End-to-End Graph Neural Networks for Real-Time Hydraulic Prediction in Stormwater Systems

This under-review paper presents GNN-SWS, an end-to-end spatiotemporal graph neural network surrogate for rainfall-driven hydraulic prediction in urban stormwater systems. The model jointly learns rainfall-runoff generation and hydraulic flow routing, predicts both node-level and conduit-level states, incorporates physics-guided constraints, and uses autoregressive forecasting with the pushforward trick to improve long-horizon stability.

Recommended citation: Zandsalimi, Z., Taghizadeh, M., Lynn, S. L., Goodall, J. L., Shafiee-Jood, M., & Alemazkoor, N. (under review). "End-to-End Graph Neural Networks for Real-Time Hydraulic Prediction in Stormwater Systems." Hydrology and Earth System Sciences.