Measuring spatial association between different spatial layers is important in spatial data modeling. Traditionally, the relationship between variables can be measured by linear regression. The assumptions of those traditional methods are hard to meet in the spatial data. Also, the traditional statistical methods do not consider Tobler's First Law of Geography which is an important spatial data property. To address these drawbacks, I propose a spatial data association estimator (termed as SPatial Association DEtector, SPADE). By comparing the spatial heterogeneity, this estimator, which evolved from a variance-based relation estimator, explicitly considers the spatial variance by assigning the weight of the influence based on spatial distribution. It also overcomes the drawback of its old version which can only measure the association between continuous and discrete variables.

This method has been applied to estimate the influence of the environmental factors and their outcome (e.g. junction angle and environmental factors). The associations between environmental factors and junction angles have been used to infer the paleoenvironment of Mars; they showed that Mars was probably "warm" and "wet" several billion years ago. The method could also be used in human geography and social science to estimate the importance of spatial factors and their outcome.

Dr. Xuezhi Cang graduated from Northern Illinois University with Ph.D. degree in Geography. His research and teaching focus on Mars Geomorphology, Spatial Analysis, and Geographic Information Systems. He is currently a Postdoctoral Scholar at UMBC.