中文摘要：

基于中国静止气象卫星FY-2C的多光谱数据，研究了由热红外亮温和可见光反照率拟合反演白天地面像素级逐时雨强分布的方法。利用降雨概率判识矩阵区分了降雨云和非降雨云，技术得分为0．5073，比简单利用阌值划分方法显著提高了雨区判识精度。对逐时降水率的分析，允许误差为±20％内，正确率为52．19％。对降水等级的分析，经1200多个实测有雨样本的检验，其对小雨、中雨、大雨和暴雨的实测命中率达62．79％。若将降水率小于0．5mm／h的微雨这种临界状态作无雨处理，则有雨样本的降水强度判识准确率达到88．17％。

英文摘要：

Based on the geostationary meteorological satellite FY-2C＇ s multi-spectral data of China, this paper studies the method using themlal infrared bright temperature and visible light albedo to imitate and demonstrate the pixel-level hourly rain- fall rate and rainfall intensity class. We firstly employ the rainfall probability analysis matrix to differentiate precipitation cloud and non-precipitation cloud with the skill score reaching 0. 507 3. This notably improves the accuracy of the rain area estimation compared with the threshold value distinguishing method. The accuracy of analyzing hourly rain rate with the method is about 52.19% if the deflection of ± 20% is permitted. And after the test of more than 1200 samples, the accuracy of this in- vestigation of small, middle and heavy rain and rainstorm reaches 62.79%. Furthermore, if we consider the tiny rain with its rain rate less than 0.5 mm/h as the non-rain case, the accuracy of the rainfall rate estimation can reach 88.17%.