Land Use Patterns and River Nitrate Dynamics in Karst Regions: Insights from High-Resolution Sentinel-2 Imagery and Partial Least Squares Structural Equation Modeling Analysis

Abstract

The effects of land use patterns on river nitrate ( ${\text{NO}}_3^{-}$ and ${\text{δ}}^{\text{15}}{\text{N-NO}}_3^{-}$ ) in the karst regions vary significantly, yet the underlying mechanisms remain inadequately understood. This study focused on the Lijiang River Basin in southern China, a typical karst river system, utilizing supervised classification based on Sentinel-2 remote sensing image data to obtain high-resolution land use data from 2016 to 2018. Partial Least Squares Structural Equation Modeling was used to investigate the relationship between land use patterns and river nitrogen concentrations. The results indicated that nitrogen concentration in the Lijiang River ranged from 0.59 to 23.69 mg L⁻¹, with isotopic ( ${\text{δ}}^{\text{15}}{\text{N-NO}}_3^{-}$ ) values varying from −3.71‰ to 18.03‰. A strong correlation was found between land use and river nitrogen concentrations. Comprehensive analysis revealed that river nitrate concentration is positively correlated with construction land and cultivated land, whereas it is negatively correlated with forest-grassland and water areas. Among the three quantitative indicators—area, area percentage, and inverse distance weighted area ratio—the last demonstrated the best predictive capability. In addition, land use was affected by gross domestic product, population, and slope, which in turn influenced the levels of river nitrate. This study elucidates the impact of different land use types on river nitrogen concentrations and identifies the “source” and “sink” of river nitrogen in karst areas under the influence of integrated land use, emphasizing the significance of considering the proximity of land parcels to the river. These insights are crucial for land use planning and water resource management.