High Arsenic Risk Distribution Prediction of Groundwater in the Hetao Basin by Random Forest Modeling

Abstract

BACKGROUND Arsenic pollution is a serious problem in shallow groundwater in the Hetao Basin, and has seriously affected the health of residents. The research on the distribution of high arsenic shallow groundwater in the Hetao Basin is limited by the sampling time and sample number. OBJECTIVES To obtain a comprehensive understanding of the risk distribution characteristics and important influencing factors of high arsenic groundwater in different seasons in the region. METHODS Based on 506 shallow groundwater samples and 9 surface environmental parameters as prediction variables, a random forest model was established to evaluate the importance of prediction variables and the impact of important variables on high arsenic groundwater. Taking the climate factors as the dynamic prediction variables, the probability distribution of high arsenic groundwater in different seasons was identified and thematic maps of risk areas were made. RESULTS The results showed that the arsenic content of 506 groundwater samples ranged from 0.05 to 916.7μg/L with an overshoot rate (>10μg/L) of 50%. Groundwater arsenic risk areas were mainly distributed in the depositional center of the Hetao Basin, but the area of groundwater arsenic risk areas decreased by 1907km2 in winter, accounting for 14.14% of the total area. Precipitation and drought index, influence of drainage and irrigation channels, potential evapotranspiration and temperature were the most important indexes affecting the high arsenic groundwater in this area. CONCLUSIONS In the Hetao Basin, climate variables (precipitation and drought index) are significantly correlated with arsenic accumulation in the aquifer, which controls the seasonal variation of groundwater with high arsenic content in the depositional center of the Hetao Basin.

Keywords

• groundwater
• arsenic pollution
• hetao basin
• random forest
• seasonal variation
• risk distribution