Multi-source data fusion for precision prevention and control of groundwater pollution in industrial brownfields

Groundwater contamination at industrial brownfields is governed by interacting sources, hydrogeochemical processes, aquifer structures, and land-use context, making concentration-based assessment insufficient. We developed a data-driven, multi-source fusion framework based on latent class analysis for a large Chinese industrial brownfield. Using 379 monitoring wells, the framework integrated land-use patterns (L), hydrogeochemical patterns (H), groundwater vulnerability (V), contaminant mobility (M), and comprehensive pollution grade (P) to delineate prevention-and-control zones via statistical clustering. Mn and NH₃-N were the principal contaminants. The area was divided into five management zones: key monitoring (15.14%), protection and conservation (23.89%), transitional control (25.87%), critical control (15.73%), and urgent remediation (19.37%). Key monitoring zones exhibited low pollution but high mobility and vulnerability. Transitional zones faced mixed land-use pressures. Critical control zones require migration interception, while urgent remediation zones demand immediate containment and targeted remediation. By integrating contaminant occurrence, migration risk, environmental sensitivity, and management demand, this framework shifts groundwater risk assessment from identifying pollution intensity toward formulating differentiated control strategies.