Spatial Pattern, Analysis of

Abstract

Spatial data consist of measurements taken at specific locations in a geographic space. Typically, such data are presented visually in the form of a map although they may also be represented in other analog forms such as aerial photographs and remotely sensed images. The relative locations of the data values in the geographic space define a spatial pattern. Spatial pattern analysis involves describing and explaining such patterns. In general, this is done by comparing characteristics of an empirical pattern with those of null models representative of spatial randomness (i.e., no discernible spatial order). Once identified, the distinctive characteristics of the empirical pattern can be used as evidence to gain a better understanding of the processes which generated the pattern. Methods of spatial pattern analysis can be classified in terms of the kinds of spatial data analyzed. This essay presents a commonly used procedure for each of the three types of spatial data most usually recognized: K function analysis for point data; semivariogram analysis for geostatistical data; and Moran's I measure of spatial autocorrelation for lattice data. Most recent developments in spatial pattern analysis have come in response to the proliferation of large spatial data sets, primarily from automated monitoring devices, coupled with increases in computing power and memory capacity. One result is increased interest in automated methods of spatial pattern analysis, such as spatial data mining. Another result is the shift away from global methods of spatial pattern analysis, which summarize the entire data set by means of a single measure, towards local methods, which search for localized patterns within the data set. This, in turn, has spurred interest in visualization of the results. Other developments include generating more sophisticated models of spatial randomness, which include some form of pre-specified spatial structure, and real time surveillance of pattern changes as new data becomes available.