The snow level, or altitude in the atmosphere where snow changes into rain, is an important variable for hydrometeorological prediction in mountainous watersheds, yet, there is no operational performance measure associated with snow-level forecasts in the U.S. In order to establish a performance measure, it is necessary to first establish the baseline performance associated with snow-level forecasts. This paper evaluates the skill of snow level forecasts produced by the California-Nevada River Forecast Center by comparing gridded point freezing level forecasts with observed freezing levels estimated by vertically pointing Doppler radars operating at 2875 MHz (S-band). The evaluation occurred at two sites, one in the coastal mountains north of San Francisco, and one in the foothills of the Sierra Mountains in the American River Watershed. The evaluation was conducted for forecasts made during the winter wet season of 2005-2006. Statistical results of forecast performance will be presented.

An automated algorithm has been developed to detect the altitude of the radar brightband, here defined as the altitude of maximum radar reflectivity in the brightband layer. This altitude can be used as a proxy for the snow level partly because it always exists below the freezing level. The algorithm has been applied to nine years of S-band radar data collected in the coastal mountains of California. The time series is examined in context of the forecast evaluation described above, where the largest forecast errors were associated with atmospheric river conditions, which have been identified as the primary cause of flooding events in this region. A four-year winter season composite analysis indicates that brightband heights are highly correlated with the amount of precipitable water, underscoring the importance of snow-level forecasts for hydrometeorological prediction.