Multi-mode fusion rainfall dry-wet time sequence change reconstruction method and system

The invention discloses a multi-mode fusion rainfall dry-wet time sequence change reconstruction method and system, and belongs to the technical field of flood control and disaster reduction and hydro meteorological analysis, and constructs a double-layer deep learning fusion architecture model by fusing CMIP6 mode libraries, superposing rainfall event occurrence probability information and combining computer parallel computing and big data mining technologies, and finally, reconstructs the rainfall dry-wet time sequence change. And the dry-wet time sequence change and magnitude of climate mode rainfall are reconstructed. According to the method, the time-airborne downscale from large-scale grid precipitation to point-scale ground precipitation is achieved, the ground precipitation event capturing capacity of mode precipitation is effectively improved, the defect of insufficient output of an existing numerical mode is overcome, foundation support is better provided for watershed hydrometeorological forecast a