Murphy, Hall, & Yesaki, 1983; Otsuki & Wetzel, 1972; Solim & Wanganeo,2007; Wetzel, 2001; White & Wetzel, 1985; Williams, 1972).Proposed Solution 1: Bypassing Ellison Lake With A Portion/All of UVCFlow Directly to MVCThe proposed bypass of VC around Ellison Lake was motivated by the idea thatEllison Lake serves as a source of nutrients to MVC and that bypassing this lakecould result in higher flows into Wood Lake without raising the water level in Elli-son Lake (BCWIB, 1974; Epp & Neumann, 2014). However, there is controversyover whether Ellison Lake serves as a source or sink for nutrients with respect toWood Lake. LAC (2010) suggests that Wood Lake is susceptible to decreased res-idency times of Ellison Lake. However, Epp and Neumann (2014) suggest thatEllison Lake is a nutrient sink. Also BC Research (1974) indicated that EllisonLake’s flushing (water) time and nutrient retention is influenced by the inflow rateand the degree to which flows short-circuit the lake. General consensus is thatEllison Lake predominantly serves as a nutrient sink.This proposition is not advised. Water demands on the VC system already ex-ceed available flow, causing MVC to become dry in sections below Ellison Lakein July/August-October. Flow in MVC is largely controlled by Ellison Lake waterlevel, which decreases significantly each year (i.e. in 2015 the water level de-creased by 0.35m despite receiving some inflow from UVC and having relativelyno outflow for several months) (Epp & Neumann, 2016). The inflow to EllisonLake from UVC is significantly reduced relative to the release from Swalwell Lakefrom June to September due to diversion for DLC, so much so that even the wa-ter level of Ellison Lake fails to be maintained despite near-zero outflow to MVC.In 2015, the average monthly UVC flows into Ellison Lake from June to Septem-ber were only: 0.21, 0.053, 0.051, and 0.217 m3/s respectively. The WinfieldOkanagan Center Water System utilizes VC as the primary source (gravity fed)and Okanagan Lake as the secondary source. Results from a detailed hydrologystudy by the BC MOE in 1977 stated that a supply of 9868e3m3 is available in49 out of 50 years on average. (These estimates are based on 1960 - 1970 data).If fish flows and operational waste are considered, then based on 1977 hydrology52estimates, only 7,698e3m3 is available on an annual basis. VC is evidently nearingcapacity. Water licenses on VC, totalling 9,436e3m3, exceed the water available ina dry year if fish flows are considered. In 2003 (high water use year) water con-sumption from VC was 5808e3m3. The projected annual water requirement (2024)for VC is 6,846e3m3. Considering fish flows and operational waste, any surplusesin wet years should be retained until a new hydrology study is performed. Losses togroundwater from UVC and Ellison Lake are also significant, complicating calcu-lation of the water volume required to ensure minimum fish flows. In 2003, the BCMinistry of Water, Land, and Air Pollution stated that a minimum flow of 0.075m3/s in UVC was adequate to support fish in UVC creek, but indicated that thismay result in low flows in MVC. DLC has committed to provide minimum flowsin UVC (but not in MVC) when sufficient water is available, and has also commit-ted to providing water for agricultural use from VC (Mould Engineering, 2004).This comparison of available water with current and future projections of annualwater consumption and authorized diversions outlines that insufficient flows areavailable in VC to supply current and future demands, while also providing suf-ficient fish flows during dry years. Therefore, implementation of a bypass fromUVC to MVC, is not a feasible option for providing sufficient fish flows in MVCand for helping to improve the water quality of Wood Lake by reducing its flushingtime.Proposed Solution 2: Generating Marling Events in Wood Lake and FixingInternal Nutrient LoadingThere is current interest in investigating if it is possible to trigger marling events(calcium carbonate (CaCO3) precipitation and deposition) in Wood Lake by mix-ing water from Kalamalka and Wood Lake as had been suggested by R. Nordin(1980) and BCWIB (1974). Kalamalka Lake is the only lake in the Okanagan thatis characterized by an annual marling cycle during summer stratification, render-ing it the least productive of the mainstream lakes (Anonymous, 1974; Williams,1972). Marling helps to improve water quality and reduce productivity. Theseevents are related to the naturally elevated concentrations of calcium and/or sul-phates in Kalamalka Lake and normally occur between mid-July and early August.The chemical reaction of calcium/magnesium with carbonate/sulphate in surface53waters results in CaCO3, magnesium carbonate, and/or calcium sulphate (gypsum)precipitates that sink to the lake bottom (LAC, 2010). This cycle increases thesedimentation rate of a marling lake (Dittrich & Koschel, 2002; Koschel, 1990;LAC, 2010). Suspended carbonate is initially colloidal. Newly formed precipitatesadsorb and complex dissolved organic compounds (reducing bioavailability of dis-solved organics), as well as metallic nutrients (particularly iron), phosphate andother macronutrients. These compounds are consequently unavailable for bacteriaand plant growth (Berg et al., 2004; G. D. Cooke et al., 2005; Dittrich & Koschel,2002; Gilbert & Leask, 1981; Hamilton et al., 2009; Koschel, 1990; Koschel etal., 1983; T. P. Murphy et al., 1983; Otsuki & Wetzel, 1972; Solim & Wanganeo,2007; Wetzel, 2001; White & Wetzel, 1985; Williams, 1972). During calcite pre-cipitation, there is a decrease in free CO2 and an increase in turbidity, which bothserve to limit primary productivity (Koschel, 1990). P-concentrations in KalamalkaLake are significantly affected by marling events. Marl precipitation influences al-gae growth via co-precipitation of P and B-vitamins (made by bacteria), removingnutrients and small algae cells from the water and causing algae productivity to de-cline (LAC, 2010; Wetzel, 2001; White & Wetzel, 1985). The reaction can occuron the surface of the algae and draw it down to the lake bottom (Epp & Neumann,2014; Wetzel, 2001). Marling events appear to follow algal-induced decrease inCO2 and increase in pH and are concurrent with a peak in surface water turbid-ity, alkalinity, conductivity, and calcium concentrations (Dittrich & Koschel, 2002;Hamilton et al., 2009; Otsuki & Wetzel, 1972). Precipitation typically occurs inthe epilimnion (warmer waters with localized increase in pH and decrease in CO2due to photosynthetic algal) and crystals may resolubilize as they settle through thehypolimnion (Dittrich & Koschel, 2002; Gilbert & Leask, 1981; Hamilton et al.,2009; R. Nordin, 1987; Wetzel, 2001; White & Wetzel, 1975; Williams, 1972). Theinternal loading (release of co-precipitated) P back into the water column from cal-careous sediments, even during anoxia, is significantly reduced due to calcite act-ing as a sediment cap, similarly to alum treatment (Robertson et al., 2007). Theseevents occur very quickly (timescale of hours) on an annual basis in KalamalkaLake and typically last for several weeks, as the precipitates remain suspended(LAC, 2010).In the past 30 years, at least four of these events have been observed in Wood54Lake (1987, 1994, 2010, and 2015), preventing late summer/fall algal blooms inthese years (LAC, 2010), and maintaining its excellent water quality and olig-otrophic state (Anonymous, 1974; Northcote et al., 1974; Williams, 1972). Bio-logical activity and the eutrophic conditions in Wood Lake beginning around 1930raised the pH of the water, encouraging the formation of carbonate (CO3-2) from bi-carbonate (HCO3-1) and subsequently the formation of calcite (Walker et al., 1993;Williams, 1972). 1940 marked the beginning of a significant increase in CaCO3deposition in Wood Lake, suggesting an increase in marling events (as indicatedby core samples studied by Walker et al. (1993)). There is a notable correlationbetween the peak in CaCO3 concentrations and trough in organic matter between1940 and 1970 in these core samples. Jasper & Gray (1982) (as discussed byR. Nordin (1987)) measured a decrease of 800 tonnes of calcium in the epilimnionand metalimnion from May to September, and a decrease of 3-7 mg/L in surfacewaters between April and October.During dry summers when evaporation and outflows exceed inflows, WoodLake surface elevation falls below Kalamalka Lake and a reversal in the flow pat-tern occurs (LAC, 2010). Predominant north winds during summer also favour thissoutherly movement of water through the canal from Kalamalka Lake into WoodLake (BC Research, 1974). This is a key factor in increasing the water hardness ofWood Lake and in initiating marling events. In 2010, in addition to being a warmdry year, there were also 16 large seiche events in Kalamalka Lake that occurred,drawing additional water from Kalamalka Lake into Wood Lake and producing asignificant change in the water chemistry of Wood Lake. A marling event in Au-gust 2010 in Wood Lake caused cyanobacteria numbers to decline significantly,preventing a fall bloom (cyanobacteria blooms typically occur during fall turnoverevery year). 2008 and 2009 had measured peak chlorophyll-a of 15.3 µg/L. In2010, surface chlorophyll-a values were very low, peaking at 3.2 µg/L in May and2.0 µg/L in October, averaging only 1.7 µg/L from May to October. These valuesare comparable with those for Kalamalka Lake (1.35 µg/L average in 2010) (LAC,2010).Given the significant marling event that occurred during the 2015 field sea-son (July-September) and the chemical and physical data collected, it has beenpossible to formulate a preliminary basis for understanding the conditions that pro-55mote marling in Wood Lake. This analysis is presented in Appendix C. An entirestudy regarding the potential for triggering these events in Wood Lake is beyondthe scope of the current study. If this proposition is possible, subsequent stepsinvolve formulating a procedural plan for assessing Wood Lake’s capacity for mar-ling annually and estimating the volume of Kalamalka Lake water (based on waterchemistry analysis) that may be required to initiate marling events in Wood Lake.Microcosm or mesocosm studies are recommended to perform this task. To thisend, it is important to understand the chemistry of Kalamalka Lake waters. LACis continuing ongoing research (began in 1998) for Greater Vernon Water (GVW)and DLC, collecting monthly data (May to October) on pH, conductivity, salinity,temperature, DO, water chemistry, and taxonomy in Kalamalka Lake at severalstations (LAC, pers. comm., 2015). This data will be integral to further futureinvestigations in this regard.A second option to attempt to achieve similar results would be to mimic pre-liminary studies performed in the 1980’s that involved fixing the nutrient loadingin Wood Lake via pickling liquor (high iron-concentration that is a key componentin P-binding) (R. Nordin, 1987) and/or spreading limestone (CaCO3) (Berg et al.,2004; Dittrich & Koschel, 2002; T. Murphy & Prepas, 1990; Prepas et al., 2001)from a local quarry on the winter ice-cover to generate a marling event (H. Larratt& M. Sokal, pers. corr., 26 September & 3 October 2014). This topic has not beenaddressed in the current study.2.8.2 Proposed Management Options Subject to DetailedInvestigation in Current StudyThe focus in this study is to evaluate the effect of the latter three scenarios (Sec-tion 4.12.1) on the thermal structure of the lake and its implications for availablekokanee habitat. The four potential solutions investigated with DYRESM involveattempting to modify the hydrodynamics of Wood Lake. For each of the three pro-posed potential solution scenarios investigated with DYRESM, there are a numberof potential regimes that are addressed. Two of these options involve altering theinflows and/or outflows of Wood Lake; additional inflow volume is sourced fromOkanagan Lake or Kalamalka Lake respectively. Depending on the inflow/outflowcontrols, some of these scenarios would reduce the residence time of the lake by56increasing both inflow and outflow, whereas other scenarios involve maintaininga similar residence time and alter the storage volume of the lake. Decreasing theresidence time would improve the trophic status of the lake by increasing the rateat which nutrients are flushed from the basin, similarly to during the period whenHWD was in operation.Preliminary tests on the effects of addition of boron, molybdenum, iron andEDTA on algal growth show that the addition of molybdenum, and a combinationof iron and EDTA have the most pronounced effect on algae growth (Anonymous,1974). Although beyond the scope of this study, the concentrations of these ele-ments and EDTA (organic chelating agent) should be considered when identifyingpotential sources of water to supplement inflows to Wood Lake. Understandinghow Kalamalka Lake responds to nutrient loading from Coldstream Creek and/orother sources (i.e. Wood Lake) is also critical since reducing the residence time ofWood Lake would initially likely flush more nutrients, phytoplankton, and otherpotential threats to water quality into Kalamalka Lake, until the water quality ofWood Lake improves significantly to reflect that of Kalamalka Lake. KalamalkaLake’s capacity for removing these elements from the water column and reducingthe bioavailability of N and P via its natural annual marling cycle must be furtherstudied. This topic is beyond the scope of the current study.Proposed Solution 3 (Modelled with DYRESM): Increasing Outflow fromWood Lake, Whilst Maintaining Current Inflows (i.e. Drawing Down WoodLake Level)Interest in the proposed solution of increasing outflow from Wood Lake to sloughoff the heat content in the surface waters of the lake stemmed from similar prac-tices on Osoyoos Lake (Paul Askey, pers. comm., March 10, 2016). One criticaldifference between these two basins is that Osoyoos Lake receives constant inflowfrom Okanagan Lake via Skaha Lake and Vaseux Lake by way of the OkanaganRiver. Wood Lake on the other hand, receives negligible inflow for approximatelyfour months of the year. Therefore, while this process will serve to slough off theheat content in the surface waters, it will also proportionately draw the lake leveldown and decrease Wood Lake’s storage volume, effectively making Wood Lakeshallower and potentially increasing nutrient concentrations. Nonetheless, several57outflow rates are investigated with DYRESM in order to assess the effect of thisscenario on the thermal structure of Wood Lake in the late summer.Proposed Solution 4 (Modelled with DYRESM): Pumping CoolerHypolimetic Water in Wood Lake Back to the Surface (Recirculation)The primary goal of recirculating water from Wood Lake’s hypolimnion to thesurface is to cool the lake’s surface layer in the late summer when the temperature-oxygen squeeze becomes critical for kokanee. This is not a favourable solutionunder current conditions given the significantly higher nutrient (TP) concentrationsin Wood Lake’s hypolimnion that result from internal loading (Anonymous, 1974;BCWIB, 1974; Epp & Neumann, 2014; R. Nordin, 1987; R. Nordin et al., 1985;(BCMOE Data)). Raising these nutrients to the surface from the hypolimnion, de-spite the cooling potential of this water, would likely trigger algal blooms in WoodLake by making these nutrients readily bioavailable for algal growth in the epil-imnion (Mike Sokal, Hillary Ward, Paul Askey, Heather Larratt, pers. corr., 18April 2016). Nonetheless, several different recirculation flow rates are evaluatedwith DYRESM in order to evaluate its cooling potential. If Wood Lake water qual-ity improved significantly enough over time, then this may become a favourableoption for lake management at a later time in the future due to reduced operatingcosts (compared to sourcing cooler water from a different lake).Proposed Solution 5 (Modelled with DYRESM): Pumping HypolimneticWater From Okanagan Lake back into Wood LakeThe primary goals with regards to pumping water from Okanagan Lake are alteringthe thermal structure of Wood Lake by cooling the epilimnion and/or metalimnionand diluting the nutrient-rich water of Wood Lake with higher quality water fromOkanagan Lake. It has been hypothesized that supplementing VC flows with waterfrom Okanagan Lake could improve the trophic status of Wood Lake and mitigatefish kills by reducing the residence time of the lake and thereby increasing the rateat which nutrients flush from the lake (reducing biological growth and reducingBOD in the hypolimnion) (Anonymous, 1974; BCWIB, 1974; R. Nordin, 1980).Alternatively, increasing inflows while controlling the outflow, could increase thestorage volume of Wood Lake during the late summer and dilute the nutrient-rich58water of Wood Lake with cooler (hypolimnetic) low-nutrient Okanagan Lake water.BCWIB (1974) proposed mixing Wood Lake water with water in the hypolimnionof Okanagan Lake through a 2500 m long tunnel with a 1200 mm diameter pipe,assuming that due to the vast difference in size between the two basins (OkanaganLake is 131 times larger than Wood Lake by volume), that the effects on Okana-gan Lake water quality would be negligible. This proposition involved drawing3.67e5 m3/day from Wood Lake’s hypolimnion into Okanagan Lake (gravity flow,as Wood Lake surface is 50 m above Okanagan Lake surface), and returning halfof this volume back to Wood Lake (1.83e5 m3/day) by way of a 2500 horsepowerpumping station. This proposition included placing a stop-log on Oyama Canal toensure water exchange only involved Wood Lake and Okanagan Lake (not Kala-malka Lake), thereby reducing nutrient transport from Wood Lake to KalamalkaLake. The tunnel could serve as a new outlet for Wood Lake and encourage returnflow from Kalamalka Lake, thereby further improving conditions in Wood Lake.The operating range would be restricted within acceptable limits (i.e. maximumrange for fill/draw-down would not exceed 0.6m). The tunnel could also providethe ability to reduce flows through lower Vernon Creek in above-average runoffyears, helping mitigate stream-bank erosion and sediment transport into OkanaganLake. In 1974, the estimated capital cost of this project was $4.75M with operatingcosts of $81,500/year. The flow though this tunnel and through the dam at the northend of Kalamalka Lake could both be controlled to maintain optimal water qualityin Wood and Kalamalka Lakes, and would provide for an additional on-demandsupply of water for Wood Lake (BCWIB, 1974).In 1971, HWD in Winfield began pumping water from Okanagan Lake to coolits stills and then discharged the water into VC north of Ellison Lake at an av-erage rate of around 13,600 m3/day (estimates vary) from 1972-1992, reducingthe residence time of Wood Lake from 25-30 years to around 12-14 years (off-setting upstream diversions) (Anonymous, 1974; BC Research, 1974; BCWIB,1974; Jensen & Bryan, 2001; R. Nordin et al., 1985). Other sources indicatethat this flow may have been closer to 22819 m3/day (Anonymous, 1974), 12269m3/day (LAC, 2010), 12233 m3/day (BC Research, 1974; BCWIB, 1974), or 22730m3/dayNordin-1985; a provisional permit under the pollution control allowed fordischarge of 22,581 m3/day (water license was for 24,094 m3/day including cool-59ing and general plant use) (Ferguson et al., 1974). This water contained less Pand N than the ambient water in Wood Lake and Ellison Lake, diluting nutri-ent levels in these lakes. Also, nitrate in this water may have served to enhancespring P uptake by phytoplankton and avoid blue-green mid-summer algal blooms(Jensen & Bryan, 2001; MacDougall, 1984). Compared to background VC flows,the water quality after discharge had higher calcium and magnesium concentra-tions, could help increase the water hardness of Wood Lake and aid in generatingmarling events. In 1992, the distillery ended operations (reducing flows to < 380m3/day) and cooling flows ceased in 1995. Since this time, the residence timeof waters in Wood Lake have increased back to 20-30 years (Jensen & Bryan,2001). There is considerable controversy as to the net impacts of HWD on thewater quality in Wood and Kalamalka Lakes (Anonymous, 1974; BC Research,1974; BCWIB, 1974; Jensen & Bryan, 2001; R. Nordin, 1987; R. Nordin et al.,1985; R. N. Nordin, 2005; Walker et al., 1993). The majority of studies indicatethat the reduction in residence time of Wood Lake significantly improved waterconditions in this lake (Anonymous, 1974; Jensen & Bryan, 2001; R. Nordin et al.,1985; R. N. Nordin, 2005). Autumn epilimnion chlorophyll-a declined from 18 µ/lto 2 µ/L from 1975 to 1979 (Walker et al., 1993). Meanwhile, water clarity as indi-cated by secchi depth was 2-2.5m (1939) and less than 2m in 1974 - 1977, rising to3.2m in fall 1977; however, from 1978 to 1984, annual secchi depth averaged from3.75m (1978) up to 9.4m (1983) (R. Nordin, 1980, 1987; R. Nordin et al., 1985)and averaged 5.8m in 1984 (MacDougall, 1984). While the water quality data from1969 to 1999 provides useful information in assessing the impact of this proposedsolution in terms of the water quality (water chemistry, temperature, flushing time)of Ellison, Wood, and Kalamalka Lakes, the proposed use of this infrastructure isslightly different from the past in that the water would be pumped directly intoMVC or into Wood Lake and would not be subject to any heat transfer from thestills at HWD or acquisition of nutrients from Ellison Lake. Therefore, the inflowtemperature and water quality into MVC or Wood Lake would be closer to that ofthe hypolimnion of Okanagan Lake (water source) than HWD discharge.Available data from HWD’s operations quantify the approximate water volumethat is required to reduce the residence time of Wood Lake from 20-30 years to lessthan 15 years (Anonymous, 1974; BC Research, 1974; BCWIB, 1974; R. Nordin,601987; R. Nordin et al., 1985). Okanagan Lake provides an enormous natural reser-voir, with a usable storage of 4.15e8 m3 (maximum depth: 244 m) (Anonymous,1974). Jensen and Bryan (2001) noted a decrease in [SO4-2] and [Na+1] and in-crease in [Ca+2] during HWD operations from 1972 - 1992 due to relative con-centrations of these elements in Okanagan Lake waters being pumped into WoodLake during this time. An assessment of the water quality from Okanagan Lakecompared to VC water quality data (recorded since 1996) shows that OkanaganLake water quality is significantly better than that of VC; the water is drawn from31m depth (below the thermocline) where temperatures (in April) were 4.60◦C, ascompared to the annual average July/August temperatures in VC of 19◦C (and anannual average of 10◦C) (Mould Engineering, 2004). The 2011 Annual Report forWater Operations for the District of Lake Country outlines that the annual averagewater temperature at distribution from the Okanagan Lake source is between 9.1◦C(conductivity: 404 µS/cm, pH: 8.2, turbidity:1.29NTU) from the Coral Beach Wa-ter System and 10◦C (conductivity: 288 µS/cm, pH: 8.0, turbidity: 0.53NTU) fromthe District of Lake Country Water System (Mitchell & Hansen, 2011).Analyses of the quantity of water available from Okanagan Lake (authorizeddiversions), and capacity of the Okanagan Lake pumping station show that theprojected future demands on the system are well below available supply. TheOkanagan Lake pump station has two 750 hp pumps and motors (currently un-used), and one rebowled 750 hp pump operating at 350 hp (to reduce electricitydemand). This latter pump has a capacity of 0.105 m3/s and supplies DLC’s cur-rent annual demands from Okanagan Lake of 808e3m3. This demand is well belowthe district’s authorized diversions (10,993e3m3). The design pumping rate of thissystem is 0.464 m3/s with two 750 hp pumps operating. A 100 hp booster pumpwas installed in 2002 when screening improvements were being made to the VCintake, enabling the Okanagan Lake pump station to provide water for the entiresystem during construction (and also during spring freshet when VC water qualitydeclines severely). Instabilities in the VC mainline and current Okanagan Lakepump station capacity limit the extent to which this booster system can be used(Mould Engineering, 2004). The water quality improvement plan for DLC utilizesthis pumping station; however, the full capacity is reduced to a maximum dailyvalue of 0.29 m3/s, capable of supplying 40% of the maximum day demand. New61pump control works would be required before the additional two pumps could beused again. The total cost of these upgrades was estimated at $500,000 in 2004(including relocating of the chlorination injection point). The pump station de-mand is conservative, and most components are capable of increased rates, but itcannot currently be upgraded to allow all three pumps to run simultaneously due tohydraulic constraints in the rest of the distribution system. A man-made reservoir(2270 m3) that receives inflow from these pumps is used to control pump oper-ation and provide balancing during fluctuating demands. In order to meet futuredemands, proposed plans to increase the reservoir volume to 2.5 M litres have beensubmitted (total cost of $810,000) (Mould Engineering, 2004). For more informa-tion regarding the restrictions imposed by the infrastructure and Okanagan Lakereservoir, as well as the costs associated with upgrading these facilities, the readeris encouraged to review Mould Engineering (2004). Despite the capacity of theOkanagan Lake reservoir, the proposed future demands on this system from DLCand Regional District of the North Okanagan (RDNO) may limit the amount ofwater actually available from this system; however, no quantified allowance wasforetold (pers. Communication, Mike Sokal, Heather Larratt, Hillary Ward, April18, 2016).A distribution system computer model using Waterworks for AutoCAD R14has been developed for the DLC water distribution system and calibrated withflows from 2001 (Mould Engineering, 2004). This model could be used (in col-laboration with Mould Engineering) for predicting the actual timing and volumeof water available and the impacts on the rest of the system of diverting waterto Wood Lake. Potentially two options exist with various volumetric flow rates(constant/time-varying) that could be modelled: (a) pumping water directly intoWood Lake; (b) pumping water into MVC downstream of Ellison Lake. Option(a) has been initialized directly in the DYRESM model as a newly defined in-flow with user defined (controlled) inflow properties. Option (b) would requirethe supplementary utilization of the Middle Vernon Creek Decision Support Sys-tem (MVCDSS) Model (described below).ESSA Technologies Ltd created a model (MVC DSS) in 2013 that utilizes wa-ter/mass balance/conservation equations along with hydrological data to predicthow various management alternatives would influence flows and water levels in62MVC. This was developed to assist water managers in supporting the success ofkokanee spawning in MVC and helping increase kokanee populations in WoodLake by identifying scenarios that result in detrimental/sub-optimal/optimal flowswithout imposing risks on other resources. The model simulates measured MVCflows and Ellison Lake levels very closely until flow is disrupted by installationof the TFCS (end of July to mid-August) or by beaver dams in MVC (end ofSeptember) (Epp & Neumann, 2014). The MVC DSS could be used in concertwith DYRESM when simulating pumping water from Okanagan Lake into VC tosupplement VC flows, thereby enabling assessment of the water volume required toprovide optimal flow rates (in terms of WUHW for kokanee) in MVC for spawn-ing. This portion of the assessment in MVC is beyond the scope of the currentstudy; however, the subsequent impacts of this supplementary flow on Wood Lakeare assessed with DYRESM.Given sufficient evidence from historical studies that supplementing inflow intoWood Lake with Okanagan Lake water would increase water quality (chemistry) inWood Lake, a number of different inflow rates (Okanagan Lake water source) andcoincident outflow controls are investigated with DYRESM in terms of the poten-tial for also altering the thermal structure of Wood Lake (Section 4.12.1). OkanganLake water parameters are based on Mould Engineering (2004) and Mitchell andHansen (2011). Several of these scenarios decrease the residence time of the lake,while other combinations of inflows and outflows would serve to increase the stor-age volume of the lake.Proposed Solution 6 (Modelled with DYRESM): Pumping HypolimneticWater From Kalamalka Lake back into Wood LakeThe primary goals with regards to pumping water from the hypolimnion of Kala-malka Lake into Wood Lake are similar to the previous management scenario(Section 2.8.2) (i.e. altering the thermal structure and dilution/flushing of WoodLake). Since these lakes lie at the same elevation, the pumping head required totransfer water between these two basins would be less than that from OkanaganLake, and the length of the conveyance system would also be less. The pumpingand conveyance system that would need to be constructed is beyond the scope ofthis study. This proposition also has the added potential for triggering marling63events in Wood Lake, given the higher calcium and magnesium ion concentrations(and higher conductivity) of Kalamalka Lake water and that Kalamalka Lake waterproffers ideal water parameters that enable it to marl annually (Anonymous, 1974;BCWIB, 1974; LAC, 2010; Northcote et al., 1974; Williams, 1972).The notion of mixing Wood Lake and Kalamalka Lake has been discussed inthe past by BCWIB (1974), R. Nordin (1980), and R. Nordin (1987). At this time,the proposition (capital cost of $340,000) was to locate a pump on the land betweenthe two lakes, with inflow/outflow pipes in the hypolimnion of both lakes(BCWIB,1974; R. Nordin, 1987). BCWIB (1974) suggested pumping 1.22e5 m3/day duringthe stratified season (200 days/year). The annual operating cost was estimated at$16,000. Estimating the size of Wood Lake’s hypolimnion (approximately 7.4e7m3), BCWIB (1974) suggested that it would take a minimum of three years to mixthis volume of water; however, likely a longer period would be required becauseof mixing throughout the water columns during spring and fall turnover. It is sug-gested that pumping nutrient-rich water from Wood Lake into the hypolimnion ofKalamalka Lake would enable this water to be cleaned (stripped of nutrients) by theannual marling process in Kalamalka Lake, making it unavailable for algal growthin Kalamalka Lake. Kalamalka Lake is approximately 7.7 times larger than WoodLake by volume. It was hypothesized that the mixing in Wood Lake would reduceTP concentrations from 74 µg/L to 16 µg/L, total soluble phosphorous (TSP) from33 µg/L to 9 µg/L, TKN from 515 µg/L to 180 µg/L, and raise Nitrate N from26 µg/L to 54 µg/L (BCWIB, 1974). Wood Lake nutrient concentration valueswere based on 1972/1973 mean values and Kalamalka Lake’s values were basedon concentrations measured in 1972 (BCWIB, 1974). It would be critical to under-stand the limiting capacity of the calcite precipitation in Kalamalka Lake to removenutrients from the water column so as to not overload the system and worsen thewater quality in this pristine lake.Similarly to using Okanagan Lake as a source, a number of different flow rates(using hypolimnetic water from Kalamalka Lake) and coincident outflow controlswill be investigated with DYRESM in terms of the potential for altering the thermalstructure of Wood Lake during the late summer. Some of these scenarios decreasethe residence time of the lake, while other combinations of inflows and outflowswould serve to increase the storage volume of the lake. Kalamalka Lake water64parameters (temperature and salinity) have been obtained from one CTD station(35m depth) measured regularly during the current study’s field program from 30June 2015 to 25 April 2016.2.9 SummaryThe principal research question to be investigated using DYRESM is to what ex-tent it is possible to mitigate fish kills in Wood Lake by providing additional habi-tat in the late summer and early fall when kokanee are threatened by an annualtemperature-oxygen squeeze by way of managing inflows and outflows, or throughan alternative “natural-design” approach. This chapter (Chapter 2) has provided acontextual setting for the current study, a summary of the motivation for research,as well as a detailed account of the historical development of current key physical,chemical, and biological conditions in Wood Lake. This information, in concertwith discussions with BCMOE, guided the selection of appropriate managementscenarios for modelling (Section 2.8.2 and 4.12.1). Given the summary of currentmanagement strategies presented (Section 2.8.2), the following chapter (Chapter 3)conveys the research methods employed to obtain the data necessary to calibrateand validate a DYRESM model of Wood Lake. Chapter 4 and Chapter 5 reveal anddiscuss results from the current study’s field observations, as well as application ofDYRESM to investigate what made 2011 a unique year, evaluate the key meteoro-logical factors that influence Wood Lake’s heat content, and the results of severalsimulated management scenarios.65Chapter 3MethodsThis chapter outlines methods involved in the current study of Wood Lake withregards to collection and analyses of past limnological data, field work employedduring the 2015-2016 field season, processing of field data, and calibration, valida-tion, and application of a one dimensional DYRESM model. The methods outlinedherein were integral to addressing the research objectives discussed in Section 2.3.3.1 Data: Previous and Ongoing Limnological Studiesand MeteorologyPrior to commencement of the fieldwork portion of this study, all available datarecords (complete and incomplete) from 1969 to present day were collected fromvarious sources from past and ongoing limnological studies. Historical observa-tions, data sets and accompanying analyses (in some cases) enabled a broader un-derstanding of the development of eutrophic conditions in Wood Lake over timeand of the factors responsible for the annual temperature-oxygen squeeze phe-nomenon. Furthermore, available historical and current meteorological and limno-logical observations enabled calibration, validation, and application of DYRESMfor investigating the thermal structure of Wood Lake and evaluating several man-agement scenarios in terms of the thermal characteristics of the lake.663.1.1 Complete RecordsComplete records of recent limnological data related to the current study of WoodLake were obtained from BCMOE, and Larratt Aquatic Consulting (LAC). Com-plete meteorological records were compiled from Environment Canada, Coral BeachFarms (weather station for solar radiation), and republished data the NorwegianMeteorological Institute and World Weather Online via CedarLakeVentures (2015)(cloud cover (CC)).Source: Ministry of Environment (BCMOE)Monthly raw profile data for 2013, 2014, and 2015 (March to October) was pro-vided by BC MOE (measured at site 0500848, near the lake’s deepest location),including temperature, DO, chlorophyll-a, and secchi depth. These data havebeen subject to a quality control process to ensure accuracy. Measurements wererecorded at 2m intervals throughout the water column. Composite water sampleswere collected at the same time as profile measurements in the epilimnion (1m,5m, 10m) and the hypolimnion (20m, 25m, 30m) and analyzed for chlorophyll-a,silica, ortho-P, ammonia, TKN, TON, TDP, NO2+1, NO3, TN, and TP in March- October 2013-2015 as well as SO4, Cl, TOC, and water hardness in spring/fall2015 only. Seasonal averages and trends in the hypolimnion and epilimnion arevisible from this chemistry data and used in comparison to the field data from thecurrent study.The 2013 - 2014 datasets provide the most consistent, thorough, and reliabletemperature profiles available to validate a calibrated DYRESM model in termsof its ability to replicate the thermal structure of the water column in Wood Lake,given measured meteorological data (Section 3.1.1). An important reason for se-lecting these years for model validation was also based on having access to reliableshort-wave radiation data (critical parameter for modelling Wood Lake’s thermalregime) from a weather station adjacent to Wood Lake that was established on 26June 2012. Additional years (i.e. 1984 and 1987-1988) (Section 3.1.2) providerelatively consistent enough data for further model validation if required; however,short wave radiation data is not available for these years, thus this parameter wouldneed to be estimated based on time of year, latitude, and the sun’s angle of inci-67dence (i.e. Brock (1981)), which would lead to an unquantifiable error.Source: Larratt Aquatic Consulting, Ltd. [LAC]LAC has collected monthly profile data on Wood Lake from May to October from2005 to 2016, typically at 1m intervals from surface level to 20 - 24m depth; how-ever, measurements were taken to 28m depth following October, 2014. The sam-pling location used by LAC in Wood Lake was located at: 50.10351◦N, 119.37994◦W(prior to October, 2014) and a deeper location was utilized following October, 2014at: 50.09360◦N, 119.38629◦W (J. Self, pers. corr., 20 March 2015). Measure-ments include temperature and DO (using a Hanna multi-parameter portable meterprobe). pH was also recorded from 2012 - 2015. In order to obtain this data, formalpermission from Pattie Meger (Water Quality Technician, DLC) and Renee Clark(Water Quality Manager, Regional District of North Okanagan) has been granted(P. Meger, pers. corr., 16 February 2015; R.Clark, pers. corr., 20 February 2015).Secchi depths in 2007 - 2010 and additional data analyses discussed in Chapter 2are also presented in LAC (2010).Meteorological DataMeteorological data was collected from various weather stations in close proximityto Wood Lake. Hourly average/maximum/minimum temperatures [◦C], hourly av-erage wind speed [km/hr] and direction [◦from North], relative humidity [%], dewpoint temperature [◦C], and air pressure [kPa] for specific years of interest wereobtained from Kelowna Airport (WMO ID: 71203; Climate ID: 1123939; TC ID:YLW) at 49◦57’26“ N, 119◦22’40” W (433.10m amsl) from Environment Canada.Parameter values were averaged over daily time steps. A more complete recordof daily precipitation (rain [mm] and snow [cm]) was obtained from KelownaMWSO Weather Station (Climate ID: 11239RO) at 49◦57’00“ N, 119◦24’00” W(456.00 m amsl) and any missing precipitation data was flagged and filled with datafrom the former Kelowna Airport weather station. These weather stations are lessthan 1.8km apart and are approximately 10.7km (Kelowna Airport) and 11.4km(Kelowna MWSO) from the south end of Wood Lake.Two meteorological datasets that were not immediately available from these68Environment Canada weather stations were daily averaged short wave (solar) radi-ation and CC. Solar radiation data [W/m2] was provided by Coral Beach Farms whohave a Ranch Systems Weather Station located at 50◦05’08.50“N, 119◦22’13.37”W(494m amsl) (Craig Dalgliesh, pers. corr., 28 Oct. 2015). This location is adja-cent to the lake, just north of the lake’s centre, and 700m from the east shore.This station records real-time solar radiation [W/m2], temperature, precipitation,wind speed and direction, relative humidity, soil moisture, and soil temperature at15minute intervals. A four-year online database for this weather station is avail-able, beginning 26 June 2012. CC data was obtained from WeatherSpark.com,which republishes data provided by met.no (The Norwegian Meteorological Insti-tute) and WWO (World Weather Online); the data was collected at the METARstation at Kelowna International Airport (Orna Kretchmer, pers. corr.,19 January2016)(CedarLakeVentures, 2015).3.1.2 Incomplete RecordsIncomplete data records (i.e. inconsistent data sets) were obtained for tempera-ture and flow data in MVC since 1970 (from BCMOE and Natasha Neumann) aswell as historical limnological data for Wood Lake from 1969 to 2012 (from BC-MOE). Note that temperature and flow records for MVC have become increasinglycomprehensive over time, particularly since 2012 as a result of the Middle VernonCreek Action Plan (MVCAP) and comprise a fairly consistent hydrological recordfor the past four years (Section 4.6).MVC Current Stream Flow and Temperature Data and Water BalanceDr. Natasha Neumann (UBC) and Dr. Hillary Ward (BCMOE) have provideddaily discharge volume and temperature data for MVC from 1970 - 2015. From1970 - 1987, a hydrometric station was located at Woodsdale Road and operatedby the Water Survey of Canada (WSC). In 2003 the station was moved to ReimcheRoad and operated by the Oceola Fish and Game Club from 2003 - 2011. From2012 - 2015, this station was maintained at Reimche Road by the Okanagan Na-tion Alliance and the MFLNRO. The Reimche Road station is located at 50◦02’51“, 119◦24’ 22” approximately 590m upstream from Wood Lake (Hillary Ward,69pers. corr., 20 April 2016; Natasha Neumann, pers. corr., 23 March 2015). Therecording interval is 30-minutes (from 2012 - 2016) with a portion of 2015-2016at 5-minute intervals, but was previously 24 hours in years prior to 2012. Someyears have complete datasets, whereas other years only have data for a portion ofthe year (i.e. April/May - October/November). Unfortunately 2011 data is onlyavailable from 1 January - 19 April 2011.Ten water level and flow monitoring sites were implemented in VC for 2013,yielding hydrometric measurements and elevation surveys along MVC, and wa-ter samples for nutrient loading. Beginning in May 2012 (May to October 2012 -2016), stream flow and lake level measurements have been recorded from SwalwellLake to Wood Lake. Hydrometric stations (recording water level and temperature)or water level loggers, stilling wells and staff gauges were installed in UVC (3),Clark Creek (1), and MVC (3). Stage-discharge relationships for each creek mea-surement site convert depth measurements into continuous records of stream flow(Epp & Neumann, 2014). MVC flow data for 2012 - 2016 comprises a relativelycomplete dataset (Section 4.6).A water balance model has been generated as part of the MVCAP project for2012 - 2015 in order to study changes in VC flows and lake volumes from Swal-well Lake to Wood Lake. The water balance was developed using a daily time step(can be averaged to a weekly/monthly time step). Data utilized included measuredlake levels and stream-flows, reported water use by DLC and Eldorado Ranch, es-timated water diversion for additional water licenses (irrigation/domestic) on VC,estimates of evapotransipiration and precipitation from FarmWest’s Ellison station,and estimates of runoff and seepage from portions of the watershed. Reservoirseepage and evaporative losses are assessed, in addition to UVC seepage losses inthe alluvial fan. MVC flows depend on the outflow from Ellison Lake controlledby a TFCS, and the presence of beaver dams (Epp & Neumann, 2014). The waterbalance model can be used to evaluate the discharge values recorded at ReimcheRoad and to provide an explanation with regards to the sources, diversions, andavailability of water in this system from Swalwell Lake to Wood Lake.70Wood Lake Physical, Chemical, and Biological Profile DataAdditional data obtained from BC MOE includes datasets analyzed as part of otherstudies discussed in this study (i.e. Anonymous (1974); BC Research (1974);BCWIB (1974); Jensen and Bryan (2001); MacDougall (1984); R. Nordin (1980)).The majority of this data was collected at three sites (typically only one/two oneach sampling date):1. 500450: “Wood Lake at Mouth of VC” (50.057◦W, 119.4028 ◦W,) 500mnorth and slightly west of the mouth of MVC; (1969 - 2007 with some yearsmissing)2. 500245: “Wood Lake Central Site” (50.0875◦N, 119.3889◦W) near deepestlocation, south of centre of lake; (1969 - 1985)3. 500848: “Wood Lake Deep Basin” (50.0749◦N, 119.3917◦W) about 1kmnorth and 3m deeper than 500245 (30m depth): location used by OkanaganBasin Study and Kalamalka and Wood Lake Basin Study; (1985 to present)BCMOE provided one record that contains annual spring and fall measure-ments from 1970 to 2014 from site 500245 and site 0500848, including averagechlorophyll-a (epilimnion), average secchi depth, as well as average spring/fallepilimnetic and hypolimnetic nutrient data. Some measurements are missing fromsome years. Spring measurements were taken in February - May, (typically inMarch/April). Fall measurements were taken in August - October (typically inSeptember). BCMOE indicated that due to the “bowl shape” of Wood Lake thereis little lateral variation in parameters. Only the deepest central site has been moni-tored since 2007 (Mike Sokal, pers. corr., 12 Feb. 2015). Contrastingly, the currentstudy reveals that there is notable lateral variation in parameters in Wood Lake, par-ticularly during seiche events and during/following turnover. BC Research (1974)also noted these differential heating/cooling patterns in Wood Lake.BC MOE provided an additional raw dataset (temperature, chemistry, nutrients,metals, pH, BOD/COD, carbon, alkalinity, hardness, turbidity, DO, chlorophyll-a,coliforms, etc) that includes an incomplete/inconsistent record (Table 3.1) from1969 to 2014, identifying: measurement site, date, and depths measured. Jensenand Bryan (2001) compiled a record from this dataset for trend analyses from 196971to 1999 (referenced during this study for comparison and discussed in Chapter 2).All provincial data discussed herein from 1973 to 1999 is archived on the provin-cial Environmental Management System (EMS). Temperature and DO were mea-sured using a YSI or Hydrolab Meter each fall. Water chemistry samples werecollected at three shallow water depths (composited), and likewise from three hy-polimnetic depths using a Van Dorn Bottle. Water chemistry samples includedtotal P [TP], total dissolved phosphorous [TDP], total N [TN], nitrate nitrogen, DO(at deepest location), total alkalinity, specific conductance, turbidity, dissolved sil-ica and sulphate, sodium [Na], total calcium [Ca], magnesium [Mg], and chloride[Cl] concentrations (Jensen & Bryan, 2001). Within these records, March 1987 toNovember 1988 presents a continuous record in which the water column was mon-itored monthly at 2m depth intervals at site 500848 to provide a representation ofthe lake’s time-varying structure over 1.5 years. However, for this dataset, gener-ally not all depths were measured in all years, not all variables were recorded at alldepths or on all days, and sites were not monitored at regular intervals. Despite anyaforementioned inconsistencies, these datasets and additional analyses by Jensenand Bryan (2001); R. Nordin (1987); R. Nordin et al. (1985) provide a wealth ofknowledge about Wood Lake in terms of the trends in chlorophyll-a, DO, temper-ature, pH, and specific conductivity over the period 1969 to 2014 (before, during,and after HWD operations).72Table 3.1: Key parameters of interest included in BC MOE raw data record(1969 - 2014).Parameter Available Record from BC MOEChlorophyll-a Reported after 1975 inconsistently at various depths, sites, and dates.After 1982, only values from 0 - 10m are reported with 1-10 measurements.DO Measurements began in 1973 - 1974. No measurements were reported in some years.Some years only report spring/fall values. Generally DO was measured throughout thewater column up to max. depth of 20 - 32m depending on site and date; however onsome dates only one epilimnion and hypolimnion depths were measured.Specific Measured at surface level (0 - 3m), at one hypolimnion (20-30m) and/or epilimnionConductivity (1-10m) depth, at regular intervals, or at discrete depths (i.e. 0, 10, 20, and 30m).Not reported in some months/years, or at all sites. Values not reported from 2003-2014.Temperature The most consistent variable measured, although it was not measured at every sitenor at all depths. There is generally sufficient data to construct a vertical profile oftemperature variation with depth for most measurement days. Temperature data was notreported from 1982 - 1984 and 1996 - 1998. Other missing data points also exist(i.e. spring/fall values or dates in which only surface/hypolimnion values were reported).Turbidity Recorded very inconsistently with values reported for various (one or more) epilimnion,thermocline, and/or hypolimnion depths. No values from October 1982 to 1989, orafter 1992.pH Reported at one or more various depths (surface, thermocline, and/or hypolimnion) onvarious measurement days/years; a fairly consistent record (1,10,20,30m) is availablefrom March/September 1999-2002. No measurements after September 2002.3.2 Important Historic Key Studies and Knowledge BaseAmongst many important studies that have been conducted on Wood Lake over thepast 45 years, this section highlights four of which involved extensive data collec-tion: The Okanagan Basin Study (Anonymous, 1974), the Kalamalka-Wood LakeBasin Water Resource Management Study (BCWIB, 1974), Water Quality of theKalamalka-Wood Lake Basin (BC Research, 1974), and The Chemical and Micro-biological Limnology of Wood Lake, B.C. (MacDougall, 1984). Other importantstudies described and/or referenced extensively in Chapter 2 (not described furtherhere) include a paleolimnological study of Wood Lake by Walker et al. (1993) ,and limnological/hydrological analyses by Jasper & Gray (1982) (as discussed inJensen and Bryan (2001) and R. Nordin (1987)) and Jasper & Gray (1980) (asdiscussed by MacDougall (1984); R. Nordin (1987)) as well asEpp and Neumann(2014, 2016); Ferguson et al. (1974); LAC (2010); Mould Engineering (2004);73R. Nordin (1980, 1987); R. Nordin et al. (1985); Northcote et al. (1974); Wiegandand Chamberlain (1987) amongst many others.The Main Report of the Consultive Board - including the Comprehensive Frame-work Plan was prepared by the Consultative Board under the Canada-British ColumbiaOkanagan Basin Agreement in 1974, detailing the results from a rigorous four yearstudy (1969-1972) of the main valley lakes in the Okanagan. At the time of thisstudy, Wood Lake was the most eutrophic lake in the entire basin. Physical char-acteristics (secchi disk, temperature, oxygen), chemical (nutrients, metals), biolog-ical (chlorophyll-a, zooplankton, phytoplankton, periphyton, bottom fauna), geo-logical (sediment analysis), and trophic state are described in detail from month-ly/seasonal data collected in 1970-1972 (Anonymous, 1974).This dataset was fur-ther analyzed by Ferguson et al. (1974); Northcote et al. (1974) and referenced byWalker et al. (1993).BC Research (1974); BCWIB (1974) provide separate analyses of a water re-source management study on the Kalamalka-Wood Lake basin wherein water sam-pling was completed over 18 months (1 April 1972 to 31 July 1973) and includestwo spring freshets. All of the data and analytical results for this study are availableat B.C. Research (library call number TD 380.H6) (BC Research, 1974). BC Re-search (1974) provide the lake sample site locations and depths of measurements.Measurements at these locations included temperature, pH, alkalinity, carbon diox-ide, DO, water clarity (using a 16-cm secchi disk), and chlorophyll-a (at selecteddepths). DO and temperature were measured using a YSI meter and pH was mea-sured using a Metrohm Model E488 portable pH meter (standardized frequentlyeach day) (BC Research, 1974).MacDougall (1984) outlines the results of a detailed limnological study ofWood Lake performed in 1984 and provides raw data. Samples were collectedat the centre of the lake (unidentified location). Measurements of temperature,DO, secchi depth, chlorophyll-a, light extinction, and biological/chemistry analy-ses were performed. DO was measured at 5 - 10m intervals and temperature at 1mintervals up to 30m depth. pH was recorded at 5m intervals. Seven discrete depths(5m intervals) were used for chemistry data and five depths for biological analysisMacDougall (1984). Two key results from 1984 were:741. Highest P concentrations were detected in the hypolimnion, providing earlyevidence of internal loading (MacDougall, 1984).2. No clear correlation between algal biomass variations and secchi depths,despite this relationship being discussed in other literature (i.e., R. Nordin(1980) over a longer timescale from 1969 to 1980) (MacDougall, 1984).Secchi depths are typically lowest in spring when algae are abundant and in-crease until October with a slight decrease in September (Epp & Neumann,2014). In 1984, the lowest secchi depths were recorded during blue-green al-gal blooms in early June despite epilimnion chlorophyll-a levels being three-fold higher in the fall (MacDougall, 1984). The current study reveals howmany other factors affect secchi depth measurements in Wood Lake as well.3.3 Field Work: Thermistor Chains and CTD ProfilesThe field campaign for the current study involved installing several thermistorchains (multiple temperature loggers on vertical anchored moorings) to collecta continuous time series of temperature at discrete depths throughout the watercolumn at specific locations in Wood Lake (Section 3.3.1 and 3.3.2) as well asweekly/bi-weekly CTD north-south transects of the lake along its longest fetch(Section 3.3.3 and 3.3.4).3.3.1 Central Thermistor Chain (SBE 56 Temperature Loggers)On 4 May 2015 a thermistor chain was installed in Wood Lake near BC MOE’ssample location (EMS ID: 500848; Section 3.1.2) near the center of the lake (spe-cific location: 50◦04.459’N, 119 degree23.545’W, depth: 32.16m). The thermistorchain’s two 8ÏCEPLAST floats (buoyancy: 2470g/each Lundsgaard, N.M., NeptusPlast AS, pers.comm, Mar.9, 2015) at the top of the chain were placed 4m belowthe surface due to boat traffic and angler activity. A 50.0lb anchor was used atthe lower end. This chain consisted of 10 SBE 56 Temperature Loggers spacedat regular 2.7m intervals from 4m depth to 32m depth. The temperature loggersare waterproof to 1500m (weight 0.05kg in water), offer resolution to 0.0001◦C,are accurate to +/- 0.002◦C from -5◦C - 35◦C. The units are stable (drift) within750.0002◦C/month, and the clock is accurate within +/- 5 seconds/month. The log-gers contain an internal memory of 15.9 million samples. Each logger containsa 3.6V Lithium battery that lasts almost two years sampling 4 times/minute (Sea-Bird Electronics [SBE], 2015). The user-specified logging interval for this studyis 10 seconds. The units have an internal USB 2.0 interface, which allow for con-figuration and downloading data to a computer (SBE, 2015). The thermistor chainlocation (deployment & extraction) was recorded (latitude and longitude) usinga Garmin etrex 30 in the field (Table 3.3) and its depth below the surface wasrecorded on four days (Table 3.2). The thermistor chain was removed once on 30June 2015 at 50◦04.458’N, 119◦23.546’W (slight drift) to add a HOBO level log-ger to the chain (positioned at an initial depth: 3.06m; height: 28.86m). The chainwas redeployed at the initial location (50◦04.459’N, 119◦23.544’W). On July 14,the location was recorded as 50◦04.460’N, 119◦23.545’W. On 25 August 2015 thedepth to the top buoy was measured to be 2.15m at 50◦04.461’N, 119◦23.546’W(slight drift). On 13 April 2016 the middle chain was removed at 50◦04.460’N,119◦23.544’W (depth to top buoy was 2.44m). The temperature data collectedfrom these temperature loggers was uploaded using SBE Seasoft v2 software, andsubsequently analyzed and plotted as a time-series using MATLAB.Table 3.2: Central Thermistor Chain: Temperature logger heights from lakebottom and Depths from surface on given days during year (Fluctuationsin depths due to lake level change).Set-up Thermistors Heights from bottom: 0.28m, 3.33m, 6.33m, 9.38m,12.43m, 15.47m, 18.51m, 21.56m, 24.60m, 27.53m, (Level Logger: 28.86m)Date Total Depth Thermistor DepthsDeploy 32.1m 4.63m, 7.56m, 10.6m, 13.65m, 16.69m, 19.73m, 22.78m, 25.83m,(4 May 2015) 28.83m, 31.88m30 June 2015 31.92m 4.39m, 7.32m, 10.36m, 13.41m, 16.45m, 19.49m, 22.54m, 25.59m,28.59m, 31.64m (Level Logger: 3.06m)25 August 2015 31.73m 4.2m, 7.13m, 10.17m, 13.22m, 16.26m, 19.3m, 22.35m, 25.4m,28.4m, 31.45m (Level Logger: 2.87m)Remove 32.015m 4.49m, 7.42m, 10.46m, 13.51m, 16.55m, 19.59m, 22.64m, 25.69m,(13 April 2016) 28.69m, 31.74m (Level Logger: 3.16m [field measurement]; 3.08m[logger data])763.3.2 Seiche Thermistor Chains (RBR Solo-T Temperature Loggers)Eleven RBR Solo-T Temperature loggers were obtained during the 2015 field sea-son and used to monitor seiching in Wood Lake. These units are waterproof to1700m (weight: 0.02kg in water), offer resolution to <0.00005◦C, and are accurateto +/- 0.002◦C from -5◦C to 35◦C. The units are stable (drift) within 0.002◦C/year,and the clock is accurate within +/-60sec/year. The loggers contain an internalmemory of approximately 25M readings (at 2Hz or slower) and approximately66M readings at 4Hz, 8Hz, or 16Hz. Each logger contains a 3.6V LiSOCl2 AA cellbattery. The user-specified logging interval for this study was 10 seconds (equal toSBE loggers on middle chain). The units have an internal USB 2.0 interface, whichallow for configuration and downloading data to a computer (RBR Ltd., 2016).These additional temperature loggers were used to monitor seiching in WoodLake along the north-south axis (longest fetch) (Figure 3.1). One thermistor chainwas installed at the south end of the lake on 19 August 2015 to 30 September 2015where the total depth was 22m (with temperature loggers at approximately 21.5m,15m, 10m, 5m depths). A second thermistor chain was installed at the north end ofthe lake on 19 August 2015 to 30 September 2015 where the total depth was 15m(with temperature loggers at approximately 15m, 10m, 5m depths). Both chainswere redeployed from 7 October 2015 to 25 April 2016 with a similar layout;however, the 5m-depth logger was re-positioned to 12.5m depth in order to be po-sitioned within the thermocline and capture the oscillatory seiche movements (thebottom of the epilimnion was over 10m depth at this time). The durations of de-ployment, locations, and exact depths at deployment/removal of these temperatureloggers were recorded (Table 3.3).In an attempt to capture water movement through Oyama Canal exchangedbetween Wood Lake and Kalamalka Lake (as expressed by thermal signals), tem-perature loggers were also installed on either end of the canal. Three loggers (Fig-ure 3.1) were deployed: one at the south end of the Oyama Canal at 1.7-1.9m depth(total depth 1.8-2.0m), one at the north end of the canal at 0.3-0.4m depth (totaldepth: 0.5-0.6m) about 226m north from the former, as well as one in the shallowbay at the south end of Kalamalka Lake at 1.7-1.9m depth (total depth: 1.9-2.1m)about 228-230m north of the logger at the north end of the canal (Figure 3.1). Due77to concern regarding freezing in the shallow waters through the canal at the be-ginning of January, these latter three temperature logger stations were removed on1 January 2016. From 8 January 2016 to 25 April 2016 these three temperatureloggers were redeployed in a slightly different manner (shallowest station at northend of canal was not redeployed). On 8 January 2016 there were two tempera-ture loggers installed at the south end of the canal at 1.1m and 1.9m depth (totaldepth: 2.0m) and one in the shallow bay at the south end of Kalamalka Lake at1.75m depth (total depth: 2.0m). These locations were nearly identical to wherethey had previously been removed. The durations of deployment, locations, andexact depths at deployment/removal of these temperature loggers were recorded(Table 3.3).780 1 kmT1T2T3T4T5T6Thermistor ChainTemperature LoggerKalamalka LakeWood LakeOyama CanalNFigure 3.1: Locations for thermistor chains and individual temperature log-gers in Wood Lake, Oyama Canal, and south end of Kalamalka Lake.79Table 3.3: Duration, location, and depths of temperature loggers in WoodLake to monitor seiche and water movement through the Oyama Canal.Station Deploy/ Location Total Logger Distance NotesName Remove Depth Depth to South(m) (m) ShoreT1:Wood 19Aug.-30Sep. 50◦03.399’N 21.92 21.66,14.92, 450mSouth 119◦23.876’W 9.92,4.92(chain) Removed 50◦03.398’N 22.03 21.77,15.03, 449m30Sep. 119◦23.875’W 10.03,5.03Re-deployed 50◦03.403’N 21.91 21.67,14.91, 456m Reposition shallow logger7Oct. 119◦23.877’W 12.41,9.91 from 5m to 12.5m depthRemoved 50◦03.402’N 22.38 22.15,15.33 455m25Apr.2016 119◦23.877’W 12.85,10.36T2: Central Deploy 50◦04.459’N 32.16 Table 3.2 2454m Removed 30 June 2015 atThermistor 4May2015 119◦23.545’W 50◦04.458’N,119◦23.456’W,Chain Removed 50◦04.460’N 32.02 Table 3.2 2454m install level logger;13Apr.2016 119◦23.544’W redeployed at 4May location.14Jul.: 50◦04.460’N,119◦23.545’W. 25Aug.:50◦04.461’N,119◦23.546’WT3:Wood Aug.19-25 50◦06.370’N 14.76 14.48,9.76 6051mNorth 119◦23.026’W 4.76(chain) Adjusted 50◦06.360’N 15.2m 14.72,10, 6031m25Aug.-30Sep. 119◦23.035’W 5Removed 50◦06.360’N 15.15 14.67,9.95, 6031m30Sep. 119◦23.034’W 4.95Re-deployed 50◦06.363’N 15.01 14.82,12.55, 6037 Reposition shallow logger7Oct. 119◦23.035’W 10.05 from 5 to 12.5m depthRemoved 50◦06.364’N 15.41 15.29,13.04 6039m25Apr.2016 119◦23.034’W 10.51T4:Wood Aug.19-25 50◦06.585’N 2.02m 1.87 6457mNorthB 119◦22.961’W(single) Adjusted 50◦06.584’N 1.93 1.78 6455m25Aug.-30Sep. 119◦22.963’WRemoved 50◦06.584’N 1.84 1.69 6455m30Sep. 119◦22.963’WRe-deployed 50◦06.584’N 1.90 1.75 6455m7Oct. 119◦22.964’WRemoved 50◦06.582’N 1.97 1.82 6451m1Jan. 2016 119◦22.965’W80Table 3.3: (Continued) Duration, location, and depths of temperature loggersin Wood Lake to monitor seiche and water movement through the OyamaCanal.Station Deploy/ Location Total Logger Distance NotesName Remove Depth Depth to South(m) (m) ShoreT4: Wood Jan.8,2016 50◦06.584’N 2.19 2.05, 6455m Redeployed withNorthB2 119◦22.964’W 1.19 two thermistors(2 Loggers) Adjusted 50◦06.583’N 2.07 1.93, 6453m20Jan.2016 119◦22.965’W 1.12Adjusted 50◦06.583’N 2.07 1.93, 6453m Drifted to: 50◦06.586’N,24Feb.2016 119◦22.965’W 1.12 119◦22.966’WRemoved 50◦06.582’N 2.57 2.43, 6451m25Apr.2016 119◦22.966’W 1.57South End 6495m Distance South to Northof Oyama Shore of Wood Lk. alongCanal ’seiche transect’ (Figure 3.1)North End 6672m Oyama Canal Lengthof Oyama ∼177mCanalT5:Oyama Aug.19-25 50◦06.704’N 0.57 0.42 6680mNorth 119◦22.934’W(single) Adjusted 50◦06.704’N 0.59 0.44 6680m Reposition block25Aug. 119◦22.934’WAdjusted 50◦06.704’N 0.48 0.33 6680m Reposition block: Block1Sep. 119◦22.934’W submerged in groundRemoved 50◦06.704’N 0.46 0.31 6680m30Sep. 119◦22.934’WRe-deployed 50◦06.704’N 0.47 0.32 6680m7Oct. 119◦22.934’WRemoved 50◦06.704’N 0.54 0.39 6680m1Jan.2016 119◦22.933’WT6:Kal Lk. Aug.19-25 50◦06.760’N 1.87 1.54 6784m(single) 119◦22.926’WAdjusted 50◦06.826’N 2.12 1.90 6908m25Aug.-30Sep. 119◦22.909’WRemoved 50◦06.826’N 2.13 1.91 6908m30Sep. 119◦22.909’WRe-deployed 50◦06.826’N 1.96 1.73 6910m7Oct. 119◦22.902’WRemoved 50◦06.826’N 2.06 1.83 6910m1Jan.2016 119◦22.902’WRe-deployed 50◦06.826’N 2.0 1.75 6910m8Jan.2016 119◦22.902’WRemoved 50◦06.826’N 2.42 2.19 6910m25Apr.2016 119◦22.902’W813.3.3 CTD and DO Transect DesignCTD and DO profiles during the 2015-2016 field season were collected in WoodLake along one north-south transect of the lake (Figure 3.2). On 4 May 2015, 24CTD profiles were conducted along two transects (along-lake [21] and across-lake[3]) of Wood Lake using a SBE 19plus v2 SeaCAT Profiler. The lake is 6450m longalong the north-south transect, and 1100m across at the mid-point. The across-lakeprofiles were conducted 300m apart, beginning and ending 100m offshore. Forthe along-lake (north-south) profiles, the first (S1) and last (S21) stations werepositioned 100m from the south and north shores respectively, with the remainingstations positioned at 315m spacing (except for stations S20 and S21, at 150m spac-ing). Only the 21 along-lake profiles (Figure 3.2) were conducted on subsequentfield days due to the insignificant lateral variation in the across-lake profiles. Dueto the shape of the valley, predominant winds are oriented along the north-southaxis (longest fetch) of the lake. 47 of these north-south transects were conductedweekly (biweekly on occasion) between 4 May 2015 and 25 April 2016. The lat-itude and longitude of each profile was recorded. With the exception of a thin icemass that prevented deployment of the CTD at precisely the correct coordinates fortwo stations on 1 January 2016 (S19 and S20: 24m and 44m east of intended lo-cations respectively), and missing two casts (S4 and S13) on 21 July 2015 and onecast (S8) on 25 November 2015, all CTD profiles were always conducted within5m of the same location on all field days.820 500 mS1 S2 S3 S4 S5 S6 S7 S8 S9 S10S11S12S13S14S15S16S17S18S19S20S21CTD and DO Sample StationCTD only (no DO) Sample StationNFigure 3.2: Wood Lake CTD and DO sample stations.3.3.4 CTD ProfilesThe SBE 19plus v2 SeaCAT Profiler measures conductivity, temperature, and pres-sure, and is capable of making measurements in depths up to 600m. The unit isequipped with three auxiliary sensors: a WET Labs ECO fluorometer sensor (fluo-rescence converts to measurement of chlorophyll-a during processing), a SatlanticPAR sensor, and a Sea-Point turbidity sensor. The CTD is lowered into the wa-83ter at 0.5m/second on a rope in “profiling mode” while running continuously toacquire samples at 4 Hz (4 scans/second) and obtain vertical profiles of the watercolumn. The unit stores and transmits averaged data. It contains an internal mem-ory capable of recording 400 hours of data. The CTD unit is capable of measuringtemperatures between -5 - 35◦C, conductivity between 0 - 9 S/m and pressure from0 - 600m. A real-time clock records the time of measurements, accurate to within+/- 1 min/year. The data is transferred to a field computer using an input/output(I/O) cable connected to the computer’s serial port, and is subsequently processedusing Seasoft v2 software (Sea-Bird Electronics [SBE], 2013). The SBE profileris extremely accurate (with high resolution) and stable (Table 3.4). The pressuresensor was recalibrated at regular intervals during the field season given the lake’selevation above sea level (391m amsl) and fluctuations in air pressure during theyear. The Gibbs-SeaWater (GSW) Oceanographic Toolbox v3.05 of TEOS-10 forMATLAB [Thermodynamic Equation of Seawater - 2010] was used to evaluatethe thermodynamic properties of the lake water (uses the International Associationfor the Properties of Water and Steam Formulation [IAPWS-09 for pure water andIAPWS-08 for the saline part]) (IOC & IAPSO, 2010; McDougall & Barker, 2011).This toolbox was utilized to derive accurate freshwater depth, salinity, and densityvalues for each data point from every cast from measured specific conductivity,temperature, and pressure.Table 3.4: Accuracy, stability, and resolution of SBE 19plus v2 SeaCAT Sen-sors (SBE, 2013).Parameter Accuracy Stability (drift) ResolutionTemperature (◦C) +/- 0.005 0.0002/month 0.0001Conductivity (S/m) +/- 0.0005 0.0003/month 0.00001 in fresh waterPressure (m) +/- 0.1% of full scale range 0.1% of full scale range/year 0.002% of full scale rangeA WET Labs ECO Fluorometer sensor determines chlorophyll-a concentra-tions based on a relationship between flouroescence and chlorophyll-a concentra-tions. This unit samples at 4Hz and can store up to 108,000 samples (14 bit dataresolution). It can measure chlorophyll-a concentrations over a range of 0 - 125µg/L with a sensitivity of 0.016 µg/L. A scale factor is determined based on signalmeter output with black tape over sensor (dark counts). This scaling factor is used84to determine the actual chlorophyll-a concentrations from raw signal output of thefluorometer (Wet Labs, 2014). Measurements were compared to those from BC-MOE and LAC throughout the water column at similar measurement locations, andalso with values obtained using a HANNA multimeter probe (November-December2015), to ensure accuracy.A Seapoint Turbidity Meter determines turbidity by measuring scattered lightfrom suspended particles within 5 cm of the sensor window (uses dual 880 nm lightsources and dual silicon photodiode detectors with filters to block visible light).Two control lines are used to determine the sensitivity of the meter by adjustingthe range and resolution. With a cable gain setting of “100X”, the sensitivity of theunit is 200 mV/FTU and the range is 0 - 25 FTU. Seasoft V2 calculates turbiditybased on a scale factor, cable gain setting, and voltage output (Seapoint SensorsInc., 2013).A Satlantic PAR sensor is also connected to the CTD. PAR is radiation withwavelength between 400 and 700 nm, is used by phytoplankton and other plants forphotosynthesis (Horne & Goldman, 1994; Wetzel, 2001). The rate at which lightdiminishes with depth is an important variable in constructing an energy balancemodel for Wood Lake. It is typically measured as Photosynthetic Photon FluxDensity (PPFD) with units of quanta (photons) per unit time per unit surface area(i.e., micromoles photon m-2s-1). The sensor continually samples and transmitstelemetry at 4Hz (Satlantic, 2012).3.3.5 Dissolved Oxygen ProfilesBeginning on 21 July 2015, DO was measured at every odd numbered station(1,3,5. . . 21) until 16 December 2015. From 21 July 2015 to 28 October 2015a YSI Pro-20 Dissolved Oxygen sensor was used to record DO (resolution: 0.1mg/L, accuracy: greater of +/-0.2 mg/L or +/-2%, range: 0 - 50 mg/L (YSI aXylem Brand, 2012)) at 0m-5m-8m-10m and at 0.5m intervals below 10m through-out the water column. This unit was generously provided by the OCEOLA fish andgame club. From 5 November 2015 to 16 December 2015 a HANNA HI 9828multiparameter-meter was used to measure DO (resolution: 0.01 mg/L, accuracy:greater of 0.1mg/L or 1.5%, range:0-50 mg/L (Hanna Instruments, 2008)) at 0m-855m-8m-10m and at 1m intervals below 10m throughout the water column. Thismeter was generously provided by Larratt Aquatic Consulting (LAC). Both unitswere calibrated in the field prior to deployment at each station. The YSI meterrequires manual recording of data, while the HANNA meter has a built in memory.All DO measurements were recorded in a field book.Note that the HANNA meter also records chlorophyll-a measurements, whichenabled cross-referencing with values measured by the SBE CTD to ensure thefluorometer readings were consistent.3.4 Processing Field DataFollowing field data acquisition, several important post-processing techniques wererequired in order to understand the important thermodynamic processes in WoodLake, as well as to facilitate modelling the lake with DYRESM and comparingfield data with modelled output. These techniques involved developing a spatiallyaveraged temperature profile for Wood Lake on each field day, establishing a com-mon datum between field data and DYRESM, and determining the light extinctioncoefficient in Wood Lake.3.4.1 Spatially Averaging Temperature Profiles for Each Field DayDuring the field season, it was realized that there was significant lateral variabilityin Wood Lake, particularly following days of steady or gusting north and/or southwinds, as a result of seiching in the lake. Other factors believed to have played arole in the lateral distribution of heat in Wood Lake that were particularly notice-able in the winter season when temperature profiles were near isothermal include:the bathymetry of the lake (i.e. shallower slope on northern end than on south-ern end), groundwater inflows, inflow from MVC, and exchange with KalamalkaLake. As a result of this lateral variability, no one single profile was representa-tive of the thermal structure of the lake on all field days. For example, consider23 September 2015 (Figure 3.3 and 3.4). Station S9 was located at 50◦04.545’N,119◦23.441’W, and was the sample location closest to BC MOE station 500848(120m NE of 500848 along a 35.5◦heading). This one profile does not serve as anaccurate representation of the temperature distribution in the lake and would pro-86vide an erroneous overestimate of the heat content of the lake. Similar observations(not shown) were made on other field days when comparing a single profile at otherlocations (i.e. S10, which was the deepest sampling location, and S11, which wasthe station closest to the geographic centre of the lake) to the overall distribution oftemperature in the lake.As a result of this lateral variation, a spatially averaged profile was developedfor every field day (similar to BC Research (1974)) that was derived as an av-erage of the five deepest CTD casts in the middle of lake (i.e. S9-S13). Thesefive profiles were used in this process because they were the deepest central fivecasts providing temperature data throughout the water column to the lake bottom(nearest to the deepest location), and the calculated spatially averaged profiles con-sistently provided the best estimate for the lake’s average temperature prAofileson all field days. Field data observations (i.e. Figure 3.3 and 3.4 and 4.10 (Sec-tion 4.3.2) and 4.43 (Section 4.9)) and past research (Wiegand & Chamberlain,1987) show that first mode seiche dominates in Wood Lake (i.e. with the nodenearest to the lake center). Temperature data from each profile was averaged intovertical 0.20m depth bins and then these 0.2m depth bins were averaged laterallyacross the five selected profiles. The resulting profiles were always compared to allfield data from the same field day (i.e. Figure 3.4) to ensure they provided a robustestimate of the average temperature distribution in the lake. This provided betterconfidence in calculating an estimate of the heat content of the lake on each fieldday and in comparing field profiles to DYRESM modelled outputs. The remain-ing 16 profiles, although not used to compute the spatially averaged temperatureprofile on each field day, enabled understanding of how the lake was behaving andresponding to wind forcing (i.e. Figure 3.3 and Figure 4.10 in (Section 4.3.2)).87Figure 3.3: Temperature contours in Wood Lake on 23 September 2015. Dis-tance along x-axis is measured from south to north along the long-axisof the lake (Figure 3.2).884 6 8 10 12 14 16 18Temperature [° C]05101520253035Depth [m]All Temperature [°C] Profiles on 23 September 2015Temperature [°C] Profile at Station S9Average Temperature [° C] vs. Depth [m] for Casts S9-S13Figure 3.4: Spatially averaged temperature profile (S9-S13, solid line) com-pared to all overlaid temperature profiles and S9 (dashed line) on 23Sept, 2015.3.4.2 Bathymetry and Establishing a Common DatumThe current study required a new bathymetric survey, since past bathymetric surveydata of Wood Lake provided insufficient accuracy and precision for a hypsograph(cross sectional area vs depth) (Figure 3.6) of the lake to use with DYRESM as wellas to plan thermistor chain deployment locations and depths. A bathymetric surveyusing a dual-beam sonar was conducted on Wood Lake by Raphael Nowak overthe course of three days (24 July, 31 July, and 7 August 2015) when weather andboat traffic conditions permitted. The sonar used enabed bottom coverage equalto depth in either direction perpendicular to the transect. A total of 68 east-westtransects were made on the lake, with an additional 28 east-west transects in thevicinity of MVC, and two north-south transects through the Oyama Canal. Thecontour depth interval attained was 0.5m (Figure 3.5). This data was then analyzedusing the software package “ReefMaster” in order to generate a hypsographic tableof depth-area-volume values for the lake (Raphael Nowak, pers. corr., 17 August892015).Figure 3.5: Bathymetric chart for Wood Lake (0.5 m intervals from 0 - 33m)(survey conducted by Raphael Nowak, 2015).900 5 10 15 20 25 30 35Height (m)012345678910Cross-Sectional Area (m2 )×106Hypsographic Curve for Wood Lake from Bathymetric DataExtrapolated Height-Area Values for Heights over 33.18mFigure 3.6: Hypsographic curve for Wood Lake showing minor extrapolationof cross-sectional area for heights above 33.18m (above lake bottom atdeepest location).It is necessary to determine the maximum lake depth on each field day in or-der to establish a common datum between field work temperature profiles (mea-sured from surface down to lake bottom) and DYRESM temperature profile outputs(measured from lake bottom at deepest point in lake to lake surface). Thus, knowl-edge of the maximum depth in the lake on each field day, allows depth-variable datapoints from CTD profiles to be converted to height-variable data points measuredfrom the lake bottom, which then facilitates direct comparison of DYRESM outputprofiles with CTD field profiles. During the period at which the bathymetric sur-vey was conducted, the maximum lake depth was 33.18m. A ratio was developedbetween this maximum depth measured (7 August 2015) and the average depth ofthe four deepest CTD casts measured on 4 August 2015 and 11 August 2015 (twofield days nearest to 7 August 2015). This ratio (average depth of four deepestcasts to maximum lake depth) could then be used to estimate the maximum depth91of the lake on any field day given the average depth of these four casts each day.This method was used prior to installing the HOBO level logger on 30 June 2015.Using this method, the maximum estimated lake depth for the year based on CTDcasts was 33.8m on 4 May 2015 (Figure 3.7). Very little extrapolation was requiredfor cross-sectional area values at depths that exceed the maximum depth measuredin July/August (33.18m, Figure 3.6) (i.e. accounts for an increase in lake surfacearea of only 1.07% over that measured on 7 August 2015).The comparison between the estimate of the maximum lake level based onthis predictive ratio method and that based on the HOBO level logger data isquite strong (RMSE: 0.12m) for the time period between 30 June 2015 and 13April 2016. Note the level logger recorded absolute pressure (kPa), which wasthen corrected for gauge pressure by accounting for atmospheric pressure mea-sured at Kelowna Airport and subsequently converted to depth based on temper-ature, salinity, and gauge pressure using TEOS-10. By 4 May 2015 68% of thetotal 2015 MVC inflow volume had entered Wood Lake along with contributionsfrom groundwater inflow, and evapotranspiration was minimal. Therefore, around6.2e6m3 of 9.08e6m3 (total inflow from MVC in 2015) had flowed into Wood Lakefrom MVC at this time, wherein the total MVC inflow only accounts for 4.6%of Wood Lake’s volume and the remaining inflow from MVC in 2015 after thistime was negligible, only accounting for ∼1.5% of Wood Lake’s total volume.Therefore, the maximum lake surface elevation in May is assumed to provide agood estimate of the crest elevation of the lake. The range of estimated maximumdepth values for the lake is 32.70m (10 December 2015) to 33.80m (4 May 2015),which are reasonable given personal observations of the lake level along the shore-line and data from past bathymetric surveys available online and/or from literature(Anonymous, 1974; R. Nordin, 1987; Northcote et al., 1974; NAVIONICS mapavailable at http://webapp.navionics.com). More information regarding convertingCTD depth-parameter values to height-parameter values is in Appendix A.Given this analysis, the maximum lake depth on 29 May 2015 was estimatedand an initial profile could be constructed for DYRESM based on the spatially av-eraged temperature profile from 29 May 2015, converted into height-temperaturedata points. Water density was dictated by temperature, and since inflows and out-flows were negligible (Section 4.6), salinity was held constant spatially averaged92value of 0.16 PSU. Field data showed negligible variation in Wood Lake salinityduring the year (not shown). Likewise, this process was done for all field days forcomparison with DYRESM outputs.04-0501-06 01-07 01-08 01-09 01-10 01-11 01-12 01-01 01-02 01-03 01-04Day [dd-mm]30.8531.0531.2531.4531.6531.8532.0532.2532.4532.6532.8533.0533.2533.4533.65Depth (m)Average of max. depths from S8 - S11 (m)Est. max. depth using ratio of S8-S11 & measured max depth on 7 August 2015 (m)Est. max. depth (m) based on HOBO level loggerFigure 3.7: Estimated maximum lake depth in Wood Lake during 2015 - 2016based on conversion factor (ratio from average depth of S8-S11, alsoshown) and from HOBO level logger.3.4.3 Light Extinction CoefficientThe PAR attenuation coefficient as light passes through the water column (Kd(PAR))is an important parameters for lakes in terms of the total energy available for photo-synthesis (Smith, 1968), and is used herein as a best estimate of the vertical atten-uation in light energy available for heating the lake (Appendix B). This parameteris a required model input parameter in DYRESM. Irradiance is measured using anauxiliary Satlantic PAR sensor attached to the CTD unit. This parameter indicatesthe amount of light available for photosynthesis and radiation transferred throughthe water column (Kirk, 2010). This quanta meter (“cosine collector”) measuresradiant flux per unit area without differentiating between photons of wavelength93400-700nm (PAR) and regardless of angle of incidence because the photosyntheticusefulness of light is a function of quanta flux rather than energy flux (independentof wavelength) (Kirk, 2010).The Beer-Lambert Law for the vertical distribution of downwelling irradianceis given by the equation 3.1 (Lund-Hansen, 2004).Iz = Ioe−Kd(PAR)∗z (3.1)where Io is the irradiance measured just below the lake surface [µmol photonsm-2s-1], Iz is the irradiance at depth (z) (where z is in [m]), and Kd(PAR) is thelight attenuation coefficient (Lund-Hansen, 2004). The light extinction coefficientKd(PAR) [m-1], or the vertical attenuation coefficient for downwards irradiance, isestimated from a linear regression of the natural logarithm PAR data (ln(PAR)) ver-sus depth (z) (R. J. Davies-Colley & Vant, 1988; Hingsamer, Peeters, & Hofmann,2014; Lund-Hansen, 2004; Vant & Davies-Colley, 1984). This Equation 3.1 canbe rewritten (Kirk, 2010) as Equation 3.2 and 3.3.ln(Iz) =−Kd ∗ z+ ln(Io) (3.2)−ln(Iz/Io) = Kd(z) (3.3)Therefore, Kd can be estimated as the slope of the line given by −ln(Iz/Io) vs.Depth (z) for each PAR profile (i.e. the linear regression coefficient of ln(Iz) withrespect to depth) (Kirk, 2010). The PAR sensor is located 0.57m above the pres-sure sensor on the CTD unit. Therefore, for each cast, PAR values for depths above0.6m are disregarded, and the first PAR value recorded at depth below or equal to0.6m is used as Io (referred to here as PARo). Although Kd(PAR) is not con-stant with depth, the variation is typically insignificant. It has been recommendedthat Kd(PAR) estimation is restricted to the euphotic zone, where light intensityis sufficient for photosynthesis to occur (Kirk, 2010). The bottom of the euphoticzone is defined as the depth at which the irradiance is 1% of that at the surface(R. J. Davies-Colley & Vant, 1988; Kirk, 2010; Vant & Davies-Colley, 1984).There are three major potential sources of error in these measurements: waves,94variation in surface-incident flux due to changes in cloud cover, and disturbanceto the light field due to the boat (Kirk, 2010). The effects of the former two arediscussed in (Appendix B). To avoid effects of the boat shadow on PAR measure-ments, care was taken in deployment to always attempt to lower the CTD on thesame side of the boat as incident sunlight. A relatively small vessel was used in thisstudy, which helps minimize this effect. The maximum relative fluctuations due towaves occur between the surface and approximately 2m depth, and then decay al-most exponentially below this depth (Dera & Gordon, 1968). The upper portion ofeach PAR profile where these fluctuations occur is ignored in order to minimize theconsequences of wave effects on light extinction (Kirk, 2010). In order to accountfor the effects of clouds, typically one would monitor the incident solar flux contin-uously on deck or at some fixed depth below surface with a reference PAR sensor,and then adjust the simultaneously recorded PAR data at depth as the sensor de-scends through the water column (Kirk, 2010). A secondary reference sensor wasnot used in this study. However, this effect can be assumed minimal since the CTDis lowered at approximately 0.5m/second, requiring only 18-30 seconds to traversethe depth interval of interest; the euphotic depth is typically between 9 - 15m, basedon the calculated light extinction in Wood Lake (Figure 4.2 in Section 4.1). Duringthis time interval, changes in cloud cover are assumed negligible.To further minimize the effects of boat shadow, waves, and cloud cover onthe estimation of the light extinction coefficient for each field day, 19 PAR pro-files from S2 through S20 were used to obtain a spatially averaged light extinctioncoefficient across the lake (Figure 4.2 (Section 4.1). Rather than selecting an ar-bitrary upper depth (i.e. 1.5 - 2m) for the depth interval of interest to avoid theshading effects of the boat and effects of waves (Dera & Gordon, 1968; Hingsameret al., 2014)), an alternative method was employed to eliminate these effects us-ing MATLAB. Firstly, the PAR data for each separate profile was averaged into0.25m depth interval bins. Subsequently approximate differentiation was used todetermine the minimum depth at which the depth-averaged PAR values (for eachprofile separately) were continuously decreasing with depth. The minimum valuebetween this depth value and 2m (depth below which wave effects typically decayexponentially with depth (Dera & Gordon, 1968) was then selected as the upperlimit for the depth interval to evaluate Kd(PAR) for each profile (Figure 3.8).950 1 2 3 4 5-ln[(PAR)/(PARo)]024681012Depth [m]-ln[(PAR)/(PARo)] From Station S12 on 30-June-2015Upper limit for evaluating linear regressionBottom of Photic ZoneLinear RegressionFigure 3.8: −ln(Iz/Io) vs. Depth (z) and linear regression of PAR profilesfrom one PAR profile from S12 on 30 June 2015.The lower depth of interest for evaluating Kd(PAR) is the bottom of the eu-photic zone (R. J. Davies-Colley & Vant, 1988; Kirk, 2010; Vant & Davies-Colley,1984). The average euphotic depth for the entire lake on a given day is calcu-lated by evaluating -ln(0.01)/Kd(PAR) (R. J. Davies-Colley & Vant, 1988) whereKd(PAR) is the average Kd(PAR) for all profiles taken on that day. Therefore, eval-uation of each of these quantities relies on knowledge of the value of the other. Amethod was employed to evaluate Kd using an iteration process in which the initiallower depth of interest was initially estimated as 15m. The average Kd(PAR) valuefor all 19 profiles on a given field day was evaluated by using 15m as the lowerlimit, and then this Kd value was used to determine a corresponding lower limit forthe euphotic zone (by evaluating -ln(0.01)/Kd) (R. J. Davies-Colley & Vant, 1988).This calculated euphotic depth was then used as the lower limit for evaluating a96new average Kd value for all 19 profiles on that day, and the process was repeateduntil the input lower limit value was equal (within 0.1m) to the calculated euphoticdepth. This Kd(PAR) value was then taken as the light extinction coefficient for thelake that day. This was an important exercise for the current study in order to en-sure a high level of accuracy for the monthly averaged light extinction coefficientfor the DYRESM model of Wood Lake. It was realized during preliminary trialsthat the modelled results were quite sensitive to this parameter (not shown).This evaluation of the euphotic depth relies on the assumption that Kd is ap-proximately invariable with depth below the initial ∼2+/- meter surface layer,which is deemed to be a reasonable assumption in most relatively turbid inlandwaters. Note that this assumption could lead to an underestimate of the euphoticdepth in very clear water columns wherein Kd(PAR) has a notable biphasic nature(Kirk, 1994, 2010) (Appendix B).3.5 Dynamic Reservoir Simulation Model (DYRESM)ConstructionWood Lake has been modelled using DYRESM (v4), a 1-D hydrodynamics modelfor lakes that models and simulates changes in vertical temperature, salinity, anddensity distributions. The model yields quantifiable predictions of the lake’s ther-mal structure over periods of weeks to decades, enabling analyses of seasonal/an-nual variability, and conducting sensitivity testing to changes in climate and/or hy-drology. The model can also be coupled with CAEDYM (Computational AquaticEcosystem Dynamics Model) to additionally study lake biology/chemistry (Imer-ito, 2015b). CAEDYM was not implemented in this study. The program waswritten in Fortran 95 and was developed by the Centre for Water Research (CWR)at the University of Western Australia. Program input provides the static and forc-ing data over the simulation period and can be configured using the graphical userinterface (GUI) or via text files (Imerito, 2015b). The names and contents of theinput text files required by DYRESM (as discussed by Imerito (2015b) (includingsome specific details related to the current study) are:1. “DYRESM Configuration File” for Simulation: simulation start day, dura-tion, output interval, light extinction coefficient, minimum and maximum97layer thickness, time interval, and output selections.2. “Physical Data and Lake Morphometry File”: latitude, elevation (amsl),number and type of inflows, streambed half angle, slope and drag coefficient(assumes CD = 0.015), lake bottom elevation, number of outlets, outlet eleva-tions, and lake morphometry (matrix of hypsographic curve data). RaphaelNowak conducted a new bathymetric survey of Wood Lake in 2015; thesoftware package “ReefMaster” was used to generate a hypsograph of WoodLake at 0.5m depth intervals (Figure 3.5, Section sec:Bathy). Given the neg-ligible net inflow and outflow to the lake observed from May-2015 untilDecember-2015 (Section 4.6) the lake was modelled as a closed system withno inflows or outflows. However the user must still specify at least one in-flow and at least one outflow into the model, and then set the flows to zero inthe “inflow” file.3. “Initial Profile File” At Beginning Of Simulation: initial values of ice/s-now depth, vertical temperature and salinity profiles. Variable values areindicated at different heights from the lake bottom, corresponding to theheight of the top of each given initial “layer”. The initial profile was takenas the spatially averaged temperature profile from 29 May 2015, averagedinto 0.2m depth layers. Given that inflows/outflows were neglected, and ex-change of salinity with the boundary layers (surface/sediment) is ignored, thesalinity profile was held constant at an average uniform value throughout thewater column (0.16 PSU). Field data showed negligible change in salinityprofiles throughout the year, and water desnity was dictated by temperature.4. “Meteorological Data” File: Meteorological data over the simulation pe-riod in the form of daily averages. The time scale used for averaging mete-orological data is used for all forcing variables. Values averaged over eachtimestep include: short wave [SW] radiation flux [W/m2], fraction of CC,air temperature [◦C] (average of max/min for daily time step), windspeed[m/s], rainfall/snowfall [m] (total over timestep), and vapor pressure [hPa](estimated from average relative humidity and average air temperature usingformulae provided by Imerito (2015b). Incident LW radiation [W/m2], net98LW radiation [W/m2], or CC fraction [0,1] can be entered to account for longwave radiation fluxes. In this case, daily average CC (Kelowna Airport datarepublished by CedarLakeVentures (2015)) was entered. The model esti-mates LW radiation based on CC and water surface temperatures throughoutsimulation.5. “Stream Inflow” File: Daily average inflow data: volume, temperature, andsalinity (as well as chemical/biological water quality data if using CAEDYM)for period of simulation for each inflow stream. Inflow volume, temperature,and salinity were set to zero for each time step in this model.6. “Withdrawal” File: Daily withdrawal volume [m3/day] for the simulationperiod for each outflow; this parameter was also set to zero for the timeperiod of interest given the nature of the exchange of minimal flows betweenWood Lake and Kalamalka Lake.7. “Parameter” File: Contains simulation parameters (pseudo-constants) thatare adjusted within certain limits to enable model to best represent the lakeof interest, and time of day of output (seconds after 00:00AM).8. “Field Data” File (optional): Measured field data includes same measuredvariables on each day (can have different depth resolutions) (Imerito, 2015b).MATLAB was used to analyze modelled results and compare to field data.Selection of model parameters is discussed in detail in Appendix A.3.5.1 Inflows and OutflowsWood Lake was modelled as a closed system in 2015 (Section 4.11.2), and like-wise validated in this manner for 2013 and 2014 (Section 4.11.4). However, giventhat the management scenarios (Section 4.12.1) for Wood Lake involve increasinginflows and/or outflows for Wood Lake, then DYRESM’s management of theseprocesses must be validated (Section 4.11.5) and understood.The user inputs values for the inflow volume (Q), temperature (T), and salinity(S), used to determine inflow density. The user also defines the drag coefficient99of the river bed, the slope angle of the inflow, the stream half-angle, and the en-trainment coefficient. The entrainment coefficient is set at 0.083 (H. B. Fischer,List, Koh, Imberger, & Brooke, 1979). DYRESM allows the user to calculate adifferent entrainment coefficient (E) at the beginning of each simulation, followingequations presented by Dallimore et al. (2001) and (H. B. Fischer et al., 1979), asdiscussed by Imerito (2015a). The inflow pushes water ahead of it in the reservoiruntil the buoyancy forces eventually stop the flow. The density of the inflow iscalculated by DYRESM using the UNESCO equation of state (Imerito, 2015a). Ifthe inflow is less dense than the surface layer then it will float across the lake’s sur-face (inflow volume is added to top layer and new properties for the surface mixedlayer are determined). If the inflow channel is a defined drowned river valley (sidesconfine flow), and the inflow is more dense than the ambient surface water then itwill plunge down into the water column fairly uniformly along a visible “plungeline” in a manner that is relatively 1-D. As the inflow proceeds into the lake, it willflow along the plunge line (determined by inflow and lake water properties) andentrain lake water from each layer in the water column (Imberger and Patterson(1981) and H. B. Fischer et al. (1979) as discussed in Imerito (2015a) based onconservation of volume, causing its density to decrease until it is neutrally buoyant(or reaches basin bottom). Once neutral buoyancy is achieved, then a layer is gen-erated in the water column that is occupied by this volume. The inflow is insertedinto a number of existing layers in the lake at various heights, and if these layervolumes become too large, then new layers are subsequently formed as necessary.As these layers grow in thickness, those above are pushed up and decrease in thick-ness in order to maintain their volume-height relationship. These processes occurinstantaneously in the model at each time step. DYRESM then re-amalgamates thelayers to maintain the layer limits set by the user (Imerito, 2015a).The model also allows the user to specify subsurface inflows. In this case theflow is either neutrally dense or denser than the reservoir water at the height of theinflow (in which case it behaves in a similar manner to that discussed above), orit is more buoyant than the ambient lake water in which case it is modelled as asingle-phase plume. The buoyancy flux and corresponding flow rate of the plumeand entrained water is computed by DYRESM following equations of H. B. Fischeret al. (1979). The inflow is inserted into a layer of neutral density in the lake or100when the plume reaches the surface. If it reaches the surface, then it forms a newsurface layer containing only plume water (Imerito, 2015a).DYRESM allows the user to specify withdrawals (height and volume) and thecrest height (which determines overflow). For overflow, any layers above the crestheight are removed down to the height of the crest. For outlets, water is removedfrom the layer at the specified height of the outflow until the given layer volumeis consumed, at which point water is extracted from the next layer up until Qoutletis satisfied. The layers are then re-structured. DYRESM outputs the water prop-erties for the resulting water column and the withdrawal aliquot volumes for eachtime step during the simulation (Imerito, 2015a). Outflow from Wood Lake occursthrough the Oyama Canal to Kalamalka Lake. Estimates of monthly flow ratesthrough the canal have been provided in past studies with data from BC Research(1974) and BCWIB (1974); however, the flow rates are minimal. Irrigation out-flows (unknown locations/volumes) directly from Wood Lake have ben considerednegligible and thus neglected in this model, although these could also be specifiedaccording to depth and volumetric flow rates.3.5.2 Lake Surface Boundary Layer Fluxes and ExchangesLake surface boundary-layer fluxes and exchanges (heat, mass, and momentum)provide the energy in this model for heating, mixing, and stratifying the lake. Thesecomprise shortwave [SW] radiation, latent heat (evaporation), sensible heat, LW ra-diation, and wind stress. All variables except SW insolation and wind speed are as-sumed in DYRESM to take on an average uniform value throughout the day. Windspeed can be uniform or sinusoidal during the day (maximum occurring aroundmidday) (Imerito, 2015a). A uniform distribution was used in the current model,based on an average daily wind speed and using a wind factor multiplication factorof 1.5. SW radiation was measured directly at Coral Beach Farms weather station(50◦05’08.50“N, 119◦22’13.37”W). LW radiation (>2800 nm) can be measureddirectly (Imerito, 2015a); however, in this case it is determined as a function of CC,air temperature [Ta], and relative humidity. CC data was obtained from KelownaInternational Airport, compiled and downloaded from CedarLakeVentures (2015).The sub-daily time-step employed must lie between 10minutes (600 sec) and three101hours (10800sec) (Imerito, 2015a); a timestep of 30minutes (1800sec was em-ployed). SW albedo was defined monthly. LW albedo is assumed to be constant(0.03) (Henderson-Sellers, 1986 as discussed in Imerito (2015a). It is assumed that55% of SW radiation is absorbed by the surface layer and 45% (PAR) penetratesdown into the water column according to the Beer-Lambert Law (Equation 3.1) inorder to account for total SW energy per unit area that penetrates each layer andis subsequently converted to heat energy (Equations are based on TVA (1972) andHenderson-Sellers (1986) as discussed by Imerito (2015a).DYRESM accounts for non-penetrative surface thermal fluxes as well as sur-face mass fluxes. Sensible heat flux, latent heat flux, and LW radiation flux are allnon-penetrative (occur at surface). DYRESM automatically calculates these fluxesand accounts for their influence on the water column at the user-specified time stepaccording to well-reviewed and commonly referenced formulae. Equations arebased on TVA (1972), Fischer et al (1979), and Swinbank (1963) as discussed byImerito (2015a). A full review of these formulae can be found in Imerito (2015a).The net LW radiation flux (QLW) (accounting for LW radiation emitted by thelake surface) is estimated based on given atmospheric conditions ([Ta]) and CC.DYRESM also accounts for surface mass fluxes (latent heat mass flux and pre-cipitation) based on the latent heat flux, latent heat of vaporization, lake surfacearea, and total daily precipitation. Daily rain and snow data was obtained fromthe Kelowna MWSO Weather Station, which offered a more consistent record ofrainfall and snowfall data than the Kelowna International Airport Weather Station(Government of Canada. Environment Canada [GCEC], 2016). Any missing datawas filled with data from Kelowna International Airport.Surface momentum flux due to wind is also simulated. The wind data used inDYRESM is a daily average value that has been computed based on hourly averagevalues from Environment Canada for Kelowna International Airport, and modifiedby a wind multiplication factor of 1.5. Only wind values that exceed 3 m/s areconsidered to generate motion in the surface layer (Imerito, 2015a). The wind-driven velocity in the surface later is calculated based on wind speed, the surfacelayer depth, and the time interval. Generally, features around the lake perimeter(i.e. topography) reduce the actual wind stress to a value below that expectedbased on the surface shear velocity, and therefore an effective lake area is used102to compute the transfer of momentum to the surface mixed layer (80% of surfacearea for Wood Lake). Further details are provided in Imerito (2015a). In order forthe model to accurately calculate and resolve surface layer thermodynamic fluxes,DYRESM automatically re-grids the uppermost portion of the water column priorto modelling energy transfers. Note that because meteorological sensors are atKelowna Airport (10-15km south) and not located within the boundary layer overthe lake surface, then the effects of air column stability and water roughness onheat and momentum transfer are neglected (i.e. non-neutral atmospheric stabilityswitch is off) (Imerito, 2015a).3.5.3 Surface Layer MixingSurface layer mixing due to turbulent kinetic energy (KE) is accomplished by con-vection, stirring due to wind action, and shear (transfers energy between adjacentlayers through the water column). Each component of KE input is parameterizedwith dependency on the efficiency of the process (i.e. wind-stirring efficiency,shear production efficiency, potential energy mixing efficiency), layer properties,wind speed, and the model timestep (Imerito, 2015a). Only wind speeds over 3 m/sare considered to impart momentum to the surface layer (the velocity in all layersis zero before this). DYRESM calculates surface mixed layer (SML) velocity andshear velocity based on formulae from H. B. Fischer et al. (1979), which continuesuntil the shear period is reached (i.e. duration that shear builds in SML). If windsover 3 m/s persist for longer than the shear period, then all layer speeds are resetto zero at the end of the shear period and the algorithm begins again. Layer ve-locities are altered as the layers amalgamate and mix momentum vertically downthrough the water column. The shear period used by DYRESM is the minimumof: 7 days (Tlimit); the period based on the coriolis effect (Tcor = Nd /sin(latitude))where Nd is the number of seconds in a day (i.e. Tcor ∼ 31 hours for Wood Lake);and one quarter of the internal seiche period (Ti/4 = AN0.5/8c) where AN is thelake surface area (9.14e6m2) and c is the phase speed of the uninodal seiche (i.e.∼3 hours for Wood Lake). The surface layer is re-gridded before these processesoccur in order achieve a higher resolution in the SML and permit a higher degreeof thermodynamic accuracy (Imerito, 2015a).103Mixing is completed first by convection and wind stirring, and subsequentlyby shear. At the beginning of each time step if the potential energy required formixing the SML with the layer beneath it is negative (i.e. unstable profile) thenthese two layers are mixed to form a new SML. This process continues downwarduntil the profile is stable. The sum of the convective KE (potential energy releasedwhen dense water descends) and wind stirring KE (momentum transferred to SMLby wind) act to deepen the surface layer until the available KE is insufficient tomix the new SML with the layer beneath it. At this time, DYRESM computes themomentum of the new SML and the resulting KE transfers momentum betweenupper and lower layers via shear, which continues deepening the SML until theavailable KE is insufficient to further deepen the SML. Any remaining energy iscarried forward to the following timestep. If the bottom is reached before the KEis consumed then any remaining energy is lost. Conservation laws (temperature,salinity, energy, momentum) are always withheld (Imerito, 2015a).After each timestep, the recalculated height and density values of the SMLand the hypolimnion are carried forward to the next timestep. To facilitate theoption of using CAEDYM to model water quality, the velocities in the surfacemixed layer and hypolimnion are calculated by DYRESM. The bottom stress forthe surface layer (epilimnion) and hypolimnion can then be determined (used forre-suspension in bottom layer) (Imerito, 2015a).3.5.4 Deep Water Column (Hypolimnion and Metalimnion) MixingDYRESM also simulates deep mixing throughout the water column below theSML. Strong density gradients in the metalimnion (depth determined by the mixedlayer algorithm) stabilize the water column and impede vertical mixing. However,mixing still occurs below the SML due to internal wave interactions, local shear(i.e., combining long and short internal waves and Kelvin-Hemholtz billowing),and gravitational overturning due to wave energy absorption in certain layers).Given the complexity and limited current state of knowledge with regards theseprocesses, DYRESM uses a parameterization approach and evaluates deep mixingas having two components:1041. Internal mixing: via molecular diffusion and shear mixing (a calculated vol-ume aliquot is exchanged between each layer from the bottom to the surface).2. Benthic boundary layer (BBL) mixing: KE available in BBL is a functionof shear velocity along sediment interface. A pseudo two-dimensional BBLstructure is generated wherein the fluid entrained into the BBL is separatedfrom the remaining water in the water column (i.e. each layer consists of aBBL cell and an internal cell).These processes are triggered in the model at midnight each day only if thewater column is not fully mixed. The volume of water transferred between adjacentlayers is based on formulas presented by Imberger and Patterson (1990) and Yeatesand Imberger (2003) as discussed in Imerito (2015a). Maximum internal mixingoccurs in the thermocline, as governed by the vertical mixing coefficient (Imerito,2015a).The volume flux between adjacent layers involves a benthic boundary layerexchange and an internal exchange. The volume exchange between internal cellsand between benthic boundary layer cells from the lake bottom up to the surface(layer by layer) simulates vertical (up and down) fluxes due to turbulent mixing inlakes. Cell volumes and the layer structure are unaltered during this process. Theheight of the BBL is updated at the beginning of each day using a parameteriza-tion approach based on the energy available at the boundary (shear stress) and theamount of water that can be entrained into the BBL. The bottom shear velocityis a function of the bottom drag, and an estimation of the hypolimnion velocity(based on surface velocity and continuity laws). The growth of the BBL is offsetby a constant eddy diffusivity (“BBL dissipation coefficient” or “BBL detrainmentdiffusivity”) at the top of the BBL that detrains water from the BBL (Gloor et al.,2000 as discussed in Imerito (2015a)). The default BBL dissipation coefficientprovided by CWR is 1.4e-5 (Imerito, 2015a); however, a value of 1.0e-6 was usedfor Wood Lake, which enabled the model to better simulate these BBL processes asvindicated by the thermal regime (compared to field data). For further informationon these mixing processes and how DYRESM parameterizes these processes, thereader is directed to Imerito (2015a).1053.5.5 Modelling Adjustments and Specifications for Wood LakeWithout having access to DYRESM’s source code, several methods were employedin order to optimize DYRESM for modelling Wood Lake and to facilitate moreaccurate comparisons with field data. On the first day (29 May 2015), the mid-point time for measuring the temperature at the middle five stations was 12PM.DYRESM assumes the initial profile is taken at midnight on the start of the firstday of the simulation. Therefore, in order to ensure model accuracy and not add anadditional 12 hours of meteorological forcing to the input data, the first segment ofthe model involves running a simulation on the first day for 12 hours and using thefinal output as the initial temperature profile for the following day (i.e. 30 March2015). Following this, due to the fact that the surface albedo and light extinctionin the lake were both observed to vary throughout the season (Section 4.11.1 and4.1 and Appendix C), the model is broken into monthly segments. Each segmenthas a different surface albedo and light extinction value. Therefore, from this point(i.e. 30 March 2015) forward, the model must be run through twice. The first sim-ulation involves having the model output profiles every day at midnight; at the endof each monthly segment, the final model output is then used as the initial profilefor the following monthly segment (i.e. midnight on the first day at the start of thesegment). This process is done for each month until the last month of interest (i.e.December). Subsequently the model is run through a second time using the initialprofiles for each month segment derived from the first simulation run. However,during the second run, the model output timestamp is based on the average timeof day at which the middle five profiles (that comprise the spatially average tem-perature profile) were obtained in Wood Lake (i.e. May/June: 11AM, July/Aug:13PM, Sept: 11AM, Oct/Nov: 12PM, Dec: 11AM). This facilitates more accuratecomparison with field data. During the second run, all other parameters remain thesame, and only the output timestamp differs.Several assumptions were necessary with regards to model parameters for modelvalidation in 2013 and 2014. For model validation (2013/2014) the salinity inWood Lake was held at a constant value (density controlled by temperature) forthese simulations. Meteorological data was obtained from the same sources andprocessed in the same manner as 2015. The maximum lake depth values from 2015106were assumed to be the same for 2013 and 2014 for each month, since these valuesare unknown and 2015 estimates provide the best available estimate for 2013/2014.Since the lake did not marl in either 2013 or 2014, the albedo for each month wascalculated based on the average value for the month as derived from the lake lat-itude and solar zenith angle (Imerito, 2015a). The light extinction coefficient foreach month was held constant at an average value outside of the period of marlingfrom 2015 (i.e. 0.33m-1). The timestamp for the output of results was taken as12PM.For modelling management scenarios (Section 4.12.1) and investigating the ef-fects of perturbations in daily/monthly/seasonal average air temperatures on WoodLake’s thermal structure (Section 5.1.1 to Section 5.1.3), 2015 field and model datawas used as a baseline. Although 2015 did not yield a severe temperature-oxygensqueeze in Wood Lake (compared to 2011), the quantity, accuracy, and precisionof field data available, as well as the accuracy with which DYRESM was able toreplicate the thermal structure of Wood Lake in 2015, lends confidence to usingthis year as the baseline for comparing and evaluating future modelled scenarios(Section 4.12.1) by eliminating additional sources of uncertainty.107Chapter 4ResultsThis chapter begins with a description of field observations made between 4 May2015 and 25 April 2016, including water clarity and light extinction (as derivedfrom PAR data), chlorophyll-a, and progression of the thermal structure and dis-solved oxygen structure (21 July to 16 December 2015) of Wood Lake. The conse-quential physical (temperature-DO) limitations on kokanee habitat are discussed.An analysis of the hypolimnetic oxygen deficit for this period is provided in Sec-tion 4.5. Following this, low flow data from MVC (only major tributary) and fieldobservations of the temperature gradient through Oyama Canal are provided tocharacterize Wood Lake with regards to its mass and/or thermal fluxes at theseboundaries. Analyses of a time series of the calculated heat content as well as epil-imnion and hypolimnion heating rates are discussed in Section 4.9 and 4.10. Thischapter concludes with a comparison of simulated 2015 DYRESM model resultswith field data (in terms of temperature profiles and total heat content estimates),model validation using 2013 and 2014 field data, and the application of this vali-dated model to simulate a variety of inflow and outflow management scenarios interms of their impact on the thermal structure of Wood Lake in the late summerand early fall when the temperature-oxygen squeeze becomes a threat for kokanee.1084.1 Water Clarity: Secchi Depth, PAR Profiles, andLight Extinction CoefficientA marling event occurred from 14 July 2015 until the end of August, having anabrupt significant impact on water colour, light attenuation and water clarity. On14 July 2015, surface water colour changed from dark blue to light green and thewater became cloudy. As will be discussed in this section, these observations of achange in colour and clarity were commensurate with a decrease in secchi depthand an increase in light extinction coefficient. This period of decreased clarityis attributable to marling (Kirk, 1994; Koschel, 1990; Koschel et al., 1983; Mor-timer, 2004; Weidemann, Bannister, Effler, & Johnson, 1985; Wetzel, 2001). Theobserved change in surface characteristics during the 2015 Wood Lake marlinggradually dissipated until the end of August as calcite settled out of the epilimnion.Chemical analyses of water samples, including calcite saturation, during this mar-ling event is discussed in Appendix C.Observed secchi depth between 4 May 2015 and 25 April 2016 ranged between2.3 and 6.9 m (Figure4.1). The minimum secchi depth was 2.3m, recorded on 29July 2015 during the peak of this event. Other local minimums occurred on 2December 2015 during fall turnover (3.15m) and on 23 March 2016 during thespring algal bloom (2.8m). The maximum secchi depth of 6.9m was recorded on16 September 2015 following the marling event. Another local maximum in secchidepth occurred on 6 January 2016 at 6.5m when the lake was inversely stratified.109[04-05-15] [01-07-15] [01-09-15] [01-11-15] [01-01-16] [01-03-16][25-04-16]Day [dd-mm-yy]2.533.544.555.566.5Secchi Depth [m]Figure 4.1: Secchi depth in Wood Lake measured on 33 field days between 4May 2015 and 25 April 2016.The average light extinction coefficient (Kd) for Wood Lake was calculated oneach field day from PAR data obtained from stations S2 to S20 (Section 3.4.3).For comparison, the light extinction coefficient was also calculated from secchidepth using the simple inverse relationship (Kd = a/ZSD) with [a] = 1.16 (French,Cooper, & Vigg, 1982) for the lower limit and [a] = 2.3 (Højerslev, 1978; Tyler,1968) for the upper limit (Figure 4.2). See Appendix B for further discussionregarding the relationship between secchi depth and light extinction coefficient.Note that Kd(PAR) falls between the minimum and maximum Kd based on secchidepth on most days, except in mid-July during the peak in the marling event when itis evident that the image attenuation of the secchi disk was greater than the diffuselight attenuation measured by PAR and shortly after fall turnover when secchi depthappears to underestimate Kd (Figure 4.2). Other than during marling, the upperlimit provided by [a] = 2.3 (Højerslev, 1978; Tyler, 1968) appears to provide abetter estimate of the light extinction coefficient in Wood Lake in 2015 than [a] =1.16 (French et al., 1982). This is particularly true during the months of Septemberand October when water clarity was at its peak until near the end of January (21110January 2016).[04-05-15] [01-07-15] [01-09-15] [01-11-15] [01-01-16] [01-03-16] [25-04-16]Day [dd-mm-yy]0.20.30.40.50.60.70.80.9Extinction Coefficient [m-1]Kd Upper Estimate: 2.3/Z SD [m-1]Kd Lower Estimate: 1.16/ZSD [m-1]Kd(PAR) Estimate [m-1]Figure 4.2: Secchi depth estimates of Kd using Kd = [a]/Zsd with [a] = 1.16(dashed line) and [a] = 2.3 (dot-dash line) compared to light Kd(PAR)estimated from CTD PAR data (solid bold line).July and August are followed by a period of improved water clarity (light atten-uation decreased) as calcite precipitated out of the epilimnion. The average lightextinction coefficient between 1 September 2015 and 7 October 2015 was 0.30m-1(photic depth average: 15.2m), compared to the annual average of 0.41m-1 (photicdepth: 11.7m). The light extinction coefficient decreased to its lowest recordedvalue on 7 October 2015 (0.28 m-1) and then began to increase again as the surfacemixed layer became deeper. The absolute maximum in light extinction occurredduring fall turnover (peak of 0.63m-1). As well a smaller peak occurred duringspring turnover and the concurrent algae bloom (peak of 0.52m-1). The light ex-tinction coefficient and secchi depth on 25 April 2016 (0.41m-1 and 5.7m respec-tively) were very similar to one year previous on 4 May 2015 (0.37m-1 and 5.4mrespectively).1114.2 Chlorophyll-aIn 2015, peak average chlorophyll-a levels at station 0500848 (BCMOE data) be-tween 5-30m depth occurred on 17 March 2015 at 15.59 mg/m3, declining to 9.04mg/m3 on 15 April 2015, and then to 2.09mg/m3 by 13 May 2015. The averagechlorophyll-a measured on 15 May 2015 throughout the lake using the CTD was1.46 mg/m3 (values ranged between 0.38 and 2.60mg/m3 throughout the lake).Following the spring algal bloom, chlorophyll-a concentrations (Figure 4.3) re-veal declining algal concentrations. On 4 May 2015, the spring algal bloom wasnearing the end of its cycle, and the highest average chlorophyll-a values occurredbetween 10-30m depth and lay between 3-4 mg/m3. During the summer, peakchlorophyll-a values occurred in the metalimnion (subsurface maximum). Peakchlorophyll-a values were observed around 11.5m on 21 July (1.3 mg/m3), around7.5m (1.4 mg/m3) and 11m (1.2 mg/m3) on 29 July (two peaks), and around 7-8m (1.3-1.6 mg/m3) on 4 August to 19 August. The highest subsurface maximumduring this time occurred on 4 August 2015 with the 1.6 mg/m3 occurring at 7mdepth. Beginning in October, chlorophyll-a values began to exhibit a subsurfacemaximum once again, with values between 1.4-1.8mg/m3 on 7 October 2015 and1.7-2.0 mg/m3 occurring between 5-15m depth on 14 October 2015. Despite re-maining low throughout the season average chlorophyll-a between 5-30m depthhad a minor rise to a peak value of 2.81 mg/m3 on 22 November 2015 (values over3.5 mg/m3 were observed between 3.2 -19.2m on 25 November 2015). This steadyrise following 1 September 2015 corresponds to the period of significant seichingin Wood Lake and the period of deep downward mixing of the thermocline. From 4May 2015 to turnover on 2 December 2015 (Figure 4.3), the average chlorophyll-ain Wood Lake between 5-30m depth from the deepest 5 CTD profile stations was1.41 mg/m3 (peak at 3.1621 mg/m3 on 4 May 2015). On 2 December 2015 thelake reached turnover and the average chlorophyll-a concentration between 5-30mdepth fell to 2.0 mg/m3. Although chlorophyll-a values declined slightly follow-ing this, values remained above 1.5 mg/m3 throughout the majority of the watercolumn between this time and the spring algal bloom.112[04-05-15] [01-07-15] [01-09-15] [01-11-15] [01-01-16] [01-03-16] [25-04-16]Date [dd-mm-yy]51015202530Depth [m]]0.50.50.51111 1 1111 1.51.51.51.51.51.51.5 1.51.522222222.52.52.52.52.52.52.5333333333.53.53.53.53.53.5444444.54.555667.57.59910.510.5Figure 4.3: Spatially averaged chlorophyll-a [mg/m3] from all stations (S1-S21) measured in Wood Lake from 4 May 2015 to 25 April 2016.The chlorophyll-a concentrations measured from January to April 2016, re-vealed the rise and fall of algal populations in the lake throughout the water columnduring the spring algal bloom (Figure 4.4), also affecting secchi depth (Figure 4.1)and light extinction (Figure 4.2). Similarly, the highest chlorophyll-a levels inWood Lake in previous years on record typically occur during the spring (BCMOEdata). The peak average chlorophyll-a concentration from 5-30m depth occurredon 16 March 2016, averaging 11.1 mg/m3. During this time, chlorophyll-a levelsthroughout the water column below 5m depth were fairly uniform (Figure 4.5).Chlorophyll-a levels in the lake declined rapidly following 16 March 2016 (Fig-ure 4.4). Secchi depth and light extinction began to improve after 23 March 2016(Figure 4.1 and Figure 4.2).11311-01-16 31-01-16 20-02-16 11-03-16 31-03-16 20-04-16Day [dd-mm-yy]05101520Average Chlorophyll-a [mg/m3 ]Average Chlorophyll-a [mg/m 3] 2016 (5-30m depth)Average Chlorophyll-a [mg/m 3] March 17,2015 (4-30m depth)Average Chlorophyll-a [mg/m 3] April 15,2015 (4-30m depth)Average Chlorophyll-a [mg/m 3] March 26,2014 (4-30m depth)Average Chlorophyll-a [mg/m 3] April 23,2014 (4-30m depth)Average Chlorophyll-a [mg/m 3] March 19,2013 (4-30m depth)Average Chlorophyll-a [mg/m 3] April 18,2013 (4-30m depth)Water Samples Taken: March 2, 2016Figure 4.4: Average chlorophyll-a [mg/m3] between 5-30m depth in watercolumn from 1 January 2016 to 25 April 2016 (with comparison to av-erage chlorophyll-a values from 2013-2015).1140 1 2 3 4 5 6 7 8 9 10 11 12 13Chlorophyll-a [mg/m 3]051015202530Depth [m]Figure 4.5: Chlorophyll-a [mg/m3.] in Wood Lake (all CTD profiles) on 16March 2016.Water were samples taken on 2 March 2016 (Figure 4.4) from 0, 10, and 20mdepths at 50◦04.720’N, 119◦23.375’W (station S10, near center of lake) using aVan Dorn sampler provided by the MOE, and were analyzed by Jamie Self (Lar-ratt Aquatic Consulting) for taxonomy (Figure 4.6 and Table 4.1). Note that thisanalysis provides the results in cell/mL, and as such cyanobacteria counts appearcomparable to diatoms in terms of numbers; however, diatoms are approximately100 times larger than cyanobacteria cells. Therefore, in terms of biomass, diatomswere the dominant algae type at all three depths (0m,10m, 20m) (J.Self, pers. corr.,4 March 2016). This spring diatom bloom supports the notion that Wood Lake’strophic status has been improving in recent years (LAC, 2010). Diatoms are in-dicative of oligo-mesotrophic conditions when abundant, and are replaced by blue-green algal as the most abundant species with increasing trophic state (Anonymous,1974). A shift in algal dominance from blue-green algal to diatoms has been ob-served in other lakes in the year following calcite precipitation and removal of P115from the water column (i.e. Koschel et al. (1998b) as discussed by Dittrich andKoschel (2002)).0 500 1000 1500 2000 2500 3000 3500 4000Cells/mL0 m10m20mDepthDiatoms Blue-Green Algal OthersFigure 4.6: Taxonomic analysis of water samples (0m,10m,20m depth) takenon 2 March 2016 (Jamie Self, LAC, pers.corr., 4 March 2016). The’Others’ category is delineated in Table 4.1.116Table 4.1: Taxonomic analysis of water samples (0m, 10m, 20m depth) takenon 2 March 2016 (Jamie Self, LAC, pers. corr., 4 March 2016).Group 0m 10m 20mDiatomsAchnanthidium (cells/mL) 10 10 10Asterionella (cells/mL) 0 40 0Aulicoseira (cells/mL) 640 0 490Cyclotella (cells/mL) 50 10 0Fragillaria (cells/mL) 30 0 50Stephanodiscus (cells/mL) 0 0 20Synedra (cells/mL) 10 0 0Tabilleria (cells/mL) 1750 1570 1530Blue-Green AlgalAnacystis (cells/mL) 0 200 50Lyngbya (cells/mL) 800 1250 1400Synechococcus (cells/mL) 10 60 0OthersCryptomonas (cells/mL) 10 0 0Mallomonas (cells/mL) 0 20 10Flagellates (cells/mL) 20 20 30In comparison to previous years (2013-2015), the chlorophyll-a peak averageconcentrations (5-30m depth) during March and April were considerably lower(Figure 4.4 and Figure 4.7). From 2013-2015, the chlorophyll-a levels in WoodLake (Figure 4.7) have been fairly low on average throughout the water columnand do not show evidence of a late summer or early fall algal bloom. Chlorophyll-a measured in 2015 was on average about 24% lower than values recorded in 2013(directly comparing month values from BCMOE data). Herein we observe againthat values in 2016 were considerably lower during the spring algal bloom, thanhave been reported in previous years during spring algal blooms.117[01-03][01-04][01-05][01-06][01-07][01-08][01-09][01-10][01-11][02-12]Day [dd-mm]246810121416Average Chlorophyll-a [mg/m3 ]between 5-30m depth2015 (current study data)2016 (current study data)2015 (MOE Data)2014 (MOE Data)2013 (MOE Data)Figure 4.7: Average chlorophyll-a [mg/m3] measured between 5-30m depthfrom 4 May 2015 to 2 December 2015 and 2 March 2015 to 25 April2016 compared to BCMOE data from March to October 2013-2015.4.3 Temperature and Dissolved OxygenThe daily-averaged time series of temperature from a thermistor chain near thecenter of Wood Lake (Figure 4.8) shows the onset of stratification in Wood Lake atthe end of May 2015. The surface layer temperature rose to above 22◦C between30 June 2015 and 25 August 2015. A natural progression in the depth of the sea-sonal thermocline is observed throughout the year until turnover in late fall. Thethermocline is defined here as the region where temperature change is greater than1◦C per meter of depth (Horne & Goldman, 1994). The thermocline at the end ofMay spans between 2.8 and 7.8m depth and deepens to between 14.2 and 17.2mdepth by 28 October 2015. The maximum depth of the 17◦C contour was 10.9mon 14 September 2015; the maximum depths of the 20◦C and 22◦C contours were8.2m and 7.6m respectively occurring on 14 August 2015. This central thermistorchain data was not relied on for further analysis in accordance with recommenda-tions by Wedderburn (1911) (as discussed in Mortimer (1952)) and by Mortimer(1952) that conclusions should not be drawn from the analysis of results made at118a single location without accounting for seiching. Instead, the spatially averagedCTD profiles and DO profiles, which accounted for tilting of the isotherms duringseiching, were relied on for analyses in this and subsequent sections.777788811111414171720222201-06-15 01-07-15 01-08-15 01-09-15 01-10-15 01-11-15 01-12-15Date [dd-mm-yy]051015202530Depth [m]Figure 4.8: Temperature distribution in Wood Lake from 4 May 2015 - 31December 2015 from Middle Thermistor Chain near the center of thelake (Section 3.3.1).Throughout the following sections, meteorological conditions and physicalconditions in Wood Lake in 2015 are compared with 2010, 2011, and 2013-2014.The latter two years provide a useful comparison because the corresponding pro-file data from MOE is the most comprehensive historical record of temperatureand DO available for Wood Lake and is also used to validate the DYRESM model(Section 4.11.4). 2010 and 2011 are also years of interest because 2010 profferedthe most recent marling event in Wood Lake in recent history (LAC, 2010) and2011 was the motivating year for this current study with an extreme temperature-oxygen squeeze. Key similarities and anomalies in air temperatures and physical119conditions in Wood Lake between these years are addressed.4.3.1 Spring and Summer 2015Air temperature is primarily responsible for changes in lake surface layer temper-atures (Schmid et al., 2014). Daily average air temperatures during 2015-2016 arecompared to recent years in order to establish a contextual setting and extract keydifferences and similarities. An anomalously warm period occurred between 18May 2015 and 11 July 2015, wherein the daily average air temperature was con-siderably warmer than the same period in 2010, 2013 and 2014. Daily mean airtemperatures were 18◦C on average during this period, which was 3.22◦C warmerthan 2014, 3.77◦C warmer than 2013, and 4.33◦C warmer than 2010 (GCEC,2016). These warmer-than-average conditions may have aided in triggering a mar-ling event that started around mid-July (Brunskill, 1969; Hamilton et al., 2009).However, the average temperature difference when compared to recent years isgreatest in 2010, despite 2010 being the last recorded marling year. This confirmsthat many other factors are also important in determining the timing and intensityof calcite precipitation (Gilbert & Leask, 1981; Solim & Wanganeo, 2007). Thismarling event, along with an analysis of water samples collected, is discussed inAppendix C. For the remainder of the summer, from July 11 to August 31, averagedaily temperatures were much more similar to the past two years (20◦C), with av-erage daily temperatures only 0.7◦C cooler than 2014 and within 0.01◦C of 2013(GCEC, 2016).With regards to Wood Lake’s spatially averaged temperature and DO profiles,the first time period (warming period) of interest is 4 May 2015 to 25 August 2015,wherein a seasonal progression of temperature and DO is observed (Figure 4.9).The lake was already beginning to stratify on 4 May 2015 when the first field visitwas conducted. The thermocline was established by 29 May 2015 spanning 2.8mto 7.8m depth. By 15 June, a well-defined surface mixed layer (epilimnion) wasestablished from the surface down to approximately 5.8m depth. The surface layerwarmed dramatically from 20.8◦C on 15 June to the maximum observed spatiallyaveraged temperature of 24.9◦C on 30 June 2015, two days after the maximumdaily average air temperature (26.9◦C) was observed. Subsequently, the surface120layer cooled to 22.1◦C, and gradually thickened to 8.2m on 25 August 2015. Atthis time, the thermocline spanned from 8.2m to 14m depth. The average bottomtemperature during this time period was 5.2◦C.Dissolved oxygen was first measured on 21 July 2015, at which time DO wasless than 2 mg/L below 25m depth and less than 4 mg/L below 21m depth (Fig-ure 4.9). A DO maximum occurred in the thermocline, suggesting that water claritywas sufficient for light to penetrate to this depth and this is where algal cells wereconcentrated. PAR data confirms that the photic depth (depth at which light is 1%of that at surface (R. Davies-Colley & Vant, 1988)) was between 10.8m and 11.7mwhile this subsurface DO peak at 10m depth occurred (21 July to 11 August).Peak chlorophyll-a values were observed around 11.5m on 21 July, around 7.5mand 11m (two peaks) on 29 July, and around 7.5-8m on 4 August to 19 August(Figure 4.3). Oxygen is produced by algal via photosynthesis; pressure (depth)and stability at the thermocline enable oxygen to become supersaturated (BCWIB,1974). Such subsurface summer metalimnetic DO maxima (typically between 3-10m depth, depending on transparency of water column) are most commonly asso-ciated with production of algal that thrive on the higher nutrient concentrations inthe metalimnion and that are capable of surviving at lower ambient temperaturesand light (relative to epilimnion) (Wetzel, 2001). This phenomenon persisted untilaround 11 - 19 August 2015.121Figure 4.9: Temperature and DO in Wood Lake from 4 May 2015 to 25 Au-gust 2015. The dark red region outlines temperatures that are equal to orgreater than 20◦C (lethal for kokanee) and where 0.5mg/L<DO<2mg/L.The grey region outlines temperatures between 17◦C to 20◦C and where2mg/L<DO<4mg/L. These regions are avoided by kokanee and mayhave harmful effects. Kokanee temperature and DO tolerances are be-low those for many other well studied fish species including SockeyeSalmon (Aku & Tonn, 1999; S. B. Brandt et al., 2009; Brett, 1971;Cherry et al., 1977; C. Coutant, 1977; Herbert & Steffensen, 2005;Jobling, 1981; Vaquer-Sunyer & Duarte, 2008; Paul Askey, pers.comm).The dark red region in the Depth-DO plot outlines water that is anoxic(DO<0.5 mg/L). The temperature profile for each day is based on aver-aging the deepest five casts in the middle of the Wood Lake (S9-S13).The given average DO profiles are based on stations 9,11, and 13.1224.3.2 Early Fall 2015Despite having a warmer-than-average spring and early summer (prior comparisonto 2013/2014), average daily air temperatures were 1.23◦C cooler than 2014 and2.06◦C cooler than 2013 during early fall (1-30 September 2015) (GCEC, 2016).Following 20 August, daily average air temperature in 2015 was cooler than 2011,when the temperature-oxygen squeeze was lethal. In fact, the period from 20 Au-gust - 30 September 2015 yielded average/maximum/minimum temperatures thatwere 2.62◦C/4.64◦C/0.67◦C colder than 2011 respectively (GCEC, 2016). Thishelped to enable the surface layer of the lake to cool more rapidly in 2015 thanin 2011 and alleviate the temperature-oxygen squeeze, despite having a signifi-cantly warmer surface layer in the early summer. September had many days withrelatively strong constant diurnal or semidiurnal winds predominantly along thenorth-south (long) axis of the lake. As a result, large amplitude seiching (i.e. 5-6mpeak-to-peak end wall displacement of the thermocline on 9 September 2015) wasobserved (Figure 4.10). Observations (i.e., Figure 4.10) show that during theselarge seiche events, the stratified thermal and DO layers of the lake move in con-cert, such that it does not serve to worsen or better the temperature-oxygen squeezeconditions in the lake in terms of available habitat.Figure 4.10: Thermal contours and DO contours on 9 September 2015:isotherms and DO-isopleths moving in concert during seiche activity.Distance along x-axis is measured from south to north along the long-axis of the lake (Figure 3.2).123Similarly to Figure 4.9, Figure 4.11 shows representative average temperatureand DO profiles in Wood Lake from 1 September 2015 to 28 October 2015. Notethat by 1 September 2015, all temperatures in the lake were below 20◦C, how-ever, the thickness of the anoxic layer and DO threshold for kokanee progressivelyincreased during this time. On September 1, the maximum temperature at the sur-face was 19.7◦C and the thermocline lay between 9.2m and 15.6m depth. At thistime, the lake was anoxic below 23.2m depth, the 2 mg/L DO threshold was at21.1m depth, and the 4 mg/L DO threshold was at 17.7m depth. In comparisonto the summer profiles, highest DO levels during September and October were inthe surface mixed layer, where DO was relatively uniform with depth. The gradualincrease in depth of this uniform DO layer with time corresponds with the pro-gressive deepening of the surface mixed layer observed in the temperature profiles.DO decreased quickly across the thermocline to hypoxia in the hypolimnion andanoxia (<0.5 mg/L) in the lower levels of the hypolimnion. During this period, theaverage temperature at the bottom of the lake was 5.9◦C.124Figure 4.11: Temperature and DO in Wood Lake from 1 September to 28October 2015.DO in the hypolimnion was progressively consumed between 1 September and28 October, decreasing the 2mg/L and 4 mg/L threshold depths. By 28 October,the surface temperature was 12.9◦C and the thermocline had deepened to between14.2 and 17.2m depth. Meanwhile the lake was anoxic below 18.6m depth, the2 mg/L DO threshold was at 15.2m depth and the 4 mg/L DO threshold was at13.6m depth (Figure 4.12). Following this date, surface layer mixing progressivelyentrained low DO waters, and by 5 November 2015 the depth of 2 mg/L DO hadincreased to 17.1m, while the depth of anoxia had increased to 19.1m. Between 28October and 5 November, the surface mixed layer had become approximately 2mdeeper such that the thermocline lay between 16.2m and 19.6m (Figure 4.13).125Figure 4.12: Temperature and DO profiles from 28 October 2015, during themost extreme DO condition in Wood Lake in 2015.126Figure 4.13: Temperature and DO profiles from Wood Lake on 5 Novem-ber 2015, one week following the most extreme DO conditions (Fig-ure 4.12) as DO began to be replenished at depth.4.3.3 Late Fall 2015By the end of October, DO conditions with regards to kokanee habitat were im-proving. By 28 October 2015 temperature was below 12.9◦C throughout the wa-ter column and DO from this point onwards increased at depth with consistentlylarge wind events in November that enabled deep mixing in the water column (Fig-ure 4.14). On 5 November 2015, temperature ranged from 11.5◦C to 6.2◦C, whileDO ranged from 10 mg/L at the surface to 0 below 20.5m depth. The lake didnot fully overturn until 2 December 2015 at which time temperature and DO werefairly uniform with depth, temperatures throughout the lake averaged between 6.16127and 6.12◦C (Figure 4.15), and DO averaged between 7.7 and 8.8mg/L (mean: 8.1mg/L) (Figure 4.14). Note the drastic change between DO values on 2 December2015 and those that were observed on 25 November prior to turnover, wherein itDO was less than 4 mg/L below 24.7m depth and was anoxic below 26.0m depth.On 2 December 2015 the lake was more strongly horizontally stratified than itwas vertically, with colder temperatures occurring at the north end of the lake (Fig-ure 4.15). Wood Lake has historically been recorded to turn over in early November(R. Nordin, 1987), thus suggesting that turnover was quite delayed in 2015.Figure 4.14: Temperature and DO in Wood Lake from 5 November 2015 to23 December 2015, showing turnover and near-isothermal conditionson 2 December 2015.128Figure 4.15: Thermal contours in Wood Lake on 2 December 2015 immedi-ately following fall turnover. Distance along x-axis is measured fromsouth to north along the long-axis of the lake (Figure 3.2).4.3.4 Winter 2015-2016 and Spring 2016Wood Lake serves as a popular ice-fishing lake most winters, and personal con-versations with local anglers reveal that the lake has historically developed a thickenough ice cover to accommodate ice fishing for at least the last 35 winters. Thelast year on record that Wood Lake did not develop permanent ice cover was dur-ing winter 1980-1981 (R. Nordin, 1987). From 1 January 2016 to 24 February2016 sections of the lake (predominantly the north end, up to about 50% of the sur-face area) developed thin (maximum of 2.5 - 15mm thick) ice cover that typicallypartially/fully melted by the following week (Table 4.2). This permitted regularboat-based fieldwork to continue through the winter season.129Table 4.2: Observations of ice cover on Wood Lake during the winter season(2015).Date Notes1 Jan 2016 Ice cover (0.5-1.5cm thick) in northern third of lake about750m from shore, migrating south during day6 Jan 2016 Slush/ice (0.5-1.5cm thick) covered north end of lake about100-200m offshore and perimeter of northern half of lake8,12,20,27 Jan 2016 No ice cover3 Feb 2016 A 25m band of ice (0-0.5cm thick) at the north end of lake10 Feb 2016 Ice (0.5-1cm thick) on northern 3.3km of lake17 Feb 2016 No ice cover24 Feb 2016 Ice (0.25cm thick) on the northern 1.8km of lakeMild 2015/2016 winter conditions with regular winds contributed but do notfully explain a lack of ice cover on Wood Lake. In 1973, permanent ice coverdeveloped after mid-January and from January to March the lake temperatures re-mained between 1-2◦C. BC Research (1974) note that during spring, water tem-peratures at the southern end of Wood Lake were warmer than the north end in1972-1973, which was also observed in 2015 until about mid-February, after whichtemperatures in the north end became warmer than those in the south until the endof the field season (25 April 2016). Table 4.3 compares average, maximum, andminimum air temperatures from winter 2015/2016 to those from 2014/2015 and2013/2014 when the lake did develop complete ice cover. On average, air tem-peratures were considerably warmer in December and especially February than2013/2014 winter season, suggesting a strong air temperature influence. However,when compared to the previous winter (2014/2015), air temperatures were slightlywarmer in December, but then cooler in January and February than in 2014/2015.By itself, average daily air temperatures alone do not fully explain why the lakefailed to freeze over in winter 2015/2016, when compared to 2014 when it did de-velop full ice cover. Other possible aiding factors include regular wind stirring ofthe surface layer, and the timing and duration of sub-freezing air temperatures (i.e.extent of diurnal variations).The subsequent period of interest in terms of temperature profiles in WoodLake occurred from 1 January 2016 to 25 April 2016 (Figure 4.16). The entirewater column throughout the lake was colder than 4◦C from 1 January 2016 to 16130Table 4.3: Comparison of average/maximum/minimum daily air tempera-tures in winter (Dec-Feb) 2013/2014,2014/2015, and 2015/2016 (GCEC,2016). Values are computed as TWinter:2015/2016 - TWinter:2014/2015 for col-umn one and TWinter:2015/2016 - TWinter:2013/2014 for column two, where“winter” is defined as the period from December to February.Months 2015/2016 - 2014/2015 2015/2016 - 2013/2014Temperature Difference [◦C] Temperature Difference [◦C]December 0.52 2.06January -0.44 -0.24February -0.68 6.10March 2016. The lake did not strongly reversely stratify for a prolonged periodduring this time; relatively strong reverse stratification was observed on 6 January2016 and 10 February 2016.1310 1 2 3 4 5Temperature [°C]051015202530Depth [m]January 1January 6January 12January 20January 27February 3February 10February 17February 240 1 2 3 4 5 6 7 8 9101112131415Temperature [°C]051015202530Depth [m]March 2March 9March 16March 23March 30April 13April 25Figure 4.16: Temperature in Wood Lake from 1 January - 25 April 2016.There were several field days wherein the lake was more strongly horizontallystratified (i.e.vertical layers) than vertically stratified during the winter season. Therelative strength of inverse (vertical) stratification and horizontal stratification is as-sessed by comparing absolute maximum and minimum temperatures in the entirelake and spatially averaged maximum and minimum temperatures from all CTDprofiles across the north-south transect of the lake (Figure 4.17). 6 January 2016shows the strongest reverse stratification (Figure 4.18) with the largest differencebetween both absolute minimum/maximum (1.54/3.67◦C) and spatially averagedminimum/maximum (2.04/3.46◦C) temperatures. The strongest horizontal strat-ification occurred on 27 January 2016 with a large difference between absolute(2.72/3.40◦C) and nearly uniform average (2.86/2.94◦C) temperatures across thelake from South (warmer) to North (colder) (Figure 4.18). Although no major132changes in DO have been reported in past years during ice cover (BC Research,1974), this pattern of fluctuating between horizontal and vertical stratification dur-ing an ice-free winter has implications for continuous mixing of the water column(DO, nutrients, etc.). Numerous observations of re-suspended partially decayedorganic matter from the lake bottom were made during this period suggesting thisvertical mixing extended to the lake bottom.02-12-15 22-12-15 11-01-16 31-01-16 20-02-16 11-03-16 31-03-16Day [dd-mm-yy]2345678910Temperature [°C]Minimum Lake Temperature [°C]Maximum Lake Temperature [°C]Average Minimum Temperature [°C]Average Maximum Temperature [°C]4°CFigure 4.17: Absolute and average minimum and maximum temperatures inWood Lake during winter 2015-2016.133Figure 4.18: (Left) Thermal contours in Wood Lake on 6 January 2016 show-ing strong inverse stratification. (Right) Thermal contours on 27 Jan-uary 2016 showing weak horizontal stratification. Distance along x-axis is measured from south to north along the long-axis of the lake(Figure 3.2).In comparison to previous years on record, spring lake temperatures in 2016were warmer on average in March and April than 2013-2015 (Figure 4.19). InMarch, 2016 temperatures throughout the water column were warmer than 2013-2015, although on March 16, 2016 temperatures were relatively close to tempera-tures from March 17, 2015. In April 2016, surface temperatures (upper 10m) weresignificantly warmer than the previous 3 years, while the temperatures at depth areslightly cooler. Table 4.4 depicts numerical values of the temperature range in thewater column from March and April 2013 - 2016 for comparison. Note that 2013-2015 data was collected by the BC MOE at one location at station 050048 and 2016data represents the average temperature profile from the five deepest CTD casts.1343 4 5 6 7 8 9 10 11 12 13 14Temperature [° C]051015202530Depth [m]March 16,2016: Average Temp ProfileMarch 23,2016: Average Temp ProfileMarch 30,2016: Average Temp ProfileApril 13,2016: Average Temp ProfileApril 25,2016: Average Temp ProfileMarch 19,2013 (0500848)March 26,2014 (0500848)March 17,2015 (0500848)April 18,2013 (0500848)April 23,2014 (0500848)April 15,2015 (0500848)May 4,2015: Average Temp ProfileFigure 4.19: Comparison of temperatures in Wood Lake in March and Aprilfrom 2013-2016 (2013-2015 field data measured at station 0500848,BCMOE). BCMOE data provided by Mike Sokal, 13 February 2015(2013-2014 data) and 13 April 2016 (2015 data)).Table 4.4: Range of temperatures in water column in March and April from2013 to 2016. 2013-2015 data collected by BCMOE at one location atstation 0500848 (50.0749◦N, 119.3917◦W). 2016 data reported as aver-age of deepest 5 CTD profiles.2013 2014 2015 2016March 19 Mar: 26 Mar: 17 Mar: 16 Mar: 4.04-4.24 ◦C3.5-3.67 ◦C 3.67-3.76 ◦C 3.98-4.20 ◦C 23 Mar: 4.21-4.80 ◦C30 Mar: 4.33-5.57 ◦CApril 18 Apr: 23 Apr: 15 Apr: 13 Apr: 4.68-9.38 ◦C4.93-6.83 ◦C 5.42-6.93 ◦C 4.68-7.90 ◦C 25 April: 4.8-13.0 ◦C4.4 Progression of Temperature-Oxygen Squeeze in 2015The average depths of (1) 20◦C, (2) 17◦C, (3) DO = 4mg/L, (4) DO = 2 mg/L, and(5) DO = 0.5mg/L were monitored during 2015. Note that DO was only measured135after 21 July 2015 until 16 December 2015. The most severe condition in the lakeduring this season was observed on 25 August 2015 (Figure 4.20) wherein theavailable “non-lethal habitat layer” (T<20◦C and DO>2mg/L) was 13.6m thickand the “ideal habitat layer” (T<17◦C and DO>4mg/L) was 8.6m thick. On thisday, the average depth of 20◦C was 9.0m, while the average depth of 17◦C was10.0m. Likewise, the average depth of anoxia was 24.4m, the average depth of2mg/L DO was 22.6m and the average depth of 4mg/L DO was 18.6m. On 16September 2015 the minimum available “non-threatening habitat layer” (T<17◦Cand DO>4mg/L) occurred and was 6.1m thick, after which all temperatures fellbelow 17◦C by 23 September 2015.01-06-15 01-07-15 01-08-15 01-09-15 01-10-15 01-11-15Day [dd-mm-yy]0246810121416182022242628303234Depth [m]Mean Depth of 17°CMean Depth of 20°CMean Depth of 4 mg/L DOMean Depth of 2 mg/L DOMean Depth of 0.5 mg/L DOFigure 4.20: Temperature-Oxygen squeeze during 2015. Available non-lethal habitat layer exists where T<20◦C and DO>2mg/L while theideal habitat layer exists where T<17◦C and DO>4mg/L. The mostsevere squeeze in 2015 occurred on 25 August 2015 wherein the non-lethal habitat layer was 13.6m thick and the ideal habitat layer was10m thick.1364.5 Hypolimnetic Oxygen DeficitThe average rate of change (demand) in hypolimnetic DO between spring turnoverand summer stratification is referred to as the hypolimnetic oxygen deficit [HOD](BC Research, 1974), areal hypolimentic oxygen deficit (D. A. Matthews & Ef-fler, 2006) or the relative oxygen deficit (Wetzel, 2001), and is considered to be astrong indicator of the lake’s hypolimnetic productivity (BC Research, 1974; Wet-zel, 2001). HOD is based on the difference in oxygen content in the hypolimnion attwo points in time (spring turnover and peak summer stratification), where the topof the hypolimnion on a given day is the average depth below the thermocline atwhich the rate of change in temperature with depth is first less than 1◦C/m. The av-erage DO concentration in the hypolimnion on each day is calculated from all DOmeasurements below this depth. The volume and surface area of the hypolimnionis obtained from hypsometric data (curves). The areal oxygen content (mg/cm2) oneach day is equal to the average DO in the hypolimnion multiplied by the volumeof the hypolimnion and divided by the surface area of the hypolimnion (Wetzel,2001). The average hypolimnetic DO is computed by dividing the hypolimnioninto a series of layers based on the resolution of the field data and then weightingthe DO contribution of each layer to the total hypolimnion average according tothe ratio of the surface area of that layer to the surface area of the hypolimnion (i.e.using the hypsograph). HOD is then equal to the difference between this computedhypolimnetic DO at peak stratification and that at spring turnover, divided by time(days) between.The HOD in Wood Lake calculated during summer stratification in 1972 was0.121 mg O2/cm2/day (June - October) and 0.148 mg O2/cm2/day (April - Septem-ber) in 1973 (BC Research, 1974). For reference, these values can be compared tothe much lower rates from Kalamalka Lake in the same years: 0.034 mg O2/cm2/day(June-Nov 1972) and 0.044 mg O2/cm2/day (June - Sept 1973) (BC Research,1974). These values are considerably higher than those calculated here for 2011- 2015 (Table 4.5), and should be evaluated with caution. There are a number offactors other than improved water quality that explain this difference, including:1. Different normalizing periods used in each calculation in 1972/1973 (i.e. anormalizing period beginning in June when the lake is already stratified)1372. Significantly improved bathymetric data available today (Section 3.4.2)3. Rather than calculating an average value for DO in the hypolimnion based onsparse data points and rough bathymetry (as done in 1972/1973), the DO inthe hypolimnion was calculated by dividing the hypolimnion into a numberof layers based on the frequency of the data points available, determining thevolume of water in each layer based on the depth-area-volume data at 0.5mdepth intervals, determining the sum of contributions of DO in each layer tothe total DO in the hypolimnion, and dividing this by the surface area of thehypolimnion (procedure in accordance with Wetzel (2001)). Furthermore,the depth of the hypolimnion could be determined more accurately with thehigher resolution of data points available in more recent studies.Therefore, despite it appearing that HOD has decreased since the early 1970’sgiven the representative values calculated for 2011 - 2015, a direct comparison withhistorical values is not recommended and hence, the following discussion focuseson more recent data for direct comparison.The HOD (Table 4.5) has been calculated for 2011 to 2015 using BCMOE data.For 2013 - 2015, DO data (2m intervals) was available from March until October;only May and October were used in these calculations since this corresponds tothe stratified period when DO is progressively being consumed (onset of stratifi-cation to the time of most severe DO conditions). In 2011 (2 - 4m intervals) and2012 (1 - 4m intervals) only data from March and September (two profiles/year)were available. In March, prior to stratification (i.e. no thermocline with a slope ≥1◦C/m), the hypolimnion depth was taken as the point below the maximum slopein the temperature profile. The rate of change in hypolimentic DO was also cal-culated from DO data collected in the current study with higher resolution (0.5 -1.0m intervals) from 21 July 2015 to 28 October 2015 during the strongly stratifiedperiod (recall DO data was only collected from 21 July to 16 December 2016).Note the difference between the calculated depletion rates of 0.0255 mg O/cm2/dayin 2015 between May - October versus 0.0541 mg DO/cm2/day between July -October. Recall 28 October 2015 proffered the most extreme hypolimnetic DOdepletion and hypolimetic DO was consumed at the greatest rates from July on-ward (Figure 4.9, Figure 4.11, and Figure 4.12 from Section 4.3). Therefore, we138Table 4.5: Hypolimnetic Oxygen Deficit in Wood Lake from 2011 - 2015.Year (period) Hypolimnetic Oxygen Deficit [mg O2/cm2/day]2011 (29 March - 13 Sept)1 0.05152012 (12 March - 13 Sept)1 0.01292013 (14 May - 15 Oct)2 0.00162014 (21 May - 15 Oct)2 0.02742015 (13 May - 14 Oct)3 0.02552015 (21 July - 28 Oct)4 0.0541 *hypolimnetic consumption rate(shorter time period)1(BC MOE data, pers. corr., Mike Sokal, 15 March 2016)2(BC MOE data, pers. corr., Mike Sokal, 13 February 2016)3(BC MOE data, pers. corr., Mike Sokal, 13 April 2016)4Current Study Dataconclude that the normalizing period is an important variable in this calculation.Extending the calculation period into the time that DO is not being consumed atsignificant rates (i.e. March - May) reduces the calculated DO consumption rate(averaging depletion over longer period). In 2015, this resulted in over a two-folddifference in the calculated oxygen depletion rate (Table 4.5). When looking backat 2011 (die-off year), note the calculated oxygen depletion rate for the periodMarch - September was nearly the same as that for 2015 from mid-July to the endof October (despite lengthening this normalizing period in 2011 into March - Maywhen the lake was not strongly stratified and DO is not typically being consumedat high rates). This confirms the severity of the extreme conditions that occurredin 2011, that as of yet, have not recurred. Overall, hypolimnetic DO consumptionhas not been nearly as severe as 2011 in the past four years.4.6 Middle Vernon Creek InflowCore sample analyses as part of the Okanagan Water Basin Study (Anonymous,1974) and more recently by (Walker et al., 1993) indicate a strong correlation be-tween the deterioration in water quality in Wood Lake following the 1930’s withan increase in agricultural land use and water diversions for irrigation upstreamfrom Wood Lake. These diversions increased the retention time of the lake therebyreducing the flushing of nutrients (Walker et al., 1993). Currently, there is insuf-139ficient flow in VC in a high demand year to satisfy water licenses from the creekwhilst maintaining required fish flows in UVC (75 L/s for 8 months of year; al-though this minimum requirement in UVC will result in near-zero flows in MVC)(Mould Engineering, 2004). The presence of beaver dams in MVC further reducesflows in the creek. Chronic low flow conditions in MVC have not been improved inrecent years and maintenance of minimum flows for generating sufficient spawninghabitat is an ongoing concern (Epp & Neumann, 2014; Webster, 2013, 2015).Minimal flow from MVC entering Wood Lake was observed from 4 May 2015onwards through spring and summer. The peak flow during this time was 1.3m3/s(4 May 2015), quickly decreasing to 0.5m3/s by 24 May 2015 and below 0.1m3/sby 21 June 2015. Comparing the discharge rates and cumulative discharge in MVCfrom 2010 through 2016 (Figure 4.21 and Figure 4.22), the freshet in 2015 wasweak compared to previous years, as well as 2016. Spring freshet has occurredearlier each year since 2012; although in 2013, there was a second spike in flowsthat occurred slightly later in June/July. 2012 and 2014 yielded similar total vol-umes of inflow into Wood Lake, while 2013 yielded the highest inflow in recentyears. The watershed in 2015 yielded far less total runoff than any of the previousthree years or 2016. Unfortunately there is insufficient data from 2011 for com-parison as this data series ends after 19 April 2011 in this year. However, from theavailable data, it appears that the freshet in 2011 occurred later in comparison toother years (noted by Heather Larratt, pers. corr., 18 April 2016). Spring freshet in2016 occurred at a similar time to that in 2015, but with much higher peak flows,resulting in nearly twice as much total inflow volume into Wood Lake over thecourse of the year.140[Jan] [Feb] [Mar] [Apr] [May] [Jun] [Jul] [Aug] [Sep] [Oct] [Nov] [Dec]012345678910Average Daily Flow [m3 /s]2010201120122013201420152016Start of Data SeriesEnd of Data SeriesNo DataFigure 4.21: Average daily discharge [m3/s] in MVC measured at ReimcheRoad in 2010 - 2015. 2010 data only available from 27 May - 31December; 2011 data series ends on 19 April 2011.141[Jan] [Feb] [Mar] [Apr] [May] [Jun] [Jul] [Aug] [Sep] [Oct] [Nov] [Dec]024681012141618Cumulative Volume [m3 ]×1062010201120122013201420152016Start of Data SeriesEnd of Data SeriesNo DataFigure 4.22: Cumulative volume discharged [m3] from MVC into WoodLake in 2010 - 2015. 2010 data only available from 27 May - 31December 2010; 2011 data series ends on 19 April 2011.There are three critical periods in MVC’s discharge hydrographs (Figure 4.21).The first of these is spring freshet, which appears to have occurred earlier in recentyears (2013 - 2016) with progressively less cumulative discharge to Wood Lake(2016 being the exception). The second is the low flow period that persists fromlate spring until late fall or early winter, which is believed to be one of the rootcauses of the collapse in the kokanee fishery witnessed in recent years (Epp &Neumann, 2014). Annual low flows in late summer and early fall have occurredin MVC since the cooling discharge from HWD ceased in 1995. Even in 2013 (ahigher runoff year) (Figure 4.22) Swalwell Lake releases in August only exceededDistrict of Lake Country [DLC] demands by 0.07 m3/s. This increased to 0.2 - 0.3m3/s from September to November (Epp & Neumann, 2014). Low summer flowsare exacerbated by the use of a temporary flow control structure [TFCS] (sandbags)142first installed in 2003 by the Oceola Fish and Game Club and the Okanagan IndianBand to store water in Ellison Lake and to control the subsequent release intoMVC during kokanee spawning in October to help ensure sufficient flows for eggdeposition and alevin emergence (Epp & Neumann, 2014). This marks the thirdperiod in the hydrographs (Figure 4.21). Discharge measurements downstream inMVC reveal that flows out of Ellison Lake into MVC are largely controlled by thewater level within the lake, by the TFCS, and by beaver dams. When Ellison Lakeoutflows are zero, MVC receives minimal flow from a stormwater detention pondand Knopf Brook (small groundwater-fed creek) (Epp & Neumann, 2014). MVCflows typically decline to approximately zero around July - August in 2012 - 2016and remain so until October (2013-2014) or November (2012, 2015/2016) (Epp& Neumann, 2014). A sandbar developed at the outlet of MVC into Wood Lake(Figure 4.23) by early summer (2015) with water passing in and out of MVC. Atthe mouth of the creek the depth was 1.0m, then decreased to less than 1m forover 50m from the shore. Note that in 2012, peak daily average flows in MVC inNovember reached just over 1 m3/s; however, in 2013 - 2015 these fall/winter dailyaverage flows typically remained well below 0.5 m3/s (in 2015 flows averaged only0.084 m3/s through November until the end of December) (Epp & Neumann, 2014,2016).Figure 4.23: Sandbar at mouth of MVC (photo taken on 16 September 2015).A synopsis of the flow conditions in 2015 extracted from Epp and Neumann143(2016) is provided herein as follows:1. Swalwell Lake outflow in June (average 0.44 m3/s) was lower than previousyears. Concurrently, the DLC diversions in June were higher than previousyears (average 0.34 m3/s) due to warm/dry conditions in May/June. Outflowsfrom Swalwell Lake exceeded DLC diversions by 0.1m3/s in June. Flow inUVC above Ellison Lake in June (average 0.21m3/s) was much lower than2013 (no data from 2012/2014). Flow in MVC near Wood Lake in June(average 0.24 m3/s) was considerably lower than 2013 and 2014.2. Average diversions for DLC increased to 0.410 m3/s in July ( 0.028 m3/s lessthan outflows from Swalwell Lake). Despite the runoff into UVC from othersources (including Clark Creek), the total flow into Ellison Lake in July waslower than the previous four years (i.e. averaged 0.053 m3/s). Subsequently,flow in MVC below Ellison Lake averaged only 0.004 m3/s in July, furtherrestricted by beaver dams. MVC flows at Reimche Road above Wood Lakein July were lower than the last four years (average 0.015 m3/s).3. DLC diversions averaged 0.328 m3/s in August and the outflow from Swal-well Lake only exceeded this volume by 0.012 m3/s. Flow into Ellison Lakewas estimated at less than 0.051 m3/s in August on average. Flow in MVCbelow Ellison Lake was not measurable in August. Ellison lake level fell be-low the height of the beaver dams that obstructed flow at the outlet of EllisonLake. Flow in MVC at Reimche Road above Wood Lake was supplementedby tributary inflows and achieved an average of 0.024 m3/s (fairly similar toother years when the TFCS was in place in August).4. Diversions for DLC averaged 0.151 m3/s in September and outflows fromSwalwell Lake exceeded this by 0.083 m3/s (highest exceedence since June),enabling flows into Ellison Lake to average 0.088 m3/s. Flow in MVC belowEllison Lake averaged only 0.012 m3/s, despite attempted removal of beaverdams. All of this flow infiltrated into the streambed within a short distancedownstream, and the creek bed remained dry at Beaver Lake Road down-stream from Ellison Lake. Complete loss of re-established outflow fromEllison Lake in September due to infiltration has only been noted to occur in1442014 and 2015. Flow at Reimche Road averaged 0.024 m3/s due to tributaryinflows and groundwater seepage.5. DLC diversions in October averaged only 0.061 m3/s, and outflows fromSwalwell Lake exceeded this by 0.144 m3/s (highest exceedence), enablingan average flow into Ellison Lake of 0.217 m3/s. Once again, beaver damremoval was ineffective at the outlet of Ellison Lake and the flow in MVCbelow Ellison Lake averaged only 0.014 m3/s. Continuous flow was notachieved in MVC between Ellison Lake and Beaver Lake Road. However,tributary inflows increased flow in MVC at Reimche Road to an average of0.020 m3/s (lower than previous three years when TFCS was removed inOctober).6. Flows increased in November to 0.05 m3/s and averaged 0.084 m3/s throughuntil the end of December, aided by the increased Ellison lake level at thistime (Epp & Neumann, 2016).Ellison Lake levels were lower than previous due to less spilling from SwalwellLake in spring, high demands for water in UVC in May/June, and the TFCS notbeing closed after freshet in 2015. Flow in MVC, particularly near Ellison Lake,remained below the minimum guideline for kokanee spawning flows in the sum-mer and fall in 2015. On 21 September and 29 September 2015 several sandbagswere removed to make the opening larger, and on 26 October 2015, a small flowwas observed through the TFCS. Despite efforts to increase inflow to Ellison Lakeand to remove the beaver dams in order to re-establish flow in MVC, continuousflow was not achieved below (down to Beaver Lake Road) during spawning. El-lison Lake level did increase slowly during this time due to the additional releasefrom Swalwell Lake, decreasing DLC demands, and minimal outflow. Eventuallycontinuous flow was achieved in MVC by December (when Ellison Lake level washigh enough). The volume of water required to re-establish continuous flow in lieuof obstructions was underestimated in 2015. Beaver dam activity in concert withdrought conditions and high demands from DLC were the leading causes for notbeing able to achieve continuous flow in MVC during the spawning season in 2015(Epp & Neumann, 2016).145Available kokanee habitat in MVC is directly related to the creek’s flowrateand temperature. In 2015, the weighted usable habitat width [WUHW] remainedbelow 0.5m throughout the spawning season (15 September - 26 October 2015).For comparison, the WUHW in 2013 ranged between 0.5m on 15 September 2013to 3.5m on October 31, 2013 and the largest flux of fish counts occurred at the endof September in response to increased flow. Despite the inclination to formulate arelationship between increased flow and fish counts (Epp & Neumann, 2014), thehighest counts on record in the previous four years surprisingly occurred in 2015when flows in MVC remained lower than the previous four years at 0.02 m3/s (Epp& Neumann, 2016).MVC temperature (Figure 4.24) was measured with an RBR Solo-T temper-ature logger installed (approximately 20cm from channel bottom at location withdepth about 1.2m) 45m upstream from Wood Lake in 2015 - 2016 to investigateif there was any change in temperature between Reimche Road and Wood Lakealong this reach at low flows. Temperature data for Reimche Road was providedby Natasha Neumann (pers. corr., 9 June 2015) and by Hillary Ward (pers. corr., 20April 2016). Despite the variation in flow in recent years (Figure 4.21), there is notsignificant overall interannual variability in flow temperatures in MVC, althoughthere are notable daily, weekly, and monthly variations (Figure 4.24). There isalso not a significant difference in temperatures between Reimche Road (585m up-stream from Wood Lake) and 45m upstream from the mouth of MVC. Comparing30 minute interval temperatures from 16:00 on 8 July 2015 to 09:00 on 31 October2015, the average temperature difference between these two stations was 0.33◦C(warmer at the downstream location).2016 MVC temperatures were warmer than recent years between January andMay (logger removed on 12 May 2016 and remaining data obtained from BCMOE). The two recent years on record with data from 1 January to 12 May are2013 and 2015; on average (30min averaging period) the temperatures in 2016during this time period were 1.56◦C warmer than 2013 and 0.81◦C warmer than2015. Following this, temperatures remained near or below average (beginning ofJuly until mid-August) for the remainder of the year until mid-October 2016.14601-01 01-02 01-03 01-04 01-05 01-06 01-07 01-08 01-09 01-10 01-11 01-12Date [dd-mm]0510152025Temperature [°C]2016 at Mouth of Middle Vernon Creek into Wood Lake2016 at Reimche Road2015 at Mouth of Middle Vernon Creek into Wood Lake2015: At Reimche Road2014: At Reimche Road2013: At Reimche Road2012: At Reimche Road2011: At Reimche Road2010: At Reimche RoadFigure 4.24: Water temperature in MVC measured at Reimche Road (2010 -2016) and about 45m upstream from Wood Lake in 2015 - 2016.The temperatures near the outlet of MVC to Wood Lake have been over 17◦Cfrom around mid-May to the end of August in the past five years as the minimalflow in the creek responds to summer shortwave and long-wave radiation fluxesprior to entering Wood Lake. During the period of spring freshet until flow ap-proaches negligible levels at the end of July, the average flow temperatures in re-cent years have been 16.45◦C (1 April - 31 July 2013), 15.86◦C (16 April - 31 July2014), 16.10◦C (1 April - 31 July 2015), and 16.51◦C (1 April - 31 July 2016).Wood Lake is relatively small (volume: 2e8 m3 and average depth: 22m), yetit’s retention time is relatively long (18 - 30 years, with a majority of estimatesaround 30 years) (Anonymous, 1974; BC Research, 1974; Jensen & Bryan, 2001).Based on available 2015 MVC flow data, the average inflow for Wood Lake in2015 was 0.29m3/s and the retention time of Wood Lake based on 2015 data was147thus around 22 years. Note the median inflow is actually only 0.084m3/s, but theaforementioned mean value is skewed by the spring freshet flows. Analyses of thehydrology of MVC shows that during the modelled period in 2013, 2014 and 2015,there is neither considerable input of water nor thermal impact from inflowing wa-ters from MVC (Table 4.6)Table 4.6: Cumulative MVC inflow and water temperature during modelledperiods in 2013, 2014, and 2015.Year Modelled Period Total Cumulative Average Inflow Temperature [◦C]Inflow During Period From Beginning of Period[m3] and %Wood Lake to End of JulyVolume (negligible inflows after July)2013 14 May-15Oct 1.21e7 (6.10%) 19.2◦C2014 21 May-15 Oct 5.37e6 (2.70%) 17.55◦C2015 29 May-23 Dec 1.13e6 (0.57%) 18.65◦CThe interaction of MVC inflows with Wood Lake is of interest since several ofthe proposed solutions for Wood Lake that have been modelled involve increasinginflow into Wood Lake. Understanding how this additional flow interacts with thewater column in Wood Lake (i.e. similarly to during spring freshet) is importantin order to assess how these solutions may impact the thermal structure of WoodLake. During the low flow season, as water passed from the creek, across thesandbar (Figure 4.23) and into the shallow waters of the lake at this location, thewater warmed up to lake surface temperatures and no noticeable temperature orturbidity plume was observed from MVC entering Wood Lake (Figure 4.25) during2015. The following profiles (Figure 4.25) were conducted on 26 May 2015 whenthe flow in the creek was 0.54 m3/s.148Figure 4.25: (left)Temperature distribution at mouth of MVC into WoodLake along “Middle Transect” in Figure 4.26 on 26 May 2015. (Right)Turbidity distribution at mouth of MVC into Wood Lake along “Mid-dle Transect” in Figure 4.26 on 26 May 2015. Locations of CTD pro-files shown with black vertical lines. Distance along x-axis is measuredfrom south to north along the middle transect, away from the mouth ofMVC into Wood Lake.1490 200 mMVCMVC West TransectMVC Middle TransectMVC East TransectNFigure 4.26: Three CTD transects conducted at mouth of MVC extendinginto Wood Lake on 26 May 2015 and 25 April 2016.Despite low flows, MVC was sampled with the CTD on six occasions between26 May 2015 and 14 July 2015. Density, specific conductivity, turbidity, temper-ature, and flow data were recorded (Table 4.7). Note that as the flow decreasesfollowing freshet, the specific conductivity of the water in the creek increases sub-stantially. This tends to indicate a stronger influence from groundwater sources(i.e. ions from soils) as opposed to the lower conductivity measured during springfreshet (3 May 2016), that is diluted by runoff from the snow packs at higher ele-vations (R. Moore, Richards, & Story, 2008; Scrivener, 1989; Steven Weijs, pers.corr., June 2016).MVC inflow was observed to mix with the surface layer of Wood Lake onall days sampled (Table 4.7 and 4.8). Densities of inflowing water and residentwater in Wood Lake were calculated using TEOS-10 (not accounting for sus-pended solids). The average surface layer density during the stratified period was998.73 kg/m3 with a minimum of 997.27 kg/m3 on 30 June 2015 and maximumof 1000.1kg/m3 on 25 November 2015. MVC parameters (Table 4.7), along withobservations of the absence of a plume during low flow conditions and of inflowwarming after passing over the 50m long sandbar at the mouth of the creek where150water depth was much less than 1m (Figure 4.25), suggests that the low flows dur-ing 2015 mixed with the epilimnion as it entered Wood Lake. Hence nutrientscarried by inflowing waters are readily available for algal growth in the epilimnionduring spring and summer. BC Research (1974) compared inflowing creek temper-ature data with thermocline temperature data in 1972 and supports the observationthat MVC inflows mix with the epilimnion in spring and summer; however, in thefall, the lower flows in the creek may be cool enough relative to ambient lake tem-peratures to plunge down below the thermocline and mix with the water columnwithin and below the thermocline. Although winds generally serve to mix thesenutrients laterally in the lake’s epilimnion, the predominant summer north windsin the valley help to retain the inflowing MVC water in the south end of the lake(BC Research, 1974).Table 4.7: Measurements of background density, specific conductivity, tur-bidity, temperature, and flow in MVC during 2015-2016 field season.Date Distance Temperature Specific Turbidity Density Flowfrom [◦C] Conductivity [FTU] [kg/m3] [m3/s]Creek [uS/cm]Mouth [m]26 May 2015 52 17.14 164.01 2.87 998.82 0.5438.5 17.12 161.08 11.35 998.5911 17.73 194.09 2.51 997.0729 May 2015 28.5 20.34 167.11 10.63 998.20 0.5515 June 2015 32.5 18.36 256.99 7.33 998.63 0.2430 June 2015 33 18.99 578.94 10.06 998.63 0.0214 July 2015 36 17.73 660.63 35.30 998.90 0.023 May 2016 60 16.69 98.57 22.13 998.87 5.35On 3 May 2016, measurements in MVC were recorded with the CTD 60m up-stream from the mouth of MVC (50◦03.097’N, 119◦24.401’W) in 1m depth overa 2 minute period to determine the background parameters during freshet and howthese compared to the background values in the lake measured on 25 April 2016(Table 4.8). On 3 May 2016 the temperature of the inflow was considerably higherthan the maximum lake temperature recorded on 25 April 2016, the specific con-151ductivity was lower than the lake’s minimum specific conductivity, and the turbid-ity was considerably higher than the mean turbidity in the lake on 25 April 2016.Table 4.8: Mean and range of values for temperature, specific conductivity,and turbidity in MVC (60m upstream from Wood Lake) compared toWood Lake (entire basin) on 25 April and 3 May 2016.Parameter Wood Lake (25 April 2016) MVC (3 May 2016)Temperature 8.25◦C (4.79 - 14.21 ◦C) 16.69◦C (16.69 - 16.70◦C)Specific Conductivity 358.57 uS/cm (229.03 - 370.95 uS/cm) 98.57 uS/cm (98.21 - 98.76 uS/cm)Turbidity 1.82 FTU (0 - 114.32 FTU) 22.13 FTU (7.97 - 119.33 FTU)4.6.1 MVC 2016 Freshet PlumeCTD profiles along three northward transects into Wood Lake from the mouth ofMVC (one along the east and west edges of the plume and one in the middle ofthe plume) (Figure 4.26) on 25 April 2016 captured the freshet plume from MVCmixing with the epilimnion of Wood Lake. The highest turbidity and the most sus-pended matter from the creek was observed along the west transect. The plumewas easily visible in the surface layer (concentrated in the upper 5m) up to ap-proximately 585m from the creek mouth along the middle transect (Figure 4.28).The average background turbidity in Wood Lake on this day was 1.82 FTU (Fig-ure 4.30). Along the west transect (Figure 4.27), turbidity values in the plumeover 4FTU were measured near the surface for 347m from the creek mouth (upto a depth of 5.7m at 147m from the creek mouth). Along the middle transect(Figure 4.28), turbidity values in the plume over 4FTU were measured near thesurface for 706m from the creek mouth (up to a depth of 5.6m at 92m from thecreek mouth). Along the east transect (Figure 4.29), turbidity values in the plumeover 4FTU were measured near the surface for 691m from the creek mouth (up toa depth of 3.8m at 240m from the creek mouth). The turbidity plume was most evi-dent in the west and middle transects. In the east transect, it appears that the plumeplunges slightly below the surface and is more concentrated in the 2.5-5m depthrange after the first 250m from the creek mouth. BC Research (1974) noted thissame flow pattern on 1 June 1972 with solids deposition occurring in the southwestquadrant of the lake and by 22 June 1972, as the flow decreased, the plume adjusted152its direction towards the middle of the lake. This pattern of lower flows moving to-wards the middle/east side of the lake following freshet was also observed visuallyin the previous spring/summer of 2015.Figure 4.27: Turbidity [FTU] distribution from mouth of MVC into WoodLake along “West Transect” on 25 April 2016 (Figure 4.26).Figure 4.28: Turbidity [FTU] distribution from mouth of MVC into WoodLake along “Middle Transect” on 25 April 2016 (Figure 4.26).153Figure 4.29: Turbidity [FTU] distribution from mouth of MVC into WoodLake along “East Transect” on 25 April 2016 (Figure 4.26).Figure 4.30: Background Turbidity [FTU] in Wood Lake on 25 April 2016.Distance measured from south end of lake to north end of lake alongCTD transect of lake (Figure 3.2).154The inflow from MVC on 25 April 2016 also had a notable low specific con-ductivity signal, indicating that this water is low in total dissolved solids and ioniccontent. The distribution of specific conductivity in Wood Lake (low conductivitysignal from MVC inflow) is visible along the same three transects of the MVCplume on 25 April 2016 (Figure 4.31, 4.32, and 4.33). The background specificconductivity in Wood Lake (Figure 4.34) on this day was fairly uniform and aver-aged 358.6 uS/cm. Along the west, middle, and east transect specific conductivityvalues in the epilimnion below 300 uS/cm were observed from the creek mouthout to 294m, 456m, and 240m from the creek mouth respectively. These specificconductivity measurements and observations (Figure 4.31, 4.32, and 4.33) confirmobservations made in analyzing turbidity plots: the plume is spread across the sur-face layer along the west and middle transects, and appears to plunge down slightlybelow the surface along the east transect after a distance of about 250m from thecreek mouth. Note that the specific conductivity in MVC measured on 3 May 2016was 98.6 uS/cm with very little variation (+/- <0.4 µS/cm) (Table 4.8).Figure 4.31: Specific Conductivity [µS/cm] distribution from mouth of MVCinto Wood Lake along “West Transect” on 25 April 2016 (Figure 4.26).155Figure 4.32: Specific Conductivity [µS/cm] distribution from mouth of MVCinto Wood Lake along “Middle Transect” on 25 April 2016 (Fig-ure 4.26).Figure 4.33: Specific Conductivity [µS/cm] distribution from mouth of MVCinto Wood Lake along “East Transect” on 25 April 2016 (Figure 4.26).156Figure 4.34: Background Specific Conductivity [µS/cm] in Wood Lake on25 April 2016. Distance measured from south end of lake to north endof lake along CTD transect of lake (Figure 3.2).The plume (as marked by lower conductivity and higher turbidity signals) trav-els along the west transect, mostly concentrated in the upper 5m of the water col-umn for about 200m and then begins to become diluted by the background watervia entrainment and is more clearly seen only in the upper 2-3m of the water col-umn. Along the middle transect, MVC water appears to flow across the upper4-5m of the water column for approximately 600m from the mouth of the creek,after which the signal is more clearly seen only in the upper 2-3m. The flow travelsalong the east transect in the upper 4m for about 250m and then is more clearlyseen along a path between 2.5 - 5m depth until over 600m away from the creekmouth.The temperature signal of the inflowing water, at 14.16◦C on 25 April 2016,(45m upstream from Wood Lake in MVC) was quickly lost and not discernibleas the water mixed with the surface layer of Wood Lake (temperature between11.63◦C and 14.19◦C) on 25 April 2016.1574.7 Seiching in Wood LakeWood Lake generally flows northward through Oyama Canal into Wood Lake;however, previous observations suggest that winds and seiching play an impor-tant role on governing the exchange of water between these two basins during thesummer and fall low inflow seasons (BC Research, 1974; BCWIB, 1974; LAC,2010). Seiche motions have been well studied in literature (Heaps, 1961; Heaps& Ramsbottom, 1966; Hutter, 1984; Mortimer, 1952), and first and second verti-cal mode oscillations as well as bottom currents were observed in Wood Lake in1981 (Wiegand & Chamberlain, 1987). Wood Lake typically responds to diurnalwinds by initially oscillating at the first mode frequency and subsequently at thesecond-mode, which persist because they have a period similar to that of the diurnalwinds. The largest currents in Wood Lake are observed with the large amplitudefirst mode seiche, and the second mode oscillations contribute to the magnitudeand duration of these currents (Wiegand & Chamberlain, 1987). These maximumcurrents must coincide with the largest currents through Oyama Canal, given thatthere is no physical structure in place to dampen or arrest these flows prior to en-tering the canal at the north end of Wood Lake. These currents are believed to beresponsible for controlling the exchange of flow between these two basins duringthe low flow season.From August through to mid-November, Wood Lake is nearly constantly inmotion at depth within the thermocline either while subjected to wind stress orthe free internal seiches that follow (Figure 4.35 to Figure 4.38). The dominantinternal seiche mode is V1H1 (Boegman, 2010), wherein isotherms at all depthswithin the thermocline move in concert during the seiche (Figure 4.10, Section 4.3)and the movement is uninodal (Mortimer, 1952). During these events in August -November (Figure 4.35 to Figure 4.38), an upward tilt of the isotherms (decreasein temperature at a particular depth) at one end of the lake is matched with a down-ward tilt of the isotherms at the opposite end of the lake (increase in temperatureat a particular depth). However, in late November, there are a number of occa-sions (Figure 4.38) where a decrease in temperature at one end (i.e. upward tiltof isotherms) is not coupled with a change in the isotherms at the opposite endof the lake. During August and the first half of September when the depth of the158thermocline is shallower in the water column, the 10m depth temperature loggersacquired the strongest reflection of the seiche movements in Wood Lake. For ex-ample, during oscillations on 21-22 August 2015, the temperature at the south endof Wood Lake at 10m depth oscillates from 21.8◦C at 19:38 on 21 August to 8.5◦Cat 03:06 on 22 August, and then back to 21.5◦C at 06:32 on 22 August. The periodof this oscillation was approximately 12 hours. Meanwhile, at the north end ofWood Lake at 10m depth, the temperature oscillates from 10.9◦C at 20:50 on 21August to 18.9◦C at 01:27 on 22 August, and then back to 10.4◦C at 06:59 on 22August. The period of this oscillation was about 10hrs. However, in late Septemberand through October and November, the thermocline is deeper in the water columnand thus the 15m depth loggers acquired the strongest signal. For example, duringoscillations on 12-13 October 2015, the temperature at the south end of Wood Lakeat 15m depth oscillates from 14.5◦C at 21:36 on 12 October to 6.9◦C at 04:41 on13 October, and then back to 14.3◦C at 13:52 on 13 October. The period of thisoscillation was about 16hrs. Meanwhile, at the north end of Wood Lake at 15mdepth, the temperature oscillates from 7.9.0◦C at 20:47 on 12 October to 13.6◦Cat 08:27 on 13 October, and then back to 7.8◦C at 12:31 on 13 October. The pe-riod of this oscillation was about 15.5hrs. In August, temperature fluctuations ofover 13◦C were observed at 10m depth; in September, October, and November,temperature fluctuations of over 10◦C, 7◦C, and 5◦C respectively were observed at15m depth. Note the decreasing amplitude in temperature fluctuations especiallyin November (Figure 4.38) as the lake nears fall turnover and the temperaturesapproach isothermal.159Figure 4.35: Overlaid thermistor chain data from 5m (top),10m (middle), and15m (bottom) depth at the north (orange) and south (blue) ends ofWood Lake from 19-31 August 2015. Refer to Table 3.3 and Figure 3.1for chain locations.160Figure 4.36: Overlaid thermistor chain data from 5m (top),10m (middle), and15m (bottom) depth at the north (orange) and south (blue) ends ofWood Lake from 1-30 September 2015. Refer to Table 3.3 and Fig-ure 3.1 for chain locations.161Figure 4.37: Overlaid thermistor chain data from 10m (top),12.5m (middle),and 15m (bottom) depth at the north (orange) and south (blue) ends ofWood Lake from 1-31 October 2015. Refer to Table 3.3 and Figure 3.1for chain locations.162Figure 4.38: Overlaid thermistor chain data from 10m (top),12.5m (middle),and 15m (bottom) depth at the north (orange) and south (blue) endsof Wood Lake from 1-30 November 2015. Refer to Table 3.3 andFigure 3.1 for chain locations.4.8 Temperature Gradient Through Oyama CanalAs discussed in Chapter 3, temperature loggers were also installed at the north andsouth ends of Oyama Canal (Section 3.3.2, Table 3.3, Figure 3.1), allowing forcalculation of the daily average range in temperature through the canal from thenorth to the south end (Figure 4.39). These temperature differentials are evaluated163against the temperature range (maximum - minimum) observed in the surface layer(upper 2m) across the north-south transect in Wood Lake on all field days (temper-atures averaged vertically for each CTD cast in upper 2m). From all field days, theaverage temperature difference across Wood Lake was 0.60◦C. The average tem-perature range from one end of the canal to the other on all days with thermistorsinstalled was 0.96◦C. The average temperature range through the canal during themodelled time period (2015) with thermistor data was 0.72◦C from 30 June 2015to 1 January 2016. The time period for modelling described in Section 4.11.2 was29 May 2015 to 23 December 2015. For each field day in which CTD profiles ofthe lake were performed, the average difference between the range of temperaturesobserved in the upper 2m of the lake as compared to the average range on thatsame day through the canal was 0.49◦C. BC Research (1974) noted that warmerWood Lake water tends to discharge along the surface of cooler Kalamalka Lakewaters near Oyama Canal. However, it was observed in the current study that tem-peratures at the north end of Wood Lake are often equal, or even cooler at times,than those in the shallow bay at the south end of Kalamalka Lake. Similarly to dis-cussion earlier regarding MVC inflow temperatures, the temperature effects of anyinflowing water from the shallow bay in the south end of Kalamalka Lake can alsobe neglected when considering the evolution of basin-wide thermal stratificationduring a single year.164[30-06] [01-08] [01-09] [01-10] [01-11] [01-12] [01-01] [01-02] [01-03] [01-04] [25-04]Day [dd-mm]123456Average Temperature Range [°C]2015 2016Partial Day Thermistor Chain RecordsContinuous T-Chain Record: Aug 19 - Sept 30Continuous T-Chain Record: Oct 7 - Jan 1Continuous T-Chain Record: Jan 8 - Apr 25Upper 2m of Wood Lake (CTD Profiles)Figure 4.39: Average daily temperature range through Oyama Canal (northand south ends) compared to average lateral temperature range acrossthe N-S transect in Wood Lake in the upper 2m of water column. Forthe canal, the “average range” refers to the daily averaged difference intemperatures between the north and south ends of the canal. For WoodLake, the average range refers to the range in surface layer (upper 2m)temperatures observed along the N-S CTD transect.Assuming a steady state condition, and ignoring evaporation from the lake sur-face, the average annual outflow can be assumed equal to the average MVC inflowfor 2015 (0.29m3/s). A relatively long residence time (20-30 years) and low annualinflows and outflows suggest these thermal and mass fluxes do not contribute sig-165nificantly to the development of seasonal thermal stratification and therefore canbe neglected in simulations of less than one year. Despite physical observations ofsurface flow through the canal suggesting wind and seiche moments control direc-tion of flow between basins, accounting for this complex transfer between basin inmodelling is unnecessary as the inflows and outflows are negligible.4.9 Total Heat ContentUsing a spatial average temperature profile for each field day based on the deepest5 central CTD casts at stations S9 - S13 (Section 3.4.1), the heat content of the lakewas estimated throughout the field season. Figure 4.40 presents the heat content[J/m2] for Wood Lake, as well as the heat content separated into the annual averagephotic (depth to which light reaches 1% of that at the surface: annual average is11.75m) and aphotic zones (water below the photic zone) of the lake. There are twoapparent anomalies in the time series of the heat content (Figure 4.40). The firstanomaly (a) is a local minimum in the heat content of the lake (on 29 July 2015)occurs in conjunction with the onset of the marling event and a local minimum indaily average air temperatures. The second anomaly (b) appears as a second localminimum in the heat content of the lake and a deviation from the overall averageslope at this time (9 September 2015).16604-05 01-06 01-07 01-08 01-09 01-10 01-11 01-12 01-01 01-02 01-03 01-04Day [dd-mm]02468101214Total Heat Content[Joules/m2 ]×10520162015(a)(b)Total Heat Content [J/m 2]Heat Content in Average Aphotic Zone [J/m 2]Heat Content in Average Photic Zone [J/m 2]Figure 4.40: Heat content per unit surface area [J/m2] in Wood Lake from 4May 2015 to 25 April 2016.While the heat content of the lake is influenced by many factors, one of thesefactors is the marling event’s effect on light attenuation in the water column (Koschelet al., 1983). Light extinction has been estimated from PAR (Section 4.1). A localminimum in the heat content of the lake on 29 July 2015 corresponds to a localminimum in the heat content of the photic zone and a local maximum (0.43m-1) inthe light extinction coefficient (Figure 4.41). Herein we see an increase in the lightextinction coefficient and a decrease in the heat content of the lake in mid-July, cor-responding to the time at which the marling was most prominent. Whitening of the167surface layer due to the calcite crystals significantly increase the surface albedo ofthe lake, thereby reducing the amount of light energy that penetrates into the lake(Kirk, 1994; Mortimer, 2004; Wetzel, 2001). Certainly there are a number of otherconcurrent factors involved in creating this local minimum in the heat content ofthe lake during this time, including the fact that calcite precipitation is an endother-mic reaction and therefore removes heat energy from the surface layer of the lake[CaCO3 + CO2 + H2O ←→ Ca+2 + 2HCO-3 + heat] (Brunskill, 1969; Williams,1972).04-05 01-06 01-07 01-08 01-09 01-10 01-11 01-12 01-01 01-02 01-03 01-04Date [dd-mm]12345678910111213Total Heat Content [Joules/m2 ]×1050.30.350.40.450.50.550.6Extinction Coefficient [m-1 ]20162015(right axis) Light Extinction Coefficient [m -1](left axis) Total Heat Content [J/m2](left axis) Heat Content in Average Aphotic Zone [J/m2](left axis) Heat Content in Average Photic Zone [J/m2]Figure 4.41: Overplot of heat content in Wood Lake and light extinction co-efficient (Kd(PAR)) from 4 May 2015 to 25 April 2016.168The meteorological trends (Figure 4.42) were also a factor during this period.The 7-day average incoming solar radiation peaks during the week of 29 June 2015to 5 July 2015 at 304 W/m2 and then slowly decreases thereafter, other than asmall secondary local maximum on 3-9 August 2015 at 247.99 W/m2. The 7-day mean daily average temperatures peak at 24.57◦C on 6-12 July 2015 and thendecrease to a local minimum of 20.69◦C (3-9 August 2015), before increasingagain to a second local maximum of 22.94◦C during the week of 10-17 August2015. This meteorological phenomenon also contributes to the local minimumin the heat content experienced during this time period of 14 July - 11 August2015 and the two local maximum values during peak summer stratification. Thishypothesis has been tested using DYRESM (Section 5.1.2).04-05-1501-06-1501-07-1501-08-1501-09-1501-10-1531-10-15Day [dd-mm-yy]050100150200250300350Short Wave Radiation Flux [W/m2 ]Average Daily Solar Radiation [W/m2]7-Day Average Solar Radiation [W/m2]04-05-1501-06-1501-07-1501-08-1501-09-1501-10-1531-10-15Date [dd-mm-yy]510152025Average Daily Air Temperature [°C]Daily Average Air Temp. [°C]7-Day Mean of Daily Average Air Temp. [°C]Figure 4.42: Short wave radiation flux (Coral Beach Farms weather station)adjacent to Wood Lake and daily average air temperature (GCEC,2016) at Kelowna Airport from 4 May - 31 October 2015.With regards to the second anomaly in heat content ((b) in Figure 4.40), reviewof CTD and thermistor chain field data from 9 September 2015 reveals that the lakewas experiencing extensive subsurface motion due to seiching during the time thatfield data was collected. Moderately strong diurnal/semidiurnal north/south alter-nating winds (peak hourly average speeds of 18-21 km/hr) occurred in the days169leading up to 9 September 2015 (GCEC, 2016). Such winds have been noted togenerate seiche motions within the thermocline (Wiegand & Chamberlain, 1987).Thermistor chain data (Figure 4.36) from Wood Lake confirms the seiching periodin the thermocline is 0.5 days during this event, which lasts for several days withdecreasing intensity. Observing the temperature profiles from 9 September 2015(Figure 4.43), if an arbitrary depth is selected (i.e. 12m), the temperature across thelake at this depth ranges from 8.71◦C to 17.58◦C. Likewise, an arbitrary tempera-ture such as 15◦C occurs between 7.4m -13.4m depth depending on location alongtransect. It is concluded that field data should not be used to estimate heat contentvia simple formulae that do not precisely account for tilting isotherms when thelake is experiencing such extreme subsurface seiching movement. Although theselected method for determining an average temperature profile and heat contentfor Wood Lake (Section 3.4.1) from CTD profiles has proven extremely reliableand robust throughout the field season, it has its limitations under extreme seichingevents. On this particular day it appears to have underestimated the heat content ofthe lake during peak seiche motions (Figure 4.40).1705 10 15 20Temperature [° C]05101520253035Depth [m]All Profiles on Given DayCasts at S9-S13Average Temperature [° C] vs. Depth [m] for S9-S13Figure 4.43: All temperature profiles (thin solid lines) from Wood Lake on 9September 2015 during seiching; also showing the temperature profilesfrom casts #9-13 (dashed lines) and the average temperature profile(thicker solid line).4.10 Epilimnion and Hypolimnion Heating RatesEpilimnion heating rates and heat content are strongly linked with ambient air tem-peratures, which vary substantially from year to year (BC Research, 1974), whilehypolimnion heating rates are influenced by several factors including vertical ex-tent of downward mixing of heat from the epilimnion, the strength of summer171stratification, the extent of upwelling, and the influence of groundwater influxes.In 2015, two peaks in Wood Lake’s total heat content were observed (Figure 4.40);one was observed on 14 July 2015 at the onset of marling, and the second wasobserved on 19 August 2015 following a local minimum. The representative aver-age temperature profiles on each field day, along with high resolution bathymetricdata, enabled accurate estimation of the energy required to raise lake temperaturesfrom an isothermal condition at 4◦C to peak stratification, known as the “total sum-mer heat income” (Table 4.9) (Hutchinson (1957) as discussed by (BC Research,1974)). The same calculations were also performed for 2013-2015 using data col-lected at 2m intervals by BCMOE at station 500848, under the assumption that thissingle temperature profile serves as a “good enough” representation of the thermalstructure of the lake. In this case, the layers of the lake were 2m thick (rather than0.2m layer thickness used in analysis of CTD data from 2015). The epilimnion,metalimnion, and hypolimnion heating rates are also provided (Table 4.9). Sinceheating rate is not constant throughout the season, more frequent higher resolutionmeasurements will result in more precise prediction of maximum heat content.The lake’s heat income was calculated using Equation 4.1 in accordance with BCResearch (1974).Q =n∑i=1CpρiAo∗ (Vi(Ti−4)) (4.1)Where Q is total heat income in J/m2, Cp is the specific heat capacity of water, ρiis layer density, Ao is surface area of the lake, and Vi and Ti are the volume andtemperature of each layer respectively. Units must be consistent. The maximumvalue obtained for Q is equal to the summer heat income (BC Research, 1974).Note: E = epilimnion; M = metalimnion; and H = hypolimnion.172Table 4.9: Total summer heat income [J/m2 and heating rates [◦C/day] inWood Lake for 2013 - 2015.Year Total Summer Income Heating Rates [◦C/day](Using Many Finite Layers) E = Epilimnion; M = Metalimnion;[cal/cm2] H = Hypolimnion2015 4 May - 14 July: 8.89e8 (4 May - 14 July)4 May-19 Aug:8.82e8 E:0.164; M:0.051 ; H:-0.0122015 15 April - 15 July: 8.42e8 (15 April - 15 July)15 April-12 Aug:8.68e8 E:0.179; M:0.112 ; H:0.0162014 23 April - 14 Aug: 8.44e8 (23 April - 14 Aug)E:0.157; M:0.101 ; H:0.0082013 18 April - 14 Aug: 8.71e8 (18 April - 14 Aug)E:0.158; M:0.088 ; H:0.012Average 8.65e8 E:0.165 ; M:0.088 ; H:0.006CTD data suggests that the absolute maximum heat content in 2015 occurredaround 14 July 2015, with a total summer heat income of 8.87e8 J/m2. The secondpeak in heat content occurred on 19 August 2015 yielding a slightly lower totalsummer heat income of 8.83e8 J/m2. The lake’s heat content did not change sig-nificantly during this time (approximately 1% from 14 July to 19 August 2015).In 2015, temperatures began to decline following August as daily average tem-peratures cooled and strong wind events became more regular. In 2014, the totalsummer heat income from 23 April to 14 August was 8.45e8 J/m2, and in 2013from 18 April to 14 August, the total summer heat income was 8.70e8 J/m2.Interestingly, BCMOE data (collected at one location at 2m intervals) wouldhave suggested that the second peak in heat content in August was higher, althoughthe difference between peaks on 15 July 2015 and 12 August 2015 from this datais small (approximately 3-4%). One of the key findings from this study was thesignificance of seiching in Wood Lake during summer stratification. As noted pre-viously with regards to temperature profiles from 9 September 2015 (Figure 4.43)using a single profile at a non-central and non-deepest location in the lake as beingrepresentative of the temperature structure of the lake can be very misleading whenthe isotherms are in motion. Combined with the poor resolution of this data (2mintervals), this would be sufficient to explain the discrepancy between the calcu-173lated heat incomes from CTD data compared to that from field data collected byBCMOE at station 500848.Based on the current study: from 4 May to 14 July 2015 the epilimnion heatingrate was 0.16◦C/day (4.9◦C/month) and the hypolimnion heating rate was-0.012◦C/day (-0.36◦C/month, i.e. cooling). The groundwater influx to Wood Lakecan be observed by hypolimnion water temperatures being maximum in April/-May and then declining thereafter while surface water temperatures continue torise (BC Research, 1974). The epilimnion, metalimnion, and hypolimnion cooledby 1.62◦C, 1.48◦C and 0.14◦C respectively from 14 July to 19 August 2015. Basedon BCMOE data, from 15 April to 15 July 2015 the epilimnion heating rate was0.18◦C/day (5.37◦C/month) and the hypolimnion heating rate was 0.016◦C/day(0.47◦C/month). BCMOE data also suggests that the epilimnion, metalimnion,and hypolimnion cooled by 0.82◦C, 2.02◦C and 0.46◦C respectively from 15 Julyto 12 August 2015. In 2014, from 23 April to 14 August, the epilimnion heatingrate was 0.16◦C/day and the hypolimnion heating rate was 0.0084◦C/day. In 2013,from 18 April to 14 August, the epilimnion heating rate was also 0.16◦C/day andthe hypolimnion heating rate was 0.012◦C/day. Note the similarity in epilimnionheating rates between these three years (0.16-0.18 ◦C/year). The cooler than ex-pected (for the lake size and depth) hypolimnetic warming rate is attributed to thestrong influence from cold groundwater sources (Northcote et al., 1974).1744.11 Dynamic Reservoir Simulation Model (DYRESM)Wood Lake thermal stratification was modelled using DYRESM (v4). The modelwas calibrated using 2015 field data from the current study and validated with 2013and 2014 data from BCMOE. The model successfully reproduced Wood Lake’sobserved thermal structure during the stratified season from May until December(2015), and was used to predict effects of changes in climate and/or hydrology.In future, the model could be coupled with CAEDYM (Computational AquaticEcosystem Dynamics Model) to additionally study lake biology/chemistry.A recently updated bathymetric map (sec 3.4.2) developed from a dual beamsonar survey of Wood Lake conducted by Raphael Nowak in July-August 2015 wasfound to improve calibration accuracy of DYRESM (not shown). In DYRESM,temperature profiles are output as temperature-height data using the elevation ofthe lake’s deepest position for reference. However, CTD temperature profiles arerecorded as temperature-depth data from the surface to the lake bottom, but notnecessarily at the lake’s deepest location. Thus, comparison of DYRESM resultswith CTD observations required establishment of a common datum, which is de-scribed in Section 3.4.2 and Appendix A.The final calibrated model has been evaluated in terms of its replication of spa-tial average temperature profiles (Section 4.3) from each field day between 29 May2015 and 23 December 2015 as well as the calculated total heat content in the lake.root mean squared error (RMSE) and mean absolute error (MAE) for both of theseevaluation techniques are provided and discussed in the following sections. RMSEand MAE are statistical parameters frequently used in evaluating modelled results.RMSE is the average deviation (error) in the simulated (predicted) values fromactual values (larger absolute errors are weighted more heavily than small errorsbecause the errors are squared in the calculation). MAE is the average absoluteerror between predicted values and actual values (all errors are weighted equally)(Budka et al., 2014; Chai & Draxler, 2014). RMSE and MAE are defined as:RMSE =√1n∗∑(Xpredicted−Xobserved)2 (4.2)175MAE = (1n)∗∑ |Xpredicted−Xobserved | (4.3)Where n is the number of observations (Xobserved) being compared to model out-puts (Xpredicted)(Budka et al., 2014; Chai & Draxler, 2014). The model has beenvalidated using profile data from mid-May to mid-October in 2013 and 2014 (Sec-tion 4.11.4). Validation of DYRESM’s simulation of inflows and outflows (neces-sary for modelling management scenarios in Section 4.12.1) was established basedon MVC flow and temperature data (Section 4.6) during spring freshet in 2015and 2016 when corresponding CTD field data in Wood Lake was available (Sec-tion 4.11.5). RMSE and MAE are also discussed with regards to model validation.4.11.1 DYRESM Model ParametersDYRESM was provided by CWR with generic values for numerous parameters(Table 4.10) that were all evaluated and either held constant or varied within rea-sonable limits during model calibration. The 2015 marling of Wood Lake, as wellas variations in concentrations of algal and suspended particulate matter, changedthe optical properties of the water throughout the year. To allow for changing lightextinction and surface albedo, DYRESM was run in monthly segments, each ini-tialized with the results of the previous segment. All other parameters were heldconstant between monthly segments. Albedo varies according to time of year andlatitude (Imerito, 2015a) and is also a function of lake surface color, thereby be-ing affected by the 2015 marling event (S. Effler et al., 1987; Richardson, 1982;Strong, 1978; Wetzel, 2001). Monthly average light extinction and surface albedovalues were selected for each segment of the model. Generic pre-calibration valuesprovided by CWR and final calibrated values for all DYRESM parameters are asfollows (Table 4.10). Additional detailed information and literature review regard-ing the calibration of each of these coefficients is provided in Appendix A.176Table 4.10: DYRESM Model Parameter CalibrationDYRESM Parameter Pre- Calibrated Value (References)CalibratedMinimum Layer Thickness N/A 0.2mMaximum Layer Thickness N/A 0.6mWind Factor 1.0 1.5Effective Surface Area 1.0e7m2 8.31e7m2[80%] (Parshotam, Özkundakci, McBride, & Hamilton, 2015),Coefficient (Özkundakci, Hamilton, & Trolle, 2011)Critical Wind Speed 3.0 m/s 3.0 m/sShear Production Efficiency 0.06 0.06Potential Energy Mixing 0.2 0.2EfficiencyWind Stirring Efficiency 0.4 0.4BBL Dissipation Coefficient 1.4e-5 1.0e-6(Shimizu & Imberger, 2008; Zohary, Sukenik, Berman, & Nishri, 2014)Emissivity of Water Surface 0.96 0.97 (Davies, Robinson, & Nunez, 1971; Derecki, 1976)citepRodgers-1961,Anderson-1954Mean Albedo of Water 0.08 Varies by month according to equation based on solar zenithangle (0.061-0.099) (Imerito, 2015a). During Marling: values in excess of0.17 reported on Lk. Michigan & Ontario (Richardson, 1982; Strong, 1978);peaks ∼ 0.1-0.16 observed in Owasco Lake, NY (S. Effler et al., 1987);but, could potentially increase reflectance 5-10 fold, consideringmeasured reflectance observed during coccolithophore blooms(phytoplankton with calcite shell) (Balch, Holligan, Ackleson, & Voss, 1991).(Brand, 2006). High reflectance observations made by:S. W. Effler, Wodka, and Field (1984); Strong (1978),Weidemann et al. (1985) & Peng and Effler (2011).Tanentzap, Hamilton, and Yan (2007) calibrate DYRESM using arange between 0.08-0.12 in Clearwater Lake, Ontario.Vertical Mixing 200 700 (Perroud, Goyette, Martynov, Beniston, & Annevillec, 2009)Coefficient (Özkundakci, Hamilton, & Trolle, 2011; Parshotam et al., 2015)Time of Day for N/A 39600s-46800s (after midnight)Output of Results [11:00AM - 13:00PM] (average time of day for field profiles)Time-Step N/A 1800s (30min)Bulk Aerodynamic Momentum 1.3e-3 1.3e-3Transport CoefficientMean Light Extinction N/A Monthly average values based on PAR data (Section 4.1)Coefficient June: 0.35, July: 0.41, Aug: 0.38,Sept: 0.31, Oct: 0.31, Nov: 0.48, Dec: 0.56. Modellingother years (2013/2014): average value outside 2015 marling: 0.33 m−1.177Table 4.10: (Continued) DYRESM Model Parameter CalibrationDYRESM Parameter Pre- Calibrated Value (References)CalibratedBubbler Entrainment 0.006 0.006 - not relevant (no artificial destratification system)CoefficientBuoyant Plume Entrainment 0.083 0.083 (H. B. Fischer et al., 1979) not relevant (no subsurface inflowCoefficient pipes)CAEDYM Switch N/A FalseActivate Destratification N/A FalseSystemActivate non-neutral N/A Falseatmospheric stabilityList of Output Selections N/A Salinity, Temperature, DensityBased on 2015 field data and evaluation of the Lake Number (LN) , Wedderburnnumber (W), Inflow Lake Number (LNi), Inflow Froude Number, Outflow FroudeNumber and Burger Number (Si) (Appendix A), the current model satisfies theone-dimensional assumption that validates DYRESM modelling a lake as a seriesof horizontal layers during the period of interest (Imerito, 2015a; Papadimitrakis,2011; Saddek & Casamitjana, 2013; Shintani, de la Fuente, Niño, & Imberger,2010). From 4 May 2015 to 25 November 2015 and from 30 March 2016 to 25April 2016 the assumptions requiring LN >> 1, W >> 1, LNi and Si > 1 (Si ∼ 1 on25 November) are strongly satisfied. Given the strength of summer stratification inWood Lake and values calculated for these parameters (Appendix A), it is reason-able to assume that the one-dimensional assumption would be satisfied from Mayto November in all years. During near isothermal condition (25 November 2015 -30 March 2016) the lake is unstable and does not satisfy the 1-D approximation.Nevertheless, the model continued to perform well (i.e. reproduce spatial averageprofiles) from 25 November 2015 to 23 December 2015. After this, ice cover beganto develop on the lake surface intermittently. The model reports an error and endsthe simulation when simulated water temperatures drop below 0◦C and thereforethe simulation period was ended on 23 December 2015.Given the minimal inflows and outflows observed relative to the total volumeof Wood Lake from May to December (Section 4.6), the lake has been simulated asa closed system with zero inflow/outflow. Review of modelled results (subsequentsections) supports the validity of this simplification.1784.11.2 2015 Modelled Results: Modelled and Observed TemperatureProfilesCTD field data was collected on 29 days between 29 May 2015 and 23 Decem-ber 2015, producing a time series of 29 representative average temperature pro-files for direct comparison with DYRESM outputs (Figure 4.44). All depth valuesfrom field profiles were converted to height above datum values (Section 3.4.2 andAppendix A) so that the profiles could be overlaid for a visual comparison. Thecalibrated model did an excellent job of replicating hypolimnion and epilimnionlayer temperatures in 2015, only differing very slightly in the epilimnion duringperiods of intense warming (i.e. June) or rapid cooling (i.e. late September andlate October). The largest discrepancy between the modelled results and the fieldprofiles that required hundreds of calibration simulations to refine was the depth ofthe surface layer and shape of the thermocline, particularly in July and September- November; nevertheless, for a one dimensional model, the overlaid profiles arevery nearly matched for these seven months in 2015.1790 10 20 300102030May290 10 20 30Jun150 10 20 30Jun300 10 20 30Jul140 10 20 30Jul210 10 20 30Jul290 10 20 30Aug40 10 200102030Aug110 10 20Aug190 10 20Aug250 10 20Sep50 10 20Sep90 10 20Sep160 10 20Sep230 10 200102030Height Above Datum (Lake Bottom) [m]Oct10 10 20Oct70 10 20Oct140 10 20Oct210 10 20Oct280 10 20Nov50 10 20Nov110 5 10 150102030Nov180 5 10 15Nov220 5 10 15Nov250 5 10 15Temperature [°C]Dec20 5 10 15Dec100 5 10 15Dec160 5 10 15Dec23Figure 4.44: Comparison of DYRESM modelled results (dashed line) withspatial average temperature profiles (solid lines) from 28 field daysbetween 29 May 2015 and 23 December 2015.In order to calculate the RMSE of temperature profiles, the representative aver-age field profiles and the DYRESM output profiles were averaged into equivalent0.5m depth bins and then directly compared to each other layer-by-layer through-out the water column for each day to develop a depth-averaged RMSE temperaturedifference for each day. The overall RMSE for 2015 was 0.51◦C (5.29%), and theMAE was 3.31% (Figure 4.45).In order to evaluate the RMSE between modelled results and field observa-180tions, RMSE values were compared to the average lateral layer temperature vari-ation (range) in Wood Lake on each field day. Using the same 0.5m layers dis-cussed above, the temperature range across the lake in each layer was calculated(i.e. maximum - minimum measured temperatures in each layer). These valueswere then averaged throughout the water column. Note this is a depth-averagedtemperature variance (range); not simply the range (maximum - minimum) of tem-peratures measured in lake. Accordingly, the average temperature range in the lakeacross all 0.5m layers on any given day during the simulation period (2015) was0.88◦C (Figure 4.45). Comparing the RMSE between the spatial averaged fieldprofiles and DYRESM output to the average layer-by-layer temperature variancein the lake on each day (Figure 4.45), it is noted that many of the days with a larger“apparent RMSE” in the modelled results correspondingly also have a larger aver-age lateral temperature range across the lake. For example, consider 14 July 2015,which is the day with the largest apparent RMSE between modelled and observedresults (RMSE = 1.0◦C). The average lateral temperature range at any given depthin the water column on this day was 1.7◦C. Therefore, the model accurately pre-dicts temperatures in the lake at every 0.5m depth interval within 1◦C when theaverage temperature range in the lake at any given depth is 1.7◦C. One of the otheroccasions with a large RMSE is 23 September 2015. Observed temperature pro-files (Figure 3.4) and thermistor chain data (Section 4.7) from 23 September revealsignificant movement of the isotherms due to seiching on this day. The RMSE is0.93◦C and the average lateral 0.5m-layer temperature range at any depth in thelake was 1.14◦C.Comparing all 29 days throughout the simulation period, on average, the RMSEbetween field data and DYRESM is only 59.8% of the average temperature rangein the lake, lending significant confidence to the modelled results (Figure 4.45).On only two occasions (21 October 2015 and 16 December 2015) is the RMSElarger than the actual average temperature range in the lake at any given depth. On21 October 2015, the RMSE is 0.45◦C whereas the average temperature range forany given depth is 0.42◦C (0.03◦C difference); on 16 December 2015 the RMSEis 0.20◦C whereas the average temperature range at any given depth is 0.18◦C(0.02◦C difference). Therefore, despite the RMSE being slightly larger than theaverage temperature range at any given depth on these field days, the difference181is very small and the RMSE itself on these occasions is also relatively small (i.e.within the annual average RMSE). Note the day with the largest lateral tempera-ture range (2.26◦C) in the lake at all depths is 9 September 2015, which was theday that the largest seiching movement in isotherms was observed in field profiles(Figure 4.43 in Section 4.9). The modelled RMSE on this day was only 0.67◦C.29-05 01-07 01-08 01-09 01-10 01-11 01-12 23-1200.511.522.5Temperature [°C]RMSE [°C]Depth Averaged 0.5m Layer Temp. Range [°C]Overall Average 0.5m Layer Temp. Range [°C]Figure 4.45: RMSE from comparing average temperature in 0.5m depth in-tervals from DYRESM with field data in 2015 compared to the depth-averaged (0.5m depth intervals) temperature range in Wood Lake on allfield days. The average RMSE was 0.51◦C and average MAE: 3.31%,whereas the depth-averaged layer temperature range was 0.88◦C.1826 8 10 12 14 16 18Temperature [°C]051015202530Depth [m]Figure 4.46: Temperature profiles on 23 September 2015. Also refer to Fig-ure 4.36 (Section 4.7) for thermistor chain data during this period forobservation of seiching in Wood Lake during September 2015.4.11.3 2015 Model Error: Heat ContentThe second method used to evaluate the model was to compare estimates of the heatcontent (Joules/unit surface area) in Wood Lake based on DYRESM compared tothat based on the representative average temperature profiles. Similarly to the fieldprofiles, the heat content stored in each layer of the modelled profiles is weightedby a factor proportional to the ratio of the average surface area of that layer di-vided by the lake’s surface area, thereby accounting for the lake’s morphometry.183The resulting computed heat content is then broken into two components: the heatcontent in the average photic zone for the simulation period, and the heat contentin the average aphotic zone for the simulation period. The average photic depthbetween 29 May 2015 and 23 December 2015 was 11.6m and the average aphoticzone is the water below this depth.The field and simulated estimates for heat content (total, photic zone, aphoticzone) are overlaid (Figure 4.47) and RMSE computed in order to help evaluatewhere the errors were occurring in the model. Depending on where the temperaturedifferences between the model and field profiles occurred throughout the water col-umn, even a simulated day with a small average temperature difference may havea large deviation in heat content (i.e. if the simulated temperature profile is slightlywarmer/colder consistently throughout the water column). Likewise, if there arelarge deviations in temperature between the simulated and real profiles that occurbut there are nearly equal volumes of water with positive/negative discrepanciesthroughout the water column, then the calculated heat content error could be verysmall, despite large errors in temperature profiles.The total RMSE for the modelled simulation period in terms of Wood Lake’sheat content for 2015 was 1.42e4J/m2 (1.47%) and the MAE was 1.09% (Fig-ure 4.48). The maximum apparent error is 4.84e4J/m2 (4.44%); however notethat this occurs on 9 September 2016 (previously referred to as anomaly [b] (Fig-ure 4.40 in Section 4.9))) when the largest average temperature range (Figure 4.45)at any given depth also occurs (2.26◦C) due to seiching. This reaffirms that dayswith such degrees of movement should not be used to verify or validate a 1-Dmodel due to the difficulty in establishing an accurate estimate of the representa-tive average temperature profile and total heat content from fieldwork during in-tense seiching and how this leads to a false impression of the model’s “apparenterror”.18401-06 01-07 01-08 01-09 01-10 01-11 01-12Day [dd-mm-yy]3456789101112Total Heat Content [Joules/m2 ]×105Field: Total Heat Content [J/m 2]Field: Avg Aphotic Zone Heat Content [J/m 2]Field: Avg Photic Zone Heat Content [J/m 2]DYRESM: Total Heat Content [J/m2]DYRESM: Avg Aphotic Zone Heat Content [J/m 2]DYRESM: Avg Photic Zone Heat Content [J/m2]Figure 4.47: Comparison of heat content [J/m2 normalized by surface area]based on DYRESM output to that based on the observed spatial aver-age temperature profiles (field work) in 2015.18501-06-15 01-07-15 01-08-15 01-09-15 01-10-15 01-11-15 01-12-15Day [dd-mm-yy]-101234Error (J/m2 )×104Heat Content RMSE [J/m2]Average RMSE [J/m2]Figure 4.48: Error [J/m2] in Heat Content between simulated results com-pared to field work in 2015. The average RMSE was 14195.m J/m2(1.47%) and the MAE was 1.09%.4.11.4 Model Validation (2013 and 2014)Once the calibrated model’s error was within acceptable limits, it was used to re-generate conditions from 2013 and 2014 based on initial profiles in May (15 May2013 and 21 May 2014) until October 15 for each year (last available temperatureprofile from these years). Field data from these years was collected by BC MOEat station 0500848 at 2m intervals from the surface to the bottom. It is importantto note this station (Figure 4.49) is not in the center of the lake, nor at the deepestlocation in the lake at any time. This location is south of the five profile stationsused in the current study to develop the spatial average temperature profiles.1860 500 mS9 S10S11S12S13CTD Sample Stations Used forDeveloping Representative AverageTemperature ProfilesMOE Sample Station 0500848NFigure 4.49: Five CTD locations used to generate representative average tem-perature profiles for field days in 2015 study compared to the geo-graphic location of MOE Station 0500848 (2.53km from south shoreand 3.95km from north shore).CTD and thermistor chain data from the current study show that uninodalV1H1 oscillations of the isotherms dominate (Figure 4.35 to Figure 4.38 in Sec-tion 4.7 and Figure 4.43 in Section 4.9); therefore, as the distance of a station fromthe geographic center of the lake increases, so to does the deviation away from atrue representative average temperature profile of the depth-temperature data. Us-ing a single profile at this location as being representative of the average tempera-ture profile in the lake can cause significant error in the field estimate of an averagetemperature profile and heat content. Recall that fortunately because DO and tem-perature contours move in concert with each other during these seiche motions187(stratification dominated by the thermal structure (Figure 4.10 in Section 4.3)),providing that DO and temperature profiles are taken at the same location simul-taneously, this will not lead to an error in estimating the available fish habitat interms of the 1-D thickness of the habitable layer in the lake. Nevertheless, in termsof evaluating the model, it is important to note this potential source for “apparenterror” in recognizing that the field data was not acquired at the geographic center,is based on a single profile on each field day, and is measured only every 2m depth(as opposed to CTD data with measurements every ∼0.12m). Despite these anno-tations, the model still performed extremely well to re-create the field profiles fromMay to October in 2013 and 2014.Similarly to 2015, the RMSE error between DYRESM’s temperature profileoutputs on selected days corresponding to MOE field data was assessed as a depthaverage difference throughout the water column on each day. The low resolutionof collected field data (i.e. 2m intervals) introduces unquantifiable uncertainty withregards to the actual shape of the profiles between these data points on each fieldday which. For determining the error (RMSE) in predicting temperatures at everydepth throughout the water column, the DYRESM layers were averaged into 2mdepth bins that correspond to the 2m intervals of the field work measurements andthen the average temperature from every 2m layer in DYRESM was compared withfield work measurements at each corresponding depth interval (for each field day).The RMSE determined based on comparing the temperatures at every 2m depthinterval throughout the water column for all field days in 2013 and 2014 was1.02◦C and 0.750◦C respectively. In 2013, the minimum RMSE was on 18 June2013 (0.50◦C) and the maximum RMSE (1.43◦C) was on 11 September 2013.In 2014, the minimum RMSE was on 16 July 2014 (0.55◦C) and the maximumRMSE (1.02◦C) was on 14 August 2014.The MAE for 2013 and 2014 was 6.43%and 4.76% respectively.The model performs very well in both 2013 and 2014, particularly during peakstratification in both years in terms of profile shape and epilimnion/hypolimniontemperatures (Figure 4.50). The largest discrepancies in terms of surface layertemperatures appear to occur in June 2014 (1.6◦C) and October 2013 (1.6◦C), onceagain, during times of significant warming and cooling respectively. However, thesurface layer depths and thermocline location are generally well predicted. The188largest discrepancies in terms of surface layer depth occur from August to October2013 and 2014 and are on the order of 1-2 meters, which is less than the precision ofthe MOE field data. The predicted thermocline shape differs from the field profilesmost notably in July 2013, but some slight variability also occurring in September- October 2013 and 2014.0 10 20 30[2013]05101520253035Height Above LakeBottom (Datum) [m]May140 10 20 30[2014]05101520253035Height Above LakeBottom (Datum) [m]May210 10 20 3005101520253035Jun180 10 20 3005101520253035Jun180 10 20 3005101520253035Jul160 10 20 30Temperature [°C]05101520253035Jul160 10 20 3005101520253035Aug140 10 20 3005101520253035Aug140 10 20 3005101520253035Sep110 10 20 3005101520253035Sep150 10 20 3005101520253035Oct110 10 20 3005101520253035Oct15Figure 4.50: Comparison of DYRESM modelled results with temperatureprofiles from field work in May, June, July, August, September, andOctober 2013 (top row) and 2014 (bottom row).189Similarly to investigations for 2015, the heat content estimates based on DYRESM’soutputs for 2013 and 2014 were compared to estimates of heat content from fielddata (Figure 4.51). In the field work estimate of heat content in every layer (madeup of water between two data points) is taken as having average properties of thebordering data points above and below. In 2013, the RMSE is 5.0e4 J/m2 (4.54%)and the MAE is 4.25%; in 2014 the RMSE is 1.81e4 J/m2 (1.69%) and the MAE is1.35%. Considering the sources of error and uncertainty in how well the field pro-files represent the thermal structure of the lake on each field day due to the effectsof seiching and low resolution of data in the water column, these comparisons arevery strong.19001-06-13 01-07-13 01-08-13 01-09-13 01-10-130.80.911.11.21.3Total Heat Content [Joules/m2 ] ×10601-06-14 01-07-14 01-08-14 01-09-14 01-10-14Day [dd-mm-yy]0.911.11.2Total Heat Content [Joules/m2 ] ×106DYRESM: Total Heat Content [J/m2]Field Data: Total Heat Content [J/m 2]Figure 4.51: Comparison of Heat Content [J/m2] in 2013 (top) and 2014(bottom) normalized by surface area based on DYRESM output to thatbased on the field temperature profiles from BCMOE data.Aside from the surface layer temperature in June 2014, the model provided acloser comparison to measured field temperature profiles (depth-average compar-ison) and heat content in 2014 than in 2013. In order to maintain confidence inthe model regarding results in 2013, we investigated two days that the model ap-peared to have a larger error in this year: 16 July 2013 (RMSE: 0.91◦C and heatcontent error of -6.68e4 J/m2) and 11 September 2013 (RMSE: 1.43◦C and heatcontent error of 4.88e4 J/m2). Considering previous discussions regarding effectsof north/south winds on the oscillation of isotherms, it is prudent to investigate191wind data from Kelowna Airport (GCEC, 2016) prior to each of these field days.Strong diurnal win