ESTIMATING THE PROBABILITY OF RAIN IN AN SSM I FOV USING LOGISTIC-REGRESSION/

The SSM/I has been used successfully to estimate precipitation and to determine the fields of view (FOV) that contain precipitating clouds. The use of multivariate logistic regression with the SSM/I brightness temperatures to estimate the probability that it is raining in an FOV is examined. The predictors used in this study are those that have bee n evaluated by other investigators to estimate rain events using other procedures. The logistic regression technique is applied to a matched set of SSM/I and radar data for a limited area from June to August 19 89. For this limited dataset the results are quite good. In one exampl e, if the predicted probability is less than 0.1, the radar data shows only 2 of 340 FOVs have precipitation. if the predicted probability i s greater than 0.9, the radar data shows precipitation in 748 of 774 F OVs. These probabilities can be used for both instantaneous and climat e timescale retrievals.