Groundwater quality model through the combined use of principal component analysis (PCA) and multiple linear regression (MLR) Case study: Maturín’s aquifers, Monagas, Venezuela

Abstract

A set of data collected during 2016 was studied in eight wells in the city of Maturin, Monagas, Venezuela, in an attempt to evaluate and determine the contributions of the source that affect water quality. An accurate multiple linear regression (MRL) technique was used as an advanced tool for modeling and forecasting groundwater quality. Likewise, principal component analysis (PCA) was used to simplify and understand the complex relationship between water and quality parameters. Six main components responsible for 86.57% of the total variation were found. It was found that the main source of contamination of Maturin aquifers was the residual discharges with high values ​​of fecal coliforms. An advanced receiver model was applied in order to identify the main sources of contamination. The result showed that the use of PCA as inputs improved the prediction of the MRL model by reducing its complexity and eliminating the collinearity of data, where the value of R2 in this study was 0.99, indicating that 99% of the variability of the water quality index (WQI) is explained by the eleven independent variables used in the model.