According to the concept of optimum space-matching between ground rainfall intensity,I, and radar reflectivity factor,Z, for a certain Z-H relationship, the paper suggests a new method for quantitative radar observation of precipitation. The height at which radar observations of I based on the given Z-H relationship are most close to the collocated raingauge measurements is determined and, then, precipitation retrieval from radar data is based on the radar observations of I at this height. By means of multi-variable regression and stepwise regression analyses, a number of models are obtained for precipitation estimation with not only satellite IR imagery information such as cloud top temperature and its gradient but also satellite VIS imagery information. It has been found that precipitation depends on lots of factors of which cloud top temperature, albedo, and temporal gradient of cloud top temperature are most significant. It is concluded that models with more variables would result in better estimation of precipitation.