In situ measurements of ice growth and melt in two lakes of eastern Canada to improve ice representation in the Canadian Small Lake Model

In cold regions, lake ice, which persists for months, influences the exchange of heat, momentum, and water between the earth’s surface and the atmosphere. Several lake models have been developed for simulating these exchanges, such as the Canadian Small Lake Model (CSLM) - a one-dimensional physical lake model that represents the thermal structure of lakes and their exchange of energy with the atmosphere, including the formation and evolution of ice and snow cover. These models do not always adequately simulate important parameters such as ice thickness, which in turns makes model evaluation an important, yet difficult task. The main challenge with model evaluation is the limited amount of observational data, since most northern lakes are in remote areas with limited accessibility.

This study presents an analysis of the temporal evolution of ice cover in two northern lakes over two winters and compares in situ measurements with the CSLM.

Field data were collected using an innovative temperature profiler. During the 2024-2025 winter, two profilers were deployed on two lakes in the boreal biome of Quebec, Canada: Piché Lake (0.15 km²; mean depth 2.2 m; ~47°N), a small lake surrounded by topography that facilitates snow accumulation leading to snow-ice formation, and Bernard Lake (4.6 km²; 13 m; ~51°N), a larger, more wind-exposed lake where thermal ice formation is dominant. During the 2025-2026 winter, only one profiler was deployed at Piché Lake. Profiler data were processed to generate a continuous dataset of ice and snow thickness for comparison with the CSLM outputs. The simulation was validated through manual measurements of ice and snow thickness at both Bernard and Piché Lakes, as well as upward-looking sonar measurements at Bernard Lake. Additional information on snow and ice properties was collected using snow pits, ice core sampling and visual observations using a GoPro camera, providing a more comprehensive basis for assessing and improving the representation of snow and ice processes in the CSLM.

Preliminary comparisons with CSLM indicate that the model generally underestimates maximum lake ice thickness. Also, the model predicts freeze-up earlier than observed, while it predicts breakup either earlier or later than observed. The contrasting ice processes at Piché and Bernard lakes, characterized by dominant snow-ice and thermal ice formation, respectively, provide a useful basis for evaluating the performance of the snow-ice production module in the model. Planned comparisons of snow cover will refine the module and improve the model’s representation of snow and ice.

Overall, this work advances understanding of atmosphere–cryosphere interactions and provides recommendations to improve CSLM performance.