Comparison of the National Water Model v2.0 snowpack model with surface snow observations for the 2020 water year

Abstract:

The National Water Model (NWM) is a hydrologic modeling framework that performs land surface and hydrologic modeling to simulate streamflow over the contiguous United States (CONUS). The current operational version (2.0) of the NWM lacks a snow data assimilation (DA) capability; consequently, errors and biases in model forcings and physics accumulate in modeled snowpack states throughout the winter months, impacting the accuracy of those states and the subsequent hydrologic forecasts, especially at times and places where snowmelt-driven runoff contributes significantly to streamflow. Development of a snow DA system for the NWM has begun, and two essential components of this system will be the ability to map spatially distributed (i.e., gridded) land surface model snowpack variables into observation space for comparison with ground-based and airborne snow observations, and an observation processing and quality control system for collecting and managing those observations. Prototype databases for fulfilling these requirements have been developed, and this presentation will describe those databases and the first working test of them: a comparison of modeled NWM v2.0 CONUS snowpack states with quality-controlled observations from the National Snow Analysis (NSA/SNODAS) for the 2020 water year. By comparing day-to-day changes in observed and modeled snow water equivalent throughout the winter, and through careful categorization and accounting of these changes, it becomes possible to distinguish between different sources of model departure from observations (e.g., bias and error when snow accumulation is observed vs. when snow ablation is observed), to compare the relative severity of these phenomena across space, and to describe how different categories of error and bias evolve throughout the winter season. This exercise is an opportunity to test our prototype database implementations while also providing valuable insights into model behavior. All these activities will help to maximize the effectiveness of future operational snow DA capabilities.