In order to use the brightness temperature to set up a rainfall nowcasting model,this paper analyzed the brightness temperature data of microwave radiometer TP/WVP-3000 in Taihu area from January to April 2009.It is found that,at 4-hr before rainfall,the brightness temperature data of seven channels have obvious increasing trends.According to the biggest difference and the biggest standard deviation difference in brightness temperature between the rainfall case and the no rainfall case,this paper chooses the observed data from channel 1,5,and 6 as the forecast factor using the discrimination analysis to set up the rainfall nowcasting model.The time points in the discrimination analysis are 15 and 30—330 min(time interval is 30 min) before rainfall.The equation of 90 min before rainfall is decided as the final model function because of the highest self verification.Using the external data to test this model,the probability of detection is equal to 87%,the false alarm ratio is 24%,and the not alarm ratio is 13%.