本研究利用貝氏方法反演海上颱風降雨率，首先由先驗機率分布(prior probability distribution) π(R)與條件機率分布(condition probability distribution) f(P|R)獲得後驗降雨機率分布(posterior probabilitydistribution) π(R|P)，最後藉由衛星微波觀測值估算海上颱風降雨。其中條件機率分布分別藉由雲模式與輻射傳送模式（radiative transfer model，RTM）求得，而先驗機率分布則由PR (Precipitation Radar)標準
降雨產品的近地面降雨強度獲得。本研究使用的RTM 所需各種物理參數的輸入是由天氣研究預報模式(The Weather Research and Forecasting, WRF)模式所提供。另外，我們利用RTM模擬亮度溫度(brightness temperature，Tb)及在降雨與晴空下的微波頻道極化差資訊轉換為削弱指數P，這指數具有隨著降雨強度增加而遞減，且不易受降雨飽和影響的優點。最後，利用TRMM/TMI 所觀測Tb 來計算削弱指數，進而利用此削弱指數來估算海上RR。本研究使用貝氏方法是估算出降雨強度的機率分布，再利用平均估算式將此機率分布轉換為單一RR，再以PR 所求得之RR 做驗證，結果顯示兩者相關係數在對流性降雨為0.68，層狀性降雨為0.75。此結果顯示結合數值天氣預報模式及衛星微波資料可有效估算海上颱風降雨。

This study is precipitation estimation for typhoons over ocean using Bayesian approach. At first, a prior probability distribution was obtained by observations from PR (Precipitation Radar), then a conditional
probability distribution was derived by microwave radiative transfer models (RTM), finally a posterior probability distribution of rainfall was calculated by combining a prior probability distribution and conditional probability distribution. The required inputs of atmospheric hydrometeors for RTM are from Weather Research and Forecast (WRF) model. In addition, we utilize an attenuation index with a difference of polarization brightness temperatures of between rainy and clear sky to estimate rainfall over ocean. This index has advantages of decreasing with rainfall rate and not being easily saturated by rainfall. The probability distribution of rainfall derived from Bayesian approach was used to estimate a single rainfall rate by Averaged Estimator (AVE) which was compared with the rainfall rate obtained from the PR. The correlation coefficients are 0.68 and 0.75 under the condition of convective and stratiform rainfall, respectively. The result shows that the model and microwave data can estimate precipitation effectively while the typhoon is over the ocean.