Surrogate-Model-Based Simulation and Management of Seawater Intrusion: Progress, Challenges, and Outlook
Abstract Seawater intrusion (SWI) remains a major challenge in the management of coastal aquifers. Although high-fidelity variably-density flow and transport models can accurately represent SWI processes, their computational cost limits their use in multi-scenario optimization, uncertainty quantification, and risk assessment. To address this limitation, surrogate modeling approaches have increasingly been adopted, aiming to approximate the input-output relationships of complex SWI models at substantially lower cost. This paper reviews recent advances in surrogate models for SWI simulation and management and proposes a classification framework based on the degree of physics integration and the adopted dimensionality-reduction strategy. Existing approaches are categorized into five groups: data-driven models, physics-integrated models, multi-fidelity models, reduced-order models, and neural operators. Furthermore, the coupling strategies between surrogate models and pumping-optimization frameworks (e.g., genetic algorithms, particle swarm optimization, Bayesian optimization) are examined, together with their roles in supporting multi-objective pumping management and uncertainty analysis.
Authors
- Zhenxue Dai (ORCID: https://orcid.org/0000-0002-0805-7621)
- Mingxu Cao
- Mohamad Reza Soltanian (ORCID: https://orcid.org/0000-0002-5126-0668)
- Lulu Xu
- Hung Vo Thanh
- Shuai Wang
- Zihao Wang
Institutions
- Waseda University (JP)
- Jilin University (CN)
- Waseda Research Institute for Science and Engineering (JP)
- Qingdao University of Technology (CN)
- Changchun University (CN)
- University of Cincinnati (US)
Publication Details
- Journal
- Water Resources
- Published
- 2026-11-18
- DOI
- https://doi.org/10.1134/s0097807826602207
- Primary Topic
- Groundwater flow and contamination studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00