A geostatistical two-phase sampling strategy to map soil heavy metal concentrations in a former war zone

Autor: Eef Meerschman, M. Van Meirvenne, L. Cockx
Rok vydání: 2011
Předmět:
Zdroj: European Journal of Soil Science. 62:408-416
ISSN: 1351-0754
DOI: 10.1111/j.1365-2389.2011.01366.x
Popis: In a recent paper it was concluded that an enrichment of copper (Cu) content by 6 mg kg-1 dry mass in the soil around Ypres (Belgium) was a legacy of shelling during the First World War. This conclusion was based on a regional database of the entire province of West-Flanders (3144 km2) containing a limited number of soil samples in the war zone. Shells partly consisted of an alloy of Cu and zinc (Zn), and shrapnel balls were made out of lead (Pb). We expanded the database with a two-phase sampling design, each of 100 samples, in the war zone surrounding Ypres (640 km2) to (i) increase the detail of the inventory and (ii) expand the database to include Cu, Pb and Zn. This article focuses on the geostatistical selection of additional sampling locations. As the enrichment was spatially continuous and our aim was to map accurately over the range of values, rather than to delineate the enriched area, conventional selection criteria based on the probability of exceeding a critical threshold were not suitable. Therefore the sampling locations were optimized using a combination of selection criteria based on the kriging variance, the conditional variance and the conditional coefficient of variation obtained with sequential Gaussian simulation. A jackknife validation with 102 independent observations indicated the improvement after each phase. Additionally, the local uncertainty maps tended to show reduced values and a more homogenous pattern as additional samples were added. In an overview of the final prediction maps for Cu, Pb and Zn it is clear that those for Cu and Pb reflect the position of the main front line.
Databáze: OpenAIRE