INTRODUCTION: At the August 1999 CAFTI meeting, TDL presented results of testing newly developed AVN-based MOS equations to predict max/min temperature, 2-m temperature, 2-m dew point, surface (10-m) wind speed, and surface wind direction for stations in the contiguous U.S., Alaska, Hawaii, and Puerto Rico. Final verification results are available on the TDL home page at:

At the February CAFTI meeting, we will review verifications presented previously, and show results of testing new AVN MOS cloud, ceiling height, and precipitation type forecast equations. Verifications of these latter elements focus on the probabilistic guidance. Categorical forecasts produced from the probabilities have not yet been evaluated. The new AVN MOS alphanumeric message will be discussed. A sample message is attached to this discussion, and a draft Technical Procedures Bulletin describing the message may be found at: TDL is now requesting permission to implement the temperature, dew point, wind, cloud, ceiling height, and precipitation type forecast equations.

NEW AVN MOS GUIDANCE: Development of a new AVN MOS package is underway for max/min temperature, 2-m temperature and dew point, wind speed and direction, cloud amount, ceiling height, probability of thunderstorms and severe thunderstorms, precipitation type, probability of precipitation, and probability of precipitation amount. Current plans are to develop a complete package of AVN MOS guidance from both the 0000 and 1200 UTC runs of the model, as well as a limited package of guidance from the 0600 and 1800 UTC runs. The complete guidance package will be analogous to the current NGM MOS guidance in terms of weather elements and forecast intervals with many of the forecasts being valid at 3-h intervals. AVN MOS guidance will, however, be available for projections out to 72 hours after 0000 or 1200 UTC and will be generated for over 1000 stations in the U.S. and Puerto Rico.

TESTING: In the results presented to CAFTI, tests are conducted on an independent sample of data after equations are developed from a dependent sample. In general, test forecasts are made for both the 0000 and 1200 UTC AVN cycles, all projections, all seasonal stratifications, and for a reasonable sample of the stations for which the guidance is to be developed. Comparisons of test forecasts are made with the NGM MOS or current AVN MOS (if appropriate) guidance packages, direct numerical model output, or a forecast based on climatic means or relative frequencies. A variety of verification measures are used to evaluate the accuracy and skill of the MOS guidance. The results are stratified by region (9 in total: CONUS and Alaska combined, Alaska, 6 regions in the CONUS, and HI/PR). Results are graphed and placed on the TDL home page listed above.

FINAL EQUATION DEVELOPMENT: The equations that are implemented operationally are developed from all available data, and are not the equations tested on independent data.

IMPLEMENTATION PLANS: Our plans are to implement the 0000 and 1200 UTC AVN MOS guidance on NCEP's IBM-SP in April 2000. The alphanumeric message will be implemented first; eventually the entire suite of guidance will be disseminated in binary (BUFR) format.

Figure 1. Sample 0000 UTC message.