A New Methodology of Spatial Crosscorrelation Analysis

TL;DR

This paper introduces a new theoretical framework and models for spatial crosscorrelation analysis, including global and local coefficients, and demonstrates their application to China's urbanization and economic development.

Contribution

It develops the first comprehensive models and methods for spatial crosscorrelation analysis, expanding beyond traditional autocorrelation techniques.

Findings

Defined global and local spatial crosscorrelation coefficients

Proposed scatterplots to visualize relationships

Applied methods to China's urbanization and economic data

Abstract

The idea of spatial crosscorrelation was conceived of long ago. However, unlike the related spatial autocorrelation, the theory and method of spatial crosscorrelation analysis have remained undeveloped. This paper presents a set of models and working methods for spatial crosscorrelation analysis. By analogy with Moran's index newly expressed in a spatial quadratic form and by means of mathematical reasoning, I derive a theoretical framework for geographical crosscorrelation analysis. First, two sets of spatial crosscorrelation coefficients are defined, including a global spatial crosscorrelation coefficient and a set of local spatial crosscorrelation coefficients. Second, a pair of scatterplots of spatial crosscorrelation is proposed, and different scatterplots show different relationships between correlated variables. Based on the spatial crosscorrelation coefficient, Pearson's correlation coefficient can be decomposed into two parts: direct correlation (partial crosscorrelation) and indirect correlation (spatial crosscorrelation). As an example, the analytical process is applied to the relationships between China's urbanization and economic development. Spatial crosscorrelation and spatial autocorrelation can complement one another, and the spatial crosscorrelation scatterplots can be used to reveal the causality inside a self-organized system. The spatial crosscorrelation models will play a useful role in future geographical spatial analysis.