Regional monthly rainfall prediction method and system based on convolution UNet and transfer learning

The invention discloses a regional monthly rainfall prediction method and system based on convolution UNet and transfer learning, and the method comprises the steps: carrying out the correlation analysis of regional monthly rainfall and rainfall influence factors, screening prediction factors, and building a regional monthly rainfall prediction model based on a convolution UNet network; generating simulated monthly-scale meteorological data by using daily-scale meteorological data and adopting a 30-day sliding window, pre-training the regional monthly rainfall prediction model, then training the regional monthly rainfall prediction model by using observed actual monthly-scale data, and performing parameter fine tuning; the trained model is used for predicting regional monthly rainfall at a future moment; according to the method, meteorological conditions, landform features, underlying surface conditions and other factors influencing rainfall are fully considered, and simulated monthly scale meteorological dat