Characterization of North America Snow Water Equivalent Uncertainty via Ensemble-Based Land Surface Modeling

Accurate analysis of the snow water equivalent (SWE) uncertainty is necessary for improving our understanding of the seasonal to decadal climate prediction systems as well as for planning future snow field and satellite missions. In this study, an ensemble-based land surface modeling approach and a latent factor analysis are applied to characterize the regions and the sources of high model uncertainty across North America. The NASA Land Information System (LIS) is used for conducting the model simulations across the domain over multiple winter snow seasons (2009-2017) at a 5 km spatial resolution. For ensemble-based land surface modeling approach, four different land surface models of varying complexity are driven using three different forcing datasets. The relatively coarse resolution meteorological inputs are then spatially downscaled using a mix of lapse-rate, slope-aspect, and climatology-based approaches. The Hydrological Modeling and Analysis Platform (HyMAP) in LIS is used to help evaluating of snow water budget by deriving estimates of streamflow. In this study, the latent factor analysis is applied to characterize the dominant factors that govern the spatial variability and seasonality of SWE uncertainty. The contributing role of various landscapes and precipitation regimes on the modeled SWE estimates will also be quantified. The results of the study are expected to guide the selection of sites for future snow field campaigns as well as provide useful guidance towards the planning of future satellite missions.