Abstract:Using the daily precipitation data of Gannan 2015-2021 automatic station, regional station, and the 2015-2021 geological disaster data, based on the precipitation data, the second generation display statistical early-warning model of geological hazards in Gannan Prefecture is established to study the possibility of geological hazards induced by dynamic precipitation, it can also provide technical reference for the prevention and control of geological disasters, and reduce the economic losses and casualties caused by sudden geological disasters such as collapse, landslide and debris flow, which are mainly induced by precipitation, is of great significance. Method 1: to quantitatively evaluate the geological background risk of Gannan Prefecture by using information quantity method, taking the effective precipitation before the disaster, the maximum and minimum precipitation on the same day, and the accumulated precipitation on the same day as the independent variables, by fitting different linear equations to different risk areas, the meteorological risk early-warning model of geological hazards in Gannan Prefecture was established. Method 2: take the slope, aspect, relief degree, stratum lithology, fault structure, river system and human engineering activity as evaluation indexes, and use machine learning method, by using BP neural network algorithm program and setting the learning step, the probability quantization value of geological hazard potential degree is obtained, and coupled with precipitation factor, the geological hazard meteorological risk early-warning model of Gannan Prefecture is established.