Using linear and non-linear kriging interpolators to produce probability maps

Probability maps are used to define areas with high and low certainty of exceeding a threshold value. The most popular methods for creating these maps are variants of indicator kriging. However, such methods are questionable when the data exhibit a trend or contain measurement errors, the latter of which is common in most data sets. This paper presents an alternative approach that maps the risk of exceeding a threshold value after detrending and/or transforming the data. These methods are otherwise known as kriging on residuals and lognormal, Gaussian, and trans-Gaussian kriging. To illustrate this point an environmental case study is used to compare the predicted results for six different kriging method. All predictions and maps were made using the Geostatistical Analyst extension to ESRI ArcGIS software.