Interpolation of Groundwater Quality Parameters Using Geological and Land Use Classification

Abstract

Groundwater quality parameters are influenced by a number of environmental factors. Experience shows that geology and land use are the most important among these. Therefore a straightforward interpolation not considering these factors might not deliver plausible results. As this additional information is not numerical, straightforward methods as Cokriging cannot be applied. This paper presents interpolation methods which can efficiently use additional information as a classification of the observed groundwater data and achieve better and more plausible interpolations. To take into account the two most important factors, a double classification is made referring to geology and land use similarly. Because most classes are too small then, and thus do not contain enough information for a plausible statistical analysis, an efficient and automatically working algorithm has been developed to combine small classes to such of sufficient size considering statistical and physical aspects of the parameters. The methods used for interpolation are Simple Updating, a modified version of the Simple Kriging, and a Bayesian type of combination of Ordinary or External Drift Kriging with prior information. These methods have been applied for more than 50 parameters monitored in extensive measurements in Baden-Württemberg at about 3000 locations. The different estimations with and without additional information are compared. The methods are also applied using indicator transformations. The results demonstrate that with additional information it is possible to achieve a significant improvement when this is considered in spatial interpolation.