US EPA

2858641 

Journal Article 

Contaminant plume and characterization using 3D geostatistics 

Ball, RO; Hahn, M; Sevcik, AE 

1998 

1 

85-90 

English 

WOS:000075845900014 

An historical remediation project at a site which was characterized by extensive data collection is revisited and analyzed with 3-D kriging techniques in an effort to test the hypothesis that the use of geostatistics can save significant effort and cost in the process of plume delineation. A site in Pennsylvania was found to have soil and ground water contaminated with chlorinated organic compounds (VOX) and benzene, toluene, ethylbenzene, and xylenes (BTEX) due to leaking underground storage tanks. The Pennsylvania Department of Environmental Protection (PaDEP) required a remediation objective for soil of 1 part per million (mg/kg) total VOX. Because removal of the source of contamination was required, the soil source area was characterized by approximately 150 discrete samples representing three depths on a regular grid. This produced an unusually dense data set ideal for evaluating sampling strategies for VOX contaminated soil. The historical data set was used as a basis for evaluating the utility of 3-D kriging techniques for identifying areas of uncertainty that require sampling and for plume delineation. Small data sets (e.g., 15 data points) were sampled randomly from the historic set and kriged in three dimensions. Areas of high variance, or high "uncertainty" were identified and "re-sampled" by selecting from the historic set. The results show that the plume could have been fully delineated to the required remediation objective within a reasonable confidence limit with as few as 30 samples (15% of the original data set). This use of 3-D geostatistics to optimize sampling for definition of soil source areas above a regulatory concentration isolevel can save up to 85% of investigative costs when compared to traditional RCRA grid sampling procedures described in SW-846, with equivalent confidence.