Successes and challenges on the road to an operational mass and energy balance snow model (Invited)

In order to meet the requirements of operational forecasters, a distributed mass and energy-balance snow model must be computationally efficient, capable of ingesting and quality checking data in real-time, and efficiently and accurately distributing the incoming data stream. Distributed snow modeling products with respectively 50 and 100-meter resolution were delivered in near real-time to operational forecasters in the Tuolumne and Boise River Basins during the 2013 snow season. These applications were carried out on approximately 1,000,000 grid elements at a one-hour timestep. The Isnobal snow model included routines for vegetation and terrain effects on radiation and inhomogeneous snow accumulation rates due to wind. A full energy balance calculation was conducted at each grid element. While this in itself represented a major modeling accomplishment, a lack of representative meteorological data had a direct bearing on model accuracy. Ground-based observations are sparse in mountainous regions. Wind and humidity data are particularly lacking. The spatially disperse data accentuated the impacts of unreliable observations. Can an observational network meet the demands of spatially explicit models or will model accuracy ultimately be tied to the advancement and accuracy of downscaled weather model output? Or will it ultimately be a combination of both?