Quantification and Reduction of Spatial Scale-Induced Uncertainty

点击翻译按钮获取中文摘要

This research project will examine spatial scale-induced uncertainties and address issues involved in assembling multi-source, multi-scale data in a spatial analysis. Many spatial studies are compromised due to a discrepancy between the spatial scale at which data are analyzed and the spatial scale at which the phenomenon under investigation operates. One consequence has been that research findings often conflict when analyses are conducted at differing scales. Decision making and policy formulation therefore may be misguided by conflicting, biased research findings brought about by spatial scale. Lacking appropriate ways to deal with this spatial problem has made many existing findings less compelling or even invalid. In the era of big data with the advancement of spatial data collection technologies, data that were once difficult or impossible to obtain are now widely available at various spatial scales and are being used to study a variety of problems. Questions regarding spatial data thus become more pressing. The strategies to be developed for addressing the spatial scale issues have the potential to be applied to many fields and applications. The research results will benefit researchers and practitioners in processing, analyzing, and presenting multi-source, multi-scale spatial data. Numerous studies have been conducted to understand how issues related to scale influence analyses and interpretation. Despite the efforts, this remains a widely recognized, complex problem with few generalizable solutions. Even when studies are conducted at the appropriate spatial scale, uncertainty may exist because data are usually collected at different scales and therefore must be aggregated or interpolated to achieve the study scale. Using a large public health surveillance dataset, the investigators will develop a measurement error-based statistical framework to quantify the space scale-induced uncertainties and provide strategies to ameliorate the issues. The research will address issues of 1) characterization of scale due to spatial scale definition and data misalignment in the measurement error framework; 2) quantification of the effects of scale issues on the estimation significance and parameter biasedness; and 3) strategies that may be used to reduce uncertainties in a multi-scale analysis.