Lin Pan ^*

• Shanghai Publish and Print College, Shanghai, China

China is obviously affected by the East Asian monsoon, with frequent rainstorm activities. The distribution of precipitation is extremely uneven. Landslides, droughts, floods and other natural disasters are closely related to precipitation. For many years, the government has attached great importance to the flood problem. Therefore, this paper studies the fusion scheme of stochastic forest algorithm based on satellite remote sensing and ground station precipitation observation. Firstly, a stochastic forest algorithm fusion scheme based on satellite remote sensing and ground station precipitation observation is constructed, which combines daily and monthly precipitation products; then, based on the comprehensive precipitation products, the spatial and temporal characteristics of long-term precipitation in Guizhou Province are analyzed, and its differentiation rules are revealed. Finally, the hourly precipitation data and radar data of surface rainfall stations are selected to analyze the spatial and temporal of small-scale precipitation under complex terrain conditions. By studying the characteristics of the precipitation environment in mountain areas, the dynamic visualization of the spatial and temporal of rainfall in the computer-generated 3D virtual environment is realized by combining the rainfall process with the 3D virtual simulation platform. Therefore, under the framework of machine learning algorithm and geostatistics theory, the multi-source precipitation data fusion model can be established using terrain, longitude and latitude and other auxiliary variables, so as to improve the accuracy of remote sensing precipitation and effectively prevent natural disasters.