Physics-Informed Machine Learning for Short-Term Flood Prediction

In three linesA PIML framework incorporates hydrological constraints into an LSTM loss function for flood prediction. The Trend Alignment constraint penalizes inconsistencies between precipitation and discharge trends. On limited data (5%), the model achieves NSE=0.23 vs 0.20 for standard LSTM, with improved stability under extreme conditions.

Summary generated by Claude — human-verified