A new approach combining principal component factor analysis and K-means for identifying natural background levels of NO3-N in shallow groundwater of the Huaihe River Basin

Abstract

Establishing natural background levels (NBLs) of nitrate‑nitrogen (NO₃-N) is crucial for groundwater resource management and pollution prevention. Traditional statistical methods for evaluating NO₃-N NBLs generally overlook the hydrogeochemical processes associated with NO₃-N pollution. We propose using a method that combines principal component factor analysis and K-means clustering (PCFA-KM) to identify NO₃-N anomalies in three typical areas of the Huaihe River Basin and evaluate the effectiveness of this method in comparison with the hydrochemical graphic method (Hydro) and the Gaussian mixture model (GMM). The results showed that PCFA-KM was the most robust and effective for identifying NO₃-N anomalies caused by human activities. This method not only considers the data's discreteness but also combines the influencing factors of NO₃-N pollution to identify anomalies, thus avoiding the influence of non-homogeneous hydrogeological conditions. Moreover, 70 % of the identified anomalies were explained by sampling survey data, geochemical ratios, and pollution percentage indices, confirming the method's effectiveness and reliability. The upper limits of NO₃-N NBLs obtained by PCFA-KM were 12.97 mg/L (CUs-I), 4.42 mg/L (CUs-V), and 5.57 mg/L (CUs-VI). This study provides a new approach for NO₃-N anomaly identification, which can guide future NO₃-N NBLs assessments and pollution prevention and control efforts.