Machine learning-powered geospatial mapping of groundwater quality and salinity: towards sustainable water management in Southern India

Groundwater quality degradation in semi-arid regions is a growing environmental and public health concern, particularly in areas with intensive agricultural and anthropogenic pressures. This study presents an integrated assessment of groundwater quality in Dindigul Taluk, Tamil Nadu, India, an area...

Celý popis

Uložené v:

Podrobná bibliografia
Vydané v:	Stochastic environmental research and risk assessment Ročník 39; číslo 11; s. 5131 - 5150
Hlavní autori:	Maluventhi, Saravana Kumar, Kulandaisamy, Prabakaran, Rajendran, Muthuramalingam, Balagurumoorthi, Dharshan Ragavendar, Veeramalai, Gopal
Médium:	Journal Article
Jazyk:	English
Vydavateľské údaje:	Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2025 Springer Nature B.V
Predmet:	Accuracy Agriculture Anthropogenic factors Aquatic Pollution Aquifers Arid climates Arid regions Arid zones Bicarbonates Chemistry and Earth Sciences Computational Intelligence Computer Science Cultural heritage Decision trees Dissolved solids Dolomite Drinking water Earth and Environmental Science Earth Sciences Electrical conductivity Electrical resistivity Environment Geographic information systems Geology Groundwater Groundwater chemistry Gypsum Hazard assessment Heterogeneity Human influences Hydrogeochemistry Intensive farming Ion exchange Learning algorithms Machine learning Mapping Math. Appl. in Environmental Science Original Paper Physicochemical properties Physics Probability Theory and Stochastic Processes Public health Regression Regression analysis Salinity Salinity effects Salinization Saturation index Semi arid areas Semiarid climates Semiarid zones Sodium Spatial heterogeneity Statistics for Engineering Support vector machines Sustainability management Total dissolved solids Uncertainty Waste Water Technology Water analysis Water chemistry Water Management Water Pollution Control Water quality Water sampling Middle East Tamil Nadu India China India Machine learning models Geospatial analysis TDS prediction Water quality index Groundwater salinity Hydrogeochemical processes
ISSN:	1436-3240, 1436-3259
On-line prístup:	Získať plný text
Tagy:	Pridať tag Žiadne tagy, Buďte prvý, kto otaguje tento záznam!

Popis
Shrnutí:	Groundwater quality degradation in semi-arid regions is a growing environmental and public health concern, particularly in areas with intensive agricultural and anthropogenic pressures. This study presents an integrated assessment of groundwater quality in Dindigul Taluk, Tamil Nadu, India, an area underlain by hard rock geology and subject to a tropical semi-arid climate. A total of 60 groundwater samples were collected during the post-monsoon season and analyzed for key physicochemical parameters, including pH, electrical conductivity (EC), total dissolved solids (TDS), and major ions such as sodium (Na + ), potassium (K + ), chloride (Cl − ), bicarbonate (HCO 3 − ), and fluoride (F − ). The research uses a mix of methods, including studying water chemistry, analyzing geographic data, applying statistics, and using machine learning (ML) models to find risks to water quality and forecast salt levels. Hydrogeochemical processes such as mineral weathering, ion exchange, and evaporation were found to significantly influence groundwater chemistry. Strong correlations were observed between TDS and Cl − (coefficient of determination, R 2 = 0.96) and between TDS and Na + (R 2 = 0.93), indicating salinization is driven by both natural and anthropogenic factors. Saturation Index (SI) calculations for minerals such as gypsum, halite, and dolomite provided information on dissolution–precipitation equilibria. Geospatial mapping revealed spatial heterogeneity in salinity risks, identifying zones vulnerable to groundwater quality deterioration. To reduce predictive uncertainty, four ML models—Multiple Linear Regression (MLR), Support Vector Machines (SVM), Decision Tree Regression (DTR), and Random Forest Regression (RFR)—were implemented. MLR showed the highest predictive accuracy (R 2 = 0.9997; Mean Squared Error, MSE = 332.14), followed by RFR. This study demonstrates a robust, integrative approach to assessing groundwater salinity hazards under environmental uncertainty. The results support evidence-based water management and emphasize the possible use of ML techniques to strengthen predictive groundwater quality monitoring in other vulnerable semi-arid regions worldwide.
Bibliografia:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14
ISSN:	1436-3240 1436-3259
DOI:	10.1007/s00477-025-03069-y