Abstract
AbstractThe spatial interpolation method of kriging, originally developed for estimation of ore reserves in mining and prediction in meteorology, is an important geostatistical tool for characterization of a Superfund site. Point kriging uses an estimated variogram to compute the best linear unbiased estimates (BLUE) of contaminant concentration at an unsampled location. When the contaminant concentration data is heavily skewed, the data are typically log‐transformed and kriging is performed on the transformed data. The results obtained from kriging the log‐transformed data are then back‐transformed. This back‐transformation formula involves not only the estimated log‐transformed values but also their kriging standard deviations (ksd). This method of kriging, referred to in the geostatistical literature as log‐normal kriging, was initially used to characterize the East Fork Poplar Creek (EFPC) Superfund site in Oak Ridge, Tennessee. The back‐transformed values for the EFPC site produced estimates of mercury contamination that were orders of magnitude higher than the maximum observed mercury concentration at the site. The method of rank variogram modeling/rank kriging was developed by the authors to estimate mercury contamination at the EFPC site. We present the method of rank variogram modeling/rank kriging, illustrate it with an example, and briefly describe the kriging results obtained for the EFPC site.The log‐normal probability model is routinely applied in statistical analysis of contaminant concentration data as well. The upper confidence limit (UCL) for log‐normally distributed data is typically calculated by using the H‐statistics based formula. The UCL of the mean at times turns out to be orders of magnitude higher than the maximum observed contaminant concentration. An example is presented to illustrate this unreasonable behavior of the H‐UCL formula. Copyright © 2002 John Wiley & Sons, Ltd.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have