Deep learning simulation and decision support system for groundwater salinity risk assessment in the lower Chao Phraya River Basin, Thailand
1
Issued Date
2025-10-15
Resource Type
eISSN
15732959
Scopus ID
2-s2.0-105018648372
Pubmed ID
41091255
Journal Title
Environmental Monitoring and Assessment
Volume
197
Issue
11
Rights Holder(s)
SCOPUS
Bibliographic Citation
Environmental Monitoring and Assessment Vol.197 No.11 (2025) , 1202
Suggested Citation
Heydarizad M., Liu Z., Pumijumnong N., Mohammadabadi H.G. Deep learning simulation and decision support system for groundwater salinity risk assessment in the lower Chao Phraya River Basin, Thailand. Environmental Monitoring and Assessment Vol.197 No.11 (2025) , 1202. doi:10.1007/s10661-025-14681-4 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/112688
Title
Deep learning simulation and decision support system for groundwater salinity risk assessment in the lower Chao Phraya River Basin, Thailand
Author(s)
Corresponding Author(s)
Other Contributor(s)
Abstract
Groundwater salinization poses a critical threat to freshwater security in coastal regions, particularly under intensified extraction and evolving hydroclimatic conditions. This study examines the spatial and temporal evolution of salinity in the lower Chao Phraya River Basin during 2008 and 2020 using a multi-method machine learning framework. SHAP-based feature attribution analysis identified groundwater extraction as the most influential driver of salinity dynamics. A Gaussian copula model was employed to quantify the conditional probability of salinity threshold exceedance under varying extraction pressures, capturing nonlinear dependence structures between total dissolved solids (TDS) and groundwater extraction. A Graph Neural Network (GNN) model was developed to simulate TDS concentrations at 212 monitoring stations, demonstrating high predictive performance across both periods. To translate model outputs into actionable insights, a scenario-based Decision Support System (DSS) was implemented, enabling interactive visualization of salinity risk zones under 20% and 40% increases in groundwater withdrawal. Results reveal a pronounced expansion of high-salinity areas over time, largely driven by anthropogenic factors. By fusing explainable machine learning with probabilistic analysis and decision support, this framework provides a novel, scalable tool for real-time groundwater salinity risk assessment and supports evidence-based management in data-scarce coastal aquifers.