Assessment of groundwater quality in a highly urbanized coastal city using water quality index model and bayesian model averaging

Abstract

Prediction and assessment of water quality are important aspects of water resource management. To date, several water quality index (WQI) models have been developed and improved for effective water quality assessment and management. However, the application of these models is limited because of their inherent uncertainty. To improve the reliability of the WQI model and quantify its uncertainty, we developed a WQI-Bayesian model averaging (BMA) model based on the BMA method to merge different WQI models for comprehensive groundwater quality assessment. This model comprised two stages: i) WQI model stage, four traditional WQI models were used to calculate WQI values, and ii) BMA model stage for integrating the results from multiple WQI models to determine the final groundwater quality status. In this study, a machine learning method, namely, the extreme gradient boosting algorithm was also adopted to systematically assign weights to the sub-index functions and calculate the aggregation function. It can avoid time consumption and computational effort required to find the most effective parameters. The results showed that the groundwater quality status in the study area was mainly maintained in the fair and good categories. The WQI values ranged from 35.01 to 98.45 based on the BMA prediction in the study area. Temporally, the groundwater quality category in the study area exhibited seasonal fluctuations from 2015 to 2020, with the highest percentage in the fair category and lowest percentage in the marginal category. Spatially, most sites fell under the fair-to-good category, with a few scattered areas falling under the marginal category, indicating that groundwater quality of the study area has been well maintained. The WQI-BMA model developed in this study is relatively easy to implement and interpret, which has significant implications for regional groundwater management.