A hybrid process-data driven framework for real-time hydrological forecasting with interpretable deep learning

doi:10.1016/j.jhydrol.2025.134082

A hybrid process-data driven framework for real-time hydrological forecasting with interpretable deep learning

2025-08-14

0

PRE

AI

DOI：10.1016/j.jhydrol.2025.134082

Original

Original request for help

Save

Abstract

En 中文

• The hybrid XAJ-LSTM model improves hydrological forecasting accuracy. • Bayesian inference enhances uncertainty quantification and risk assessment. • Shapley Additive Explanations improve model interpretability by revealing key variables. • The framework is adaptable to both process-based and data-driven models. • The hybrid model outperforms standalone models in accuracy and reliability.

Keywords:

hybrid XAJ-LSTM model

Bayesian inference

Shapley Additive Explanations

hydrological forecasting

uncertainty quantification

A hybrid process-data driven framework for real-time hydrological forecasting with interpretable deep learning

Abstract

Journal

Organization