Multi-physics Data Assimilation Framework for Remotely Sensed Snowpacks to Improve Water Prediction

Snowpacks provide a natural storage mechanism of freshwater resources in the cold season, producing snowmelt runoff in the warm season. High uncertainty in snowfall amount predictions and high spatial variability are the primary challenges to quantify water availability in the snowpack. Airborne and field campaigns, like NASA SnowEx provide valuable snow measurements for data assimilation and to improve snow hydrology models with different complexity. And data assimilation of snow depth and snow water equivalent (SWE) retrievals from remotely sensed high resolution Synthetic Aperture Radar data (airborne campaigns, and future satellite missions) with snow hydrology models enables correcting mass errors as well as to introduce realistic spatial heterogeneity.

In this study, we present the development and demonstrate the application of a multi-physics data assimilation framework (MPDAF) over Grand Mesa in Colorado using SnowEx’17 data. Because of the complexity added by snow stratigraphy, we assimilate integrated quantities such as total snow depth and total SWE and use a repartition algorithm to vertically redistribute the increments. First, we use data assimilation runs with perturbed observed meteorological forcings and snow depth measurements at multiple Grand Mesa SnowEx sites to assess/improve the multi-layer snow hydrology model (MSHM) and the repartitioning algorithm for DA at multiple time-scales. Second, we transfer the snow hydrology model to a distributed spatial domain encompassing the Grand Mesa, and assimilate retrievals (snow depth and density) from airborne SnowSAR measurements in February 2017 data and evaluate the improvement in recovering snow cover spatial heterogeneity and associated vertical profiles using MPDAF. Finally, the transferability of the MPDAF to operations is explored.