...
Postcleanup sampling is required to verify the level of cleanup under §761.125(c)(2) through (4). The responsible party may use any statistically valid, reproducible, sampling scheme (either random samples or grid samples) provided that the requirements of paragraphs (a) and (b) of this section are satisfied.
(a) The sampling area is the greater of (1) an area equal to the area cleaned plus an additional 1-foot boundary, or (2) an area 20 percent larger than the original area of contamination.
(b) The sampling scheme must ensure 95 percent confidence against false positives.
(c) The number of samples must be sufficient to ensure that areas of contamination of a radius of 2 feet or more within the sampling area will be detected, except that the minimum number of samples is 3 and the maximum number of samples is 40.
(d) The sampling scheme must include calculation for expected variability due to analytical error.
(e) EPA recommends the use of a sampling scheme developed by the Midwest Research Institute (MRI) for use in enforcement inspections: “Verification of PCB Spill Cleanup by Sampling and Analysis.” Guidance for the use of this sampling scheme is available in the MRI report “Field Manual for Grid Sampling of PCB Spill Sites to Verify Cleanup.” Both the MRI sampling scheme and the guidance document are available on EPA's PCB website at https://www.epa.gov/pcbs, or from the Program Implementation and Information Division, Office of Resource Conservation and Recovery (5303T), 1200 Pennsylvania Ave. NW, Washington, DC 20460–0001. The major advantage of this sampling scheme is that it is designed to characterize the degree of contamination within the entire sampling area with a high degree of confidence while using fewer samples than any other grid or random sampling scheme. This sampling scheme also allows some sites to be characterized on the basis of composite samples.
(f) EPA may, at its discretion, take samples from any spill site. If EPA's sampling indicates that the remaining concentration level exceeds the required level, EPA will require further cleanup. For this purpose, the numerical level of cleanup required for spills cleaned in accordance with §761.125(b) is deemed to be the equivalent of numerical cleanup requirements required for cleanups under §761.125(c)(2) through (4). Using its best engineering judgment, EPA may sample a statistically valid random or grid sampling technique, or both. When using engineering judgment or random "grab" samples, EPA will take into account that there are limits on the power of a grab sample to dispute statistically based sampling of the type required of the responsible party. EPA headquarters will provide guidance to the EPA regions on the degree of certainty associated with various grab sample results.
[72 FR 57241, Oct. 9, 2007; 74 FR 30234 June 25, 2009; 89 FR 59693, Aug. 29, 2023]