Water-Richness Zoning Technology of Karst Aquifers at in the Roofs of Deep Phosphate Mines Based on Random Forest Model

The development of fractures and conduits in karst aquifers and the strength of their water richness are key factors in determining whether a water intrusion will occur in a mine. In the phosphorus mining process, if the mining of water-rich areas is carried out, sudden water disasters can easily occur. Therefore, water-richness zoning of the karst aquifer on the roof of the phosphate mine is very important to protect against the incidence of water disasters in the mine. This paper proposes a random-forest-based partitioning model of the water richness of phosphate mine roofs in karst areas based on the random forest intelligence algorithm in machine learning. Taking a productive phosphate mine in southern China as a typical case, seven main assessment indicators affecting the water richness of the phosphate mine roof aquifer were determined. The proposed random forest model was utilized to determine the weight of each evaluation index, and the water richness of the karst aquifer on the roof of this phosphate mine was studied by zoning. The whole structure of the mine is highly water-rich, with strongly water-rich areas mainly concentrated in the central and northeastern part of the mine. The water-richness fitting rates (WFP) introduced for validation were all in agreement with the evaluation results, and the constructed model met the accuracy requirements. The study's findings can serve as a guide for mine design and water-disaster warnings in karst regions.