Several heavy rainfalls in Changjiang River Basin are considered to be highly related to the moving out of the mesoscale convective systems (MCSs) from the Tibetan Plateau to the east. To discover the rules of MCSs moving and promulgating, the relationship is established between moving trace of MCSs and its surrounded environmental physical field. And the database of spatial data mining is designed to the MCSs moving out to east from Tibetan Plateau. In the database of spatial data mining, there are not only 18 attribute terms brought by 9 environmental physical variables, but also some spatial character terms of MCSs, such as area, position, shape, and intension of MCSs. By analyzing the cloud-top’s temperature of black body (TBB) from the hourly GMS infrared images and the data of the High Resolution Limited Area Forecast System (HLAFS) of the Tibetan Plateau from June to August in 1998, the database of spatial data mining is built. It is studied that the database is reduced using correlation analytical method. The study reveals that the height(H), the vortex (VOR), the divergent(DIV), the water vapor flux divergent（IFVQ）, the wind aplomb speed（W） and the K-index are relatively independent parameters among the nine environment physical variables around the MCSs, while the temperature（T）, the fake correspond potential temperature (θse) and the relative humidity(RH) are highly related to each other，as well as to the Height. Hence, we suggest that the temperature(T), the fake correspond potential temperature(θse) and the relative humidity(RH) be not considered for the construction of the data base of MCSs spatial data mining, so as to increase the efficiency of the MCSs spatial data mining and to reduce the data redundancy.