Analyzing the Effects of Spatial Autocorrelation in Geospatial Databases

University Of Kentucky Research Foundation, Lexington KY

Investigators

Abstract

NATIONAL SCIENCE FOUNDATION GEOGRAPHY SPATIAL SCIENCES (GSS) PROGRAM ABSTRACT This research project will provide a better understanding of the spatial patterns of natural resources by analyzing the potential effects of spatial autocorrelation on the modeling and analyses of geospatial data. The investigators will provide new insights for improving geospatial modeling procedures by quantifying the influences of spatial autocorrelation on a wide range of environmental predictor-response variables. By analyzing and modeling multiple natural resources as they relate to a range of environmental factors across different ecosystems, the investigators will identify the interdependent relationships among soil, water, biodiversity, and environmental factors at multiple scales. Project findings will inform government agencies and policy makers by providing new information and approaches for improved prediction and management for high-demand but scarce natural resources. Outcomes from this spatial methodological framework will be useful in examining other ecological systems globally and potentially will serve as a springboard for establishing platforms for comprehensive management recommendations across diverse ecosystems. In order to model and predict spatial phenomena, spatial scientists rely on multiple geocoded data sets that contain information about physical and biological resources. Because of the interdependency among biophysical variables however, it is challenging to determine the relationships among the various the spatial processes and the predictor-response variables associated such phenomena. The investigators will focus their analyses on the potential effects of spatial autocorrelation on the modeling and interpretation of the distribution of natural resources across landscapes. They will analyze multiple global spatial data sets representing a variety of geographic and biophysical gradients, and they will develop a framework to better understand the different yet related geographic phenomena at multiple scales. Spatial eigenvector mapping and environmental modeling will be used to analyze the complex relationships among predictor and response variables. The quantification of the influences of spatial autocorrelation will be catalytic for the understanding of potential networks and spatial processes across a broad spectrum of natural resources and geographic phenomena.