Anthropogenic Interference in Nitrate-Based Validation of Intrinsic Groundwater Vulnerability: Evidence from Covered and Buried Karst Aquifers

Groundwater vulnerability assessment (GVA) in northern China karst regions remains challenging because existing models are often developed for exposed karst systems, whereas nitrate-based validation may be biased by anthropogenic disturbance. Using the covered–buried karst system of Tai’an, China, this study developed a modified concentration of flow–overlying layers–precipitation–karst development (COPK) model and reassessed the reliability of NO3−-based validation. The model incorporated region-specific hydrogeological controls, including aquifer media and productivity, groundwater depth, pore-aquifer productivity, faults, karst subsidence, karst development degree, and karst type. Compared with the original COPK model, the modified model produced a more spatially coherent groundwater vulnerability (GV) pattern, with very high and high GV areas accounting for 10.01% and 20.15% of the study area, respectively. Two complementary screening approaches, full-background (FB) and pollution-source-density (PSD) screening, were applied to reduce anthropogenic interference in the validation dataset. Direct validation using 61 karst groundwater samples yielded Pearson correlation coefficients of 0.344 and −0.268 for the modified and original COPK models, respectively. Following FB screening, the corresponding coefficients were 0.382 and −0.354; following PSD screening, they were 0.537 and −0.383. These results suggest that the modified COPK model better represents intrinsic GV controls in northern China karst systems, while part of the mismatch between GV indices and NO3− observations may reflect anthropogenic interference. This study highlights the distinction between intrinsic GV and NO3− spatial prediction and cautions against using NO3−–GV correlation as the sole benchmark for model validation in anthropogenically disturbed aquifers.