, Volume 15, Issue 3, pp 399-416

First online:

Controlling patterns of geospatial phenomena

  • Tomasz F. StepinskiAffiliated withLunar and Planetary Institute
  • , Wei DingAffiliated withDepartment of Computer Science, University of Massachusetts Boston Email author 
  • , Christoph F. EickAffiliated withDepartment of Computer Science, University of Houston

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access


Modeling spatially distributed phenomena in terms of its controlling factors is a recurring problem in geoscience. Most efforts concentrate on predicting the value of response variable in terms of controlling variables either through a physical model or a regression model. However, many geospatial systems comprises complex, nonlinear, and spatially non-uniform relationships, making it difficult to even formulate a viable model. This paper focuses on spatial partitioning of controlling variables that are attributed to a particular range of a response variable. Thus, the presented method surveys spatially distributed relationships between predictors and response. The method is based on association analysis technique of identifying emerging patterns, which are extended in order to be applied more effectively to geospatial data sets. The outcome of the method is a list of spatial footprints, each characterized by a unique “controlling pattern”—a list of specific values of predictors that locally correlate with a specified value of response variable. Mapping the controlling footprints reveals geographic regionalization of relationship between predictors and response. The data mining underpinnings of the method are given and its application to a real world problem is demonstrated using an expository example focusing on determining variety of environmental associations of high vegetation density across the continental United States.


Predictors–response relationship Association analysis Mapping predicting relationship Vegetation density Data mining