SWE-Fusion: Multi-Source, Real-Time Snow Water Equivalent Data to support water resource management in the Western US

Spatially distributed snow water equivalent (SWE) estimates have a variety of potential applications in water resources, land-surface hydrology, and ecosystem science. The past decade has seen an explosion in spatial snow-water data availability and an associated expansion of water management utilization of these data. Historical satellite-based observations of fractional snow-covered area now exceed 20+ years (e.g. from Landsat & MODIS). These data sets have enabled evaluations of real-time spatial SWE estimates in the context of the past. This is essential for water management which is often focused on evaluating current conditions as a percentage of long-term historical averages. This webinar presents the spatial characteristics of Western US SWE based on a SWE data fusion system in both a historical and real-time context. SWE-Fusion uses a Generalized Linear Model (GLM), which relates observed in situ SWE from snow stations (dependent variable) to independent variables that include spatial SWE patterns from historical Landsat-based SWE reconstructions (1985 – 2021), daily mean satellite-observed fSCA from MODIS, and 18 physiographic variables such as slope, aspect, elevation, distance to ocean, etc. Future directions for SWE-Fusion are discussed, including an experimental product that utilizes the same independent variables but using Machine Learning (ML) instead of GLM and using SWE data from forthcoming NISAR data as the dependent variable.