Assessment of optimum snowmelt model complexity, Cariboo Highlands, British Columbia, Canada.

The goal of this study is to assess the performance of melt models of varying complexity to simulate snowmelt under different aspect, forest cover, and input data conditions in the British Columbia interior. Observed snow water equivalent data from the 2008 melt season were used to compare the performance of a basic temperature-index (TI) model, three modified TI models, and a simple energy balance model (EBM). The largest (smallest) values of NSE (RMSE) for the snow course sites were 0.81 (0.0243 m) for the TI models and 0.58 (0.0362 m) for the EBM. At the automatic snow pillow (ASP) the largest (smallest) values of NSE (RMSE) were 0.54 (0.0055 m) for the TI models and 0.65 (0.0048 m) for the EBM. At the snow course sites all TI models performed better than or equivalent to the EBM. At the ASP one EBM version performed better than the TI models. --P.ii.