Modelling spatially distributed snowmelt and meltwater runoff in a small Arctic catchment with a hydrology land‐surface scheme (WATCLASS)

Abstract

The fully distributed hydrology land‐surface scheme WATCLASS is used to simulate spring snowmelt runoff in a small Arctic basin, Trail Valley Creek, dominated by open tundra and shrub tundra vegetation. The model calculates snowmelt rates from a full surface energy balance, and a three‐layer soil model is used to simulate the infiltration into and the exchange of heat and moisture within the ground. The generated meltwater is delivered to the stream channel network by overland flow, interflow, and baseflow and subsequently routed out of the catchment. Subgrid spatial variability is handled by the model through the use of grouped response units (GRUs). The GRUs in WATCLASS are chosen according to vegetation land cover. Five spring snowmelt periods with a variety of initial end‐of‐winter snow cover and melt conditions were simulated and compared with observed runoff data. In a second step, the model's ability to simulate spatially variable snow covered area (SCA) within the basin was tested by comparing model predictions to remotely sensed SCA. WATCLASS was able to predict runoff volumes (on average within 15% over five years of modelling) as well as timing of snowmelt and meltwater runoff for open tundra fairly accurately. However, the model underestimated melt in the energetically more complex shrub tundra areas of the basin. Furthermore, the observed high spatial variability of the SCA at a 1‐km resolution was not captured well by the model. Several recommendations are made to improve model performance in Arctic basins, including a more realistic implementation of the gradual deepening of the thawed layer during the spring, and the use of topographic information in the definition of land cover classes for the GRU approach.