Using Data Assimilation to Extend Snow Observations From a Snow-Focused Satellite into Forested Grid Cells

Seasonal snow provides water for agriculture, local ecology, hydropower, and municipal water supplies. However, global snow water equivalent (SWE) is difficult to model, and in-situ observations of SWE are sparse. This suggests that the best estimates of global snow volume and distribution would likely come from an approach that combines models and observations, including observations that could come from a future snow-focused satellite. Here, we test this using an Observation System Simulation Experiment (OSSE), assessing the degree of improvement to modeled SWE that comes from assimilating synthetic observations representative of a Synthetic Aperture Radar (SAR) snow satellite. We perform these tests at 250 m spatial resolution across the Western United States and portions of Canada in water year 2019, focusing on regions where the accuracy of SWE remote sensing retrievals may be obstructed by forest canopies. We perform two different data assimilation (DA) experiments: 1) a simulation excluding synthetic SWE observations in forested regions, and 2) a simulation inferring snow distribution in forested regions using synthetic SWE observations from the nearest unforested grid cells. Our results show that assimilating synthetic SWE observations improves average SWE biases at peak snowpack timing in shrub, grass, crop, bare-ground, and wetland land cover types from 14% to within 1% for both DA simulations. However, the DA simulation inferring forest snow distribution using the nearest unforested grid cells significantly improves average SWE biases from 150% to 18%, and SWE mean absolute error from 111 mm to 27 mm. These SWE improvements persist into the snowmelt season, reducing spring SWE biases by 62% and improving the Nash Sutcliffe Efficiency of Runoff in the Upper Colorado River basin from -2.59 to 0.22. These results demonstrate the combined value of SWE remote sensing and land surface models, even in densely forested landscapes.