People, pixels and weights in Vanderburgh County, Indiana: toward a new urban geography of human–environment interactions

Abstract

This research examines the social-spatial dynamics of human–environment interactions in Evansville, Indiana, USA as well as the surrounding Vanderburgh County. Employing geographically weighted regression, this paper models the observed relationship between the Normalized Difference Vegetation Index (NDVI) and key sociodemographic parameters (housing value, median household income, percent of residents in poverty, population density and percent of population white). Further, this paper demonstrates that geographically weighted regression utilized within a GISci framework can be effectively used to visualize urban human–environment interactions and that the spatial distribution of environmental resources co-varies with socioeconomic conditions. Finally, the paper demonstrates that greenness indicators derived from remote sensing data can be used as proxy measures for observed sociodemographic variables.

Keywords:

• Geographic Information Science
• Urban geography
• Remote sensing
• Geographically weighted regression

Acknowledgements

The authors acknowledge the use of the software package GWR3.0 in preparation of this article.

Notes

¹‘Greenness’ is broadly conceptualized to include the urban forest, green space, and other types of land uses such as parks. While no doubt the specific role of each type of ‘green amenity’ in an urban system is distinct, this paper deals with urban environmental conditions only as generally assessed by NDVI.

²While the paper utilizes the principles of the expansion method terminal models only are presented in the interest of economy and brevity.

³See Fotheringham (1997, 1998, 1999) and Fotheringham et al. 2002) for a full discussion on GWR's capabilities and limitations.

⁴Quintiles are used in both map sets.