Improving Snowfall Accumulation Forecasting in Colorado

Forecasting snowfall accumulation is challenging anywhere, but the topographic diversity of Colorado makes it a particularly complex task for meteorologists at the Boulder Weather Forecast Office. The County Warning Area (CWA) includes much of north central Colorado, where the precipitation reporting sites used in this study range in elevation from around 3,500 to over 10,000 feet. This study identified snowfall events that occurred from November 2017 through February 2018 and analyzed the skill of three forecast models and two model blends in predicting snow amount with a 24-hour lead time: GFS, NAM12, ECMWF, SuperBlend, and the National Blend of Global Models (NBM). SuperBlend uses a combination of the NAM, NAM nest, GFS, and Canadian models, and National Blend adds the HRRR and RAP to that list.

Of the 47 verification sites used for observational data, 30 are in the plains region of the state at elevations lower than 7,000 feet. The remaining 17 sites are in the mountain region at elevations above 7,000 feet. Sites were grouped into more specific regions for analysis based on topographic similarity, and departures and errors were then calculated to compare the performance of the five models both site-by-site and regionally. Overestimations and underestimations of snowfall by each model at each site were mapped and geographic trends were identified. Forecasters can use the results of this study during upcoming snow seasons to better inform their decisions about which models can be used with higher confidence in different regions of Colorado.