Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Climate change impacts on a eutrophying lake : Cultus Lake, British Columbia, Canada Sumka, Mark Gregory 2017

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

Item Metadata

Download

Media
24-ubc_2017_september_sumka_mark.pdf [ 3.21MB ]
Metadata
JSON: 24-1.0354396.json
JSON-LD: 24-1.0354396-ld.json
RDF/XML (Pretty): 24-1.0354396-rdf.xml
RDF/JSON: 24-1.0354396-rdf.json
Turtle: 24-1.0354396-turtle.txt
N-Triples: 24-1.0354396-rdf-ntriples.txt
Original Record: 24-1.0354396-source.json
Full Text
24-1.0354396-fulltext.txt
Citation
24-1.0354396.ris

Full Text

Climate change impacts on aeutrophying lake: Cultus Lake, BritishColumbia, CanadabyMark Gregory SumkaB.Sc., University of Alberta, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF APPLIED SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Civil Engineering)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2017© Mark Gregory Sumka 2017AbstractThis study characterizes the thermal dynamics of Cultus Lake, BritishColumbia, Canada, and assesses the impacts of climate change on criti-cal habitat for species-at-risk Sockeye Salmon (Onchorynchus nerka) andCultus Lake Pygmy Sculpin (Cottus aleuticus). Historical field data span-ning 1920s–1930s, 2001–2003, and 2009–2016 were analyzed and a one-dimensional hydrodynamic model (General Lake Model) was calibrated us-ing field data collected in 2016, and validated using field data from 2001–2003and 2009–2016. The thermal structure of the lake was simulated to 2100using outputs from the downscaled Canadian Regional Climate Model (Can-RCM4) for two climate change scenarios (RCP4.5, moderate emissions sce-nario; and RCP8.5, extreme emissions scenario). Historically (1923–2016),the total lake heat content increased at a rate of 0.80 MJ m-2 a-1 and isprojected to warm by 2.0 MJ m-2 a-1 (4.2 MJ m-2 a-1) for RCP4.5 (RCP8.5).Historically, the Schmidt stability increased by 2.2 J m-2 a-1 and is projectedto increase by 2.6 J m-2 a-1 (6.5 J m-2 a-1) for RCP 4.5 (RCP8.5). The dura-tion of stratification has historically been increasing at a rate of 0.18 d a-1and is projected to increase by 0.18 d a-1 (0.50 d a-1) for RCP4.5 (RCP8.5).The onset of stratification is now two weeks earlier than the 1920s–1930s andcurrently occurs around 23 March while the breakup date has not changedand occurs around 15 December. However, it is predicted that there willbe no change in the date of onset of stratification while breakup will bedelayed to 12 January (25 January) for RCP4.5 (RCP8.5). Lake surfacetemperature and outflow temperature is most important for salmon survivalin August through November corresponding with the salmon run. Histori-cal change in mean monthly temperature ranged from 0 °C a-1 in Novemberto a maximum of 0.016 °C a-1 in August. This is predicted to increase to0.016 °C a-1 (0.046 °C a-1) in November and 0.031 °C a-1 (0.069 °C a-1) inAugust for RCP4.5 (RCP8.5). Projections indicate fundamental changes tothe thermal characteristics of Cultus Lake, which may further degrade wa-ter quality, particularly in conjunction with ongoing eutrophication, elicitingfundamental changes in the structural and functional attributes of criticalhabitat for species-at-risk.iiLay SummaryCultus Lake is a popular recreational lake located in the eastern Fraser Val-ley, British Columbia, Canada. Previous studies indicate increased nutrientloading to the lake, resulting in ongoing cultural eutrophication. Changesin the ecosystem are changing habitat for two species at risk of extinction:Sockeye Salmon and Cultus Lake Pygmy Sculpin. The objectives of thisstudy are to 1) characterize the thermal characteristics of this warm mo-nomictic lake; 2) track changes since the early 20th century; and 3) forecasthow Cultus Lake may respond to future climate change. Field data collectedin 2016 were used to calibrate a numerical model, which was validated onhistorical measurements and used to forecast the thermal characteristics ofthe lake under different warming scenarios. Increased warming of the lakeis expected to result in habitat modification and loss for two Canadian fishspecies at risk of extinction.iiiPrefaceThis dissertation is an original, unpublished work by the author, MarkSumka. The field campaign was designed and organised by myself. I alsoconducted all field data analysis, and data analysis including the model cal-ibration, validation, evaluation, and forecasting scenarios. Assistance forthe field campaign was provided principally by members of the UBC Envi-ronmental Fluid Mechanics Group: Samuel Brenner, Kelly Graves, DavidHurley, and Christopher Young, as well as Dr. Daniel Selbie of Fisheriesand Oceans Canada. Dr. Selbie, Kelly Malange, Garrett Lidin, and KristyGrant from Fisheries and Oceans Canada provided historical lake and riverdata, a vessel, and equipment for this research. Bathymetric data were pro-vided by the Canadian Hydrographic Service. Historical lake water levelswere provided by the Cultus Lake Park Board. Alain Royer, Norm O’Neill,Ihab Abboud, and Vitali Fioletov establish and maintain the air qualitymonitoring station at Saturna Island.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . ix1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Seasonal Thermal Stratification of Lakes . . . . . . . . . . . 21.2 Climate Change and Lakes . . . . . . . . . . . . . . . . . . . 31.3 Cultus Lake . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.4 Thesis Objectives . . . . . . . . . . . . . . . . . . . . . . . . 62 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.1 Site Description . . . . . . . . . . . . . . . . . . . . . . . . . 82.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . 112.2.1 Historical Data . . . . . . . . . . . . . . . . . . . . . 112.2.2 Meteorological Data . . . . . . . . . . . . . . . . . . . 112.2.3 Limnological Data . . . . . . . . . . . . . . . . . . . . 142.2.4 Water Clarity and Extinction Coefficient . . . . . . . 182.2.5 Inflows and Outflow . . . . . . . . . . . . . . . . . . . 182.3 General Lake Model (GLM) . . . . . . . . . . . . . . . . . . 202.3.1 Model Description . . . . . . . . . . . . . . . . . . . . 202.3.2 Model Setup . . . . . . . . . . . . . . . . . . . . . . . 222.4 Regional Climate Model . . . . . . . . . . . . . . . . . . . . . 242.4.1 Forecast Data . . . . . . . . . . . . . . . . . . . . . . 24v3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.1 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.2 Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 323.3 Water Clarity and Chlorophyll-a . . . . . . . . . . . . . . . . 333.4 Inflows and Outflow . . . . . . . . . . . . . . . . . . . . . . . 343.5 Modelling Cultus Lake . . . . . . . . . . . . . . . . . . . . . 383.5.1 Calibration Period . . . . . . . . . . . . . . . . . . . . 383.5.2 Model Evaluation . . . . . . . . . . . . . . . . . . . . 403.5.3 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . 453.5.4 Hindcast (2009–2016) . . . . . . . . . . . . . . . . . . 463.5.5 Forecast Data Evaluation (2001–2003) . . . . . . . . . 493.6 Historical and Forecast Results (1923–2100) . . . . . . . . . 533.6.1 Heat Content . . . . . . . . . . . . . . . . . . . . . . 533.6.2 Stability and Stratification . . . . . . . . . . . . . . . 553.6.3 Surface and Outflow Temperatures . . . . . . . . . . 614 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634.1 Potential Changes to the Lake Mixing Regime . . . . . . . . 634.2 Implications of Increased Climate Warming on EcosystemStructure and Functioning . . . . . . . . . . . . . . . . . . . 654.3 Trophic Status of Cultus Lake . . . . . . . . . . . . . . . . . 674.4 Managing the Impacts of Climate Change . . . . . . . . . . . 705 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Appendix A Inflow and Outflow Descriptions . . . . . . . . . 84viList of TablesTable 2.1 Accuracy, resolution, and drift of HOBO weather sta-tion instruments . . . . . . . . . . . . . . . . . . . . . . . . . 13Table 2.2 Bird clear-sky model atmospheric parameters . . . . . . 14Table 2.3 Cultus Lake station information . . . . . . . . . . . . . 15Table 2.4 Accuracy, resolution, and drift for SBE 19plus V2 CTD 16Table 2.5 Stream gauging station information . . . . . . . . . . . 19Table 2.6 Stream temperature coefficients . . . . . . . . . . . . . 20Table 3.1 Calibration parameters for GLM setup . . . . . . . . . 39Table 3.2 Inflow parameters for GLM . . . . . . . . . . . . . . . . 40Table 3.3 Sensitivity analysis results and error between observedand synthetic meteorological data . . . . . . . . . . . . . . . . 46Table 3.4 Comparison of three methods used to calculate the du-ration of stratification . . . . . . . . . . . . . . . . . . . . . . 58viiList of FiguresFigure 2.1 Map of Cultus Lake . . . . . . . . . . . . . . . . . . . 10Figure 2.2 Hypsographic curve . . . . . . . . . . . . . . . . . . . 17Figure 2.3 Lake water levels . . . . . . . . . . . . . . . . . . . . . 17Figure 3.1 Limnological field data results . . . . . . . . . . . . . . 30Figure 3.2 Thermistor moorings . . . . . . . . . . . . . . . . . . . 31Figure 3.3 Wind speed, gust, and direction . . . . . . . . . . . . . 32Figure 3.4 Discharge and temperature of Frosst Creek and SweltzerRiver, and total daily precipitation at Abbotsford Airport . . 35Figure 3.5 Discharge for Watt Creek and Smith Falls Creek, andspecific conductivity and turbidity for Frosst Creek, WattCreek, and Smith Falls Creek . . . . . . . . . . . . . . . . . . 37Figure 3.6 Overlaid observed and simulated temperature profiles 41Figure 3.7 RMSE for 2016 GLM calibration . . . . . . . . . . . . 43Figure 3.8 Observed and simulated heat content for 2016 calibration 44Figure 3.9 RMSE for 2009–2016 hindcast . . . . . . . . . . . . . . 48Figure 3.10 Observed and simulated heat content for 2009–2016hindcast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Figure 3.11 RMSE for 2001–2003 hindcast . . . . . . . . . . . . . . 52Figure 3.12 Observed and simulated heat content for 2001–2003hindcast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Figure 3.13 Historical and projected heat content . . . . . . . . . . 54Figure 3.14 Historical and projected Schmidt Stability Index . . . 56Figure 3.15 Thermal stratification calculation method comparison 57Figure 3.16 Historical and predicted duration of thermal stratifi-cation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Figure 3.17 Historical and predicted onset and breakup of thermalstratification . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Figure 3.18 Historical and predicted outflow temperatures . . . . . 62viiiAcknowledgementsI would like to thank my supervisor, Dr. Bernard Laval, and Dr. DanielSelbie for their support and guidance throughout my research. Thank youto the UBC Environment Fluid Mechanics group (Samuel Brenner, KellyGraves, David Hurley, and Christopher Young) for your assistance with fieldwork. Also, thank you to the Cultus Lake Salmon Research Laboratory staff,in particular Dr. Selbie, Kelly Malange, Garrett Lidin, and Kristy Grant,for providing limnological data, and research equipment, as well as TroyDavis from the Cultus Lake Park Board for providing me with lake leveldata. I am also grateful to Steve Marks from the Cultus Lake Marina forgranting me permission to install a weather station on his roof.ixChapter 1IntroductionCultural eutrophication, the artificial fertilization of water bodies with ex-cess nutrients, is the most pervasive problem affecting the water quality oflakes and rivers worldwide (Chambers et al. 2001; D. W. Schindler 2012).Classic symptoms of lake eutrophication include increased aquatic plantand algal growth, and deterioration of fisheries and recreational suitability(Bartsch 1970; Anderson et al. 2002). While eutrophication occurs naturallyover very long time scales, many lakes are experiencing rapidly increasedeutrophication rates due to nutrient loading from anthropogenic sourcessuch as agriculture, urban runoff, and industrial sources (Bartsch 1970; An-derson et al. 2002; V. H. Smith and D. W. Schindler 2009). Increases inalgal-limiting nutrient loading, especially phosphorus and nitrogen, com-bined with sufficient sunlight penetration in lakes, results in excessive plantand algae growth (cultural eutrophication; Vallentyne (1974) and Ander-son et al. (2002)). When plants and algae in lakes senesce, they settle tothe bottom where bacterial aerobic decomposition of organic matter occurs(Kankaala et al. 2002), consuming oxygen which can have impacts on deep-water resident fish and benthos (Ficke et al. 2007).Climate change is amplifying the symptoms of lake eutrophication (Jeppe-sen et al. 2010; B. Moss et al. 2011). For instance, warmer water temper-atures lead to reduced water column mixing, which isolates lake epilimniafrom hypolimna over a more protracted period, enhancing hypolimnetic oxy-gen depletion (Matzinger et al. 2007). Hypoxic deepwater conditions can beaccentuated by decomposition of greater amounts of autochthonous organicmatter arising from nutrient-stimulated growth of macrophytes (Kankaalaet al. 2002) and algae, such as cyanobacteria (O’Neil et al. 2012). Suchfundamental abiotic and biotic changes in lakes can elicit upward cascadingeffects on the lake ecosystem to valued ecosystem components such as fish(Vincent 2009).This thesis begins with a review of the mechanisms of seasonal thermalstratification of lakes, followed by a review of the effects of climate changeon stratification and the potential effects of climate change on cultural eu-trophication, narrowing in on the trophic status of Cultus Lake. Chapter 21provides a description of the site and the fieldwork involved in the devel-opment of a one-dimensional Lagrangian hydrodynamic model, the GeneralLake Model (GLM; Hipsey, Bruce, et al. 2014), along with a descriptionof the forecast model used, Canadian Regional Climate Model (CanRCM4,Environment and Climate Change Canada 2017). Results of the 2016 fieldcampaign are provided in Chapter 3, in addition to modelled results forthe 2016 season and model validation using historical data (2001–2003 and2009–2016). Chapter 3 concludes with a comparison of historical (1923–2016) thermal observations to model hindcasts, and future projections un-der two climate change scenarios. Chapter 4 describes the possible futuremixing conditions in the Cultus Lake under climate change, and associatedpotential changes to lake ecosystem structure and functioning, with effectson resident and anadromous species, as well as the potential effects of cli-mate change on the trophic status of Cultus Lake. Options to mitigatethe ecosystem impacts of climate change through nutrient abatement areexplored.1.1 Seasonal Thermal Stratification of LakesThermal stratification is a physically forced phenomenon that develops sea-sonally in most deep temperate lakes (Wetzel 2001). Lakes are classifiedbased on their seasonal stratification and mixing regimes where dimicticlakes mix twice annually (in the spring and fall) and monomictic lakes un-dergo a single annual mixing event (Wetzel 2001). Monomictic lakes can befurther classified as warm or cold, with lake temperatures at mixis exclu-sively above or below 4 °C, respectively (Wetzel 2001). This study focuseson a warm monomictic system, Cultus Lake.Warm monomictic lakes are common in tropical and moderate temperateclimates, such as coastal regions that are influenced by oceanic climates(Wetzel 2001; Kalff 2002). Warm monomictic lakes stratify vertically intolayers following insolative heating in the spring, a pattern that becomes moreintense through the summer, characterized by a warmer, less dense upperlayer (epilimnion) overlying a cooler, more dense bottom layer (hypolimnion)(Wetzel 2001). These layers are separated by a region of rapidly changingwater temperature with depth, known as a metalimnion, which containsthe thermocline, defined as the plane of maximum rate of decrease of watertemperature with respect to depth (Wetzel 2001).Once seasonal stratification is established, the epilimnion is exposedto the atmosphere where oxygen can be dissolved into the water surface2by diffusion. The epilimnion contains the upper euphotic zone (regionof photosynthetically-active light penetration) and is associated with au-totrophic production by phytoplankton, which adds oxygen and seasonallydepletes nutrients from this lake region (Wetzel 2001). The density gra-dient at the thermocline generally inhibits mixing between the epilimnionand hypolimnion, but not the settling of senescent organic matter. Aerobicdecomposition of settling organic matter and internal loading from lake sed-iments can result in somewhat lower hypolimnetic dissolved oxygen concen-trations towards the end of the stratified period, and comparatively higherhypolimnetic nutrient concentrations, the magnitudes of which scale withlake trophic status (Wetzel 2001; Søndergaard et al. 2003).Thermal stratification typically breaks down in the fall and winter assurface water temperatures cool and wind stresses gradually mix the twolayers, ultimately resulting in isothermal conditions and complete verticalmixing of the water column, known as lake turnover (Wetzel 2001). In mo-nomictic systems, this complete mixing is maintained until the spring whensolar heating is sufficient to thermally re-stratify the lake. This completemixing of a lake is necessary to re-oxygenate the hypolimnion while at thesame time redistributing hypolimnetic nutrients to the epilimnion for use byprimary producers during the growing season.1.2 Climate Change and LakesThe direct primary influences of climate change on lake thermal structureare mediated by heat exchanges with the atmosphere via solar radiation,longwave radiation, latent (evaporative) heat, and sensible (convective) heat(Imboden and Wu¨est 1995). Abiotic and biotic dynamics in lake ecosystemsare often tightly coupled with atmospheric processes and relatively smallchanges in resultant lake temperatures could potentially elicit dispropor-tionately large effects on the physical and ecological characteristics of lakes(Adrian et al. 2009). Thus, any significant changes in climatic forcings oflakes are expected to influence lake dynamics, and potentially be interactivewith other lake stressors (Christensen et al. 2006).Past site-specific studies on lakes (e.g. Sahoo et al. (2015)) and lakes inspecific climatic regions (e.g. George et al. (2007)) have demonstrated thatclimate change is altering the physical characteristics of lakes which has thecapacity to alter water quality and affect whole lake ecosystems, includ-ing reducing habitat for cold-water fish. Additionally, comprehensive globalstudies show an increasing trend in lake surface temperatures over recent3decades (O’Reilly et al. 2015) along with air and ocean surface tempera-tures (IPCC 2013). However, lake surface temperature and air temperaturecorrelations can vary regionally, and warming (or cooling) rates are depen-dent on local climate and geomorphology (O’Reilly et al. 2015). Thus, thereis a need to assess the effects of climate change on lakes on a case-by-casebasis.The thermal structure of lakes is a sensitive indicator of climate changebecause of their responsiveness to climate forcings (Adrian et al. 2009). Epil-imnetic temperatures are an indicator of local climate trends: warming inthe water surface increases as air temperature increases (Adrian et al. 2009).Conversely, hypolimnetic temperatures do not necessarily increase with airtemperature and may in fact exhibit cooling trends, depending on lake mor-phometry (Gerten and Adrian 2001; Adrian et al. 2009). Increased warmingis predicted to force stronger temperature gradients in lakes, intensifying andprolonging stratification (Ficke et al. 2007; Adrian et al. 2009). Globally,the strength and duration of lake stratification has been increasing in recentdecades with the largest increases found in warm, deep lakes (Kraemer et al.2015). Stronger and prolonged stratification is particularly problematic forwarm monomictic lakes, where there is only one mixing season per year, asreductions in thermal exchange, nutrient redistribution, and dissolved oxy-gen regeneration occur. With sufficient warming, there is a possibility forwarm monomictic lakes to shift mixing regimes to oligomixis (i.e. incon-sistent annual overturn) or even towards a non-mixing state (meromixis)(Livingstone 2008), thereby limiting exchanges between the epilimnion andhypolimnion. This non-mixing state reduces hypolimnetic dissolved oxygen,and the decomposition of sedimented organic matter can increase hypolim-netic salinity, further inhibiting mixing due to chemically-mediated increasesin density and stability, leading to meromixis (Kalff 2002).Climate change can also elicit complex and often unpredicted lake eco-logical responses through interactions with a variety of lake and watersheddrivers and stressors (Christensen et al. 2006). For instance, climate changeis accelerating the rate of natural and cultural eutrophication in lakes (B.Moss et al. 2011). The increased frequency and strength of storms, predictedwith climate change, is expected to increase runoff and nutrient loading intolakes (Adrian et al. 2009; Robertson and Rose 2011). Key limiting nutrients,like phosphorus and nitrogen, can come from regions with high industrial,agricultural, residential, or geological sources, while dissolved organic car-bon fluxes can increase from runoff in heavily forested regions (Anderson etal. 2002; Schelker et al. 2014; McGinley et al. 2016). The resultant increasesin nutrient inputs and changing stoichiometry (i.e. nitrogen to phosphorus4ratio) can lead to algal blooming, and structural and functional changes inalgal and reliant zooplankton communities, including increases in problem-atic algal taxa, such as cyanobacteria (O’Neil et al. 2012) and increasedgrowth rates of invasive plant species (Ficke et al. 2007). Moreover, warmerwater temperatures from increased heating may reduce oxygen concentra-tion in lakes (Ficke et al. 2007). Prolonged seasonal lake stratification canenhance organic matter delivery to lake hypolimnia, resulting in acceler-ated depletion of hypolimnetic dissolved oxygen concentrations (Ficke et al.2007). Warming hypolimnetic waters reduce the solubility of oxygen, fur-ther diminishing deep water dissolved oxygen concentrations (Kalff 2002).Together, increased nutrient loading and warmer water temperatures willsignificantly change food webs and habitat for fish, likely enhancing fishmortality for many cold-water species, and generally changing fish commu-nity composition (Ficke et al. 2007). It is therefore important to study theeffects of climate change and eutrophication on lake ecosystems in order toimprove watershed planning, and develop mitigation strategies that mini-mize the impacts on ecosystems and ecosystem services.1.3 Cultus LakeCultus Lake, located in British Columbia’s Fraser Valley region, is expe-riencing increased nutrient loading that is leading to lake eutrophication(Shortreed 2007; Putt 2014; Chiang et al. 2015). Nitrogen and phosphorusare being loaded into the lake from several key sources: wet and dry at-mospheric deposition from major urban, industrial, and agricultural centresupwind, upstream agricultural runoff from Columbia Valley and its uncon-strained aquifer, septic leaching, and migratory gull guano (Putt 2014). Thelake has historically been classified as oligotrophic (Shortreed 2007), but ob-servational data and state-based modelling indicate that increased nutrientloading is transitioning the lake to mesotrophy (Putt 2014). Such funda-mental changes in the ecosystem may be increasing threats to two residentspecies-at-risk for which Cultus Lake is critical habitat: the Cultus LakePygmy Sculpin (Cottus aleuticus; SARA 2002; COSEWIC 2010) and Sock-eye Salmon (Oncorhynchus nerka; COSEWIC 2003). As the population ofBritish Columbia’s Lower Mainland grows, use of Cultus Lake by humans isexpected to increase along with associated anthropogenic nutrient loading.The climate in the Fraser Valley has been warming since at least 1900, atrend that is projected to continue changing in the future (Pacific ClimateImpacts Consortium 2012; British Columbia Ministry of Environment 2016).5The mean air temperature in Southern Coastal regions of British Columbiahas increased by + 0.8 °C per century (British Columbia Ministry of En-vironment 2016). A downscaled ensemble of climate change models undervarious greenhouse gas emissions scenarios predict an expected mean tem-perature increase of + 1.0 °C, + 1.8 °C, and + 2.8 °C for the 2020s, 2050s, and2080s, respectively above a thirty year historical average in 1961–1990 (Pa-cific Climate Impacts Consortium 2012). Annual precipitation has histori-cally increased by + 14 % per century since at least 1900 (British ColumbiaMinistry of Environment 2016) and is expected to continue increasing by+ 4 %, + 7 %, and + 9 % for the 2020s, 2050s, and 2080s, respectivelyabove the same thirty year baseline with wetter winters and drier summers(Pacific Climate Impacts Consortium 2012). Climate change is thereforeexpected to significantly affect the thermal characteristics of Cultus Lake asair temperatures increase and the watershed hydrological regime changes.1.4 Thesis ObjectivesCultus Lake has been studied extensively since the 1920s because of itsSockeye Salmon population (Foerster 1925; Ricker 1937). While more re-cent studies have focused on nutrient loading (Shortreed 2007; Putt 2014),climatically-forced physical limnological processes are likely to play a rolein enhancing the eutrophication of Cultus Lake. This study investigates thefollowing questions in more detail:1. How are changing physical processes in the lake affecting eutrophica-tion and critical habitat for species-at-risk?2. How has climate change historically affected physical processes in Cul-tus Lake?3. How will Cultus Lake’s seasonal patterns of thermal stratificationevolve as climate change continues?These questions are addressed through development and parameterization ofa one-dimensional Lagrangian hydrodynamic model (General Lake Model,GLM; Hipsey, Bruce, et al. 2014) calibrated using field data from CultusLake. The numerical lake model is coupled with the fourth iteration of theCanadian Regional Climate Model (CanRCM4; Environment and ClimateChange Canada 2017) to predict how physical processes in the lake maychange in the future through 2100 AD. Additionally, the model is hindcast6to assess how the lake has changed since the 1920s to validate more simplisticstate-based models (e.g. BATHTUB; Putt 2014).7Chapter 2Methods2.1 Site DescriptionCultus Lake is situated in the eastern Fraser Valley, British Columbia,Canada, approximately 95 km east of Vancouver and due south of Chill-iwack. The lake is a part of the ancestral territory of the Soowahlie Bandof the Sto´:lo¯ peoples. The lake is traditionally known by the Sto´:lo¯ peoplesas Sw´ı:lhcha and was the location of the community of The’wa´:l´ı (Trimble2016). The lake is characterized as a single basin with a length of approxi-mately 5.0 km and a width of 1.5 km. A recent bathymetric survey by theCanadian Hydrographic Service (Section 2.2.3, Figure 2.1) in 2016 resultedin a surface area of 6.3 km2, a mean depth of 32 m, and a maximum depth of43 m. Cultus Lake is currently classified as a warm monomictic lake with along summer stratification period (Shortreed 2007), and has previously beenobserved to inversely stratify but never for a prolonged period with rare icecoverage (Ricker 1937). The lake is situated in the hanging Columbia Val-ley above the Chilliwack River at an elevation of 45 m a.s.l., bounded byVedder Mountain to the west and International Ridge to the east. The totalcatchment area is 75 km2 (Shortreed 2007), of which approximately 16 km2is located in the United States state of Washington. A municipal park andcommunity is situated on the north shore of the lake and is overseen by theCultus Lake Park Board. The southern shore of the lake has been developedas the community of Lindell Beach. Both the east and west shores of thelake are within the boundaries of Cultus Lake Provincial Park which extendsup the slope of Vedder Mountain and International Ridge toward the UnitedStates border. Cultus Lake is a popular recreation site due to its proximityto major urban centres in the Lower Mainland. Between 2 and 3 millionpeople visit Cultus Lake every year with this number projected to grow aspopulation in the Lower Mainland increases (Delcan 2012 as cited by Putt2014).The Cultus Lake watershed has 11 major tributaries (Putt 2014), ofwhich the three largest were included in this study to parametrize GLM:Frosst Creek, Watt Creek, and Smith Falls Creek. Inflows were deemed8important for this study because of the short residence time of the lake(1.8 years, Shortreed 2007) and the long time-scale of this study. The outlet,Sweltzer River (sometimes referred to as Sweltzer Creek), drains CultusLake northward to the Chilliwack River, a tributary of the Fraser Riverwhich drains into the Strait of Georgia at Vancouver, British Columbia. Adetailed description of the inflows and outflow is provided in Appendix A.9000000FROSSTCREEK WATTCREEKSMITHFALLSCREEKANEMOMETERCL1CL2CL3CL4CL5SALMONRESEARCHLABORATORYSWELTZERRIVER10005000METRESCOLUMBIAVALLEYHIGHWAYFR122.033° W 122.017° W 122.000° W 121.983° W 121.967° W49.033° N49.050° N49.067° NCULTUS LAKEPROVINCIAL PARKCULTUS LAKEPROVINCIAL PARK122.033° W 122.017° W 122.000° W 121.983° W 121.967° W49.033° N49.050° N49.067° NFigure 2.1: Map of Cultus Lake. Bathymetric survey was performed byCanadian Hydrographic Service in UTM Zone 10N with NAD83 Datum.Depths are at 5 m contour intervals. The waterline is estimated at 44.5 me-tres above sea level. National Hydro Network and National Road Networkwere obtained from Natural Resources Canada’s GeoGratis web-service.102.2 Data CollectionVarious forms of data were collected to evaluate the thermal changes inCultus Lake with climate change. Historical observations dating to the1920s were obtained to provide insight into changes in Cultus Lake overthe past century. More recent data were used to calibrate and validate(hindcast) GLM which was used to investigate changes to the lake underdifferent warming scenarios projected through 2100.2.2.1 Historical DataThe first major limnological surveys of Cultus Lake were performed in the1920s and 1930s to study the propagation of Sockeye Salmon (e.g. Foer-ster 1925). The historical data are useful to provide insight into how thethermal characteristics of the lake have changed over the past century. Tem-perature profiles were available from April 1923 through October 1923, andJanuary 1924 near the north end of the lake (marked “F” on Figure 2.1;Foerster 1925). Monthly profiles were taken in June 1927–September 1929by Dr. Foerster at a “central station” (Ricker 1937); further semi-monthlyprofiles were taken at a central station (marked “R” on Figure 2.1) January1932–January 1937 (Ricker 1937). In 1923, temperature measurements wereperformed using a modified deep-sea thermometer equipped with a one-wayvalve that let in water on descent. The thermometer was kept at the desireddepth for several minutes before being rapidly hauled to the surface (Fo-erster 1925). For later studies, temperature measurements were performedusing a Negretti and Zambra deep-sea reversing thermometer with gradua-tions to 0.2 °C, and readings typically made to the nearest 0.1 °C (Ricker1937).Additional historic conductivity, temperature, and depth (CTD) datawere provided by Fisheries and Oceans Canada (DFO). Monthly data werecollected from 18 April 2001–13 March 2003 and 17 February 2009–presentday as part of ongoing limnological surveys. The primary DFO study stationis at approximately the same location as station CL2 (Figure 2.1). CTDcasts were performed using an Applied Microsystems Ltd. MicroCTD andan Applied Microsystems STD-12 Plus recording at 1 Hz.2.2.2 Meteorological DataMeteorological data were collected from several locations near Cultus Lake.An Onset HOBO H21 weather station was installed on the roof of the Cultus11Lake Marina at the northeast end of the lake on 12 August 2016 (Figure 2.1).This site was chosen to minimize tampering while also capturing wind speedsalong the lake’s long-axis. The weather station was equipped with an Onsetbarometric pressure sensor and anemometer set to measure the 10-minuteaverage pressure [mbar], and wind speed [m s-1] and wind direction [degreesfrom North]. Table 2.1 provides the accuracy, resolution, and drift of theweather station components. The elevation of the anemometer was esti-mated to be 5 m above the ground so a correction to 10 m was requiredassuming a log profile:U10 = Uzln(10z0)ln(zz0) (2.1)where U10 is the wind speed at 10 m, Uz is the observed wind speed atheight z, and z0 is the aerodynamic roughness length (Wieringa 1992). Forwinds between 140° and 310°, a roughness length of 0.005 m was assumedfor the smooth lake surface. Between 310° and 140°, a roughness of 2.0 m,described as ‘chaotic’, was assumed due to the irregular topography and theheight of the surrounding trees. Due to the relatively late installation ofthe weather station in the study, correlations of wind speed were requiredfor nearby Environment Canada reporting stations. Wind speed measure-ments obtained at the marina provided insight into the dominant wind di-rections and to determine the best fit for historical wind data and to fillin the missing dataset. No strong correlations were found for wind speedsbetween Cultus Lake and nearby weather stations. However, it was deter-mined that wind speeds reported at the Abbotsford International Airport(49.025°N 122.360°W, Elevation: 59.1 m, Climate ID: 1100030 & 1100031,WMO ID: 71108, ICAO: CYXX) provided the best fit, though were slightlystronger than those measured at the lake. Barometric pressure obtained atthe marina was used to develop a linear trend with nearby weather stations.A strong correlation was found with the barometric pressure measured atAgriculture Canada’s Agassiz Research and Development Centre (49.243°N121.760°W, Elevation: 19.3 m, Climate ID: 1100119, WMO ID: 71113).12Table 2.1: Accuracy, resolution, and drift of HOBO weather station instru-ments (Onset Computer Corporation 2008, 2017a)Instrument Measurement Accuracy Resolution DriftS-BPA-CM10Barometricpressure[mbar]Maximum ± 5 0.1Typical 1 peryearS-WCA-M003Windspeed/gust[m s-1]± 0.5± 3 % 17–30 m s-1± 4 % 30–44 m s-10.19 N/AWind direction[degrees]± 5 1.4 N/AHourly temperature [°C] and relative humidity [%] were obtained fromthe Environment Canada weather station located at the Cultus Lake SalmonResearch Laboratory (49.079°N 121.979°W, Elevation: 45.7 m, Climate ID:1102221). Total daily precipitation [mm] was obtained from the Abbotsfordstation. This measurement includes both rain and snow, where 1 cm of snowis equivalent to 1 mm of rain. Cloud cover fraction was obtained from theAbbotsford International Airport Meteorological Terminal Aviation RoutineWeather Reports (METARs). Three letter METAR abbreviations for cloudcover were converted to oktas, or eighths of the sky that is occupied byclouds, where CLR = 0, FEW = 2, SCT = 4, BKN = 6, OVC = 8 oktas.METARs report cloud cover with increasing altitude, so the maximum valuewas obtained for each measurement. Finally, hourly photosynthetically ac-tive radiation (PAR [mol m-2 s-1]) obtained at the Cultus Lake Laboratorywas provided by DFO. Incident shortwave radiation [W m-2] was calculatedfrom PAR by dividing by 2.11 (Kalff 2002). Gaps in the shortwave radi-ation data where the sensor malfunctioned were filled with the clear-skymodel described by Bird and Hulstrom (1981) and multiplied by a cloudfactor of (1 − 0.65C2), where C is the relative cloud cover of sky (Barryand Chorley 1976). Atmospheric parameters used in the clear-sky modelare shown in Table 2.2.Some historical meteorological data were obtained from different stationsthan those described above. Temperature and relative humidity data priorto 5 June 2012 (commissioning date of the Cultus Lake station) was obtainedfrom the Abbotsford Airport station. The air temperature correlation be-tween Abbotsford and Cultus Lake is given by TCultus = 1.034TAbbotsford −13Table 2.2: Bird clear-sky model atmospheric parametersParameter Value SourceAtmospheric pressure 1009 mbar Average from Onset weather stationOzone concentration 0.242 atm-cm 3-year average 52 ppb concentrationmeasured at Chilliwack air qualitymonitoring station (EnvironmentalReporting BC 2015) integrated overthe 10 km troposphereAerosol optical depth at500 nm0.094 2009–2014 average at SaturnaIsland (48.783°N 123.133°W;AERONET 2017)Aerosol optical depth at380 nm0.129 2009–2014 average at SaturnaIsland (48.783°N 123.133°W;AERONET 2017)Surface albedo 0.2 (Bird and Hulstrom 1981)Precipitable water mass 1.26 atm-cm 2009–2014 average at SaturnaIsland (48.783°N 123.133°W;AERONET 2017)0.792 with a coefficient of determination, r2, of 0.960. The correlation ofthe relative humidity data at Abbotsford to Cultus Lake is approximatelyone-to-one. Additionally, a weather station located in Chilliwack (49.172°N121.925°W, Elevation: 11.0 m, Climate ID: 11015300) was used for totaldaily precipitation values, because of its proximity to Cultus Lake, beforeit was decommissioned after August 2014. Hourly solar radiation and cloudcover were measured at Abbotsford Airport between 1953–2005 and sourcedfrom Environment Canada’s Canadian Weather Energy and EngineeringDatasets (CWEEDS; Environment and Natural Resources Canada 2017).2.2.3 Limnological DataWater temperature data were collected beginning 28 April 2016 at five sta-tions in the lake (Figure 2.1, Table 2.3). The stations were set along thelong-axis of the lake with a spacing of 660 m. Both high resolution tem-poral and spatial temperature data were obtained. High resolution tem-poral data were collected by installing subsurface moorings of thermistorchains equipped with Onset HOBO TidbiT v2 water temperature sensors14(UTBI-001, ± 0.21 °C accuracy, 0.02 °C resolution, 0.1 °C per year drift;Onset Computer Corporation 2017c) at 4 m depth intervals and logging at15 minute intervals at stations CL1, CL3, and CL5. The moorings at sta-tions CL3 and CL5 were installed on 28 April 2016, while the third mooringat CL1 was installed on 25 July 2016. The mooring at station CL3 was dis-turbed by a recreational boater on 5 June 2016 and redeployed by DFO to anearby location. This mooring was then returned to its original location on17 June 2016 after reducing the distance between the top two thermistorsby 2 m to shorten the line to reduce tampering. All three moorings wereretrieved and redeployed on 9 May 2017.Table 2.3: CTD cast and thermistor mooring station informationStation Location Mooring Deployment Mooring RetrievalCL1 49.062°N 121.981°W 25 July 2016 9 May 2017CL2 49.057°N 121.986°W N/A N/ACL3 49.052°N 121.990°W 28 April 2016 9 May 20171CL4 49.047°N 121.995°W N/A N/ACL5 49.042°N 122.000°W 28 April 2016 9 May 20171 Station CL3 mooring was removed by a boater on 5 June 2016 and was redeployedby DFO to 49.053°N 121.989°W. Mooring was returned to Station CL3 on 17 June2016.Higher resolution spatial data were collected at semi-monthly intervalsby performing conductivity, temperature, and depth (CTD) casts beginning20 May 2016 and continuing through 8 December 2016 at stations CL1–CL5with a Seabird SBE 19plus v2 SeaCAT Profiler (Table 2.4). Additional sen-sors were equipped on the unit: a WET Labs ECO fluorometer (to providechlorophyll-a readings), a Satlantic PAR sensor, and a Sea-Point turbiditysensor. Additional CTD casts were performed at the five stations on 28April 2016 and 5 May 2016 with DFO’s Applied Microsystems Ltd. Mi-croCTD. Vertical profiles were captured by lowering the CTD unit at arate of 0.5 m s-1 while recording at 4 Hz for the Seabird unit and 1 Hz forthe DFO unit. The Seabird CTD unit was turned on prior to the casts toobtain a 30 second average barometric pressure correction factor that wasapplied in post-processing. The data were processed in MATLAB using theGibbs-SeaWater (GSW) Oceanographic Toolbox v3.05 containing the Ther-modynamic Equation of Seawater 2010 (TEOS-10) functions which assumesthe chemical composition of seawater (IOC 2010). The GSW toolbox was15used to convert measured conductivity, temperature, and pressure to accu-rate freshwater depth, salinity, and density values. Data points from eachcast were averaged into 1 m layers to reduce minor observational errors. Forwater temperature, these layers were then spatially averaged across the fivestations to calculate an average profile for the entire lake. Additional filter-ing was performed on the turbidity profiles to remove random spikes in thedata: after examining several vertical profiles, it was determined that anyturbidity readings greater than 5 FTU in the epilimnion (depth less than10 m) were anomalous and were removed.Table 2.4: Accuracy, resolution, and drift for SBE 19plus V2 CTD (SeabirdScientific 2016a)Measurement Accuracy Resolution DriftTemperature [°C] ±0.005 0.0001 0.0002 per monthConductivity[S m-1]±0.00050.00001 (fresh-water)0.0003 per monthPressure±0.1 % of fullscale range0.002 % of fullscale range0.1 % of full scalerange per yearBathymetric data were obtained from a Canadian Hydrographic Serviceacoustic survey performed on Cultus Lake in 2016 using a 2 m grid spacing.AutoCAD Civil 3D was used to create a contour map (Figure 2.1). A stage-storage calculation was performed in AutoCAD Civil 3D at 1 m verticalintervals to create a hypsograph (Figure 2.2). The water level in CultusLake is controlled by means of stop logs to ensure a fairly constant waterlevel throughout the summer months when precipitation is low. The waterelevation has historically remained between 44 m and 45 m a.s.l. The averagewater elevations was assumed to be 44.5 m based on the 2016 water leveldata provided by the Cultus Lake Park Board (Figure 2.3), which results ina standardized maximum lake depth of 42.5 m.160 1 2 3 4 5 6 7x 106051015202530354045Area [m2]Depth [m]Figure 2.2: Hypsographic curve for Cultus Lake from 1 m stage-storageintervals.Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec4444.144.244.344.444.544.644.744.8Lake Water Level Elevation [m]Figure 2.3: Cultus Lake water levels for 2016. Data provided by the CultusLake Park Board is in metres above sea-level.172.2.4 Water Clarity and Extinction CoefficientLight penetration into the water column is an important driver for shortwaveradiation flux and the total available energy for photosynthesis (Kalff 2002).Light penetration decays with depth following the Beer-Lambert Law:Iz = I0e−kz (2.2)where Iz is the irradiance [µmol m-2 s-1] at depth z [m], I0 is the irradiancejust below the water surface, and k is light extinction coefficient [m-1]. Thelight extinction coefficient can therefore be calculated from the slope of−ln(Iz/I0) with depth z. The 1 m averaged PAR profile from each semi-monthly excursion was used to calculate k. This calculation has been limitedto the euphotic zone which can be defined as the region that receives greaterthan one percent of the surface light penetration (Kalff 2002). The surfaceincident light was taken as the value in the second layer due to a 0.6 m offsetbetween the pressure sensor and PAR sensor. To complement this methodof calculating k, a Secchi depth measurement was taken from the shadedside of the vessel at station CL2 to coincide with DFO measurements. Thelight extinction coefficient can be calculated by dividing the measured Secchidepth by 1.7 (Kalff 2002).2.2.5 Inflows and OutflowThree inflows for Cultus Lake were regularly measured for discharge, tem-perature, conductivity, and turbidity. The three largest inflows by drainagearea were determined from Putt (2014): Frosst Creek, Watt Creek, andSmith Falls Creek. Gauging station location and drainage area informa-tion are provided in Table 2.5. Temperature, conductivity, and turbiditywere measured using the Seabird CTD initially on 17 June 2016 and semi-monthly beginning 25 July 2016 until 16 November 2016. Semi-monthlydischarges were measured in the three inflows using a Swoffer 3000 flowmeter (3 % accuracy; Instruments 2016) beginning 25 July 2016 until 16November 2016. A HOBO TidBiT v2 logger was installed in Frosst Creekon 17 June 2016 just upstream of the Columbia Valley Highway to measurewater temperatures every 10 minutes alongside a HOBO water level log-ger (U20-001-01; Pressure: ±0.3 % accuracy, <0.02 kPa resolution, <0.3 %stability; Temperature: ±0.44 °C accuracy, 0.1 °C resolution, 0.1 °C stabil-ity (Onset Computer Corporation 2017b)) that measured water pressure at10 minute intervals. The water pressure measured by the logger was baro-metrically adjusted using the corrected pressure measured at Agassiz and18the onshore weather station. This adjustment was used to calculated waterdepth which was used to obtain values for discharge based on a rating curvedeveloped from measurements obtained by Putt (2014), assuming a powerrating curve equation (ISO 2010):Q = β(h− h0)α (2.3)where Q is the discharge [m3 s-1], h is the stage [m]. β, h0, and α arecalibration coefficients: β is the discharge when (h − a) is 1 m, h0 is thegauge height when flow is zero, α is the rating curve slope. This equationallows a shift in values for h0 between locations, assuming the creek hasthe same cross-sectional shape at those locations. Since pressure loggerswere not installed on Smith Falls and Watt Creeks, rating curves for thosecreeks were not created, and a linear trend was assumed between dischargemeasurements. As a result, instantaneous discharges were not obtained forSmith Falls and Watt Creeks for 2016. Instantaneous discharges for thesethree inflows for 2011–2013 were obtained from Putt (2014).Table 2.5: Stream gauging station informationStream Name Location Drainage Area [km2]Frosst Creek 49.033°N 122.027°W 36.3 (Water Survey ofCanada Station 08MH072)Watt Creek 49.031°N 122.005°W 6.58 (Putt 2014)Smith Falls Creek 49.060°N 121.959°W 6.49 (Putt 2014)Sweltzer River 49.078°N 121.978°W 75.0 (Shortreed 2007)A linear regression was established between the air temperature andcreek temperatures of the form:Ts = aTa + b (2.4)where Ts is the stream temperature [°C], Ta is the air temperature [°C],and a and b are stream specific coefficients. Table 2.6 provides the calcu-lated coefficient values. This correlation was used to determine the dailyaverage creek temperature for Watt and Smith Falls creeks rather than in-terpolating between stream temperature measurements. For Frosst Creek,the hourly average stream temperature from the HOBO TidBiT was used tocorrelate with the hourly air temperature. The temperature of the smallerwatercourses (Watt Creek and Smith Falls Creek) have higher r2 values19than Frosst Creek because of their stronger response to changes in air tem-perature from a smaller thermal inertia (Stefan and Preud’homme 1993).Turbidity and salinity values derived from the CTD data in each creek werelinearly interpolated between measurements.Table 2.6: Stream temperature coefficients for Equation 2.4Stream Name a b r2Frosst Creek 0.167 6.885 0.629Watt Creek 0.422 3.626 0.885Smith Falls Creek 0.344 6.395 0.905Outflow data on Sweltzer River were obtained by installing a HOBOwater level logger (U20-001-01) recording at 10-minute intervals on 5 Octo-ber 2016 downstream of the fish counting fence outside the Salmon ResearchLaboratory and barometrically adjusting the water pressure to obtain depth.Flow measurements were obtained on 5 October 2016 and 8 December 2016to adjust a power rating curve fitted to data from Putt (2014). DFO pro-vided daily staff gauge readings and temperature for the period of 11 July2016–20 October 2016, which were also adjusted to the rating curve. How-ever, the staff gauge is located upstream of the fish counting fence, thus theconversion between readings and discharge is only applicable when sectionsof the fence are open because the closed fence acts as a weir.2.3 General Lake Model (GLM)Cultus Lake was modelled using the open-source one-dimensional Lagrangianhydrodynamic General Lake Model (GLM, v2.2β) developed by the AquaticEcoDynamics Research Group at the University of Western Australia (Hipsey,Bruce, et al. 2014). GLM is used to simulate long-term changes in the ver-tical distribution of temperature, salinity, and density by accounting for theeffects of surface heating and cooling, and inflows and outflows. The modelcan be coupled with the Aquatic EcoDynamics library (AED) to simulatelake biogeochemistry; however, AED was not incorporated in this study.2.3.1 Model DescriptionGLM follows a Lagrangian structure with homogeneous layers that can con-tract and expand from changes in the layer density due to surface heating,20vertical mixing, inflows, and outflows. Layers are sorted by stability accord-ing to density and with varying thicknesses to resolve the density gradient.The density of each layer is computed at each timestep based on tempera-ture and salinity from the equation of state described in UNESCO (1981).Maximum and minimum layer thicknesses are set by the user; the maximumthickness should be at least twice the minimum thickness to ensure properlayer merging. The primary drivers for the model, and changes in layerdensity, are from meteorological forcings that alter the water temperature.Surface boundary layer fluxes provide energy for heating and mixing thelake. The driving fluxes are shortwave radiation, longwave radiation, andsensible (convective) and latent (evaporative) heat fluxes. Solar radiationis the primary driver for lake thermodynamics. Solar radiation data maybe specified directly or GLM can calculate the theoretical surface irradiancebased on the clear-sky model from Bird and Hulstrom (1981) which can becorrected for cloud cover, if available. The reflected fraction of shortwave ra-diation is dependent on the lake albedo. Several options for the calculationof albedo are given for the user (see Hipsey, Bruce, et al. 2014). Shortwaveradiation penetrates into the lake following the Beer-Lambert Law (Equa-tion 2.2). The Beer-Lambert Law is only applied to the PAR fraction (45 %in GLM) of incident light, while the surface layer accounts for all incidentlight that is not reflected. Longwave radiation can be specified as a net flux,incoming flux, or an incoming flux computed based on cloud cover frac-tion and air temperature. Several options described in Hipsey, Bruce, et al.(2014) are provided to calculate longwave emissivity accounting for reflec-tion from clouds. If no longwave radiation data are available, the model willapproximate the cloud fraction from solar irradiance. Sensible and latentheat fluxes are calculated from common bulk aerodynamic formulae. Thesensible heat flux is dependent on wind speeds and air temperatures, whilethe latent heat flux is most strongly dependent on wind speeds. The sensibleand latent heat fluxes account for a minor portion of the net heat flux andcan vary greatly day-to-day depending on wind speeds.Surface mixing in GLM is accomplished by comparing the available en-ergy and required energy to undergo mixing at each time-step. The balancebetween the turbulent kinetic energy (TKE) and the required energy to un-dergo mixing (potential energy) provides the surface mixed layer deepeningrate. Wind stirring, convective overturning, and Kelvin-Helmholtz (K-H)billowing contribute to the TKE, while the potential energy is the energyrequired to lift up water at the bottom of the mixed layer and accelerated itto the mixed layer velocity in addition to energy dissipated in K-H produc-tion (Hipsey, Bruce, et al. 2014). The premise for the model’s algorithm is21described in Imberger and Patterson (1981); a complete outline for surfacemixing is available in Hipsey, Bruce, et al. (2014).The layers are subject to a mass balance at the end of each day thatare dependant on evaporation, precipitation, inflows, and outflows. Layervolumes are calculated from the layer thickness and interpolated areas fromthe user-defined hypsographic curve. Evaporation and precipitation onlyoccur in the surface layer, but all layers are subject to mass conservationfrom inflows and outflows. Both surface runoff and submerged inflows (e.g.groundwater) can be simulated. For surface inflows, the inflow intrudes themodel layers and entrains water from them until the inflow reaches a neutralbuoyancy, where a new layer is formed. The rate of entrainment is adaptedfrom Fisher et al. (1979). The user specifies the stream half-angle and theincoming slope. Submerged inflows are inserted at a user specified depth andhave no entrainment, but instead mix with adjacent layers until a neutralbuoyancy is reached. Three types of withdrawals may be specified: out-flows, seepage, and overflow. Withdrawals are all calculated from a similaralgorithm; however, a depth must be specified for outflows, whereas seepageis removed from the bottom layer, and overflow is the volume of water inexcess of maximum storage.2.3.2 Model SetupThe physical model is configured in a text file split into several blocks:ˆ glm setup: general simulation information including number of layers,maximum and minimum layer thickness, minimum layer volume, lightextinction coefficient, and mixing parametersˆ morphometry: lake latitude and longitude, basin length and width,and bathymetry input as a hypsograph.ˆ time: simulation start and stop times, timestep, and time zoneˆ output: output file details for depth specific outputs and bulk file, dailylake heat flux and water balance summary file, outlet and overflowoutput file specificationsˆ init profiles: initial values of depth, temperature, and salinity.ˆ meteorology: information on surface forcing and meteorology data.Options for calculating albedo and emissivity are specified. A sepa-rate text file containing data for shortwave radiation, cloud cover, air22temperature, relative humidity, wind speed, daily rainfall. Sub-dailyvalues can be provided, except for rainfall because the mass balanceoccurs only once per day.ˆ inflows: inflow morphometry consisting of the streambed half-angle,slope, and drag. Inflow data for each inflow are provided in its own fileconsisting of daily discharge, stream temperature, and stream salinity.ˆ outflows: outlet elevation, basin length and width at outlet. Similarlyto inflows, outflow discharge for each outflow is provided in its ownfile. Outflow temperature is calculated by the model depending on theelevation of the outlet.GLM was run for several periods to simulate the thermal characteristicsof Cultus Lake: a calibration in 2016, a validation in 2009–2016, a sec-ond validation and climate model evaluation in 2001–2003, and two forecastscenarios for 2009–2100. The initial model calibration was completed us-ing the semi-monthly field data obtained in 2016. Because GLM is a one-dimensional model, spatially averaged thermal profiles were used for modelevaluation. Lake conditions from 28 April 2016 at 17:00 PST were used asthe initial profile. Both temperature and salinity were included in the modelfor the calibration period, but only thermal profiles were used to evaluatethe model. The model was run at an hourly timestep until 12 December2016 at 12:00 PST and data from every timestep was saved. Only thermalprofiles from a single station were available for the 2001–2003 and 2009–2016periods, so the initial salinity was assumed to be the same as the observa-tions in 2016. The model was run at an hourly timestep from 17 February2009 12:00 to 8 December 2016 12:00. A daily profile at noon was saved formodel validation. For the 2001–2003 hindcasts, the model was initialized at18 April 2001 12:00 and run hourly until 14 March 2003 12:00. Again, onlya daily profile was saved to file. Finally, for the 2009–2100 forecasts, themodel was run at an hourly timestep, and only a daily profile was saved.The forecast simulations were initialized at 17 February 2009 12:00. Theregional climate forecast data are only available at a daily timestep (Sec-tion 2.4), but because the model was calibrated and validated using hourlymeteorological data, the model forecast scenarios were run at an hourlytimestep using daily average data. However, to prevent excess heating fromshortwave radiation, the model automatically assumes an idealized diurnalcycle based on the daily average shortwave radiation (Hipsey, Bruce, et al.2014).232.4 Regional Climate ModelGLM was coupled with the fourth generation Canadian Regional ClimateModel (CanRCM4; Environment and Climate Change Canada 2017) to fore-cast the effects of climate change on Cultus Lake. A regional climate modelwas used because it reduces the need for bias correction to regional climatessuch as Cultus Lake. CanRCM4 employs the same physics and setup as thesecond generation Canadian Earth System Model (CanESM2) but down-scaled to a 0.22° horizontal grid (Scinocca et al. 2016). CanESM2 employsan atmospheric-ocean general circulation model, a land-vegetation model,and terrestrial and oceanic carbon cycling. The first generation model isdescribed in Christian et al. (2010) with updates to the second generationdescribed in Arora et al. (2011). CanESM2 was used as the Canadian contri-bution for the Intergovernmental Panel on Climate Change Fifth AssessmentReport (AR5) as part of the fifth phase of the Coupled Model Intercompar-ison Project (K. E. Taylor et al. 2012; Flato et al. 2013). A result of AR5was the development of new climate change scenarios called RepresentativeConcentration Pathways (RCP). The four RCPs in increasing severity areRCP2.6, RCP4.5, RCP6.0, and RCP8.5 named after the radiative forcingvalues for 2100 above pre-industrial levels (R. H. Moss et al. 2010). ForCanRCM4, three experiment scenarios are available: a historical simulationfor 1950–2005, and two climate change scenarios RCP4.5 and RCP8.5 for2006–2100. The historical scenario is used as a control run to evaluate theclimate model with observed data (K. E. Taylor et al. 2012). RCP4.5 isa moderate warming scenario with an increase of radiative forcing stabi-lizing near 4.5 W m-2 and an atmospheric CO2-equivalent concentration of650 ppm after 2100 (R. H. Moss et al. 2010). RCP8.5 is an extreme warmingscenario with an increase of radiative forcing greater than 8.5 W m-2 and anatmospheric CO2-equivalent concentration greater than 1370 ppm in 2100(R. H. Moss et al. 2010).2.4.1 Forecast DataClimate forecast data were obtained from CanRCM4 for the grid point near-est Cultus Lake located 10 km to the south. (48.960°N 122.000°W, Eleva-tion: 589.75 m). Daily outputs obtained from CanRCM4 were the following:surface temperature [K], surface downwelling shortwave radiation [W m-2],total cloud cover [%], precipitation [kg m-2 s-1], near-surface specific humid-ity [%], pressure at the grid point [Pa], sea level pressure [Pa], eastward andnorthward near-surface wind [m s-1], and total runoff [kg m-2 s-1].24Some minor adjustments were required to optimize the CanRCM4 out-puts from the grid point to Cultus Lake. Precipitation and total runoff wereconverted to m s-1 by dividing by the density of water (1000 kg m-3); totaldaily precipitation was calculated by subsequently multiplying by 86400 s d-1.East and north components of wind speed were combined using Pythagoras’theorem to obtain values for the daily average wind speed.Air temperature was adjusted to the elevation of Cultus Lake by themoist adiabatic lapse rate following Stull (2015). First, ε = 0.622 [unitless]is the ratio of the molecular weight of moist air to the molecular weight ofdry air. The saturated vapour pressure, es [kPa], is then calculated fromthe Clausius-Clayperon equation:es = e0 exp[Lv<v(1T0− 1T)](2.5)where e0 = 0.611 kPa, T0 = 273.15 K, <v = 461 J kg-1 K-1 is the gas constantfor water vapour, and T is the temperature in Kelvin. Lv is the latent heatof vapourization, which for water is 2.5x106 J kg-1. The saturated mixingratio, rs [unitless], is calculated from pressure in kPa:rs =εesP − es (2.6)Finally the moist adiabatic lapse rate, Γs [K m-1], is given as:Γs =|g|Cp1 + rsLv<dT1 + L2vrsεCp<dT 2(2.7)where g is the gravitational constant, <d = 0.622<v is the gas constantfor dry air, and the specific heat of air Cp ≈ Cpd(1 + 1.84rs) where thespecific heat of dry air, Cpd, is assumed to be 1004 J kg-1 K-1. The moistadiabatic lapse rate was then used to adjust the air temperature data froman elevation of 589.75 m to 44.5 m.Relative humidity was calculated from specific humidity following Stull(2015). First, the air pressure at Cultus Lake, P [kPa], was calculatedfollowing the barometric formula using the new air temperature [K]:P = P0 exp(−0.0342zT)(2.8)The sea level pressure [kPa] was used as P0 because the elevation and airtemperature at Cultus Lake is closer to that at sea level than that at the25grid point. A new value for es was calculated for Cultus Lake from Equa-tion 2.5 using the adjusted temperature. The water vapour pressure, e [kPa],at Cultus Lake was calculated from the adjusted pressure and the specifichumidity, q, assuming no change in specific humidity with altitude:e =qPε+ (1− ε)q (2.9)Finally the relative humidity [%] is given as:RH = 100ees(2.10)26Chapter 3ResultsResults from the 28 April 2016–8 December 2016 field campaign are pre-sented in this chapter along with results of the GLM simulations. The fielddata include the limnological data obtained from the thermistor mooringsand the semi-monthly CTD casts, specifically the temperature and conduc-tivity, and results relating to water clarity (light penetration, turbidity, andchlorophyll-a). The inflow data from Frosst Creek, Watt Creek, and SmithFalls Creek are described in addition to the outflow data from Sweltzer River.The Cultus Lake specific tuning of GLM is outlined and evaluated to the2016 field data. Two hindcast scenarios are presented: the first from 2009–2016 to validate the 2016 model calibration, and the second from 2001–2003to evaluate climate change data that will be used to forecast the model.Historical thermal data dating to 1923 are compared to more recent data(2001–2003 and 2009–2016) to evaluate changes in heat content and thestratification season. These historical data are also compared to projectionsin the forecast model to 2100 AD.3.1 TemperatureHigh temporal resolution temperature data were obtained from a set of ther-mistor moorings installed at three locations in the lake, while high spatialresolution temperature data were obtained from semi-monthly CTD castsat five locations in the lake; the results are shown in Figures 3.2 and 3.1a,respectively. The lake was already stratified on the first date of the field cam-paign (28 April 2016). The surface temperature was initially above 13 °C,while the hypolimnion temperature was near 6.5 °C. The thermocline, de-fined here as the plane of maximum decrease in temperature (Wetzel 2001),was initially at a depth of 8 m. The surface temperature rose at a steady ratethroughout the summer. A strong heating period was observed in mid-Juneat the same time as the development of a secondary thermocline. Secondarythermoclines occur during days with high surface heating by radiation andwhen winds are insufficient to mix this warm water with the cooler water be-low, and can persist for many days if wind energy remains low (Lewis 1979).27The secondary thermocline is shown by a divergence in temperatures of theupper two thermistors on moorings CL3 and CL5: the surface thermistors(2.2 m and 2.7 m at CL3 and CL5, respectively) recorded temperatures of21 °C, but the second shallowest thermistors (6.2 m and 5.5 m at CL3 andCL5, respectively) did not record this sharp increase in temperature. AfterJune, the epilimnetic thermistors heated in tandem indicating the lake sur-face layer was well mixed. A maximum surface temperature of 22 °C wasobserved in late August. The thermocline continued to deepen throughoutthe summer to a depth of 13 m by 5 October 2016. The thermistors at 10.6,10.2, and 9.5 m for stations CL1, CL3, and CL5, respectively measured ahigh temperature variability because these thermistors were located near thethermocline. In the fall, temperatures began to decline in the epilimnion andhomogenize as wind began to mix the lake. Sudden increases in temperatureat mid-depth in October occured as the warm upper layers mix downwards.The lake finally became de-stratified on 8 December 2016 which was thesame day that de-stratification was observed with CTD casts. The lake re-mained relatively isothermal throughout the winter. In late January 2017,surface temperatures cooled to below 3.5 °C while temperatures at depthremained slightly below 4 °C, and the lake briefly inversely stratified. Inmid-February, the lake was isothermal at roughly 2.7 °C. Ice rarely forms onCultus Lake because air temperatures generally remain at or above 0 °C; thewater surface therefore remains open for wind to continue mixing the lakeat cooler temperatures which reduces the mixed lake temperature to belowthe temperature of maximum density without inversely stratifying. Finally,increased warming in the spring caused the lake to stratify in April 2017.Remarkably, the hypolimnetic temperatures recorded in April and May 2017are roughly 1 °C cooler than hypolimnetic temperatures observed in Apriland May 2016. The cooler hypolimnetic temperatures may be caused byincreased cooling in the winter of 2016–2017 relative to the year before.Strong wind events dominated along the long-axis of Cultus lake causeshigh spatial variability due to internal seiching. Higher temperature vari-ability throughout the summer is seen in the upper five thermistors in theoff-centre moorings compared to station CL3 (Figure 3.2), which is evidenceof seiching. The highest temperature variability is observed near the ther-mocline where the temperature gradient is largest. A sudden increase intemperature at 14.6 m in Figure 3.2a coincided with a sudden decrease intemperature at 9.5 m in Figure 3.2c on 7 October 2016 at 11:15 a.m. Thisevent also coincided with a sudden brief spike in the outflow discharge inSweltzer River (Figure 3.4), suggesting a sudden surge of water from thelake. A sudden inflow was not observed by the logger in Frosst Creek; how-28ever, the event occurred at the same time as a large wind event orientedalong the long-axis of the lake (220°) with gusts up to 18 m3 s-1 (Figure 3.3).Because the lake was still strongly stratified, there was a large downwellingof warm water at the downwind end of the lake to a depth of at least 14.6 m,along with a large upwelling of cold water at the upwind end of the lake to adepth of at least 9.5 m, and a surge of outflowing water. However, the windsetup was brief and when winds subsided, large internal waves from seichingwere not captured by the thermistors. Post event, there was high variabilityin the thermistor temperatures at mid-depth at CL5 that was not seen inthe other moorings. This variability occurred as inflow discharges continueto be elevated, suggesting the cool water from Frosst Creek was plunging.29Depth [m]22 20 18 16 14 12 10877820181614 12 10(a)0102030Depth [m] 182182182 184 186 188190 192192190 188186186186184188(b)010203001020Depth [m](c)Depth [m] 0.510.5 1 0.50.51210.50.510.50.512 3 210.5(d)0102030Depth [m]0.52430.5110.5130.5 1 1(e)May Jun Jul Aug Sep Oct Nov Dec0102030Figure 3.1: Results of limnological field data: (a) water temperature [°C]at station CL3, (b) specific conductivity [µS m-1] at CL3, (c) Secchi depth(solid) at station CL2 and spatially averaged euphotic depth (dashed),calculated from PAR measurements, (e) turbidity [FTU] at CL3, and (f)chlorophyll-a concentration [mg m-3] at CL3.300510152025Temperature [°C]  (a)3.3 m6.6 m10.6 m14.6 m18.6 m22.6 m26.6 m30.6 m34.6 m38.6 m0510152025Temperature [°C]  (b)Removed → ← Returned2.2 m6.2 m10.2 m14.2 m18.2 m22.2 m26.2 m30.2 m34.2 m38.2 mMay 2016 Jul 2016 Sep 2016 Nov 2016 Jan 2017 Mar 2017 May 20170510152025Temperature [°C]  (c)2.7 m5.5 m9.5 m13.5 m17.5 m21.5 m25.5 m29.5 m33.5 m37.5 m40.5 mFigure 3.2: Temperature at incremental depths on thermistor moorings CL1 (a), CL3 (b), and CL5 (c). Dashedlines at 4 °C indicate temperature of maximum density. Note: CL3 mooring was repositioned by a boater on 5June 2016, and returned to the original location on 17 June 2016. Depth of top thermistor in (b) changed from2.2 m to 4.2 m on 17 June 2016.31051015Wind speed [m s−1 ](a)0102030Wind gust [m s−1 ](b)Aug Sep Oct Nov Dec090180270360Wind direction [°](c)Figure 3.3: 10-minute average wind speed (a), gust (b), and direction (c)at the Cultus Lake Marina.3.2 ConductivityConductivity in Cultus Lake was measured throughout the 2016 field cam-paign at the five stations using a CTD. Raw conductivity measurementswere converted to specific conductivity (conductivity referenced to 25 °C)from the SBE data processing toolbox (Seabird Scientific 2016b):κ =C[1 + 0.02(T − 25)] (3.1)32where κ is the specific conductivity [S m-1], C is the measured conductivity[S m-1], and T is the measured temperature [°C]. On 5 October 2016, castswere performed at station CL2 with both the Seabird CTD and the Mi-croCTD to compare results and to determine an offset between data sources.The specific conductivity values calculated from the MicroCTD data wereincreased by 7.903 µS cm-1 to match the Seabird CTD data; this offset wasused for the measurements taken on 28 April 2016 and 4 May 2016. The spe-cific conductivity for the central station, CL3, is shown in Figure 3.1b. Thespecific conductivity is rather constant at depth in late spring at roughly182 µS cm-1. The epilimnetic specific conductivity increased to a maxi-mum of 192 µS cm-1 in late August through early October, and began todecrease in the fall to 186 µS cm-1. This epilimnetic increase of conductiv-ity may be caused by atmospheric deposition of ions and particles from airpollutants and sea spray (Wetzel 2001) originating in the Lower Mainland.Another source may be the increase in inflow conductivity in the summer(Section 3.4). The hypolimnetic specific conductivity increased to slightlyabove 187 µS cm-1 by November. A sharp gradient developed throughoutthe summer at the thermocline. Conductivity measurements were convertedto absolute salinity following the TEOS-10 MATLAB toolbox assuming thelake water has the same chemical composition as seawater. Absolute salinityranged between 0.081 g L-1 and 0.091 g L-1.3.3 Water Clarity and Chlorophyll-aSecchi depth measurements were obtained at station CL2 to determine waterclarity and light extinction coefficient. The results are shown in Figure 3.1c.The Secchi depth was greater than 8 m in May and decreased into earlysummer to 6 m. The Secchi depth began to decrease at the beginning ofAugust, and decreased steadily throughout the fall to a depth of 11.4 m on16 November 2016. It should be noted that high surface waves and heavyrain on 2 November 2016 obscured observations of the Secchi disk. Theaverage Secchi depth over the study period was 8 m. This corresponds to alight extinction coefficient, k, from Equation 2.2 of 0.22 m-1 by dividing theaverage Secchi depth of 8 m by 1.7 (Kalff 2002).A comparison to the light extinction coefficient obtained from the Secchidepth was performed using PAR data according to Equation 2.2. The eu-photic depth roughly follows the trend seen in the Secchi depth (Figure 3.1c).Light penetration on 8 December may have been reduced by high waves af-fecting the light’s angle of incidence and acting as a lens (Dera and Gordon331968), low angle of the sun’s incidence and clouds cover (Reynolds 2009) dueto the time of year, and partial obscuration of the sun by mountains to thesouth along International Ridge. A value for k was calculated for each layerand averaged spatially through the euphotic zone resulting in an averagevalue of 0.23 m-1 for the study period, excluding 8 December 2016—similarto k calculated using the Secchi depth.The turbidity in Cultus Lake was relatively low at 0.5–1.0 FTU. Therewas an increase in turbidity to above 1.0 FTU near the thermocline in thesummer, during the same period as the reduction in Secchi depth (Fig-ure 3.1d) suggesting that the increase in turbidity reduced water clarity. Thegreatest increase in turbidity occured near the bottom of the lake through-out the summer. A near-benthic turbidity of 1.5 FTU was observed in lateMay at 40 m, which increased beyond 5.0 FTU in September. This increaseis not likely caused by bottom disturbance from the CTD since only thedowncast data were analyzed.Chlorophyll-a is a surrogate for standing algal biomass (Kalff 2002). Aclear increase in chlorophyll-a was observed throughout the summer nearthe thermocline to above 4.0 mg m-3 (Figure 3.1e). A strong negative cor-relation was observed between Secchi depth and turbidity, suggesting thatbiological turbidity from phytoplankton productivity decreased water clarityat metalimnetic depths. Unlike turbidity, chlorophyll-a was not detected inthe benthic region of the lake. Presumably, the increase in benthic turbiditywas from sedimentation of senescent organic matter (eg. algae and plants)where the chlorophyll-a has broken down to prevent fluorescence.3.4 Inflows and OutflowInflows were monitored beginning 17 June 2016. A water level logger wasinstalled in Frosst Creek to measure the hydrostatic pressure, which wasused to calculate the discharge following Equation 2.3 (Figure 3.4). Theflow rate remained low throughout the summer near 0.2 m3 s-1. Streamdischarge increased in the fall as rainfall increased, and exhibited a strongresponse to rainfall. The maximum discharge observed was 5.7 m3 s-1 onOctober 8. The discharge remained steady in the fall at approximately2.0 m3 s-1. The temperature of Frosst Creek showed a strong diurnal cycleof 2.0 °C in the summer when flows were low. As flows increased in the fallwith increased precipitation, the diurnal cycle disappeared as the streamtemperature became less responsive to air temperatures because the thermalinertia of the stream increased (Stefan and Preud’homme 1993).34Discharge [m3  s−1 ](a)  0246Frosst CreekSweltzer RiverStream Temperature [°C](b)0102030Precipitation [mm](c)Jun Jul Aug Sep Oct Nov Dec0204060Figure 3.4: 10-minute average discharge derived from rating curve (a)and temperature (b) of Frosst Creek and Sweltzer River, and total dailyprecipitation (c) at Abbotsford Airport for 17 June 2016 to 8 December 2106.Circles indicate measured values; black diamond on 2 November indicatesflow was estimated because stream velocities were too high to safely measure.Daily staff gauge and water temperature readings on Sweltzer River wereprovided by DFO for dates prior to 5 October 2016.35Stream flow data for the minor tributaries, Watt Creek and Smith FallsCreek are provided in Figure 3.5. The minor tributaries had significantlyless flow than Frosst Creek. The flow regime of Watt Creek was found tobe intermittent: drying up in the summer and increasing to 0.2 m3 s-1 inthe late fall. Smith Falls Creek had a summer baseflow of 0.03 m3 s-1 andalso increased to 0.2 m3 s-1 in the late fall. The stream temperatures ofFrosst Creek and Watt Creek were relatively similar; both were observedat 12 °C on 17 June and decreasing to 7 °C in November. However, on25 July, Watt Creek was observed to be 5 °C warmer than Frosst Creek,likely due to the low flow rate being quickly warmed by ambient summerair temperatures. The temperature of Smith Falls Creek was on average3.6 °C warmer than Frosst Creek throughout the year, with the largestdifference on 12 August (6.7 °C). The higher temperature of Smith FallsCreek was possibly caused by warming of extensive wetlands in the upperreaches of this catchment. Specific conductivity followed a similar trend.The average specific conductivity was 204 µS m-1, 194 µS m-1, and 318 µS m-1for Frosst Creek, Watt Creek, and Smith Falls Creek, respectively. Thespecific conductivity for Smith Falls Creek was consistently the highest ofall three inflows and is consistent with results from Putt (2014). All threeinflows saw a decrease in specific conductivity in the fall because of theincrease flow rates diluting the water. Increased flows also likely contributedto freshening of the lake in the fall (Section 3.2). Finally, the turbidityin the inflows was 22.4 FTU, 12.6 FTU, and 12.5 FTU for Frosst Creek,Watt Creek, and Smith Falls Creek, respectively. However, a large spike to103 FTU in Frosst Creek was observed on 25 July of unknown cause and ispotentially an anomalous reading. Frosst Creek saw a decrease in turbidityfrom the summer into the fall, while levels for Watt Creek remained constantrelative to 25 July, and Smith Falls Creek saw an increase in turbidity.Flow rates and temperatures of the outflow, Sweltzer River, are providedin Figure 3.4. Summer discharges ranged between 0.5 m3 s-1 and 1.9 m3 s-1,and summer outflow temperatures ranged between 20 °C and 23 °C. Flowrates increased in the fall as precipitation increased, peaking at 4.8 m3 s-1.Increases in discharge were not responsive to rain events due to the control-ling aspect of the lake. Stream temperatures steadily declined in the fall,with a decrease in lake surface temperatures, to a temperature of 7 °C.36Jun Jul Aug Sep Oct Nov Dec00.20.4Discharge [m3  s−1 ]Jun Jul Aug Sep Oct Nov Dec5101520Temperature [°C]Jun Jul Aug Sep Oct Nov Dec0200400Sp. Conductivity [µS m−1 ]Jun Jul Aug Sep Oct Nov Dec050100150Turbidity [FTU]  Frosst CreekWatt CreekSmith Falls CreekFigure 3.5: Discharge for Watt Creek and Smith Falls Creek, and specificconductivity and turbidity for Frosst Creek, Watt Creek, and Smith FallsCreek. Watt Creek was dry between 12 August 2016 and 5 October 2106.373.5 Modelling Cultus LakeSeveral modelling periods were run in GLM to simulate the thermal char-acteristics of Cultus Lake. First, the model was calibrated using field dataobtained in 2016. A sensitivity analysis was also done for the 2016 calibratedmodel. Secondly, a hindcast was done for 2009–2016 to validate the cali-brated model. Thirdly, another two hindcasts were performed for 2001–2003to evaluate the regional climate model data by comparing them to observeddata. Finally, GLM was run for 2009–2100 for two climate change scenarios.3.5.1 Calibration PeriodWater temperature data collected in 2016 were used to calibrate GLM. Theinitial profile was set as the spatially averaged temperature profiles binnedinto 1 m layers as calculated from CTD casts on 28 April. The initial lakedepth was set to 42.5 m as provided by the Cultus Lake Park Board. Anhourly timestep was used in the model. A summary of the calibration pa-rameters used in the model are shown in Table 3.1. Minimum and maximumlayer thicknesses of 0.3 m and 0.9 m, respectively, to ensure that the mini-mum layer was not so small as to form ice in the winter and that three smalllayers could mix together to form a larger layer. The extinction coefficientof 0.22 m-1 was calculated from the average Secchi depth and PAR profilesover the study period. All aerodynamic and mixing coefficients were setto the default values. Good results were found using albedo mode 3 andcloud mode 4 (see Hipsey, Bruce, et al. 2013 from Yajima and Yamamoto2015). Wind data were reduced by 5 %, due to higher wind values observedat Abbotsford Airport than at Cultus Lake, likely the result of topographicand predominant atmospheric flow differences. Shortwave radiation was in-creased by 30 %. This is an increase of the PAR portion of solar radiationfrom 47 % to 63 %. The conversion between solar radiation and PAR is notdirect nor constant. Water vapour, clouds, and the annual solar cycle cancause an increase in the PAR portion of solar radiation. Past observationsin College Station, Texas (30.583°’N 96.350°W) resulted in a PAR fractionranging between 47 % and 58 % of the solar radiation (Britton and Dodd1976). It is therefore not unreasonable to see a slight increase at higherlatitudes where the photoperiod is shorter. Further, the calibrated PARfraction value of 63 % is within the ±5 % accuracy of the LI-COR PARsensor. Other meteorological factors did not require adjustment.38Table 3.1: Calibration parameters for GLM setup on CultusLake.ParameterDefault (Hipsey, Bruce,et al. 2014)Cultus Lake Value(Comments)Maximum number of layers N/A 500Minimum layer volume [m3] 0.025 0.025Minimum layer thickness [m] 0.5 0.3Maximum layer thickness [m] 1.00.9 (At least twice theminimum layerthickness)Extinction coefficient [m-1] 0.20.22 (From averageSecchi depth)Bulk aerodynamic coefficientfor sensible heat transfer0.0013 0.0013Bulk aerodynamic coefficientfor latent heat transfer0.0013 0.0013Bulk aerodynamic coefficientfor transfer of momentum0.0013 0.0013Mixing efficiency - convectiveoverturn0.2 0.2Mixing efficiency - wind stir-ring0.23 0.23Mixing efficiency - shear pro-duction0.3 0.3Mixing efficiency - unsteadyturbulence (acceleration)0.51 0.51Mixing efficiency - Kelvin-Hemlholtz turbulent billows0.3 0.3Mixing efficiency of hypolim-netic turbulence0.5 0.5Albedo mode N/A 3Cloud mode N/A 4Wind factor 1.00.95 (Correction forvalues from AbbotsfordAirport)Shortwave factor 1.01.3 (Correction forshortwave radiationcalculated from PAR)39Table 3.1 – continued from previous pageParameter DefaultCultus Lake Value(Comments)Longwave factor 1.0 1.0Air temperature factor 1.0 1.0Relative humidity factor 1.0 1.0Rain factor 1.0 1.0The inflow parameters used for GLM are shown in Table 3.2. The streamhalf-angle set to 60° for all streams. The slope for Frosst Creek is 4 %, or2.29° (Zubel 2000) and the same slope was assumed for Watt Creek. Theslope for Smith Falls Creek was estimated to be higher (4.5 %, 2.58°) becausethe creek discharges through a culvert beneath the Columbia Valley highwayand onto the exposed banks of the lake. The streambed drag coefficient wasset to 0.016 (Fisher et al. 1979) for all three streams. Outflow data were notincorporated into the model because the Cultus Lake Park Board attemptsto maintain a stable water level throughout the summer by means of stoplogs. Therefore, overflow discharges from GLM were used as surrogates forthe outflow.Table 3.2: Inflow parameters for GLMInflowStream HalfAngle (degrees)Streambed Slope(degrees)Streambed DragCoefficientFrosst Creek 60 1.54 0.016Watt Creek 60 1.54 0.016Smith Falls Creek 60 2.58 0.0163.5.2 Model EvaluationObserved average lake temperatures and simulated result profiles are shownin Figure 3.6 for each field day. The model replicates the hypolimnetic tem-peratures very well; however, the model struggles in the spring to replicatethe epilimnion. On 20 May, the simulated epilimnion is deeper and thethermocline sharper than observations. Interestingly, the secondary ther-mocline observed in June was replicated in the model. There is some delay40in the warming of the epilimnion throughout July and August; the observedepilimnetic temperature is up to 1.6 °C warmer than the model on July 25.However, the model performs well in the late summer as the epilimnion be-gins to cool. The bottom of the metalimnion in the model is not as sharpas what was observed in October and November, this is likely due to anincrease in inflow discharge in the fall which is eroding the simulated ther-mocline. Finally, on December 8, the observed and simulated results arevery similar as the lake begins to completely mix.Temperature [°C]Depth [m]0 10 2002040Apr 280 10 2002040May 040 10 2002040May 200 10 2002040Jun 020 10 2002040Jun 170 10 2002040Jul 060 10 2002040Jul 150 10 2002040Jul 250 10 2002040Aug 090 10 2002040Aug 240 10 2002040Sep 070 10 2002040Sep 220 10 2002040Oct 050 10 2002040Oct 190 10 2002040Nov 020 10 2002040Nov 160 10 2002040Dec 08Figure 3.6: Model evaluation for the 2016 calibration period. Solid linesindicated the spatially average temperature profile. Dashed lines are thesimulated results.41Root-mean-square error (RMSE) between the average observed and sim-ulated temperatures was calculated to provide a single number for the errorin the profile for each field day. RMSE is given by:RMSE =√∑ni=1(xi − xi)2n(3.2)where xi and xi are the simulated and observed values at each point i,respectively. This equation can also be normalized to give the normalizedroot mean square error (NRMSE):NMRSE =RMSExmax − xmin (3.3)where xmax and xmin are the maximum and minimized observed valuesfor each day. A time-series of the RMSE for the 2016 calibration periodis presented in Figure 3.7. The maximum observed spatial temperaturevariation for the lake is shown alongside the RMSE. This variation wascalculated by determining the maximum difference between casts for each1 m layer and taking the maximum of these values from each field day; itrepresents a metric for the acceptable level of error for the model. The RMSEaverages 0.46 °C throughout the simulation and is lower than the maximumobserved spatial temperature variation in the lake for each profile, whichhas an average of 1.64 °C. For the final field day on 8 December 2016, theRMSE is larger than the temperature variation. Only a single profile wascompleted to the bottom of the lake due to difficult weather conditions,therefore a complete survey of the lake was unattainable.42Apr May Jun Jul Aug Sep Oct Nov Dec Jan00.511.522.53 Temperature [°C]  RMSEMaximum observed temperature variationAverage RMSEAverage observed temperature variationFigure 3.7: RMSE for each 2016 temperature profile compared to themaximum observed temperature variation for each field day.Further model evaluation was done by comparing the observed and sim-ulated heat content. Total lake heat content was determined by summingthe heat content of the 1 m layers:H =1A0zm∑z0cp(T − TTMD)ρA∆z (3.4)where A0 is the lake surface area, zm is the maximum depth, cp is theheat capacity of water, T is the temperature, TTMD is the temperature ofmaximum density, ρ is the density, and A is the area of each layer. TheMATLAB TEOS-10 toolbox was used to calculate the values for cp, TTMD,and ρ assuming the chemical composition of seawater. The heat contentindicates the amount of heat released upon cooling to the temperature ofmaximum density (Wetzel 2001) (roughly 4 °C, depending on salinity anddepth). A negative value of H indicates the amount of heat required to raisethe temperature of the lake to the temperature of maximum density (Wetzel432001). The modelled heat content roughly follows the same increasing trendas the observed data throughout the spring (Figure 3.8). A minor increasein heat content was simulated in early June that was not captured in theCTD casts, but is seen in the mooring data. The observed peak heat contentin late August was not simulated in the model. The results correlate wellthroughout September, but the simulated heat content is lower throughoutthe fall but does not reach the minimum observed on 8 December.Heat content was also calculated assuming pure water (i.e. no salinity)to determine the effects of salinity and potential error. The heat contentRMSE over the calibration period was 2.4 MJ m-2, or a NRMSE of 0.4 %.The effects of salinity are therefore negligible and are ignored in the modelhindcasts (Sections 3.5.4 and 3.5.5) and forecasts (Section 3.6); however, fu-ture studies incorporating nutrient loading scenarios should include salinityin their calculations.Apr May Jun Jul Aug Sep Oct Nov Dec Jan456789101112x 108Heat Content [J m−2 ]  ObservedSimulatedFigure 3.8: Observed and simulated total heat content of Cultus Lake for2016.443.5.3 Sensitivity AnalysisThe finalized model was subject to an elementary effects sensitivity analysison the most uncertain parameters and those parameters that would likely beaffected by climate change: extinction coefficient, meteorological multiplica-tion factors, inflow half-angle, and inflow slope. A simple Morris screeningmethod was used to evaluate the effects of the inputs on the model. Themodel inputs were evaluated between ± 50 % of their calibrated 2016 values(Tables 3.1 and 3.2) and compared to their effects on the total NRMSE forwater temperature. The Morris method is described in detail in Campolongoet al. (2007) but gives the elementary effects as:EEi(X) =Y (X1, . . . , Xi−1, Xi + ∆, Xi+1, . . . , Xk)− Y (X)∆(3.5)where X = (X1, X2, . . . , Xk) is any value within± 50 %, and ∆ = p/[2(p−1)]with p equal to the number of repetitions computed per elementary factor,here p = 20 was chosen to ensure enough calculations were performed. Acriterion to rank the effects of the inputs is defined as the mean of the ab-solute values of the elementary effects, µ∗i . The values for µ∗i are shown inTable 3.3. The model is highly sensitive to changes in the longwave, short-wave, and air temperature factors. These parameters account for the bulkof the heat flux in the model, so it is expected that they would contributeto model sensitivity. The model is moderately sensitive to wind speeds,and has little sensitivity to the extinction coefficient and relative humidity.There is very little sensitivity to the rain factor and inflow slopes and halfangles.45Table 3.3: Sensitivity analysis results, and normalized root-mean-squareerror of the observed and CanRCM4 historical data for 2001–2003. GLMinput parameters are ranked by the mean of their elemental effects (µ∗i ).Parameter, i µ∗i NRMSE of Input DataLongwave factor 0.703 0.1791Shortwave factor 0.430 0.271Air temperature factor 0.404 0.121Wind factor 0.122 0.1932Extinction coefficient 0.085 N/ARelative humidity factor 0.063 0.2692Frosst Creek half angle 0.013 N/AFrosst Creek slope 0.006 N/ARain factor 0.004 0.1602Smith Falls Creek slope 0.004 N/AWatt Creek half angle 0.004 N/ASmith Falls Creek half angle 0.004 N/AWatt Creek slope 0.003 N/A1 Calculated longwave radiation from observed cloud cover data at Ab-botsford Airport following Yajima and Yamamoto (2015).2 Calculated after bias correction for CanRCM4 data.3.5.4 Hindcast (2009–2016)The calibrated model was hindcast to validate the model using monthly DFOdata obtained near station CL2 at monthly intervals beginning 17 February2009. Hourly air temperature and relative humidity from Abbotsford Air-port were used for the period between 17 February 2009 until 5 June 2011when the Cultus Lake weather station was installed. Daily precipitation wasobtained from the Chilliwack weather station until it was decommissionedon 31 August 2012, subsequently the Abbotsford precipitation data wereused. Inflow data for 2011–2013 were obtained from Putt (2014). Becausethe inflow data are not consistent, the water level drops between 2009 and2011 until the data from Putt (2014) are used, and again between 2013 and2016. However, the water level drop is minimal and less than 2 m. Valida-tion techniques were the same as those for the 2016 calibration. The averagetemperature RMSE over this period is 1.10 °C (Figure 3.9). A yearly cycleis seen in the RMSE, with peaks in the summer and minima in the win-ter. The RMSE shows large spikes, near or above 2 °C in the summers of462009–2011 and 2014. Overall, the average RMSE is less than the observedtemperature variation limit for 2016. The total lake heat content was alsoused to evaluate the model (Figure 3.10). The trends roughly follow thatof the observed heat content, but some peaks do occur in the summers of2009, 2010, 2011, and 2014. These are the same years where the RMSE wasalso at a maximum.There are clearly some heating biases in the model. Possible errors maybe from the air temperature data being obtained from Abbotsford Airportprior to the commissioning of the Cultus Lake weather station in the summerof 2012. Additionally, observed temperature profiles were only obtainedat one location prior to 2016 near the upstream end of the lake, so therewould be some error expected from internal seiching, as determined fromthe spatial variability in 2016. Furthermore, inflow data were provided foronly 2011–2013 and again in 2016 which may account for some error. Whilethe model is not able to perfectly resolve the temperature in the lake—partially due to disparities in meteorological measurements—it does a goodjob at reproducing seasonal trends. This reinforces the statement in Hipsey,Bruce, et al. (2014) that the model is best intended for use in long-terminvestigations.472009 2010 2011 2012 2013 2014 2015 2016 201700.511.522.53RMSE [°C]  RMSEAverage RMSEFigure 3.9: RMSE for each field day of the hindcast model (2009–2016).2009 2010 2011 2012 2013 2014 2015 2016 2017−4−202468101214x 108Heat Content [J m−2 ]  ObservedSimulatedFigure 3.10: Daily observed and simulated heat content (2009–2016).483.5.5 Forecast Data Evaluation (2001–2003)Corrected CanRCM4 meteorological data from the historical run were com-pared to observations taken at Abbotsford Airport in 2001–2003. TheNRMSE was calculated between the observed and synthetic data and com-pared to the sensitivity analysis results (Table 3.3). While the NRMSE ofall parameters was within 30 %, it was found that the seasonality of thesynthetic relative humidity did not match that of the observed data: theobserved relative humidity was much lower in the summer than that of thesynthetic data. A bias correction was performed to adjust the seasonal-ity differences following the quantile mapping (Panofsky and Brier 1958) ofthe cumulative frequency distributions for the parameters. For any climatevariable, this can be written as:xm-corr(t) = F−1o-ref (Fm-ref(xm-raw(t)) (3.6)where xm-corr is the corrected modelled value at each timestep, xm-raw is theraw modelled value at each timestep, and F is the distribution function forthe observed (o) and modelled (m) data over the reference period (a Weibulldistribution was used in this case). A time period of 1953–2005 was usedfor the reference data. The bias correction was calculated for each monthto ensure the seasonal cycle was properly corrected. Precipitation and windspeeds were also corrected in this manner. The correction of precipitationeliminated the so-called drizzle effect (Gutowski Jr et al. 2003) found inclimate models where the models produce too many low intensity rain eventscompared to observations. Other meteorological parameters were not biascorrected due to their low NRMSE and relatively similar seasonality anddistribution curves to the observed data. This seasonal similarity can beattributed to the small grid spacing of the CanRCM4 model that is ableto refine micro-climates better than global models. The model neverthelessremains highly sensitive to longwave radiation. In this study, observed cloudfraction was used to calculate longwave radiation, while incoming longwaveradiation was obtained for the CanRCM4 model runs. Cloud feedbacks arethe largest source of uncertainty in climate models and remain difficult toquantify (von Salzen et al. 2013). The model is also sensitive to shortwaveradiation which also has a high NRMSE. However, this can be attributedto the cloud fraction uncertainty. Air temperature, for which the model hashigh sensitivity, had the lowest NRMSE for the synthetic data. Wind speedshad a high NRMSE, but the model is less sensitive to this parameter. Forthe bias corrected relative humidity, the NRMSE is still high but the modelhas low sensitivity. The model is largely insensitive to rain which also has a49low NRMSE.Highly porous soil in the vicinity of Cultus Lake results in much of theprecipitation (roughly 63 %) draining into groundwater (Zubel 2000), whileapproximately a third of the rainfall becomes surface runoff (Scibek andAllen 2006). Total runoff data from CanRCM4, which includes drainageto groundwater, were therefore halved to determine surface runoff, thensubsequently multiplied by each stream’s drainage area to determine dailyinflows. A baseflow of 0.2 m3 s-1 was added to the inflow for Frosst Creek,while a baseflow of 0.03 m3 s-1 was added to the inflow for Smith Falls Creek,as determined from 2016 flow data, to account for groundwater runninginto each creek. No baseflow was added to the intermittent Watt Creek.Forecast stream temperatures were determined using the correlation betweenair temperatures and stream temperatures for the 2016 season.A verification of the CanRCM4 output adjustments was done by runningtwo simulations for 2001–2003: one using the CanRCM4 historical simula-tion and the other using observed data at Abbotsford Airport. Observed andbias corrected synthetic wind speeds at Abbotsford Airport were again re-duced by 5 %, but a shortwave radiation correction was not required. Again,the water temperature RMSE and heat content were used to evaluate themodel. The RMSE follows a similar trend to the 2009–2016 hindcast (Fig-ure 3.11). The RMSE increases in the summer and returns to a minimum inthe winter. Unfortunately, observed inflow data were not available for thisperiod, but from Table 3.3 the model has a low sensitivity to inflows. Theaverage RMSE over the simulation period was similar for both scenarios:0.90 °C and 0.81 °C for the observed and synthetic scenarios, respectively.These results are comparable to those of the 2009–2016 hindcast and lessthan the observed maximum temperature variation in the lake for 2016.Heat content was also evaluated for both scenarios and compared to the ob-served data (Figure 3.12). The scenario using observed data at AbbotsfordAirport typically has a higher heat content throughout the simulation pe-riod than the observed data. This result is similar to the 2009–2016 hindcastyears where temperature data were obtained from Abbotsford, reinforcingthat the model is sensitive to air temperature. The simulation using syn-thetic CanRCM4 historical data has a heat content that roughly followswhat was observed. However, the heat content increases at a reduced ratein the spring, but follows a similar decline in the fall to what was observed.The heat content also reaches a minimum earlier in the year than what wasmeasured.Overall, the bias corrected CanRCM4 data provide similar simulationresults to those obtained using the observed meteorological data. The high-50est errors are observed in the summer; however, they are generally less thanthe spatial variation in the lake as determined from the 2016 data. The biascorrection technique described here can therefore be applied to the forecastCanRCM4 meteorological data to be confidently used to forecast future lakeconditions.51Apr 2001 Jul 2001 Oct 2001Jan 2002Apr 2002 Jul 2002 Oct 2002Jan 2003Apr 200300.511.52RMSE [°C]   Abbotsford (Observed)CanRCM4 Historical (Synthetic)Abbotsford Average (Observed)CanRCM4 Historical Average (Synthetic)Figure 3.11: RMSE for each field day of the hindcast model using observedand synthetic meteorological data (2001–2003).Apr 2001 Jul 2001 Oct 2001Jan 2002Apr 2002 Jul 2002 Oct 2002Jan 2003Apr 2003024681012x 108Heat Content [J m−2 ]   ObservedSimulated from AbbotsfordSimulated from CanRCM4 HistoricalFigure 3.12: Daily observed and simulated heat content using observed andsynthetic meteorological data (2001–2003).523.6 Historical and Forecast Results (1923–2100)Two forecast scenarios were run in GLM for 2009–2100. The eight yearoverlap between the forecast scenarios and the observed data provides agood evaluation for the early years of the CanRCM4 forecast scenarios. Theresults from the forecast years were compared to historical data dating tothe 1920s.3.6.1 Heat ContentDaily heat content was calculated for the historical data. To calculate theheat content, the historical data obtained from Foerster (1925) and Ricker(1937) for the periods of 1923, 1927–1929, and 1932–1936; and from DFO for2001–2003 and 2009–2016 were linearly interpolated to obtain daily values.This was combined with the semimonthly data obtained in 2016. A lineartrend was fit to the data which indicates the yearly average heat content(Figure 3.13. This trend increases at a rate of + 0.80 MJ m-2 per year overthe observed period. The trend may not in reality be linear; however, itshould be noted that increased heating compared to the 1930s was observedin the 1960s–1970s (Goodlad et al. 1974). Additionally, measurements from1923 results in an anomalously high heat content, peaking at 1.47 GJ m-2.Hypolimnetic temperatures were recorded at or above 9 °C in 1923, which issignificantly warmer than those observed in 2016. Summer air temperatureswere not unusually warm for this year (Foerster 1925) that would causeelevated hypolimnetic temperatures, so it is possible that the temperatureof the water volume in the modified deep-sea thermometer warmed whenthe thermometer was hauled to the surface. Winter heat content minimaare historically continually approaching zero. This shows that the lake wasbecoming isothermal and approaching the temperature of maximum den-sity. For years where the heat content was negative, the lake temperatureshad cooled below the temperature of maximum density and the lake waspotentially inversely stratified.Future climate scenarios indicate a continued increase in total lakeheat content (Figure 3.13). The RCP4.5 scenario indicates a heating ratemore than double the historical trend (+ 2.0 MJ m-2 per year compared to+ 0.80 MJ m-2, respectively). The RCP8.5 scenario shows a sharp increasein heating rate to + 4.2 MJ m-2 per year, over twice that of the RCP4.5scenario. In both scenarios, the winter heat content minima are predictedto continually increase above zero, with some anomalous years at or belowzero. For the RCP8.5 scenario, the years where the heat content is below53zero occur near the beginning of the simulation, whereas RCP4.5 has twoyears (2042–2043, 2044–2045) where the heat content minimum is belowzero. This trend means that lake temperatures are on average remainingabove the temperature of maximum density throughout the year. There-fore, there would be no possibility of inverse stratification.1950 2000 2050 2100−5051015x 108Heat Content [J m−2 ](a)1950 2000 2050 2100−5051015x 108Heat Content [J m−2 ](b)  ObservedSimulatedFigure 3.13: Historical observed heat content and projected heat content ofCultus Lake for RCP4.5 (a) and RCP8.5 (b). Linear trend lines are shownin bold.543.6.2 Stability and StratificationLake stability is the amount of work required to completely mix the lake toa uniform temperature (Wetzel 2001). It is therefore a good indicator of thestrength of stratification. The stability increases throughout summer as theepilimnion warms and stratification intensifies; when the lake is completelymixed, the stability returns to zero. Unlike heat content, stability is not ref-erenced to a specific temperature, therefore a stability of zero could indicatemixing above or below the temperature of maximum density. The lake sta-bility was calculated using the Schmidt Stability Index (SSI) as formulatedby Idso (1973):SSI =gA0zm∑z0(z − zρ)(ρ− ρ)A∆z (3.7)where g is the gravitational constant, ρ is the average density, and zρ is thedepth of average density calculated from:ρ =zm∑z0ρzA∆zzm∑z0ρA∆z(3.8)zρ =zm∑z0ρA∆zzm∑z0A∆z(3.9)The historical SSI was calculated for all observed historical data and inter-polated to obtain daily values. A linear trend was fit to the historical andprojected data which indicates the yearly average SSI (Figure 3.14). Histori-cally, the SSI increased at a rate of + 2.19 J m-2 per year with summer peaksbetween 2000 J m-2 and 3000 J m-2. For the climate change scenarios, theSSI is expected to increase at a rate of + 2.6 J m-2 per year and + 6.5 J m-2per year for climate change scenarios RCP4.5 and RCP8.5, respectively. ForRCP4.5, the SSI begins to approaches but does not consistently seasonallypeak above 3000 J m-2 by 2100; whereas the seasonal peaks shift to above3000 J m-2 by 2100 for RCP8.5 and begin to approach 4000 J m-2. This isa significant increase to what was observed historically and further isolatesthe hypolimnion from the epilimnion due to an increase in stratificationstrength.Stratification season was determined using three methods: (i) periodswhen the difference between the surface and bottom temperatures was greater551950 2000 2050 2100010002000300040005000SSI [J m−2 ](a)1950 2000 2050 2100010002000300040005000SSI [J m−2 ](b)  ObservedSimulatedFigure 3.14: Historical observed Schmidt Stability Index and Schmidt Sta-bility Index of Cultus Lake for RCP4.5 (a) and RCP8.5 (b). Linear trendlines are shown in bold.than 1 °C, (ii) the SSI was greater than 30 J m-2 (Engelhardt and Kirillin2014), and (iii) a thermocline was established based on the traditional defi-nition of a vertical temperature change greater than 1 °C m-1 (Ricker 1937;Wetzel 2001). The surface temperature was also required to be at or greater56than 4 °C to remove periods where the lake may be inversely stratified. Afterpreliminary calculations, it was noticed that the lake was remaining strati-fied beyond 31 December after some years of warming had elapsed. A lakenew year was therefore established as February 20 and defined as the datewhere Cultus Lake was typically completely unstratified following the abovedefinitions of stratification. By shifting the new year, calculations could thenbe done by finding the first and last dates of the year where the thresholdvalues were crossed. All three methods saw an increase in duration of strat-ification for both warming scenarios (Figure 3.15). A linear trend was fit tothe data and is summarized in Table 3.4. Overall, the thermocline methodgives a more conservative result but requires a vertical grid spacing of 1 m,which is not available in the historical data. The SSI threshold method washenceforth chosen as the representative method to calculate historical andfuture length of stratification season, but the top to bottom temperaturedifference method could equally have been chosen because the two methodsproduce similar results.2000 2050 2100180200220240260280300320340360Duration of stratification [days] (a)2000 2050 2100180200220240260280300320340360 (b)  Top to Bottom Threshold = 1°CSSI Threshold = 30 J/m2Thermocline Threshold = 1°C/mFigure 3.15: Comparison of the three methods used to evaluate thermalstratification season for future climate scenarios RCP4.5 (a) and RCP8.5(b). Linear trend lines are shown in bold.57Table 3.4: Comparison of the duration of stratification for the three meth-ods. Days shown are for the linear trend lines from Figure 3.15.(i) (ii) (iii)Temperaturedifferencesurface–bottomSSI ThermoclineRCP 4.5Days stratifiedin 2009273 263 193Days stratifiedin 2100285 279 209Rate of change[days/year]0.13 0.18 0.17RCP 8.5Days stratifiedin 2009269 260 194Days stratifiedin 2100305 305 214Rate of change[days/year]0.39 0.50 0.22A comparison of the forecast results to the historical duration of stratifi-cation was done to evaluate the effects of climate change. Historical temper-ature data were linearly interpolated between cast dates to obtain a dailySSI in order to calculate when stratification occurs and breaks down. A lin-ear trend was calculated for the total duration of stratification. This trendexcluded the total duration calculated from 1923, 1924, and 1927–1929 be-cause complete monthly data sets are not available which resulted in largeinterpolation steps. The linear trend increases from 249 days to 266 days, ora rate of 0.18 days per year (Figure 3.16). With the SSI threshold method,The rate of increase in duration of stratification is expected to stay the samewith climate change at 0.18 days per year for RCP4.5 and but expected tomore than double to 0.50 days per year for RCP8.5.Further investigation was done determine the onset and breakup of ther-mal stratification (Figure 3.17). Historically, the onset of stratification oc-curred around 10 April in the 1920s–1930s, and has shifted 18 days earlierto 23 March in the 2010s. For RCP4.5, the onset of stratification remainsrelatively steady and occurs around 5 April, while for RCP8.5, the onset of58stratification increases slightly from 4 April to 27 March. There are there-fore no substantial changes to the onset of stratification predicted for futurewarming scenarios out to 2100 AD. Stratification breakup has historicallybeen relatively steady and occurred around 15 December. However, majorchanges in stratification breakup are seen in the climate forecast: a shift to12 January is predicted for RCP4.5 and a shift to 25 January for RCP8.5.Furthermore, breakup of stratification did not occur in some forecast years.These years were the winters of 2045–2046 and 2050–2051 for RCP4.5, and2063–2064, 2077–2078, 2090–2091, and 2092–2093 for RCP8.5. For thesewinters, complete lake mixing did not occur but this does not necessarilymean the lake was stratified for 365 days as in Figure 3.15. If the onsetof stratification of the preceding year occurred after 20 February (lake newyear), the duration of stratification would therefore be the difference be-tween the date of onset of stratification and 20 February of the followingyear. The total duration of stratification would then be the sum of the du-ration of stratification of the two years affected (e.g. 2045 and 2046), whichcould be, but is not necessarily, greater than 365 days.591920 1940 1960 1980 2000 2020 2040 2060 2080 2100180200220240260280300320340360Duration of stratification [days]   ObservedSimulated (RCP4.5)Lake did not destratify (RCP4.5)Simulated (RCP8.5)Lake did not destratify (RCP8.5)Figure 3.16: Historical and predicted duration of thermal stratification forRCP4.5 and RCP8.5. Linear trend lines are shown in bold.1920 1940 1960 1980 2000 2020 2040 2060 2080 2100FebMarAprMayJunOnset of stratification(a)1920 1940 1960 1980 2000 2020 2040 2060 2080 2100NovDecJanFebMarBreakup of stratification (b)  ObservedSimulated (RCP4.5)Lake did not destratify (RCP4.5)Simulated (RCP8.5)Lake did not destratify (RCP8.5)Figure 3.17: Historical and predicted onset (a) and breakup (b) of thermalstratification. Linear trend lines are shown in bold.60A small disparity in the onset of stratification is seen in trendlines ofthe 2009–2016 overlap for the observed and simulated data. It is importantto note that the observed data were linearly interpolated between monthlymeasurements so an error on the order of several days to weeks is expected.Interestingly, the simulated onset of stratification is similar to the date ofonset in the early 20th century. The error in the linear interpolation ofthe measurements by Ricker (1937) is expected to be on the order of daysbecause of the semi-monthly observation interval. There is very little changein the projected onset of stratification in both climate scenarios, indicatingthat the onset of stratification is not expected to significantly shift withclimate change. The bulk of the increase in the duration of stratificationtherefore comes from a delayed breakup of stratification. Stormy eventswith high winds are frequent in the late fall and early winter on the westcoast. The lake is remaining stable for a longer period from the increase andretention of heat and preventing wind mixing rather than becoming stableat an earlier period. Winds in the late winter are keeping the lake well mixedand preventing the lake from becoming thermally stratified.Inverse stratification was also investigated using SSI threshold method.This was constrained by requiring the surface temperature to be below 4 °C.Using this method, inverse stratification was neither detected for the histor-ical data nor the forecast results. Therefore, the potential effects of inversestratification are not a concern and Cultus Lake can be classified as warmmonomictic.3.6.3 Surface and Outflow TemperaturesRiver temperatures in Sweltzer River were analyzed for potential fish mor-tality impacts on Sockeye Salmon. Surface temperatures from Cultus Lakewere used as surrogates for river temperature because they have been previ-ously shown to have similar values (Shortreed 2007). In recent years, SockeyeSalmon mortality has increased due to an earlier migration time into watercourses with warm temperatures (Cooke et al. 2004). Sockeye Salmon havebeen shown to have increased mortality in migration temperatures greaterthan 18 °C (Brett 1971; Crossin et al. 2008). The Cultus Lake populationtypically runs between early August and early December (Shortreed 2007),so mean monthly river temperatures have been calculated for the monthsof August–November (Figure 3.18). Historically, the Sweltzer River tem-perature has been above the 18 °C threshold in August and September andhas been increasing at a rate of 0.016 °C and 0.013 °C per year for Augustand September, respectively since 1923. October river temperatures have61been increasing at a rate of 0.010 °C per year, while river temperatures inNovember have not seen a significant historical increase. River temperaturesare projected to increase at a much higher rate in both warming scenarios.For RCP4.5, the river temperatures are projected to increase at a rate of0.031 °C per year, 0.025 °C per year, 0.018 °C per year, and 0.016 °C peryear for August, September, October, and November, respectively. Heatingrates for RCP8.5 are 0.069 °C per year, 0.069 °C per year, 0.059 °C per year,and 0.046 °C per year for August, September, October, and November, re-spectively. The October river temperatures in RCP4.5 approach and areoccasionally greater than 18 °C by 2100, whereas the RCP8.5 October rivertemperatures cross the 18 °C threshold in 2072 and reach 20 °C by 2100.For both warming scenarios, the November river temperatures remain be-low the mortality threshold. The predominant salmon migration into CultusLake typically occurs in late September and early October (Cultus SockeyeRecovery Team 2005), so temperature increases beyond the 18 °C thresholdduring this period, such as what is projected for RCP8.5, would have direimpacts on the Cultus Lake Sockeye Salmon population.1900 1950 2000 2050 210051015202530Temperature [°C](a)  AugustSeptemberOctoberNovember1900 1950 2000 2050 210051015202530Temperature [°C] (b)Figure 3.18: Historical and predicted surface and outflow temperatures forRCP4.5 (a) and RCP8.5 (b). Linear trend lines are shown in bold. Dashedline at 18 °C indicates temperature threshold for Sockeye Salmon.62Chapter 4DiscussionClimate change has the capacity to significantly affect the physical processeswithin lakes by modifying heat exchanges between the atmosphere and thewater surface. Lakes act as sentinels of climate change because they elicitstrong responses in their thermal and ecosystem structures (Adrian et al.2009). The analysis of the thermal characteristics of Cultus Lake presentedhere provides a good understanding of the historical response of Cultus Laketo climate change since the 1920s. Two potential climate change scenarios(RCP4.5 and RCP8.5) indicate that the thermal characteristics of the lakeare projected to change significantly through 2100 AD. The lake ecosystemis expected to respond to these changes, likely further exacerbating threatsto extinction for Sockeye Salmon and Cultus Lake Pygmy Sculpin.4.1 Potential Changes to the Lake MixingRegimeEpilimnetic and surface water temperatures often exhibit a strong responseto warming air temperatures (Adrian et al. 2009). Air temperatures in theFraser Valley are projected to increase by + 2.8 °C through the 21st cen-tury (Pacific Climate Impacts Consortium 2012). Resultant heating of thewater column from climate change can increase the total lake heat contentand lake stability thereby increasing the duration of thermal stratification(Weinberger and Vetter 2014; Sahoo et al. 2015).The current mixing period of Cultus Lake typically begins in Decemberwhen the lake becomes isothermal at temperatures around 7 °C. Duringcirculation, the lake cools to the temperature of maximum density (roughly4 °C). In some years the lake has been observed by Ricker (1937), DFO,and in the winter of 2016–2017 to cool, isothermally, below 4 °C withoutinversely stratifying. Under current conditions, wind continuously mixesthe lake throughout the winter below the temperature of maximum density.Winter mixing typically lasts until late March to early April at which pointthe lake becomes thermally stratified. The heat content projections for63the lake indicate that complete mixing will occur at temperatures at orabove 4 °C with no possibility of inverse stratification. So, at the onsetof stratification, hypolimnetic temperatures would already be above 4 °Cleading to an oligomictic state, described as having rare turnover at irregularinterannual intervals and temperatures remaining above 4 °C (Wetzel 2001).Warm monomictic lakes are particularly sensitive to climate changes be-cause their single mixing period in the winter can be shortened with increas-ing water temperatures in a shift towards an oligomictic state (Livingstone2008; Adrian et al. 2009). A shift in the mixing regime of Cultus Lake frommonomictic to oligomictic is possible in the future under more extreme cli-mate change scenarios (e.g. RCP8.5) as increased heating is expected tocontinue affecting the timing and duration of thermal stratification. Theonset of thermal stratification currently occurs 18 days earlier than histori-cal observations in the 1920s and 1930s, while the breakup of stratificationhas not seen a significant historical shift, based on a linear interpolation ofsemi-monthly to monthly thermal profiles. These trends are unique com-pared to larger lakes in North America, namely Lake Tahoe (Sahoo et al.2015) and Lake Simcoe (Stainsby et al. 2011) in which the onset of strat-ification is occurring earlier and the breakup of stratification is occurringlater in the year compared to historical values. Nevertheless, the durationof summer stratification in all cases has historically been increasing. Pro-jections for Cultus Lake show that the onset of thermal stratification isexpected to remain steady, whereas the breakup of thermal stratification isprojected to occur a month later by 2100. Comparatively, a climate changestudy on Lake Tahoe resulted in an onset of stratification 16 days earlierand a breakup of stratification 22 days later compared to present valuesby 2100 (Sahoo et al. 2015). Thus, regionality and geomorphology are ex-pected to influence the timing of onset and breakup of thermal stratificationwith increased heating from climate change. Model projections show CultusLake is shifting towards an oligomictic mixing regime, but not yet becomingoligomictic by 2100. Only two years were identified as having no mixing inthe RCP4.5 warming scenario, which occurred at seemingly random times(2045–2046 and 2050–2051). By the end of the study period, the length ofstratification season was 279 days; a mixing regime shift would thereforenot be expected to fully occur for moderate warming scenarios. However, inthe RCP8.5 scenario, turnover did not occur in four years towards the endof the simulation (2063–2064, 2077–2078, 2090–2091, and 2092–2093) withthe duration of stratification increasing to 305 days by the end of the studyperiod. This suggests that with increased warming beyond 2100, the lakewill become oligomictic and possibly shift to meromixis (incomplete water64column mixing) under an extreme climate change scenario.4.2 Implications of Increased Climate Warmingon Ecosystem Structure and FunctioningFreshwater fish are poikilotherms that cannot regulate body temperature,and thus are dependent upon and sensitive to the temperature of their sur-roundings (Ficke et al. 2007). Cold-water fish species have evolved to sur-vive within specific thermal tolerances, and their physiological processes(e.g. growth, metabolism, and reproduction) and behaviours (e.g. diur-nal vertical migration) are directly influenced by changes in surroundingtemperature (Po¨rtner and Farrell 2008; Healey 2011). Cultus Lake hoststwo fish species at-risk: the Cultus Lake population of Sockeye Salmon,and the Cultus Pygmy Sculpin (SARA 2002; COSEWIC 2003, 2010); anychanges in their environment are predicted to have drastic consequences(SARA 2002; COSEWIC 2003, 2010). The modelling efforts presentedhere indicate that the total lake heat content has increased since the early1920s by + 0.80 MJ m-2 per year and is projected to markedly increase by+ 2.0 MJ m-2 per year and + 4.2 MJ m-2 per year for RCP4.5 and RCP8.5,respectively, with isothermal temperatures remaining above 4 °C, and po-tentially deleterious lake ecosystem alterations for species-at-risk.Increased lake heating can have significant effects on Sockeye Salmonthroughout their life cycle (Crozier et al. 2008; Po¨rtner and Farrell 2008;Healey 2011). In the egg and alevin stage of the Sockeye Salmon life cycle,warmer water temperatures increase metabolic rates, leading to the devel-opment of smaller fry, as well as a reduction in disease resistance (Healey2011). Embryos may also develop faster leading to earlier hatching andemergence, which can present a phenological shift relative to zooplanktonprey availability (D. E. Schindler et al. 2005; Healey 2011). Juvenile SockeyeSalmon exhibit a diel vertical migration within the water column, feedingon zooplankton in the epilimnion and metalimnion at night and returningto the bottom of the lake during the day, presumably to evade predators(Clark and Levy 1988; Levy 1990; Scheuerell and D. E. Schindler 2003;Cultus Sockeye Recovery Team 2005). Warmer water temperatures and anincreased metabolic demand will require Sockeye Salmon to feed more fre-quently and may increase their risk of predation (Healey 2011), otherwisean increased metabolic rate would lead to smaller fry sizes and potentialreductions in winter survival when food is less abundant (Healey 2011). Areduction in fry size can lead to either smaller smolts which have reduced65survival rates in a marine environment, or fry may delay smolting by a yearresulting in fewer smolts migrating to the ocean (Healey 2011).Recent decades have been marked by a decrease in the number of SockeyeSalmon returning to Cultus Lake (Shortreed 2007). The salmon that are re-turning are making their run earlier in the season which has had great effectson pre-spawn mortality (Cultus Sockeye Recovery Team 2005; Shortreed2007). Sockeye Salmon have been shown to have an increased mortality ator above temperatures of 18 °C (Brett 1971; Crossin et al. 2008). Increasedoutflow temperatures from Cultus Lake, coupled with migration earlier inthe season could have detrimental impacts to the returning salmon, as highstream temperatures lead to an increase in bacteria and fungal growth as wellas reductions in dissolved oxygen concentrations, impeding migration andpotentially resulting in enhanced pre-spawning mortality (COSEWIC 2003).Warmer outflow temperatures could also lead to increased recreational useof the lake outlet, which may interrupt the migration of Sockeye Salmonand increase stress levels and energy use influencing subsequent mortality(Cultus Sockeye Recovery Team 2005).Dissolved oxygen is essential for the metabolism of all aerobic aquaticspecies (Wetzel 2001). Dissolved oxygen concentrations decrease non-linearlywith increasing water temperature (Wetzel 2001). However, the biologicaloxygen demand of poikilothermic species increases with increasing watertemperature as their metabolic rates increase (Ficke et al. 2007). The pri-mary sources of dissolved oxygen in a lake are through atmospheric diffu-sion and from photosynthesis (Kalff 2002). Oxygen is mixed throughout thelake by wind during mixis. During stratification, however, oxygen is unableto mix significantly into the hypolimnion. Increasing hypolimnetic watertemperatures would therefore have a decrease in oxygen solubility and anincrease in oxygen demand by fish. An oxygen squeeze can occur when theoxygen availability is unable to keep up with the oxygen demand of aquaticorganisms (Ficke et al. 2007). It is possible that increased water tempera-tures from climate change would decrease the amount of oxygen availablein Cultus Lake. As epilimnetic temperatures increase, another phenomenoncalled a temperature-oxygen squeeze can occur when the hypolimnetic oxy-gen concentrations are insufficient to keep up with fish oxygen demands andepilimnetic temperatures are too warm for sustained cold-water fish expo-sure (Ficke et al. 2007). As Cultus Lake shifts towards an oligomictic regime,this squeeze is expected to greatly reduce the available habitat for aquaticspecies forcing them into a smaller volume which in turn increases oxygenconsumption, stress, and disease transmission, and reduces prey availability(Ficke et al. 2007).66The Cultus Lake Pygmy Sculpin is also expected to be affected by in-creased lake warming. It is presumed that the sculpin exhibits a diel verticalmigration by feeding on zooplankton in the metalimnion at night and re-turning to depths in the day to evade predators (McPhail 2007; Woodruffand E. B. Taylor 2013). Disproportionately large amounts of energy arelikely required for the Cultus Lake Pygmy Sculpin to maintain position ormove vertically within the water column because it lacks a swim bladder(Woodruff and E. B. Taylor 2013). As water temperatures increase, in-creased energy expenditure is expected, due to reduced dissolved oxygenlevels and a higher metabolic rate. A temperature-oxygen squeeze couldhave deleterious effects for the Cultus Lake Pygmy Sculpin as their profun-dal habitat becomes more hypoxic or anoxic and they are unable to persiston the lake bottom.4.3 Trophic Status of Cultus LakeCultus Lake is experiencing increased nutrient loading which is leading toeutrophication. The lake has historically been oligotrophic and is presentlyclassified as oligo-mesotrophic (Putt 2014). A shift in the algal communitycomposition and production is expected as eutrophication continues underclimate change (Vincent 2009). Cultus Lake has seen a decrease in waterquality and increased presence of macrophytes and algae in recent years(Putt 2014) and an increase in primary productivity during warmer years(Shortreed 2007). In the spring, high epilimnetic nitrogen concentrationsare quickly depleted as productivity increases with increased temperaturesand available light (Shortreed 2007). Primary production quickly shifts tothe metalimnion, owing to the high water clarity of Cultus Lake and ac-cess to hypolimnetic nutrient reserves. As there is reduced mixing in themetalimnion and upper hypolimnion, an algal community structural shiftis expected, with non-motile algae likely to sediment to the bottom wherethey are aerobically decomposed (decreasing hypolimnetic oxygen levels),and motile algae are expected to dominate productivity (Vincent 2009). Apotential shift to less nutritious algal assemblage can affect zooplankton di-versity and lead to upward cascading influences in the food web for SockeyeSalmon and Cultus Lake Pygmy Sculpin. High nitrogen levels in the plung-ing summer baseflow from Frosst Creek (Putt 2014) will deliver additionalnutrients to the metalimnion, likely sustaining continuous primary produc-tion. As the hydrological regime shifts to drier summers and wetter winters(Pacific Climate Impacts Consortium 2012), a larger portion of the summer67inflow in Frosst Creek will be expected to come from the cooler groundwater(Putt 2014), increasing metalimnetic nitrogen concentrations. Hydrologicalconcentration of nutrients within the lake would be expected, due to de-creased nutrient flushing through Sweltzer River and increased evaporationfrom warmer air temperatures in the summer. A longer growing seasonis also expected with more growing degree days as climate change contin-ues (Pacific Climate Impacts Consortium 2012; British Columbia Ministryof Environment 2016), resulting in a higher overall algal biomass withinCultus Lake, and associated deep water decomposition. This impact couldbe exacerbated by expected increases in fish planktivory, and a subsequentdecrease in zooplankton biomass, reducing grazing pressures on algae (B.Moss et al. 2011). However, although the nutrients in the lake would stag-nate throughout the summer under climate change, projections of increasedwinter precipitation could both increase annual nutrient loading, but alsoflushing. It is unclear at this time what the net primary productivity effectsof these oppositional forcings for Cultus Lake primary productivity may be,although it is likely to be a net increase in productivity.Changes in algal community structure and function arising from in-creased heating and nutrient loading may create conditions where cyanobac-teria dominate Cultus Lake algal communities for substantial periods of thegrowing season (O’Neil et al. 2012). Cyanobacteria are unicellular organisms(prokaryote) that can create large blooms rendering the water unsuitablefor recreation if substantial (Paerl, Fulton, et al. 2001). Some cyanobacteriaspecies develop hepatotoxins and neurotoxins making the water toxic foranimals and humans (O’Neil et al. 2012). Dominance of the algal commu-nity by cyanobacteria can lead to an increase in toxins in the flesh of fishrendering them unsuitable for consumption (Ficke et al. 2007). Localizedcyanobacteria blooms have occurred previously in Cultus Lake in 2012 (Putt2014). More frequent cyanobacteria blooms are expected with increasednutrient loading and climate change, as certain problematic cyanobacteriathrive during long periods of stratification and enhanced epilimnetic nitrogendepletion, coupled with warm water temperatures (Ficke et al. 2007; O’Neilet al. 2012). Warm water has a reduced viscosity which is preferential forcyanobacteria that can regulate their buoyancy over larger non-motile phy-toplankton that will sink in less viscous water (O’Neil et al. 2012). Certainspecies of cyanobacteria would therefore have a competitive advantage toaccess nutrients trapped in the hypolimnion (O’Neil et al. 2012), particu-larly in Cultus Lake, since light can penetrate deeper than the thermocline.Another selective advantage heterocystous cyanobacteria have in periods oflong stratification is their ability to survive in nitrogen-limited conditions68through fixation of atmospheric nitrogen (Paerl, Fulton, et al. 2001; Ficke etal. 2007; O’Neil et al. 2012). However, the formation of large cyanobacterialmats on the lake surface under nitrogen-limitation could create shading ef-fects further limiting deep water productivity and out-competing other algalspecies (Paerl and Huisman 2009), but generally leading to the degradationof water quality and food webs for species at risk.The propagation of invasive Eurasian watermilfoil (Myriophyllum spica-tum L., EWM) increases with eutrophication (C. S. Smith and Barko 1990).EWM was first observed in Cultus Lake in 1977 (Mossop and Bradford2004). Growth rates of EWM increase with increasing water temperatures:a maximum growth rate occurs at relatively high water temperatures (30 °C–35 °C) (C. S. Smith and Barko 1990). In high clarity lakes, such as CultusLake, EWM is able to grow at greater depths, increasing its footprint andpotentially expanding its coverage to all littoral regions of the lake. EWMgrows rapidly and tends to form canopies or mats, often extending to thewater surface. Increased growth of EWM can increase the rate of eutroph-ication because of their access to phosphorus sequestered in the sediments.Sequestering of sediment-bound phosphorus is the primary uptake pathwayfor EWM (C. S. Smith and Barko 1990; Ficke et al. 2007). Upon dying, thisphosphorus that was otherwise unavailable to other producers in the lake isreleased into the water column where it can stimulate algae growth and fur-ther growth of EWM and other macrophytes (C. S. Smith and Barko 1990;Ficke et al. 2007), in addition to the depletion of hypolimnetic dissolvedoxygen concentrations as bacteria aerobically decompose the organic matter(Kankaala et al. 2002). Additionally, salmonids are particularly affected byEWM invasions because of the growth of EWM in spawning grounds (C. S.Smith and Barko 1990; Mossop and Bradford 2004; Ficke et al. 2007). EWMcolonies may also provide habitat for juvenile Northern Pikeminnow (Pty-chocheilus oregonensis); adult Northern Pikeminnow are a natural predatorof Sockeye Salmon. It has been hypothesized that EWM provides shelter forjuvenile Northern Pikeminnow from being cannibalistic prey of adult North-ern Pikeminnow (Mossop and Bradford 2004). So, an increase in survivalfor juvenile Northern Pikeminnow could mean an increased predation of ju-venile Sockeye Salmon. It has also been speculated that increased refugefor Northern Pikeminnow could increase predation on Cultus Lake PygmySculpin (COSEWIC 2010).694.4 Managing the Impacts of Climate ChangeA changing climate cannot be ameliorated at the regional level and requiresglobal participation. However, given the interactivity of climate warmingand eutrophication on lake ecosystems (B. Moss et al. 2011), regional nutri-ent mitigation measures can be undertaken to reduce the impacts of climatechange on Cultus Lake to protect Sockeye Salmon and Cultus Lake PygmySculpin. As watershed loading of nitrogen to Cultus Lake is elevated fromatmospheric and landscape sources, Putt (2014) recommends targeting phos-phorus to mitigate eutrophication trajectories in Cultus Lake. Heavy septicusage in the summer may be leading to high phosphorus loading in CultusLake (Putt 2014). Water quality may be improved by limiting nutrient in-puts by modifying sewage disposal. Additionally, migratory gulls overnight-ing in Cultus Lake during the fall-through-spring period has led to a rise inphosphorus loadings from guano (Putt 2014). Non-lethal mitigation meth-ods may be employed in regional agricultural areas to reduce the numberof gulls returning to Cultus Lake (Cook et al. 2008). Finally, agriculture inthe Frosst Creek watershed represents a large source of nutrients for CultusLake (Putt 2014). Care should be taken to alter agricultural managementpractices to reduce nutrient runoff, especially with an expected change inhydrological regime. The application of these phosphorus mitigation strate-gies would likely reduce the effects of eutrophication on Cultus Lake, makingthe system less reactive to climate change over the next century.70Chapter 5ConclusionsClimate change and accelerated eutrophication are threatening the habi-tat for at-risk species and affecting the water quality in Cultus Lake. Thetotal lake heat content has increased since the early 20th century and is pro-jected to accelerate through the 21st century under climate change scenariosRCP4.5 and RCP8.5. Similarly, the stability and intensity of stratificationof the lake has historically been increasing, and is expected to continueincreasing through 2100 under the two climate change scenarios explored.Along with an increase in stratification strength, Cultus Lake is remainingstratified for a longer period. The onset of summer stratification is expectedto continue occurring in late March and early April, while the breakup ofstratification is expected to occur later, shifting from early December to midto late January with some years remaining stratified throughout the win-ter. These projections suggest the lake is shifting from a warm monomicticregime toward an oligomictic regime. Reduced winter whole-lake mixingwould therefore reduce the amount of hypolimnetic dissolved oxygen, thusthreatening the available habitat for cold-water species. Finally, surface andoutflow temperatures during the Sockeye Salmon run (August–November)have also been increasing with climate change. Surface temperatures havea direct response to air temperatures (Adrian et al. 2009), indicating thatair temperatures have been increasing since the early 20th century. Surfaceand outflow temperatures during late summer and the fall are projected toincrease at a higher rate in the future. Increased water temperatures canaccelerate the effects of eutrophication, such as more frequent cyanobacteriablooms, which owing to their enhanced organic matter production, couldaccelerate annual hypolimnetic water oxygen depletion.It is important that first-order lake stresses that are interactive with cli-mate change, such as eutrophication, are minimized to reduce the impactsof climate change on Cultus Lake and its species-at-risk. Regional measurescan be undertaken to abate nutrient inputs into Cultus Lake to reduce orreverse eutrophication. It is recommended that further investigation of theinteractive effects of climate change and eutrophication on Cultus Lake beconducted by coupling this physical lake model with a suite of biogeochem-71ical models, such as AED or CAEDYM (Hipsey and Hamilton 2008). Suchan approach will likely afford quantitative insight on future lake outcomes,to inform lake and watershed management and improve outcomes for CultusLake and its species-at-risk.72ReferencesAdrian, Rita, Catherine M. O’Reilly, Horacio Zagarese, Stephen B. Baines,Dag O. Hessen, Wendel Keller, David M. Livingstone, Ruben Som-maruga, Dietmar Straile, Ellen Van Donk, et al. (2009). “Lakes as sen-tinels of climate change”. In: Limnology and oceanography 54.6, p. 2283.doi: 10.4319/lo.2009.54.6_part_2.2283.AERONET (2017). Aerosol Robotic Network: Saturna Island. url: https://aeronet.gsfc.nasa.gov/cgi- bin/type_one_station_opera_v2_new?site=Saturn_Island&nachal=2&level=3&place_code=10(visited on 03/29/2017).Anderson, Donald M., Patricia M. Glibert, and Joann M. Burkholder (2002).“Harmful algal blooms and eutrophication: nutrient sources, composi-tion, and consequences”. In: Estuaries 25.4, pp. 704–726. doi: 10.1007/bf02804901.Arora, V. K., J. F. Scinocca, G. J. Boer, J. R. Christian, K. L. Denman,G. M. Flato, V. V. Kharin, W. G. Lee, and W. J. Merryfield (2011).“Carbon emission limits required to satisfy future representative concen-tration pathways of greenhouse gases”. In: Geophysical Research Letters38.5. doi: 10.1029/2010GL046270.Barry, Roger G. and Richard J. Chorley (1976). Atmosphere, weather, andclimate. English. 3d. Vol. 208. London: Methuen Publishing Ltd. isbn:0416704204.Bartsch, A. F. (1970). “Accelerated eutrophication of lakes in the UnitedStates: ecological response to human activities”. In: Environmental Pol-lution (1970) 1.2, pp. 133–140. doi: 10.1016/0013-9327(70)90013-3.Bird, R. E. and R. L. Hulstrom (1981). Simplified clear sky model for directand diffuse insolation on horizontal surfaces. Tech. rep. Solar ResearchInstitute. doi: 10.2172/6510849.Brett, John R. (1971). “Energetic responses of salmon to temperature. Astudy of some thermal relations in the physiology and freshwater ecologyof sockeye salmon (Oncorhynchus nerka)”. In: American Zoologist 11(11), pp. 99–113. doi: 10.1093/icb/11.1.99.73British Columbia Ministry of Environment (2016). Indicators of ClimateChange for British Columbia: 2016 Update. Ed. by Thomas White, Jo-hanna Wolf, Faron Anslow, Arelia Werner, and Reber Creative. isbn:0-7726-4732-1.Britton, C. M. and J. D. Dodd (1976). “Relationships of photosyntheticallyactive radiation and shortwave irradiance”. In: Agricultural Meteorology17.1, pp. 1–7. doi: 10.1016/0002-1571(76)90080-7.Campolongo, Francesca, Jessica Cariboni, and Andrea Saltelli (2007). “Aneffective screening design for sensitivity analysis of large models”. In: En-vironmental modelling & software 22.10, pp. 1509–1518. doi: 10.1016/j.envsoft.2006.10.004.Chambers, P. A., M. Guy, E. S. Roberts, M. N. Charlton, R. Kent, C.Gagnon, G. Grove, and N. Foster (2001). “Nutrients and their impacton the Canadian environment”. In: Agriculture and Agri-Food Canada,Environment Canada, Fisheries and Oceans Canada, Health Canada andNatural Resources Canada, p. 241.Chiang, E., L. Pon, D. T. Selbie, J. M. B. Hume, P. Woodruff, and G.Velema (2015). “Identification of critical habitat for Coastrange Sculpin(Cultus Population)(Cottus aleuticus)”. In: DFO Canadian Science Ad-visory Secretariat Research Document 33.Christensen, Michael R., Mark D. Graham, Rolf D. Vinebrooke, David L.Findlay, Michael J. Paterson, and Michael A. Turner (2006). “Multipleanthropogenic stressors cause ecological surprises in boreal lakes”. In:Global Change Biology 12.12, pp. 2316–2322. doi: 10.1111/j.1365-2486.2006.01257.x.Christian, J. R., V. K. Arora, G. J. Boer, C. L. Curry, K. Zahariev, K. L.Denman, G. M. Flato, W. G. Lee, W. J. Merryfield, N. T. Roulet,and J. F. Scinocca (2010). “The global carbon cycle in the CanadianEarth system model (CanESM1): Preindustrial control simulation”. In:Journal of Geophysical Research: Biogeosciences 115. doi: 10.1029/2008JG000920.Clark, Colin W. and David A. Levy (1988). “Diel Vertical Migrations by Ju-venile Sockeye Salmon and the Antipredation Window”. In: The Amer-ican Naturalist 131.2, pp. 271–290. doi: 10.1086/284789.Cook, Aonghais, Steven Rushton, John Allan, and Andrew Baxter (2008).“An Evaluation of Techniques to Control Problem Bird Species on Land-fill Sites”. In: Environmental Management 41.6, pp. 834–843. doi: 10.1007/s00267-008-9077-7.Cooke, Steven J., Scott G. Hinch, Anthony P. Farrell, Michael F. Lapointe,Simon R. M. Jones, J. Steveson Macdonald, David A. Patterson, Michael74C. Healey, and Glen Van Der Kraak (2004). “Abnormal Migration Tim-ing and High en route Mortality of Sockeye Salmon in the Fraser River,British Columbia”. In: Fisheries 29.2, pp. 22–33. doi: 10.1577/1548-8446(2004)29[22:AMTAHE]2.0.CO;2.COSEWIC (2003). COSEWIC Assessment and Status Report on the SockeyeSalmon Oncorhynchus nerka, Cultus population in Canada. Tech. rep.Ottawa: Committee on the Status of Endangered Wildlife in Canada.url: http://www.registrelep-sararegistry.gc.ca/virtual_sara/files/cosewic/sr%5Fcultus%5Fsockeye%5Fsalmon%5Fe%2Epdf (vis-ited on 10/13/2016).— (2010). COSEWIC Assessment and Status Report on the CoastrangeSculpin Cottus aleuticus, Cultus Population in Canada. Tech. rep. Ot-tawa: Committee on the Status of Endangered Wildlife in Canada. url:http://www.registrelep- sararegistry.gc.ca/virtual_sara/files/cosewic/sr%5FCoastrange%20Sculpin%5F0810%5Fe1%2Epdf(visited on 10/13/2016).Crossin, Glenn Terrence, S. G. Hinch, S. J. Cooke, D. W. Welch, D. A.Patterson, S. R. M. Jones, A. G. Lotto, R. A. Leggatt, M. T. Mathes,J. M. Shrimpton, G. Van Der Kraak, and A. P. Farrell (2008). “Exposureto high temperature influences the behaviour, physiology, and survivalof sockeye salmon during spawning migration”. In: Canadian Journal ofZoology 86.2, pp. 127–140. doi: 10.1139/Z07-122.Crozier, L. G., A. P. Hendry, P. W. Lawson, T. P. Quinn, N. J. Mantua,J. Battin, R. G. Shaw, and R. B. Huey (2008). “Potential responses toclimate change in organisms with complex life histories: evolution andplasticity in Pacific salmon”. In: Evolutionary Applications 1.2, pp. 252–270. doi: 10.1111/j.1752-4571.2008.00033.x.Cultus Sockeye Recovery Team (2005). National conservation strategy forthe sockeye salmon, (Oncorhynchus nerka), Cultus Lake population, inBritish Columbia. Tech. rep. Fisheries and Oceans Canada. url: http://www.dfo-mpo.gc.ca/Library/334716.pdf (visited on 11/16/2016).Delcan (2012). Cultus Lake Traffic and Transportation Study. Tech. rep.Vancouver: Delcan Corporation.Dera, Jerzy and Howard R. Gordon (1968). “Light field fluctuations in thephotic zone”. In: Limnology and Oceanography 13.4, pp. 697–699.Engelhardt, Christof and Georgiy Kirillin (2014). “Criteria for the onsetand breakup of summer lake stratification based on routine temperaturemeasurements”. In: Fundamental and Applied Limnology/Archiv fu¨r Hy-drobiologie 184.3, pp. 183–194. doi: 10.1127/1863-9135/2014/0582.75Environment and Climate Change Canada (2017). Canadian regional cli-mate model. Government of Canada. url: http://www.ec.gc.ca/ccmac - cccma / default . asp ? lang = En & n = 7CC1E8B6 - 1 (visited on06/26/2017).Environment and Natural Resources Canada (2017). Engineering Cli-mate Datasets: Canadian Weather Energy and Engineering Datasets(CWEEDS). Government of Canada. url: http://climate.weather.gc.ca/prods_servs/engineering_e.html (visited on 06/26/2017).Environmental Reporting BC (2015). Status of Ground-Level Ozone in B.C.(2011-2013). Tech. rep. Government of British Columbia: Ministry ofEnvironment. url: http://www.env.gov.bc.ca/soe/indicators/air/print_ver/envreportbc_ozone_August2015.pdf (visited on03/29/2017).Ficke, Ashley D., Christopher A. Myrick, and Lara J. Hansen (2007). “Po-tential impacts of global climate change on freshwater fisheries”. In: Re-views in Fish Biology and Fisheries 17.4, pp. 581–613. doi: 10.1007/s11160-007-9059-5.Fisher, Hugo B., E. John List, Robert C.Y. Koh, Joerg Imberger, and Nor-man H. Brooks (1979). Mixing in Inland and Coastal Waters. AcademicPress. 302 pp. isbn: 0122581504.Flato, G., J. Marotzke, B. Abiodun, P. Braconnot, S. C. Chou, W. Collins,P. Cox, F. Driouech, S. Emori, V. Eyring, C. Forest, P. Gleckler, E.Guilyardi, C. Jakob, V. Kattsov, C. Reason, and M. Rummukain (2013).“Evaluation of Climate Models”. In: Climate Change 2013: The PhysicalScience Basis: Working Group I Contribution to the Fifth AssessmentReport of the Intergovernmental Panel on Climate Change. CambridgeUniversity Press, pp. 741–866. doi: 10.1017/CBO9781107415324.020.Foerster, R. E. (1925). “Studies in the ecology of the sockeye salmon (On-corhynchus nerka)”. In: Contributions to Canadian Biology 2, p. 335.George, Glen, Margaret Hurley, and Dian Hewitt (2007). “The impact ofclimate change on the physical characteristics of the larger lakes in theEnglish Lake District”. In: Freshwater Biology 52.9, pp. 1647–1666. doi:10.1111/j.1365-2427.2007.01773.x.Gerten, Dieter and Rita Adrian (2001). “Differences in the persistency ofthe North Atlantic Oscillation signal among lakes”. In: Limnology andOceanography 46.2, pp. 448–455. doi: 10.4319/lo.2001.46.2.0448.Goodlad, J. C., T. W. Gjernes, and E. L. Brannon (1974). “Factors af-fecting sockeye salmon (Oncorhynchus nerka) growth in four lakes of theFraser River system”. In: Journal of the Fisheries Board of Canada 31.5,pp. 871–892. doi: 10.1139/f74-106.76Gutowski Jr, William J., Steven G. Decker, Rodney A. Donavon, Zaitao Pan,Raymond W. Arritt, and Eugene S. Takle (2003). “Temporal–spatialscales of observed and simulated precipitation in central US climate”.In: Journal of Climate 16.22, pp. 3841–3847. doi: 10 . 1175 / 1520 -0442(2003)016<3841:TSOOAS>2.0.CO;2.Healey, Michael (2011). “The cumulative impacts of climate change on FraserRiver sockeye salmon (Oncorhynchus nerka) and implications for man-agement”. In: Canadian Journal of Fisheries and Aquatic Sciences 68.4,pp. 718–737. doi: 10.1139/f2011-010.Hipsey, M. R., L. C. Bruce, and D. P. Hamilton (2013). Aquatic Ecodynamics(AED) Model Library Science Manual.— (2014). General Lake Model: Model Overview and User Information; v2.0β. url: http://aed.see.uwa.edu.au/research/models/GLM/downloads/AED_GLM_v2_0b0_20141025.pdf (visited on 09/21/2016).Hipsey, M. R. and D. P. Hamilton (2008). “Computational aquatic ecosys-tem dynamics model: CAEDYM v3, Science Manual”. In: University ofWestern Australia.Holding, S. and Diana A. (2012). Cultus Lake Watershed Numerical Ground-water Flow Model. Tech. rep. Department of Earth Sciences Simon FraserUniversity. url: https://www.psf.ca/sites/default/files/FSWP_11_34_FR_Attachment_Cultus_Lake_Groundwater_Model_Report.pdf (visited on 11/01/2016).Idso, Sherwood B. (1973). “On the concept of lake stability”. In: Limnologyand Oceanography 18.4, pp. 681–683. doi: 10.4319/lo.1973.18.4.0681.Imberger, J. and J. C. Patterson (1981). “Dynamic Reservoir SimulationModel- DYRESM: 5”. In: Transport Models for Inland and Coastal Wa-ters. New York: Academic Press.Imboden, Dieter M. and Alfred Wu¨est (1995). Mixing mechanisms in lakes.In: Physics and chemistry of lakes. Springer, pp. 83–138.Instruments, Swoffer (2016). Swoffer Instruments Inc. Model 3000 CurrentM eter Flow Calculator: Data Recorder. url: http://www.swoffer.com/pdf/3000bulet16.pdf (visited on 02/14/2017).IOC, SCOR (2010). IAPSO: The international thermodynamic equation ofseawater–2010: Calculation and use of thermodynamic properties, Inter-governmental Oceanographic Commission, Manuals and Guides No. 56,p. 196.IPCC (2013). Climate Change 2013: The Physical Science Basis. Contribu-tion of Working Group I to the Fifth Assessment Report of the Inter-governmental Panel on Climate Change. Ed. by T. F. Stocker, D. Qin,77G. K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia,V. Bex, and P. M. Midgley. Cambridge, United Kingdom and New York,NY, USA: Cambridge University Press, p. 1535.ISO (2010). ISO 1100-2:2010 Hydrometry – Measurement of liquid flowin open channels – Part 2: Determination of the stage-discharge rela-tionship. Tech. rep. Geneva, Switzerland: International Organization forStandardization.Jeppesen, Erik, Brian Moss, Helen Bennion, Laurence Carvalho, Luc DeMeester, Heidrun Feuchtmayr, Nikolai Friberg, Mark O. Gessner, MarietHefting, Torben L. Lauridsen, Lone Liboriussen, Hilmar J. Malmquist,Linda May, Mariana Meerhoff, Jon S. Olafsson, Merel B. Soons, and JosT. A. Verhoeven (2010). “Interaction of Climate Change and Eutrophi-cation”. In: Climate Change Impacts on Freshwater Ecosystems. Ed. byMartin Kernan, Richard W. Battarbee, and Brian Moss. Blackwell Pub-lishing, Oxford. Chap. 6, pp. 119–151. doi: 10.1002/9781444327397.ch6.Kalff, Jacob (2002). Limnology: inland water ecosystems. Vol. 592. PrenticeHall New Jersey.Kankaala, Paula, Anne Ojala, Tiina Tulonen, and Lauri Arvola (2002).“Changes in nutrient retention capacity of boreal aquatic ecosystemsunder climate warming: a simulation study”. In: Hydrobiologia 469.1,pp. 67–76. doi: 10.1023/A:1015563224554.Kraemer, Benjamin M., Orlane Anneville, Sudeep Chandra, MargaretDix, Esko Kuusisto, David M. Livingstone, Alon Rimmer, S. GeoffreySchladow, Eugene Silow, Lewis M. Sitoki, Rashid Tamatamah, YvonneVadeboncoeur, and Peter B. McIntyre (2015). “Morphometry and aver-age temperature affect lake stratification responses to climate change”.In: Geophysical Research Letters 42.12, pp. 4981–4988. doi: 10.1002/2015GL064097.Levy, David A. (1990). “Reciprocal Diel Vertical Migration Behavior inPlanktivores and Zooplankton in British Columbia Lakes”. In: Cana-dian Journal of Fisheries and Aquatic Sciences 47.9, pp. 1755–1764.doi: 10.1139/f90-199.Lewis, William M. (1979). Zooplankton Community Analysis. Springer NewYork. doi: 10.1007/978-1-4612-9986-8.Livingstone, David M. (2008). “A change of climate provokes a change ofparadigm: taking leave of two tacit assumptions about physical lake forc-ing”. In: International Review of Hydrobiology 93.4-5, pp. 404–414. doi:10.1002/iroh.200811061.78Matzinger, Andreas, Martin Schmid, Elizabeta Veljanoska-Sarafiloska,Suzana Patceva, Dafina Guseska, Bernd Wagner, Beat Mu¨ller, MichaelSturm, and Alfred Wu¨est (2007). “Eutrophication of ancient Lake Ohrid:Global warming amplifies detrimental effects of increased nutrient in-puts”. In: Limnology and Oceanography 52.1, pp. 338–353. doi: 10 .4319/lo.2007.52.1.0338.McGinley, Paul M., Kevin C. Masarik, Madeline B. Gotkowitz, and DavidJ. Mechenich (2016). “Impact of anthropogenic geochemical change andaquifer geology on groundwater phosphorus concentrations”. In: AppliedGeochemistry 72, pp. 1–9. doi: http://dx.doi.org/10.1016/j.apgeochem.2016.05.020.McPhail, John Donald (2007). The freshwater fishes of British Columbia.University of Alberta.Moss, Brian, Sarian Kosten, Mariana Meerhoff, Richard W. Battarbee, ErikJeppesen, Ne´stor Mazzeo, Karl Havens, Gissell Lacerot, Zhengwen Liu,Luc De Meester, Hans Paerl, and Marten Scheffer (2011). “Allied attack:climate change and eutrophication”. In: Inland Waters 1.2, pp. 101–105.doi: 10.5268/IW-1.2.359.Moss, Richard H., Jae A. Edmonds, Kathy A. Hibbard, Martin R. Manning,Steven K. Rose, Detlef P. Van Vuuren, Timothy R. Carter, Seita Emori,Mikiko Kainuma, Tom Kram, Gerald A. Meehl, John F. B. Mitchell,Nebojsa Nakicenovic, Keywan Riahi, Steven J. Smith, Ronald J. Stouf-fer, Allison M. Thomson, John P. Weyant, and Thomas J. Wilbanks(2010). “The next generation of scenarios for climate change researchand assessment”. In: Nature 463.7282, pp. 747–756. doi: 10 . 1038 /nature08823.Mossop, Brent and Michael J. Bradford (2004). Review of Eurasian water-milfoil control at Cultus Lake and recommendations for future removals.Tech. rep. Fisheries and Oceans Canada Fraser River Stock Assessment.O’Neil, J. M., T. W. Davis, M. A. Burford, and C. J. Gobler (2012). “The riseof harmful cyanobacteria blooms: The potential roles of eutrophicationand climate change”. In: Harmful Algae 14, pp. 313–334. doi: 10.1016/j.hal.2011.10.027.Onset Computer Corporation (2008). Wind Speed/Direction Smart Sensor(S-WCA-M003) Manual. url: http://www.onsetcomp.com/files/manual_pdfs/6858-G-MAN-S-WCA.pdf (visited on 04/06/2017).— (2017a). Barometric Pressure Smart Sensor (S-BPA-CM10) Manual.url: http://www.onsetcomp.com/files/manual_pdfs/6122- E%20MAN-S-BPA.pdf (visited on 04/06/2017).79Onset Computer Corporation (2017b). HOBO U20 Water Level Log-ger (U20-001-0x and U20-001-0x-Ti) Manual. url: http : / / www .onsetcomp.com/files/manual_pdfs/12315-H%20MAN-U20.pdf (vis-ited on 03/01/2017).— (2017c). TidbiT v2 Temp (UTBI-001) Manual. url: http : / / www .onsetcomp.com/files/manual_pdfs/10385-H%20MAN-UTBI-001.pdf(visited on 03/01/2017).O’Reilly, Catherine M. et al. (2015). “Rapid and highly variable warming oflake surface waters around the globe”. In: Geophysical Research Letters42.24, pp. 10, 773–10, 781. doi: 10.1002/2015GL066235.Pacific Climate Impacts Consortium (2012). Plan2Adapt. url: https://plan2adapt.ca (visited on 07/13/2017).Paerl, Hans W., Rolland S. Fulton, Pia H. Moisander, and Julianne Dy-ble (2001). “Harmful freshwater algal blooms, with an emphasis oncyanobacteria”. In: The Scientific World Journal 1, pp. 76–113. doi:10.1100/tsw.2001.16.Paerl, Hans W. and J. Huisman (2009). “Climate change: a catalyst forglobal expansion of harmful cyanobacterial blooms”. In: EnvironmentalMicrobiology Reports 1 (1), pp. 27–37. doi: 10.1111/j.1758-2229.2008.00004.x.Panofsky, Hans A. and Glenn W. Brier (1958). Some Application of Statis-tics to Meteorology. University Park, Pennsylvania: College of MineralIndustries, Pennsylvania State University.Po¨rtner, Hans O. and Anthony P. Farrell (2008). “Physiology and ClimateChange”. In: Science 322.5902, pp. 690–692.Putt, Annika Elsie (2014). “Spatiotemporal nutrient loading to Cultus Lake:Context for eutrophication and implications for integrated watershed-lake management”. MA thesis. Simon Fraser University. url: http ://rem- main.rem.sfu.ca/theses/PuttAnnika_2014_MRM591.pdf(visited on 05/02/2016).Reynolds, C. S. (2009). “Biological–Physical Interactions”. In: Encyclopediaof Inland Waters. Ed. by Gene E Likens. Vol. 1, pp. 515–521.Ricker, William E. (1937). “Physical and Chemical Characteristics of CultusLake, British Columbia.” In: Journal of the Biological Board of Canada3.4, pp. 363–402. doi: 10.1139/f37-022.Robertson, Dale M. and William J. Rose (2011). “Response in the trophicstate of stratified lakes to changes in hydrology and water level: potentialeffects of climate change”. In: Journal of Water and Climate Change 2.1,pp. 1–18. doi: 10.2166/wcc.2011.026.80Sahoo, G. B., A. L. Forrest, S. G. Schladow, J. E. Reuter, R. Coats, and M.Dettinger (2015). “Climate change impacts on lake thermal dynamicsand ecosystem vulnerabilities”. In: Limnology and Oceanography 61.2,pp. 496–507. doi: 10.1002/lno.10228.SARA (2002). “Species at Risk Act. SC 2002, c 29”. In: url: http://laws-lois.justice.gc.ca/PDF/S-15.3.pdf (visited on 08/09/2017).Schelker, Jakob, Karin O¨hman, Stefan Lo¨fgren, and Hjalmar Laudon(2014). “Scaling of increased dissolved organic carbon inputs by forestclear-cutting–What arrives downstream?” In: Journal of hydrology 508,pp. 299–306. doi: 10.1016/j.jhydrol.2013.09.056.Scheuerell, Mark D. and Daniel E. Schindler (2003). “Diel vertical migra-tion by juvenile sockeye salmon: empirical evidence for the antipreda-tion window”. In: Ecology 84.7, pp. 1713–1720. doi: 10.1890/0012-9658(2003)084[1713:DVMBJS]2.0.CO;2.Schindler, Daniel E., Donald E. Rogers, Mark D. Scheuerell, and Caryn A.Abrey (2005). “Effects of changing climate on zooplankton and juve-nile sockeye salmon growth in southwestern Alaska”. In: Ecology 86.1,pp. 198–209. doi: 10.1890/03-0408.Schindler, David W. (2012). “The dilemma of controlling cultural eutroph-ication of lakes”. In: Proceedings of the Royal Society B: Biological Sci-ences. The Royal Society. doi: 10.1098/rspb.2012.1032.Scibek, J. and D. Allen (2006). “Comparing modelled responses of twohigh-permeability, unconfined aquifers to predicted climate change”. In:Global and Planetary Change 50.1-2, pp. 50–62. doi: 10 . 1016 / j .gloplacha.2005.10.002.Scinocca, J. F., V. V. Kharin, Y. Jiao, M. W. Qian, M. Lazare, L. Solheim,G. M. Flato, S. Biner, M. Desgagne, and B. Dugas (2016). “CoordinatedGlobal and Regional Climate Modeling”. In: Journal of Climate 29.1,pp. 17–35. doi: 10.1175/JCLI-D-15-0161.1.Seabird Scientific (2016a). SBE 19plus V2 SeaCAT Profiler CTD User Man-ual: Conductivity, Temperature, & Pressure Recorder with RS-232 Inter-face.— (2016b). Seasoft V2: SBE Data Processing Software Manual. url:http : / / www . seabird . com / sites / default / files / documents /SBEDataProcessing_7.26.4.pdf (visited on 04/12/2017).Shortreed, K. R. S. (2007). Limnology of Cultus Lake, British Columbia.Tech. rep. Fisheries and Oceans Canada.Smith, Craig S. and J. W. Barko (1990). “Ecology of Eurasian watermilfoil.”In: Journal of Aquatic Plant Management 28.2, pp. 55–64.81Smith, Val H. and David W. Schindler (2009). “Eutrophication science:where do we go from here?” In: Trends in ecology & evolution 24.4,pp. 201–207.Søndergaard, Martin, Jens Peder Jensen, and Erik Jeppesen (2003). “Roleof sediment and internal loading of phosphorus in shallow lakes”. In:Hydrobiologia 506.1, pp. 135–145. doi: 10.1023/B:HYDR.0000008611.12704.dd.Stainsby, E. A., J. G. Winter, H. Jarjanazi, A. M. Paterson, D. O. Evans, andJ. D. Young (2011). “Changes in the thermal stability of Lake Simcoefrom 1980 to 2008”. In: Journal of Great Lakes Research 37, pp. 55–62.doi: 10.1016/j.jglr.2011.04.001.Stefan, Heinz G. and Eric B. Preud’homme (1993). “Stream temperatureestimation from air temperature”. In: Journal of the American WaterResources Association 29.1, pp. 27–45.Stull, Roland (2015). Practical Meteorology: An Algebra-based Survey of At-mospheric Science. University of British Columbia. isbn: 978-0-88865-176-1. url: https : / / www . eoas . ubc . ca / books / Practical _Meteorology/prmet/PracticalMet_WholeBook-v1_00b.pdf (visitedon 02/21/2017).Taylor, Karl E., Ronald J. Stouffer, and Gerald A. Meehl (2012). “AnOverview of CMIP5 and the Experiment Design”. In: Bulletin of theAmerican Meteorological Society 93.4, pp. 485–498. doi: 10.1175/BAMS-D-11-00094.1.Trimble, Sabina (2016). “Storying Sw´ı:lhcha: Place Making and Power ata Sto´:lo¯ Landmark”. In: BC Studies 190, p. 39. url: http://ojs.library.ubc.ca/index.php/bcstudies/article/viewFile/187241/186087 (visited on 05/02/2017).UNESCO (1981). Tenth report of the joint panel on oceanographic tables andstandards. url: http://unesdoc.unesco.org/images/0004/000461/046148eb.pdf (visited on 11/18/2016).Vallentyne, John R. (1974). The Algal Bowl: Lakes and Man. Dept. of theEnvironment, Fisheries and Marine Service.Vincent, W. F. (2009). “Effects of Climate Change on Lakes”. In: Encyclo-pedia of Inland Waters. Ed. by Gene E Likens. Vol. 3, pp. 55–60.von Salzen, Knut, John F. Scinocca, Norman A. McFarlane, Jiangnan Li,Jason N. S. Cole, David Plummer, Diana Verseghy, M. Cathy Reader,Xiaoyan Ma, Michael Lazare, and Larry Solheim (2013). “The CanadianFourth Generation Atmospheric Global Climate Model (CanAM4). PartI: Representation of Physical Processes”. In: Atmosphere-Ocean 51.1,pp. 104–125. doi: 10.1080/07055900.2012.755610.82Weinberger, Stefan and Mark Vetter (2014). “Lake heat content and sta-bility variation due to climate change: coupled regional climate model(REMO)-lake model (DYRESM) analysis”. In: Journal of Limnology73.1. doi: 10.4081/jlimnol.2014.668.Wetzel, Robert G. (2001). Limnology: lake and river ecosystems. 3rd. Aca-demic Press. isbn: 0127447601.Wieringa, Jon (1992). “Updating the Davenport roughness classification”.In: Journal of Wind Engineering and Industrial Aerodynamics 41.1-3,pp. 357–368. doi: 10.1016/0167-6105(92)90434-c.Woodruff, Patricia E. and Eric B. Taylor (2013). “Assessing the distinc-tiveness of the Cultus pygmy sculpin, a threatened endemic, from thewidespread coastrange sculpin Cottus aleuticus”. In: Endangered SpeciesResearch 20.2, pp. 181–194. doi: 10.3354/esr00493.Yajima, Hiroshi and Shigetomo Yamamoto (2015). “Improvements of Radia-tion Estimations for a Simulation of Water Temperature in a Reservoir”.In: Journal of Japan Society of Civil Engineers, Ser. B1 (Hydraulic En-gineering) 71.4. doi: 10.2208/jscejhe.71.i_775.Zubel, Marc (2000). Groundwater Conditions of the Columbia ValleyAquifer Cultus Lake , British Columbia. Tech. rep. Ministry of En-vironment, Lands & Parks, p. 98. url: http : / / a100 . gov . bc .ca / appsdata / acat / documents / r6477 / 1287 _ 1143682706883 _8b7182b483ed458aad9da29ff7cef555.pdf (visited on 11/01/2016).83Appendix AInflow and OutflowDescriptionsHydrological and morphological characteristics of the three largest streamsfrom the Cultus Lake watershed are described in detail: Frosst Creek, WattCreek, and Smith Falls Creek.Frosst Creek is the main inflow to Cultus Lake, entering at the southend of the lake, north of Lindell Beach. Frosst Creek flows from the forestedregion around the United States border into the Columbia Valley which ispredominantly developed for agriculture. The downstream stream coursewas diverted from the middle of Lindell Beach northward to its presentcourse around the 1940s (Cultus Sockeye Recovery Team 2005). Water Sur-vey of Canada (WSC) gauge 08MH072 (Frosst Creek near Lindell Beach)located near the Columbia Valley Highway bridge regularly recorded flowrates between 1960 and 1964. The drainage area is approximately 36 km2.The upper reaches of the stream are fairly steep with a gradient of 12 % anddecreasing to 4 % near the mouth (Zubel 2000). The bankfull discharge, ortwo-year flood return, is approximately 8.5 m3 s-1 based on a log-PearsonType III distribution for the WSC data with annual extremes typically oc-curring in November and December. At the 2016 gauging location, thestream morphology consists of rapids and pools, with measurements beingtaken in a pool. The width and maximum depth at the gauging locationwere 6.2 m and 0.82 m, respectively at high flow on 16 November. A sum-mer baseflow was found to be 0.2 m3 s-1. The D50 grain size is estimatedat 70 mm, but large boulders measuring 300 mm–500 mm are present. Thepresence of large boulders and the steep slope indicate that Frosst Creek iscapable of moving large volumes of water at great force.Watt Creek is an intermittent creek, drying in the summer during periodsof low precipitation. The creek mouth is located east of Lindell Beach. Thedrainage area is 6.6 km2 (Putt 2014) and consists predominantly of forestedareas from Cultus Lake Provincial Park. Upstream of the Columbia ValleyHighway bridge, the slope is relatively mild because a log weir is installedat the foot of the bridge to prevent scouring of the foundations. The stream84undergoes a series of falls downstream of the log weir before leveling out priorto entering Cultus Lake. The width of the creek at the gauging location was2.5 m and the maximum depth was 0.3 m at a discharge of 0.2 m3 s-1 on 16November 2016. The bed material ranges from gravel to small cobble withthe presence of some large rocks. Watt Creek is expected to have flashy flowcharacteristics due to the small drainage area and steep slopes.Smith Falls Creek is a perennial stream that enters Cultus Lake at thenortheast shore with a drainage area of 6.5 km2. Smith Falls Creek hastwo arms: the southern arm is situated in Cultus Lake Provincial Park,the northern arm consists of wetlands created in the flat topography whichlikely is the result of the summer baseflow of 0.03 m3 s-1. The creek consistsof many chutes and pools. A waterfall (Smith Falls) is located immediatelyupstream of the Columbia Valley Highway where the stream enters a culvertthat flows to the mouth at Cultus Lake. At the stream gauging station, thecreek had a width of 2.7 m and a maximum depth of 0.4 m with a dischargeof 0.3 m3 s-1 on 16 November. The bed material consists of sand ranging tolarge gravel. Similarly to Watt Creek, Smith Falls Creek likely has flashyflows.Sweltzer River is the only outlet of Cultus Lake. The river drains northtoward the Chilliwack River at a gradient of 0.005 (Holding and A. 2012).A fish counting fence has existed on Sweltzer River outside the SalmonResearch Laboratory with data extending back to the 1920s (Cultus SockeyeRecovery Team 2005). WSC gauge 08MH033 (Sweltzer River at CultusLake) was active downstream of the Columbia Valley Highway near the fishcounting fence between 1947–1956 and 1960–1964. The drainage area tothe gauge location as indicated by WSC is 65 km2; however, delineation byShortreed (2007) indicates a drainage area of 75 km2, which upon furtherinvestigation will be taken as correct. The flow is artificially controlled bymeans of stop logs and a low head weir that maintain a constant water levelin Cultus Lake. The two-year flood flow rate is approximately 14.3 m3 s-1based on a log-Pearson Type III distribution with annual extremes occurringbetween October and February. The width and maximum depth were 18.3 mand 0.89 m, respectively with a discharge of 2.8 m3 s-1 on 2 November. Thebed material consists of coarse sand and gravel.85

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0354396/manifest

Comment

Related Items