Evaluating the Noah-MP Snow Depletion Curve Against Observed and Simulated Data in Western U.S. Basins

The snow depletion curve, which relates fractional snow-covered area (fSCA) to snow depth, plays a significant role in the snowpack and land surface energy budgets in Noah-MP and similar land surface models. Errors in fSCA based on a poorly-represented snow depletion curve propagate into errors in snow melt timing, as well as simulated evapotranspiration, vegetation green-up, and other variables. Historically, snow depletion curves have been empirically derived—often from relatively coarse remotely-sensed and/or in situ observations. Indeed, the snow depletion curve currently used in Noah-MP was derived from remotely-sensed fSCA at 1 degree spatial resolution and in situ snow depth data interpolated to 0.25 degree spatial resolution. Although the relationship between fSCA and snow depth is known to be scale dependent, this and similar formulations have not seen significant improvements that account for the finer spatial resolutions required for applications like snow and streamflow forecasting. Here, we derive empirical snow depletion curves from a number of modeled and observational sources at high spatial resolution and use these to evaluate the validity of the standard Noah-MP depletion curve. We leverage data from the Airborne Snow Observatory (ASO), high-resolution SnowModel simulations, SNODAS (Snow Data Assimilation System), SWANN (Snow Water Artificial Neural Network), and MODSCAG (MODIS Snow Cover and Grain Size) to develop snow depletion relationships over several basins in the western U.S. Basin selection was largely dictated by the availability of ASO data, which is limited to certain basins (e.g. Tuolumne River Basin, California; East River Basin, Colorado). Beyond resolution, we investigate factors influencing the shape and coherence of the snow depletion curve relationships, including snow regime, elevation, and land cover type.