A scheme to produce high-resolution gridded snowfall estimates over the conterminous United States

For the casual weather consumer served by The Weather Channel, there is great interest in reports of actual snowfall amounts associated with specific storms, or integrated over longer timescales encompassing an entire winter season or an entire climatological year. Thus, it is desirable to have timely, accurate, frequent, and high-resolution estimates of snowfall for a domain that covers - at minimum - the Conterminous U.S. (CONUS).

Unlike rainfall, there is no automated snowfall measurement gauge or instrument in widespread use within the CONUS observation network, so nearly all surface observations of snow accumulation are provided manually by trained NWS employees, cooperative observers, or storm spotters. The reports themselves are often non-standard, may cover different time intervals, are often significantly delayed, and may have problems with quality control or consistency. This makes the collection, assimilation, and synthesis of a large domain of near real-time high-resolution snowfall information, based upon the loose confederation of manual snowfall reports, a very difficult task. We've found the results are typically unsatisfactory when performed on a large scale, or performed without a great deal of human oversight and intervention.

One proposed alternative to building gridded estimates of snowfall from manual reports is to transform the much denser, higher fidelity and more automated melted precipitation information compiled by NOAA/NWS into equivalent snowfall data. An example of these high-resolution precipitation fields is the so-called Stage IV gridded 24-hour precipitation produced by NOAA's River Forecast Centers (RFC) and mosaicked and published by the Hydrometeorogical Prediction Center (HPC) – a department within NCEP. Another alternative starting point is the 1-km gridded snowfall precipitation (in melted form) published by the National Operational Hydrologic Remote Sensing Center (NOHRSC) as part of their cryospheric analyses for CONUS.

The transformation of Stage IV melted precipitation to actual snowfall estimate is quite involved because Stage IV includes no information about precipitation type occurring over the 24-hour measurement period. In the case of NOHRSC, precipitation type is inherent in the format of the product - but the actual estimated melted precipitation amounts appear to be less accurate than those provided by the Stage IV analysis.

In this paper, we describe transformation methods that employ Stage IV and/or NOHRSC gridded melted precipitation estimates in determining daily, monthly, and seasonal snowfall estimates at high resolution across CONUS. We discuss the trade-offs involved with the two methods. Some operational results will be shown for each approach, and simple comparisons between the two resulting fields are also provided. We conclude with a discussion of future work needed to refine this synthetic field of snowfall information.