ESRI is focusing on spatial techniques that are robust, scalable, and have broad application for our GIS user community. We are particularly interested in techniques specific to spatial data.

Our geostatistical work focuses on providing interactive tools to examine spatial data and generate interpolation models. Many different interpolation techniques are supported, and recent work has focused primarily on those that provide optimal estimates for unsampled locations, while also providing information on the distribution of possible values at these locations. An important component of this is the addition of Gaussian geostatistical simulation in ArcGIS 9.3.  Also in 9.3, we exposed the functionality of Geostatistical Analyst in ArcGIS Server. These tools now also run on Linux and Solaris and allow the execution of geostatistical tools through Internet services.

Over the last five years we have been growing a collection of statistical tools for spatial pattern analysis. These tools provide functionality to describe and model spatial distributions, patterns, and processes.  (How dispersed are features and is there a directional trend?  Where are the hot spots?  How intense is spatial clustering and is that intensity increasing over time?)  With 9.3 we have added Regression Analysis tools to examine spatial relationships, allowing us to not only identify spatial patterns (such as hot spots or cold spots), but also to begin to explore their possible causes.  New tools in ArcGIS 9.3 include Ordinary Least Squares and Geographically Weighted Regression.

Statistics is a broad field with many techniques that can be leveraged by the GIS community, so we also support integration of ArcGIS with 3^rd party statistical software packages. ESRI has a long-standing partnership with SAS, and the two companies have collaborated to develop the “SAS Bridge,” a software utility to facilitate data exchange so our mutual users can work effectively in both software packages simultaneously.  With the new Geoprocessing Resource Center, clients will begin to see new developer samples illustrating ArcGIS integration with SAS, R, and other packages.