Machine learning surrogates for efficient hydrologic modeling: Insights from stochastic simulations of managed aquifer recharge

2024-07-30

Journal of Hydrology

AI Summary

High-fidelity hydrological modeling for Managed Aquifer Recharge (MAR) systems faces prohibitive computational costs, hindering uncertainty quantification and real-time decision support. To address this, we propose a novel multi-fidelity machine learning surrogate modeling framework integrated with stochastic MAR scenario generation. Our method synergistically combines stochastic hydrological simulation, Bayesian optimization, and hybrid surrogate architectures—XGBoost and neural networks—to enable collaborative modeling of heterogeneous, multi-source simulation data and adaptive fidelity scheduling. The resulting surrogate achieves an R² of 0.982 while reducing computational time by over 99%. This significantly accelerates MAR scheme evaluation, enhances risk quantification, and enables dynamic optimization. The framework establishes a scalable, generalizable paradigm for real-time decision support in complex hydrological systems.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Problem

• Enhancing hydrologic model computational efficiency
• Using ML surrogates for aquifer recharge simulations
• Comparing ML architectures for accuracy and speed

Innovation

• Machine learning surrogate models
• Hybrid modeling workflow
• Deep convolutional networks