Abstract ID: 3.12428 | Accepted as Talk | Talk/Oral | TBA | TBA

Lateral redistribution can play a significant role in determining patterns of snow spatial variability, which in turn influences temporal signatures of snowmelt-driven runoff. However, processes influencing spatial variation operate at different scales, and their representation in a unified modeling framework is challenging. On the other hand, high-mountain experimental catchments provide a rich source of information on many aspects of the snowpack mass and energy fluxes and have led to conceptual improvements in many modeling contexts. However, reconciling observations from experimental catchments and large-scale hydrological models remains elusive. In this work we evaluate the performance of the Canadian Hydrological Model (CHM), a variable-resolution, physically based snowpack model, against different sources of information obtained at the Estero Las Bayas experimental catchment and at the Rio Yeso headwater basin, in the Andes Cordillera of central Chile. We explore the impact of different sources of uncertainty and modeling decisions on the model’s ability to reproduce the observed snowpack variability and suggest directions for the application of small-scale insights toward large-domain implementation. CHM snow cover (SCA) simulations were evaluated against remotely sensed SCA across two seasons in a local scale domain (~150 km2). CHM was found to be skillful at predicting peak-season SCA extent, although a consistent overestimation was observed, most notably during melt periods where simulated SCA depleted more slowly than satellite retrievals . This effect was more pronounced during snow drought conditions, where multiple melt cycles were observed during winter. This application also helped in identifying challenges regarding wind field generation modeling approaches, showing important wind speed distribution biases at the point measurement sites available.