Prediction of Spatiotemporal Pollution of Soil Heavy Metal in Mining Areas Based on Grey Neural Network Algorithm

Abstract

The temporal and spatial prediction and early warning of soil heavy metal pollution are crucial for preventing and controlling soil environmental contamination and optimizing the utilization of regional soil resources. This study investigates the spatiotemporal prediction and early warning of soil heavy metal pollution in a lead–zinc mining area in Chifeng City, Inner Mongolia. Soil samples were collected at various depths and times across the mining area and its surroundings. A combination of BP neural network and grey prediction models was used to forecast the distribution of heavy metals, providing a basis for soil pollution control and remediation. The BP neural network model showed that As, Cu, Zn, and Cd concentrations exceeded the risk screening values set by the Soil Environmental Quality Risk Control Standard for Agricultural Land (GB15618-2018), with significant enrichment of As and Cd. Pb showed slight contamination. Spatial analysis indicated that contamination was most severe near the mine and decreased with distance and depth. Grey prediction results suggested that As and Cu levels in the mine restoration area would decline over the next three years, with Cu potentially falling below risk levels by 2024. However, As and Cu levels are expected to increase in surrounding agricultural and unremediated areas. The study concludes that the combined use of BP neural network and grey prediction models is effective for predicting and managing soil heavy metal contamination, supporting targeted remediation efforts in mining regions.