Fig. 3: Performance comparison of graph neural networks on LamaH-CE dataset under different connectivity scenarios. | npj Natural Hazards

Fig. 3: Performance comparison of graph neural networks on LamaH-CE dataset under different connectivity scenarios.

From: Accelerating flood warnings by 10 hours: the power of river network topology in AI-enhanced flood forecasting

Fig. 3

a and b Illustrate the performance of six different GNNs on LamaH-CE. We compare the NSE trends of the models under three scenarios: dense adjacency, EA-LSTM, and isolated adjacency. Overall, the NSE across different discharge, dense adjacency significantly outperforming both isolated adjacency and EA-LSTM.

Back to article page