Spatio-Temporal Semivariograms as Neighborhood Definers



Journal Title

Journal ISSN

Volume Title



SPATIO-TEMPORAL SEMIVARIOGRAMS AS NEIGHBORHOOD DEFINERS. Brendan J. Hurley George Mason University, 2022 Dissertation Director: Dr. Timothy F. Leslie Spatial neighborhood definitions are a consistent source of disagreement among geographic scholars. This dissertation will focus on the implementation and evaluation of spatio-temporal semivariograms (STVs) as a source of spatial neighborhood definition. STVs show the similarity, measured by semivariance, of spatial events to each other when separated by time and space. Over both time and space, there should exist distances over which pairs of points become “independent” of one another. This dissertation seeks to answer two research questions in relation to STVs and their use as neighborhood definitions: (1) What data and process adjustments are necessary to implement STVs to provide neighborhood search definitions in time and space, and (2) How do spatio-temporal variograms perform when compared to other neighborhooddefinitions using spatially autocorrelative evaluation metrics? Given that there are many ways to define a neighborhood, STVs may provide a comprehensive method that uses the data themselves to inform the size and scope of neighborhoods, with the added advantage to evaluating both spatial and temporal axes at once. A well-defined neighborhood that accounts for temporal variation as well as spatial is a needed addition a limited set of tools designed for defining neighborhoods.



Semivariography, Spatial autocorrelation, Spatial neighborhoods, Spatial statistics, Spatio-temporal