| Abstract: | Contour or isopleth maps of regionalized variables are widely used in decision making without regard for the inherent uncertainty of interpolated contours. One concept for addressing the accuracy of contours given the data is the computation of the conditional probability of the value being lower than the contour. Two simple techniques for an approximate computation, indicator kriging and approximation by normal distribution quantiles, are applied to immission data. Since the assumption of a stationary underlying process does not seem justifiable, a moving windows technique is adopted. |
