Construction of Isolinear Maps of Soil Attributes with Empirical Confidence Limits

Abstract

AbstractMoisture‐deficit data were obtained at 398 point locations in the Mander area in the Netherlands by simulation using soil‐profile attributes. From this data set, a test subset of 75 points was selected at random. The test set was used to select the best of four prediction procedures: ordinary and universal kriging with the original data set, and ordinary and universal kriging with a log‐transformed data set. The mean square error of prediction (MSEP) was used to evaluate prediction quality. The lowest MSEP of prediction was obtained using ordinary kriging of the untransformed data. Ordinary kriging was then used to predict moisture deficits and estimate their associated kriging variances at points on a 50 by 50 m grid. The average ratio of actual square errors of prediction to the estimated kriging variances at the 75 test locations was used to adjust the kriging variance estimates on the regular grid to get more realistic estimates. These empirically derived, more realistic estimated kriging variances were then used to construct an isolinear map of moisture deficit with confidence limits. The resulting map showed that, in most of the region under study, the 90% confidence interval for moisture deficit contains 15 mm. This type of map allows the user to obtain confidence limits as well as the predicted value for any point on the map.