中文摘要：

本文应用单位特征空间归类方法进行了MTSAT多光谱卫星图像降水强度场的反演研究.该方法将MTSAT多光谱卫星测值与2007年华东地区梅雨季节高密度地面站实测小时降水率结合起来,进行协同分析,较好的确定了各降水概率和降水强度等级在不同的二维光谱特征空间的分布特点.在此基础上,分别建立了相应的不同光谱组合的降水概率和降水强度类属矩阵用于全天时的连续降水监测：白天利用IR1（10.3~11.3μm）I、R3（6.5~7.0μm）、VIS（0.55~0.90μm）;夜间利用IR1、IR3、IR4（3.5~4.0μm）以及亮温差（BTD）资料,即IR24（IR4-IR2）来构建降水概率和强度类属矩阵.对比实测降水资料和反演结果,发现20%的降水概率可较好的划分降水区与非降水区;降水强度等级的分析也基本与实况有较好的匹配.但是,卫星进行的是每小时一次的瞬时观测,雨量计是一小时雨量的累计。这两种观测方式的不同是产生比较误差的重要原因.

英文摘要：

The retrieval of MTSAT multi-spectral satellite image rainfall intensity field was discussed by using ＂Unit-Feature Spatial Classification＂（UFSC） method.In this method,MTSAT multi-spectral satellite measured data and measured precipitation rate that is from high density ground stations of Mei-Yu peroid in east China in 2007 are combined to conduct the cooperative analysis,and the distribution features of precipitation probability and precipitation intensity are well established on different two-dimensional spectral feature spaces.On the basis,the genus matrices containing the information of precipitation probability and rain intensity derived from different spectral combinations are correspondingly established for simulation of the precipitation area and intensity at all time continuously during the daytime.The data IR1（10.3~11.3 μm）,IR3（6.5~7.0 μm）,and VIS（0.55~0.90 μm） are used to establish the precipitation probability and intensity genus matrixes for the daytime.Furthermore,IR1,IR3,IR4（3.5~4.0 μm） and brightness temperature difference（BTD）,i.e.,IR24（IR4-IR2） are used for the night.The contrast test between the observed data and the retrieval results shows that 20% precipitation probability can ideally distinguish precipitation area from non-precipitation area;and the analysis of rain intensity category also matches well with the actual situation.However,the monitoring of satellite is instantaneous once per hour while the rain gauge observation is an accumulative process during an hour.The difference between these two observation methods is the crucial cause for the relative error.