February 3, 2000

REQUIREMENTS: The National Weather Service Strategic Plan (2000-2005) includes the extension of weather forecasts to 7 days and the provision of forecasts in probabilistic terms as two performance measures in the area of expanding and improving public services. TDL is updating the MRF MOS package to provide objective guidance which can be used to assist HPC's medium-range forecast group and WFO forecasters in this effort.

NEW MRF MOS GUIDANCE: Several changes will occur in the new MRF MOS package. New definitions for the wind, cloud, and precipitation type elements have been made to increase the utility of the guidance. Also, for the first time, the medium-range MOS message will contain forecasts of categorical precipitation amounts, temperature, dew point and thunderstorms. In a manner similar to the currently operational MRF MOS, the new guidance will be available daily based on output from the 0000 UTC MRF. The message format will be expanded to include the new and revised weather elements and will be generated for over 1000 stations. A sample alphanumeric message is included in this package, and a draft Technical Procedures Bulletin which describes this message is available at . Eventually, the guidance will also be transmitted in binary formats. At the February 3 CAFTI meeting, results of testing the MRF MOS maximum/minimum temperature, 2-m temperature and dew point, probability of precipitation, and probability of precipitation amount will be presented. We plan to implement these elements by early April.

MOS APPROACH: In the MOS approach, predictand data (local hourly observations) are correlated to predictors derived from direct model output, geographic or climatic values, and observations. The model forecasts are interpolated to the location of the observing site before use as predictors. For this development, we used a new archive of MRF data established by TDL in April 1997. Initial and forecast fields for projections every 12 hours out to 192 hours generated from the 0000 UTC forecast cycle are saved on a polar stereographic grid with a resolution of 95.25 km at 60N. This resolution is substantially finer than that available when the current operational MRF MOS system was developed (381 km resolution at 60N). To extend the sample of forecasts available for our development, we augmented our archive with forecasts from NCEP's re-analysis archive. Data from January 1992 through March 1997 were available from the T62 resolution MRF every 5th day for projections of 12 to 192 hours after 0000 UTC. These data were archived on a 190.5 km grid, and add approximately 1 year to our MRF developmental sample.

TESTING: In the results discussed here, test equations were developed on a portion of our archived data, and tested on an independent sample withheld from the development. Test equations were developed for both cool (October - March) and warm (April - September) seasons, all projections, and a sample of stations. Comparisons of the test forecasts are made with the operational MRF MOS forecasts, direct model output, and a forecast based on climatic predictors or relative frequencies. The sample station lists and test results can be found at .

Figure 1. Sample 0000 UTC message.