Evaluating a hierarchy of snowmelt models at a watershed in the Canadian Prairies

Three semidistributed snowmelt models (SDSM) were developed and applied to the Paddle River Basin (PRB) in the Canadian Prairies: (1) A physics-based, energy balance model (SDSM-EBM) that considers vertical energy exchange processes in open and forested areas, and snowmelt processes that include liquid and ice phases separately; (2) a modified temperature index model (SDSM-MTI) that uses both near surface soil temperature (T-g) and air temperature (T-a); and (3) a standard temperature index (SDSM-TI) method using T-a only. Other than the regulatory'' effects of beaver dams that affected the validation results on simulated runoff, both SDSM-MTI and SDSM-EBM simulated reasonably accurate snowmelt runoff, snow water equivalent, and snow depth. For the PRB, where snowpack is shallow to moderately deep and winter is relatively severe, the advantage of using both T-a and T-g is partly attributed to T-g showing a stronger correlation with solar radiation than T-a during the spring snowmelt season and partly attributed to the onset of major snowmelt which usually happens when T-g approaches 0 degrees C. After resetting model parameters so that SDSM-MTI degenerated to SDSM-TI (the effect of T-g is completely removed), the model performance worsened, even after recalibrating the melt factors using T-a alone. It seems that if reliable T-g data are available, they should be utilized to model the snowmelt processes in a prairie environment, particularly if the temperature-index approach is adopted.