Testing the Limited Area Fine Mesh Model for Probability of Precipitation Forecasting

Abstract

Experimental objective forecasts of probability of precipitation (PoP) were made and verified for a large number of United States cities in order to make preliminary tests of the usefulness of the limited area fine mesh (LFM) model. The Model Output Statistics (MOS) technique was used to derive the forecast equations for the winter and summer seasons. Predictors were selected from forecast output of the Primitive Equation (PE), Trajectory (TJ) and LFM models, and from the sine and cosine of the day of the year. Forecast equations were developed with data from only one winter and one summer season (?small sample equations?) from PE predictors only, TJ predictors only, LFM predictors only, and various groupings of these. The small-sample equations were compared on independent data with each other and with operational equations developed on a much larger data sample. It was found that the small-sample combinations that included the LFM gave better results than the operational equations in winter. However, in summer, all small-sample combinations were significantly worse than the operational equations. It was also found that adding the harmonic terms of the day of year consistently improved the forecasts. Based on these preliminary results, the LFM will be incorporated into the operational PoP equations starting in the winter of 1975?76. Further tests will be made when a larger developmental sample is available for the LFM.