Document Type


Degree Name

Doctor of Philosophy (PhD)


Geography & Environmental Studies


Faculty of Science

First Advisor

Dr. Philip Marsh

Advisor Role


Second Advisor

Dr. Brent B. Wolfe

Advisor Role


Third Advisor

Dr. Oliver Sonnentag & Prof. Dr. Julia Boike

Advisor Role

Doctoral advisory committee members


There are tens of thousands of thermokarst lakes in the Inuvik-Tuktoyaktuk region, located in the northwest corner of the Northwest Territories, Canada. These lakes formed following the last glacial period in areas where ice-rich permafrost thawed and created depressions in the landscape. The Inuvik-Tuktoyaktuk region is one of the fastest warming regions in the world, leading to changing precipitation patterns, permafrost thaw and deciduous shrub expansion, all of which are affecting the water balance of thermokarst lakes. During the past several decades, lake expansion and contraction have been observed in response to fluctuations in precipitation. While these changes in lake surface area and number have been documented, less is known about how varying meteorological conditions, lake and landscape features, and hydrological processes have regulated these changes in thermokarst lake water balances. Many studies documenting fluctuations in lake surface area often observe some lakes expanding in area while others contract during the same period of time, suggesting that lake and watershed properties regulate how lakes react to climate change. Rapid lake drainage, which can be initiated by extremely high lake levels, is occurring at an increasing rate.

The main objective of this thesis is to quantify the drivers of variability in thermokarst lake water balance components (e.g. inflow, evaporation, lake level, lake water source composition), so that we may better understand how lakes will respond to ongoing climate change. Multiple properties of the environment have the potential to influence thermokarst lake water balances: seasonal and inter-annual variation in meteorological conditions (i.e. air temperature and precipitation), watershed properties (e.g. surface area, vegetation, topography, permafrost), and lake properties (e.g. depth, surface area, outlet channel presence). Achieving the objective of this thesis involves measuring lake water balances and quantifying how the environmental properties described above affect lake water balance components. The four main research chapters of this thesis divide this task into smaller parts, with each chapter quantifying a subset of lake water balance components or environmental conditions at different spatial and temporal scales.

The second chapter explores the interaction between snow and shrubs and their ultimate impact on frost table depth, which influences runoff from lake watersheds. Shrubs are expanding across the Arctic and affecting snow depth, snowmelt timing and soil shading, which in turn influences frost table depth. Hummocks, which are mineral-earth domes 0.5 - 1.0 m across formed over hundreds of years via the cyclical freezing and thawing of the uppermost soil layer (active layer), are common features on hillslopes in the Inuvik-Tuktoyaktuk region. Between hummocks lie a mesh-like network of inter-hummock zones that are filled with peat and form a preferential flow network on hillslopes. When hummocks degrade due to permafrost thaw, they collapse and mineral soils invade inter-hummock zones and reduce the hydraulic conductivity. Frost table depth, snow depth and the snow-free date were measured in hummock and inter-hummock zones at the Siksik Creek watershed throughout the summer of 2015 in shrub-covered and shrub-free locations. Areas of birch shrubs had earlier snowmelt dates, and experienced greater hummock frost table depths. In inter-hummock zones, frost table depth was shallower when adjacent hummocks were taller, indicating that inter-hummock frost table depths increase when hummocks collapse. Future birch shrub expansion may accelerate permafrost thaw, leading to hummock collapse and a reduction in the ability of hillslopes to convey runoff.

Chapter three focuses on the degree to which freshet runoff mixes with lake water, as the snowmelt period represents the largest volumetric input of water to lakes. Previous studies have observed a phenomenon called "snowmelt bypass" whereby water flowing into ice-covered lakes flows underneath lake ice and out of the lake without mixing with the entire water column. Snowmelt bypass occurs when the freshet runoff flowing into lakes (approx. 0C) is less dense than deeper lake waters (<4C), a condition that is typically present at the start of snowmelt runoff; however, lakes generally become more mixed towards the end of the snowmelt period. Using lake water isotope data from before and after the freshet, the percentage of lake water replaced by freshet runoff and the average lake source water isotope composition (𝛿I) was quantified for seventeen lakes and compared to lake and watershed properties. Lake depth significantly influenced the amount of lake water replaced by freshet runoff, with deeper lakes retaining less freshet runoff because a larger portion of the lake volume was isolated from mixing with freshet runoff. Additionally, isotope data showed that the source of freshet runoff remaining in lakes contained a mixture of snow-sourced and soil-sourced water. The snowmelt bypass effect was likely stronger earlier in the freshet when runoff was more snow-sourced, while later in the freshet when runoff was likely more soil water-sourced, stronger vertical mixing in the lake was likely present and the snowmelt bypass effect would have been weaker. Earlier snowmelt relative to lake-ice melt caused by shrubification could lead to greater snowmelt bypass in the future, with the freshet runoff remaining in lakes becoming even more soil water-sourced. A shift to more soil-sourced freshet may impact lake chemistry, as soil-sourced runoff has lower concentrations of dissolved organic carbon, lower conductivity and higher pH than snow-sourced runoff. These results are relevant for open-drainage lakes, which can experience snowmelt bypass since any excess water is able to flow through the lake outlet.

The fourth chapter investigates how lake and watershed properties mediate the response of lakes to seasonal shifts in meteorological conditions using an isotope hydrology approach. Twenty-five lakes along the Inuvik-Tuktoyaktuk Highway were water sampled five times during 2018 for isotope analysis, with sampling starting before snowmelt and ending in early September. Lake water isotope compositions were used to estimate the ratio of evaporation-to-inflow (E/I) and (𝛿I). Four distinct seasonal phases of lake water balance were identified from the isotope data and prevailing meteorological conditions. The initial Freshet Phase occurred during snowmelt, with lakes experiencing a reduction in E/I and shift in 𝛿I towards the average isotope composition of precipitation. An Evaporation Phase followed, a period of typically warm air temperatures and minimal precipitation, during which E/I increased and 𝛿I remained stable due to minimal inflow. As air temperatures declined and precipitation increased, the Soil Wetting Phase began, where E/I and 𝛿I did not respond to initial rainfall as dry soils had not yet reached a moisture level sufficient to generate runoff. As rainfall persisted and soils become wetter, the Recharge Phase initiated, during which E/I declined and 𝛿I became more rain-like as inflow to lakes increased and evaporation decreased as solar radiation and air temperature declined. Variability in E/I among lakes was strongly correlated with the ratio of watershed area to lake area (WA/LA), where lakes with smaller WA/LA had larger E/I ratios because they received less inflow relative to evaporation. The majority of lakes in the region have a WA/LA

Chapter five analyzes year-to-year differences in lake water balance components for a single lake in the region. Three years of lake water balance measurements were made at Big Bear Lake near the Trail Valley Creek research station. The water balance was calculated daily between May 1 and October 30, such that ΔLL = P + Qin - Qout - E, where LL is lake level, P is precipitation, Qin is inflow, Qout is outflow, and E is evaporation. Lake level was measured using a pressure transducer type water level recorder and outflow was estimated using a stage-discharge relationship derived from manual discharge measurements made at the lake outlet. Evaporation was estimated using the Priestley-Taylor method, while precipitation was measured using a shielded weighing gauge. Inflow was calculated as the unknown variable in the lake water balance equation, but was set to 0 if the estimated inflow was a negative value. During the freshet period, the runoff ratio (i.e. the percentage of snowpack and precipitation converted into inflow) was highest when the snowmelt occurred rapidly or when the previous year experienced higher levels of summertime precipitation. Rapid snowmelt also caused higher maximum lake levels at the time of snow dam failure. Summertime runoff ratios were higher in wetter years, but were also affected by the previous year’s summertime precipitation, with higher runoff ratios when the preceding year was wetter. Evaporation losses varied between 226 to 296 mm, with the length of the ice-free period largely controlling the amount of evaporation that occurred. There was a large potential for evaporation (4 to 6 mm day-1) in the days leading up to the lake becoming ice-free, implying that the ice-free date has a strong impact on total evaporative losses from lakes. Predicting whether lake water balances become wetter or drier in the future requires understanding how increasing active layer depths, degrading hummocks and shrub expansion will affect runoff ratios under predicted wetter summers, while ice-free seasons lengthen and lake evaporative losses increase.

Overall, this thesis improves our understanding of the controls on thermokarst lake water balances in the western Canadian Arctic. Specifically, this thesis outlines the strong predictive power of lake and watershed attributes in describing variability in lake water balances, a concept that could be further developed by integrating more variables from remote sensing data and more advanced statistical methods to quantify the influence of less influential lake and landscape attributes on lake water balance. This thesis also advances our understanding of the freshet period, including the influence of different shrub species on snow depth, snowmelt timing and frost table depth, the drivers of lake level maximums and snow dam failure, and the impact of lake depth on the mixing of snow- and soil-sourced water into ice-covered lakes.

The complexity of interactions between climate change and spatially variable landscape elements make it difficult to foresee whether lakes will experience a wetter or a drier future water balance, as climate-induced changes may increase runoff (e.g. more rainfall, wetter soils and increased runoff ratios), or decrease runoff (e.g. shrub expansion, greater rainfall interception, drier soils and decreased runoff ratios), which will either offset or amplify increases in lake evaporation losses. Given our observations of increasing E/I ratios with decreasing WA/LA in Chapter 4, we may expect under drier future conditions that lakes with smaller WA/LA are the first to decrease in surface area, while under wetter conditions lakes with larger WA/LA may be more vulnerable to rapid drainage as they experience greater increases in lake level. These hypotheses could be tested with multi-decadal remote sensing time series and observing how the lake surface area of lakes with different WA/LA behave during wetter and drier periods. Future scenarios for lake water balances may also be investigated with hydrological models, as the capability of models for representing relevant cold-regions processes such as snow redistribution and ground subsidence after permafrost thaw is improving.

Convocation Year


Convocation Season