Spatially Varying Relationships Between Land Use and Water Quality Across an Urbanization Gradient Explored by Geographically Weighted Regression
Geography & Anthropology
Significant relationships between land use and water quality have been found in watersheds around the world. The relationships are commonly examined by conventional statistical methods, such as ordinary least squares regression (OLS) and Spearman’s rank correlation analysis, which assume the relationships are constant across space. However, the relationships often might vary over space because watershed characteristics and pollution sources are not the same in different places. This study applies an exploratory spatial data analysis (ESDA) technique, geographically weighted regression (GWR), to analyze the spatially varying relationships between six land use and fourteen water quality indicators across watersheds with different levels of urbanization in eastern Massachusetts, USA. The study finds that the relationships between water quality and land use and the abilities of land use indicators to explain water quality vary across the urbanization gradient in the studied watersheds. Percentages of commercial and industrial lands have stronger positive relationships with the concentrations of water pollutants in less-urbanized areas than in highly-urbanized areas. Percentages of agricultural land, residential land, and recreation use show significant positive relationships with the concentrations of water pollutants at some sampling sites within less-urbanized areas, whereas they have significant negative relationships at some sampling sites within highly-urbanized areas. Thus, the adverse impact of land use changes on water quality is more substantial in less-urbanized suburban areas than that in highly-urbanized central cities. Pollution control policies should be adjusted in different areas based on the spatially varying pollution sources and good predictors of water quality.
Tu, J. (2011). Spatially varying relationships between land use and water quality across an urbanization gradient explored by geographically weighted regression. Applied Geography, 31(1), 376-392.