This paper investigates the non-stationary relationship between metrics of urban greenness and socio-economic conditions across Philadelphia, Pennsylvania, U.S.A. at multiple scales using Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) techniques. The analysis integrates environmental data on vegetation cover, using Normalized Difference Vegetation Index (NDVI) values derived from 30 m/pixel Landsat Thematic Mapper imagery, with land use data derived from the 2000 City of Philadelphia Licenses and Inspections Department Vacancy Survey and socioeconomic data at the U.S. Census Tract level from the 2000 U.S. Census. City-wide OLS, sub-city zone OLS and local GWR models were developed by regressing mean NDVI against three independent variables for each Census Tract: population density, median household income, and percentage of vacant lots. These models indicate that the strength and nature of the relationship among the variables varies spatially, with highly localized relationships not evident with the global regression models alone. Results suggest that while wealth is a strong predictor of vegetation vigor in some neighborhoods in Philadelphia, this relationship changes drastically across a heterogeneous urban environment. This paper contributes to the growing body of academic research on GWR and the urban ecology of dynamic urban human-environmental contexts.
All Science Journal Classification (ASJC) codes
- Geography, Planning and Development
- Environmental Science(all)
- Tourism, Leisure and Hospitality Management