Water balance models calculate water storage and movement within drainage basins, a primary concern for many hydrologists. A Thornthwaite water balance model (H2OBAAS) has shown poor accuracy in predicting low flows in the Petawawa River basin in Ontario, so lake storage and winter snow processes were investigated to improve the accuracy of the model. Lake storage coefficients, represented by the slopes of lake stage vs. lake runoff relationships, were estimated for 19 lakes in the Petawawa River basin and compared on a seasonal and inter-lake basis to determine the factors controlling lake runoff behaviour. Storage coefficients varied between seasons, with spring having the highest coefficients, summer and fall having equal magnitude, and winter having the lowest coefficients. Storage coefficients showed positive correlation with lake watershed area, and negative correlation with lake surface area during summer, fall, and winter. Lake storage was integrated into the H2OBAAS and improved model accuracy, especially in late summer, with large increases in LogNSE, a statistical measure sensitive to low flows. However, varying storage coefficients with respect to seasonal lake storage, watershed area, and surface area did not improve runoff predictions in the model. Modified precipitation partitioning and snowmelt methods using monthly minimum and maximum temperatures were incorporated into the H2OBAAS and compared to the original methods, which used only average temperatures. Methods using temperature extremes greatly improved simulations of winter runoff and snow water equivalent, with the precipitation partitioning threshold being the most important model parameter. This study provides methods for improving low flow accuracy in a monthly water balance model through the incorporation of simple snow processes and lake storages.
Author Keywords: Lake Storage, Model Calibration, Monthly Water Balance, Petawawa River, Precipitation Partitioning, Snow Melt