John N. Rayner

Abstract

In recent years the geographer has been exposed to a bewildering variety ofquantitative techniques the usefulness of many of which have yet to be demonstrated.Furthermore, discussions of the lesser known techniques tend tobe very technical or, alternatively, are limited in scope. In consequence therelative values of these procedures are difficult to assess by the majority ofgeographers and are quickly dismissed. One such technique is spectral analysiswhich is often wrongly classified as being too complicated or being applicableonly to periodic data sets of which there are few. This paper attempts toreview briefly at a relatively non technical level the scope of the technique,or better, group of techniques, which may be labelled "spectral" or "Fourier",and to describe in detail the simple though long calculations involved. Centralto these is the Fourier transformation which is nothing more than a particularform of curve fitting by least squares. At the outset it should be noted thatthese techniques usually apply to data which have equally spaced coordinatesin space and/or time. Other arrangements of data are possible but they willnot be included in the present discussion.

John N. Rayner, The Ohio State University, Columbus