On the quality of contour maps

Abstract

Contour maps, also called isopleth maps, are used fairly often in environmental statistics. Examples are maps showing the levels of pollutants or depositions over a certain area. Kriging and other estimation techniques can be used to produce the data for such maps, and prediction (and other) errors can then be associated with each point. It is shown in a simple case that the values selected for the contour levels should differ from each other by several multiples of the prediction error (or some average prediction error). If not, the probabilities for the values shown on the map to be in error will be too high. Limitations and extensions of the ideas are briefly discussed.