Heterogeneous Dynamic Graph Convolutional Networks: A Deep Learning Framework for Spatiotemporal Flood Forecasting

Accurate and timely flood forecasting is crucial to mitigate the damage and loss of life caused by floods. Specifically, Multi-step flood forecasting is always difficult due to the constantly changing conditions of the surrounding environment. In this study, we propose a Heterogeneous Dynamic Temporal Graph Convolution Network (HD-TGCN) for flood forecasting. Specifically, we designed a Dynamic Temporal Graph Convolution Module (D-TGCM) to generate a dynamic adjacency matrix by incorporating a multi-head self-attention mechanism, enabling our model to capture the dynamic spatiotemporal features of flood data by utilizing temporal graph convolution operations on the dynamic matrix. Moreover, to reflect the impact of multiple meteorological and hydrological features on heterogeneity of flood data, we propose a novel approach that utilizes multiple parallel D-TGCM for processing heterogeneous graph data and implements fusion mechanism to capture varied flood patterns influenced by multivariable. Experiments conducted on a real dataset in Wuyuan County, Jiangxi Province, demonstrate that HD-TGCN outperforms the state-of-the-art flood prediction models in MAE, NSE, and RMSE, with improvements of 80.32%, 0.15%, and 73.99%, respectively, providing a more accurate flood forecasting method that will play a critical role in future flood disaster prevention and control.