Short-Term Intensive Rainfall Forecasting Model Based on a Hierarchical Dynamic Graph Network

Accurate short-term forecasting of intensive rainfall has high practical value but remains difficult to achieve. Based on deep learning and spatial–temporal sequence predictions, this paper proposes a hierarchical dynamic graph network. To fully model the correlations among data, uses dynamically constructed convolution operator spatial correlation, recurrent structure time architecture built with pooling extract fuse multi-level feature spaces. Experiments two datasets, based measured cumulative data at ground station in Fujian Province, China, corresponding numerical weather grid product, show that method can various more effectively than baseline methods, achieving further improvements owing reversed enhancement low-rainfall removal.