Enhancing the efficiency of geochemical quality assessment and prediction in Holocene aquifers in Central Vietnam

Abstract

This study used 2-D electrical resistivity tomography (ERT), exploratory drilling, and physico-chemical analysis combined with machine learning models to analyze the geological and groundwater distributions in the Holocene aquifers in Central Vietnam. The dataset included 534 sets and nine features from 2019–2022, and the groundwater quality index (GWQI) was computed based on eight inorganic elements. The best parameter set for modeling the GWQI was found using the pattern search algorithm in R language. The prediction performance of the models was determined using the root mean square error and mean absolute error metrics. The study found that some research areas demonstrated organic pollution and alum contamination, and these pollutants were found in shallow geological layers close to the ground (within the second geological layer). The model performance varied by district, with Cubist outperforming other models with an R2 value of 0.900 for the Son Tra District and 0.992 for the Lien Chieu District.