Prediction of Groundwater Resources Quality with a Hybrid Approach of Principal Component Analysis, Spatial Clustering, and RF and SVR Machine Learning Models Optimized with RSA and DSA Algorithms

Rafati, Maryam; Kardan Moghaddam, Hamid; Haddadi, Mohadeseh; Malekzadeh, Maryam

doi:10.22059/ije.2026.418026.1919

Prediction of Groundwater Resources Quality with a Hybrid Approach of Principal Component Analysis, Spatial Clustering, and RF and SVR Machine Learning Models Optimized with RSA and DSA Algorithms

Articles in Press

Document Type : Research Article

Authors

Maryam rafati ¹

Hamid Kardan moghaddam ²

Mohadeseh haddadi ¹

Maryam Malekzadeh ¹

¹ Department of Environment, NT.C., Islamic Azad University, Tehran, Iran

² Assistant Professor, Water Research Institute, Ministry of Energy Water Research Institute, Tehran

10.22059/ije.2026.418026.1919

Abstract

،The groundwater quality of Qazvin aquifer was investigated with a hybrid approach including principal component analysis, spatial clustering, groundwater quality index, and machine learning models. Using the PCA method, it was shown that the first two components explain 78% of the total variance of the data. The first component was interpreted as a factor of salinity, salt accumulation, and hardness, and the second component was interpreted as a factor related to pH and HCO₃. K-means clustering based on PC1, PC2 and UTM coordinates separated the wells into three quality zones including low, medium and high salinity. The average GQI was 83 and most of the wells were in the good class during the measurement period. The results of the saltier cluster indicated low values of the GQI quality index. In the modeling section, RF and SVR were used to estimate the GQI, and their hyper-parameters were optimized with RSA and DSA. The best performance was obtained for the DSA-SVR model, which had a coefficient of determination of 0.99, a mean square error of 0.15 and an absolute mean error of 0.1. The results showed that the proposed framework is an effective tool for analyzing, zoning and estimating groundwater quality.

Keywords

groundwater

GQI

PCA

spatial clustering

DSA

Subjects