Characterizing the spatial variations of the relationship between land use and surface water quality using self-organizing map approach

Characterizing surface water quality is important for water resources management and protection. Numerous studies have addressed the fact that water quality is often impacted by land use, but the relationship between the two remains obscure. This paper explores the spatial dynamics of the relationship between land use and surface water quality and offers new insights into water pollution control practices. A self-organizing map (SOM) method and Pearson correlation testing were adopted to evaluate the associations between land use and surface water quality in Zhejiang Province, China. Selected water quality indicators included: dissolved oxygen (DO), permanganate index (CODMn), total phosphorus (TP), and ammonia nitrogen (NH3-N). All the sampling sites were classified into four groups (Group A, B, C, and D) according to water quality using the SOM approach. Correlations between 16 land use parameters and the four water quality indicators were calculated at both the sub-watershed and 500m buffer zone scales. The results showed that surface water quality was closely related to percent land use type, but was weakly related to spatial pattern (i.e., configuration) metrics. For percent land use type, the sub-watershed scale appeared to be more important than the 500m buffer scale when explaining water quality variability. For landscape pattern parameters, buffer zone was superior to the sub-watershed for interpreting adjacent water quality. The correlation analysis for SOM clustered groups showed that land use-water quality relationships varied over space, suggesting that poorer water conditions had less significant correlations. The findings and the method used in this study can help improve water management as well as land use planning.