The description of spatial pattern using two-dimensional spectral analysis

Abstract

Two-dimensional spectral analysis is a general interrogative technique for describing spatial patterns. Not only is it able to detect all possible scales of pattern which can be present in the data but it is also sensitive to directional components.

Four functions are described: the autocorrelation function; the periodogram; and, the R- and Θ-spectra which respectively summarize the periodogram in terms of scale and directional components of pattern. The use of these functions is illustrated by their application to a simple wave pattern, a wave pattern with added noise, and patterns simulating competition and invasion processes.