Mapping Of Soil Contaminants Using Spartan Spatial Random Fields: A Comparative Study

Spartan Spatial Random Fields (SSRF?s) provide a new geostatistical framework (Hristopulos, 2003) with applications targeting environmental risk assessment and natural resources estimation. SSRF?s have the ability to incorporate constraints based on physical laws in the statistical description and have proved to be computationally fast in the analysis of synthetic data sets. This paper presents an application of a specific SSRF model and a new local-error spatial estimator (Hristopulos, 2005) in the mapping of environmental pollutants using a real data set. A detailed comparison of the local SSRF estimator with a classical kriging method is presented. It is shown that the SSRF approach provides a competitive alternative for spatial estimation.