Coping with spatial heterogeneity effects on sampling and analysis at an HMX-contaminated antitank firing range

Short-range and mid-range (grid size) spatial heterogeneity in explosives concentrations within surface soils was studied at an active antitank firing range. Intensive sampling was conducted adjacent to two target tanks by establishing sixteen 6-m-square grids. Each grid was subdivided into four quadrants, and in each quadrant an area-integrated surface sample was formed into a pile that included about 10% of the top 5 cm of soil in the quadrant. After in situ homogenization, random aliquots were combined to form replicate representative samples. Grid composites were also prepared by combining equal portions of soil from the four quadrants for each grid. In nine of the quadrants, a second area-integrated sample was prepared. On-site analysis showed concentrations of HMX ranging from as high as 2160 mg/kg near one target to ≤1 mg/kg at a distance of 20 m from the target. TNT concentrations, ranging from ≤1 to 23 mg/kg, were much lower than would be expected based on the 70 : 30 composition ratio of HMX to TNT in the melt-cast explosive used on site. On-site concentration estimates for HMX and TNT were in excellent agreement with laboratory HPLC results; correlation coefficients were 0.992 and 0.975, respectively. Spatial heterogeneity of HMX concentrations was large on both short- and midrange scales, and this factor dominated the overall uncertainty associated with site characterization. Greater emphasis on sampling is urgently needed to improve the representativeness of explosives residue determinations in soil. ©1999 John Wiley & Sons, Inc. Field Analyt Chem Technol 3:19–28, 1999