source conditions or meteorology during the measurement period are incorporated into the 24-hour average, and are not discernible in the data except insofar as they affect the average.
The measurements also include all amounts of the material in ambient air regardless of the source. They do not directly measure emissions from the burn pit or from any other particular source. There are likely to be multiple sources of all the measured chemicals either on or in the vicinity of JBB, and the measurements include background concentrations due to those other sources as well as concentrations from the burn pit. The background concentrations from those other sources vary between sample times as a result of variations in time, number of sources, or location of those sources, as well as variations in meteorological conditions.
Several details of the sampling design and methodology affect the committee’s ability to analyze the sampling data. Such details have different effects on different analyses, so the following comments should not be interpreted as criticisms of the sampling design or methodology, since the samples were collected for a different purpose.
- Samples were not collected simultaneously at all locations for each sampling event. An effort was made to sample all the individual pollutant types (VOCs, semivolatiles, PM, PCDDs/Fs) on the same day, and there is substantial overlap in the sampling start times and the length of time for each sample, but they vary by as much as several hours both within and between pollutant types. This limits the comparability of samples, since the weather conditions and sources of pollutants are likely to be different in character, number, and strength at different times of day. Consequently, comparisons between the different sites are made more difficult, and identification of distinct sources from the sample data is compromised to some extent.
- Samples were not collected on any planned schedule. Instead a “convenience sampling” approach was used in which samples were collected when logistically possible. As a result, averages of the sample results may be unrepresentative of long-term average exposures because of unrecognized patterns in exposure (e.g., daily, weekly, monthly, or seasonal variations) not compensated by averaging the measurements.
- The PM methodology used is not suited for collecting samples with high concentrations of PM, such as occur during dust storms, or even on many non-dust-storm days at JBB (NRC 2010). As a result, some of the filter samples probably suffer from sampling artifacts (underestimates or overestimates of PM concentration). The 2009 samples, where PM10 and PM2.5 samples were collected simultaneously at the same locations, show evidence of such problems. The ratio of PM2.5 to PM10 was between 0.25 to 0.6 for most samples, but a few showed ṣubstantially higher or lower ratios. In particular, for some samples the ratio exceeded unity (the ratio should always be less than unity except for measurement errors, since PM2.5 is a component of PM10). Similar problems are likely to have occurred with PM10 sampling in 2007, and may have been masked in some 2009 samples by occurring simultaneously in both PM10 and PM2.5 samples. Further evidence of these problems is provided by some large discrepancies in measurements of the same parameter (PM2.5 or PM10) at the different sampling sites on the same day.
- Monitoring at different areas of the site would have been helpful. A site closer to the burn pit might have been more strongly and clearly affected by this source, allowing better characterization of the burn pit emissions. Sites in all the housing areas would allow better evaluation of exposures to all personnel on base rather than just those initially thought to be most highly exposed. Use of paired sites relatively close together, rather than single sites, may have compromised the analytic methods the committee attempted to use, since the paired sites were sufficiently far apart to be affected differently by local sources. Moreover, since these closely spaced sites were not sampled simultaneously, they do not provide any useful information on the variation of exposures within the housing area.
- The TO-14/TO-15 methodology (in which a sample of air is collected in an initially evacuated stainless steel flask) is not well suited to measure polar or reactive organics such as acrolein and 1,3-butadiene. Such materials may react on the walls of the stainless steel flask or in the gas phase during storage and transport of the sample. For example, 1,3-butadiene will decompose inside the canister during storage, mostly by reactions with nitrogen oxides. For risk assessments this is an important issue, because acrolein and butadiene are often the major risk contributors in screening risk assessments.