Exploration of Satellite Images in the Dynamically Linked Arcview/xgobi/xplore Environment 1

The dynamic link among the Geographic Information System (GIS) ArcView, the dynamic statistical graphics (DSG) program XGobi, and the statistical computing environment XploRe has been successfully used to explore and analyze all kinds of spatially referenced data | from forest health data over precipitation data to precision agricultural data. In this paper, we will focus on the exploration of satellite images using the linked ArcView/XGobi/XploRe software environment. INTRODUCTION Multispectral satellite image data, i. e., remote sensing data gathered through earth observation satellites attracted the attention of statisticians working in the eld of dynamic statistical graphics (DSG) during the last decade. Many prototypes of software systems have been developed over time that focus on many interesting software features but yet lack the full functionality that is required to geographically explore satellite images. The ArcView/XGobi/XploRe software environment consists of the Geographic Information System (GIS) ArcView 3.0 TM , the dynamic statistical graphics (DSG) program XGobi 25], and the statistical computing environment XploRe 13]. These three programs, linked together into one single environment, provide features such as linked brushing among multiple windows of diierent types (e. g., map views and statistical plots), grand tour 1], 2], projection pursuit guided tour 6], and standard clustering and classiication methods, that are very powerful when jointly used for the exploration and classiication of satellite images. In the following sections we will look at examples from the literature where satellite images have been analyzed using dynamic statistical graphics. We describe the ArcView/XGobi/XploRe software environment, and we explain how it can be used for remote sensing data. We nish with a discussion and an overview on future directions.