WMOD00564 - Laserfiche WebLink

Home Browse Search

Reclamation's NEXRAD Snow Accumulation Algorithmfor Estimating Snow Water Equivalent, by Arlin B. Super and Edmond W. Holroyd, ill, made at a GCIP conference held 17-18 May 1999, at the University of Maryland. The NEXRAD SAA (Snow Accumulation Algorithm) was developed for Level II data. It was thereby limited in application to past events using Level II data tapes and to a few real-time operations at those radar sites using the WDSS (Warning Decision Support System) computer software. Forecasters who were fortunate to have the use of the SAA appreciated its help in understanding regional snowfalls and were reluctant to yield the system to other users. Distribution of the algorithm to all northern (snowy) sites is waiting for future improvements to the NEXRAD hardware and software systems everywhere. In the interim, Reclamation wanted to study SAA performance in the Missouri and Mississippi River Basins. To do so, the SAA was modified to accept Level ill data in near real time from a NIDS (NEXRAD Information Dissemination System) vendor. The resolution of Level ill data is only 4 or 5 dBZ, depending on volume coverage pattern, while Level II data are at 0.5 dBZ resolution. Angular and range resolutions are the same, but no more than the first four tilts are available with NIDS data. Occultation and hybrid scan adjustments and the Ze-S relation remained the same as with Level II data. Use of the NIDS data did not seriously degrade the snow accumulation estimates. During the first winter of testing, SAA accumulations (S and SD products) were provided for five radars (Bismark, KBIS; Aberdeen, KABR; Grand Forks, KMVX; Minneapolis, KMPX, the required site; and Duluth, KDLH). Accumulation updates were provided hourly for all five radars according to the approximate 4 km HRAP grid specified by the National Weather Service (NWS) National Operational Hydrologic Remote Sensing Center (NOHRSC), in Chanhassen, MN. A five-radar composite was also available. For each site, S and SD products were made available via FTP and on the Internet for the past I-hr, 2-hr, 3-hr, and 6-hr periods ending at the top of the hour and for fixed 6-hr and 24-hr periods. The merged composite S and SD products were available for the past 3-hr and 6-hr periods ending at the top of the hour and for fixed 24-hr periods. The fixed 6-hr products ended at 00, 06, 12, and 18 UTC, and the 24-hr products ended at 12 UTe. The fixed 6-hr and 24-hr products were available for 7 days. Such products were made available for a half year but were suspended in May for the warm season. To examine the products, go to the Internet URL <http://www.usbr.gov/rsmg>, then select NEXRAD Snow Algorithm Products. The operational, real-time use of NIDS data revealed some areas of concern. Tuning the SAA to be most accurate in major snowstorms has resulted in a sensitivity to virga, which was being reported at far ranges as precipitation. Virga is recognized by rings of remote accumulations, with none at close ranges. Techniques for excluding virga are being refined and tested (see Section 7.4). The SAA, during a cold arctic airmass snowstorm (see Section 3.2), seriously underestimated snow accumulations because of the shallow nature of the storm and some microphysical considerations. Surface observations of snowfall throughout the areas of the five radars were collected for the entire winter, along with upper air soundings. Serious errors were produced by the network of cooperative gages in reporting snowfall and varying exposures to the wind, resulting in noisy data. Therefore, it was not possible to have precise surface data, especially in windy conditions, for performance verification of the SAA. Further testing is being conducted to see if the surface data, from sites inferior to the sheltered sites used for SAA development, are useful for verification of accumulations. 2