Dual-constraint physics-informed LSTM-XAJ hybrid model for 10-minute flood forecasting in small-to-medium basins: A case of the Xiang River Basin, China
- 1School of Hydraulic and Ocean Engineering, Changsha University of Science and Technology, Changsha 410114, China; Key Laboratory of Dongting Lake Aquatic Eco-Environmental Control and Restoration of Hunan Province, Changsha 410114, China; Corresponding author at: School of Hydraulic and Ocean Engineering, Changsha University of Science & Technology, Changsha 410114, China.
- 2School of Hydraulic and Ocean Engineering, Changsha University of Science and Technology, Changsha 410114, China; Key Laboratory of Dongting Lake Aquatic Eco-Environmental Control and Restoration of Hunan Province, Changsha 410114, China
- 3Power China Zhongnan Engineering Corporation Limited, Changsha 410019, China
Abstract
Study region: Luolingqiao (326 km²) and Wenjiashi (164 km²), two small-to-medium sub-basins of the Xiang River Basin in Hunan Province, China. Study focus: Reliable flash flood forecasting in small-to-medium basins (SMBs) requires both high temporal resolution and physically constrained long-lead prediction. This study develops a dual-constraint physics-informed framework that couples the Xin'anjiang (XAJ) model with an encoder-decoder LSTM to produce 10-minute runoff forecasts at lead times of up to 6 h, forced exclusively by quality-controlled X-band radar quantitative precipitation estimation (QPE). XAJ-derived physical trajectories are supplied to the LSTM decoder, and a dual-constraint loss penalizes deviations from both observed discharge and XAJ-simulated runoff. The framework is evaluated on 28 flood events under blocked chronological cross-validation, in hindcast mode with observed QPE over the forecast horizon. New hydrological insights for the region: At the 6-h lead time, the hybrid model significantly outperforms the standalone XAJ model: relative to XAJ, NSE increases by 20%, RMSE decreases by 25%, and peak error decreases by 46%. Continuous evaluation from 1 to 6 h reveals a crossover near 3 h: the LSTM dominates at shorter horizons, whereas the hybrid model becomes optimal beyond 3–4 h, where physical regularization suppresses recursive error accumulation. Coupling process-based constraints with deep learning at 10-minute resolution is therefore feasible and valuable for flash flood early warning in SMBs with concentration times shorter than 3 h.
Related
- FAST-ML: A Hybrid Physics-Machine Learning Framework for Tropical Cyclone Intensity Forecasting
- A physics guided hybrid modeling framework for rainfall runoff forecasting in data scarce regions
- Hybrid machine learning framework for drought forecasting: integration of LSBoost, RQNN, CNN-RNN, CNN-LSTM and advanced data decomposition
- Is Seismic Forecasting Possible with Physics-based AI?
- Scaling-law-informed neural point processes for earthquake sequence forecasting
BibTeX
@article{wang2026dualconstraint,
title = {{Dual-constraint physics-informed LSTM-XAJ hybrid model for 10-minute flood forecasting in small-to-medium basins: A case of the Xiang River Basin, China}},
author = {Xinkui Wang and Changbo Jiang and Yuannan Long and Xinyi Song and Xingyi Zhang},
journal = {Journal of Hydrology: Regional Studies},
year = {2026},
month = nov,
doi = {10.1016/j.ejrh.2026.104183},
url = {https://doi.org/10.1016/j.ejrh.2026.104183},
note = {CC BY},
}
RIS
TY - JOUR
TI - Dual-constraint physics-informed LSTM-XAJ hybrid model for 10-minute flood forecasting in small-to-medium basins: A case of the Xiang River Basin, China
AU - Xinkui Wang
AU - Changbo Jiang
AU - Yuannan Long
AU - Xinyi Song
AU - Xingyi Zhang
PY - 2026
DA - 2026/11/05
JO - Journal of Hydrology: Regional Studies
DO - 10.1016/j.ejrh.2026.104183
UR - https://doi.org/10.1016/j.ejrh.2026.104183
AB - Study region: Luolingqiao (326 km²) and Wenjiashi (164 km²), two small-to-medium sub-basins of the Xiang River Basin in Hunan Province, China. Study focus: Reliable flash flood forecasting in small-to-medium basins (SMBs) requires both high temporal resolution and physically constrained long-lead prediction. This study develops a dual-constraint physics-informed framework that couples the Xin'anjiang (XAJ) model with an encoder-decoder LSTM to produce 10-minute runoff forecasts at lead times of up to 6 h, forced exclusively by quality-controlled X-band radar quantitative precipitation estimation (QPE). XAJ-derived physical trajectories are supplied to the LSTM decoder, and a dual-constraint loss penalizes deviations from both observed discharge and XAJ-simulated runoff. The framework is evaluated on 28 flood events under blocked chronological cross-validation, in hindcast mode with observed QPE over the forecast horizon. New hydrological insights for the region: At the 6-h lead time, the hybrid model significantly outperforms the standalone XAJ model: relative to XAJ, NSE increases by 20%, RMSE decreases by 25%, and peak error decreases by 46%. Continuous evaluation from 1 to 6 h reveals a crossover near 3 h: the LSTM dominates at shorter horizons, whereas the hybrid model becomes optimal beyond 3–4 h, where physical regularization suppresses recursive error accumulation. Coupling process-based constraints with deep learning at 10-minute resolution is therefore feasible and valuable for flash flood early warning in SMBs with concentration times shorter than 3 h.
SN - 2214-5818
ER -