HEALTH CONSULTATION

Air Pathway Evaluation

SIERRA ARMY DEPOT
HERLONG, LASSEN COUNTY, CALIFORNIA


APPENDIX A: ATSDR COMPARISON VALUES

Following are definitions of the various health-based comparison values that ATSDR used in this health consultation to put the measured and modeled levels of environmental contamination into perspective:

CREG: Cancer Risk Evaluation Guide, a highly conservative value that would be expected to cause no more than one excess cancer in a million persons exposed over time.
EMEG: Environmental Media Evaluation Guide, a media-specific comparison value that is used to select contaminants of concern. Levels below the EMEG are not expected to cause adverse noncarcinogenic health effects. These have been developed for acute exposure scenarios (EMEG-a), intermediate exposure scenarios (EMEG-i), and chronic exposure scenarios (EMEG-c).
NAAQS: National Ambient Air Quality Standard, an ambient air concentration that EPA has established to characterize air quality. The standards are health-based and were designed to be protective of many sensitive populations, such as people with asthma and children. The standards have been developed only for a small subset of pollutants, and the averaging time and statistical interpretations of the standards vary among the regulated pollutants.
PRG: Preliminary Remediation Goal, according to EPA, are "risk-based concentrations that are intended to assist risk assessors and others in initial screening-level evaluations of environmental measurements." The PRGs are EPA guidelines that are used to identify contaminants of concern at sites with environmental contamination.
RBC: Risk-based Concentration, a contaminant concentration that is not expected to cause adverse health effects over long-term exposure. These have been developed for both cancer outcomes (RBC-C) and noncancer outcomes (RBC-N). The RBCs in this table were published by EPA Region 3. Even though SIAD is located in EPA Region 9, ATSDR considered RBCs published by EPA Region 3 because they often times are lower than screening values (i.e., PRGs) published by EPA Region 9. Thus, consideration of EPA Region 3 RBCs makes our analysis more protective of human health.
REL (California): Reference Exposure Level, an air concentration that the California Office of Environmental Health Hazard Assessment says poses no significant health risk to individuals indefinitely exposed to that level. ATSDR evaluated California's RELs when preparing this document, but, due to our selection process, all of the comparison values used in our report ultimately drew from other sources.
REL (NIOSH): Recommended Exposure Level, an air concentration that the National Institute for Occupational Safety and Health (NIOSH) recommends should not be exceeded. RELs are designed primarily for occupational settings and exposures. The RELs used in this focused PHA are all based on 8-hour time weighted average exposures.
RfC: Reference Concentration, an ambient air concentration developed by EPA that people, including sensitive sub-populations, likely can be exposed to continuously over a lifetime without developing adverse noncancer health effects. RfCs typically have uncertainty factors built into them to account for any perceived limitations in the data on which they are based.


APPENDIX B: ATSDR PLAIN LANGUAGE GLOSSARY OF ENVIRONMENTAL HEALTH TERMS

Acute Exposure:
Contact with a chemical that happens once or only for a limited period of time. ATSDR defines acute exposures as those that might last up to 14 days.


Adverse Health Effect:
A change in body function or the structures of cells that can lead to disease or health problems.


ATSDR:
The Agency for Toxic Substances and Disease Registry. ATSDR is a federal health agency in Atlanta, Georgia that deals with hazardous substance and waste site issues. ATSDR gives people information about harmful chemicals in their environment and tells people how to protect themselves from coming into contact with chemicals.


Background Level:
An average or expected amount of a chemical in a specific environment. Or, amounts of chemicals that occur naturally in a specific environment.


Chronic Exposure:
A contact with a substance or chemical that happens over a long period of time. ATSDR considers exposures of more than one year to be chronic.


Completed Exposure Pathway:
See Exposure Pathway.


Comparison Value (CVs):
Concentrations or the amount of substances in air, water, food, and soil that are unlikely, upon exposure, to cause adverse health effects. Comparison values are used by health assessors to select which substances and environmental media (air, water, food and soil) need additional evaluation while health concerns or effects are investigated.


Concern:
A belief or worry that chemicals in the environment might cause harm to people.


Concentration:
How much or the amount of a substance present in a certain amount of soil, water, air, or food.


Contaminant:
See Environmental Contaminant.


Dose:
The amount of a substance to which a person may be exposed, usually on a daily basis. Dose is often explained as "amount of substance(s) per body weight per day".


Dose / Response:
The relationship between the amount of exposure (dose) and the change in body function or health that result.


Duration:
The amount of time (days, months, years) that a person is exposed to a chemical.


Environmental Contaminant:
A substance (chemical) that gets into a system (person, animal, or the environment) in amounts higher than that found in Background Level, or what would be expected.


Environmental Media:
Usually refers to the air, water, and soil in which chemicals of interest are found. Sometimes refers to the plants and animals that are eaten by humans. Environmental Media is the second part of an Exposure Pathway.


Exposure:
Coming into contact with a chemical substance.(For the three ways people can come in contact with substances, see Route of Exposure.)


Exposure Pathway:
A description of the way that a chemical moves from its source (where it began) to where and how people can come into contact with (or get exposed to) the chemical.

ATSDR defines an exposure pathway as having 5 parts:

  1. Source of Contamination,
  2. Environmental Media and Transport Mechanism,
  3. Point of Exposure,
  4. Route of Exposure, and
  5. Receptor Population.


When all 5 parts of an exposure pathway are present, it is called a Completed Exposure Pathway. Each of these 5 terms is defined in this Glossary.


Frequency:
How often a person is exposed to a chemical over time; for example, every day, once a week, twice a month.


Health Effect:
ATSDR deals only with Adverse Health Effects (see definition in this Glossary).


Indeterminate Public Health Hazard:
The category is used in Public Health Assessment documents for sites where important information is lacking (missing or has not yet been gathered) about site-related chemical exposures.


Inhalation:
Breathing. It is a way a chemical can enter your body (See Route of Exposure).


No Apparent Public Health Hazard:
The category is used in ATSDR's Public Health Assessment documents for sites where exposure to site-related chemicals may have occurred in the past or is still occurring but the exposures are not at levels expected to cause adverse health effects.


No Public Health Hazard:
The category is used in ATSDR's Public Health Assessment documents for sites where there is evidence of an absence of exposure to site-related chemicals.


PHA:
Public Health Assessment. A report or document that looks at chemicals at a hazardous waste site and tells if people could be harmed from coming into contact with those chemicals. The PHA also tells if possible further public health actions are needed.


Plume:
A line or column of air or water containing chemicals moving from the source to areas further away. A plume can be a column or clouds of smoke from a chimney or contaminated underground water sources or contaminated surface water (such as lakes, ponds and streams).


Point of Exposure:
The place where someone can come into contact with a contaminated environmental medium (air, water, food or soil). For examples:
the area of a playground that has contaminated dirt, a contaminated spring used for drinking water, the location where fruits or vegetables are grown in contaminated soil, or the backyard area where someone might breathe contaminated air.


Public Health Hazard:
The category is used in PHAs for sites that have certain physical features or evidence of chronic, site-related chemical exposure that could result in adverse health effects.


Public Health Hazard Criteria:
PHA categories given to a site which tell whether people could be harmed by conditions present at the site. Each are defined in the Glossary. The categories are:
  1. Urgent Public Health Hazard
  2. Public Health Hazard
  3. Indeterminate Public Health Hazard
  4. No Apparent Public Health Hazard
  5. No Public Health Hazard

Route of Exposure:
The way a chemical can get into a person's body. There are three exposure routes:
- breathing (also called inhalation),
- eating or drinking (also called ingestion), and
- or getting something on the skin (also called dermal contact).


Source (of Contamination):
The place where a chemical comes from, such as a landfill, pond, creek, incinerator, tank, or drum. Contaminant source is the first part of an Exposure Pathway.


Urgent Public Health Hazard:
This category is used in ATSDR's Public Health Assessment documents for sites that have certain physical features or evidence of short-term (less than 1 year), site-related chemical exposure that could result in adverse health effects and require quick intervention to stop people from being exposed.

APPENDIX C: REVIEW OF AIR SAMPLING STUDIES

Air sampling results are measurements of the levels of air contamination that people might actually breathe. Sampling results are critical elements to this health consultation, because they are direct measures of exposure point concentrations and do not involve the inherent uncertainties of modeling studies. ATSDR consulted several agencies, databases, and reports to obtain all ambient air sampling data that might be relevant to air quality issues for SIAD.

This appendix presents ATSDR's review of the three air sampling studies identified for this site. These reviews document the number and locations of sampling stations, sampling frequencies, number of samples collected, pollutants measured, comparisons of measured concentrations to health-based comparison values, and information on data quality. Section V of this health consultation indicates how ATSDR interpreted the air sampling data when evaluating the public health implications of exposure to SIAD's air emissions.

C-1. 2000 Air Sampling Study at SIAD

In July 2000, TetraTech NUS and Advanced Infrastructure Management Technologies prepared a final plan for an ambient air sampling project to be conducted later that year in the vicinity of SIAD (TetraTech NUS 2000). A primary goal of this project was to "address the public concerns regarding the accuracy of the assessments" conducted at the time, such as the health risk assessment and draft environmental impact statement (TetraTech NUS 2000). DTSC reviewed and approved the sampling plan.

Because this sampling program generated the only air quality measurements in the immediate vicinity of SIAD when OB/OD operations occurred, ATSDR critically reviewed all aspects of the program before deciding whether the concentrations should be used in this health consultation. Our critical review follows, in which we first document the program's scope and initial findings, and then present our interpretations. ATSDR disagrees with some critical findings documented in the summary report for this sampling program.

Scope of the 2000 Air Sampling Study at SIAD. Several features of ambient air sampling programs must be considered before interpreting trends and patterns among measured concentrations. The following bulleted items document these features, with ATSDR's interpretations presented at the end of this section:

  • Level of OB/OD Activity. The air sampling study was conducted over approximately 2 months, from August 13, 2000, to October 5, 2000. Overall, 22 sampling days were considered. These included 4 "background" days, or days when SIAD did not conduct any waste treatment activities in the OB/OD area; and 18 days when SIAD conducted various combinations of waste treatment operations, including OD, OB of propellants in pans, and OB of rocket motors. On these 18 days, SIAD treated, on average, 127 tons (gross weight) of waste material per day.


  • Sampling Schedule. As noted previously, air sampling occurred between August 13, 2000, and October 5, 2000. Sampling took place both on days when OB/OD operations occurred and on days without waste treatment operations, such that trends among the data might indicate the impacts that OB/OD emissions have on local air quality. A confounding factor in this analysis was contributions from large wildfires that burned in the nearby Feather River Canyon, at locations roughly 40 miles southwest of SIAD. These fires were reportedly burning out of control between August 20 and August 29.

    • Ambient air samples were collected on 22 of the 57 days between August 13 and October 5. There were several reasons why field personnel did not collect samples on every day during this time frame:

    • As planned before the program began, no samples were collected on Fridays or Saturdays and during the entire Labor Day weekend, presumably because these were days off for the field sampling personnel.


    • No samples were collected between August 22 and August 24, because an air quality advisory was issued for the Honey Lake Valley due to air emissions from the local wildfires.


    • Sampling was originally scheduled to end on September 15, but the program was extended through October 5 such that sampling could be conducted during at least one OB treatment of a rocket motor. The summary report does not indicate why some days were selected for sampling after September 15, and why others were not.


    On the designated sampling days, air samples were collected over two different intervals. First, at the unpopulated sampling locations, which the following paragraphs define, 3-hour average air samples were collected. This averaging time was used to characterize maximum short-term air quality impacts near the OB/OD area and to collect data that could be compared to predictions from dispersion modeling analyses. The time frame for these samples was from 3:00 PM to 6:00 PM, which corresponded to when the OB/OD activities occurred. At the remaining three sampling locations, which were located in populated areas near SIAD, 24-hour average air samples were collected. This averaging time was selected such that sampling results could be compared to health-based air quality standards.

  • Sampling Locations. This program involved 12 sampling locations. The sampling locations fall into two different categories. First, three sampling locations were placed in populated areas: Susanville (CA), Patton Village (CA), and Pyramid Lake (NV). The Pyramid Lake station was located along the western boundary of the Pyramid Lake Indian Reservation. The three sampling locations in populated areas were all at least 10 miles away from the OB/OD area.

    • Second, the remaining nine locations(8) were selected to characterize air pollution impacts in closer proximity to the OB/OD area. Eight of these stations were in California, and one was in Nevada. Most of these stations were located within 4 miles of where OB/OD occurred, including several stations placed in areas that models predicted would be most heavily impacted by the waste treatment air emissions. The nine sampling locations in unpopulated areas included two upwind stations (west of the OB/OD area) and seven downwind stations (east of the OB/OD area). Most of these stations were located in rugged terrain, without access to electricity.

  • Contaminants Measured, Sampling Methods, and Analytical Methods. In 1996, contractors to the Army completed human health and ecological risk assessments for SIAD (Brown and Root Environmental 1996a, 1996b), which found that several metals accounted for the majority of estimated human health and ecological risks. Consequently, the 2000 air sampling program focused on measuring concentrations of metals by collecting particles on air filters, and having the filters analyzed in a laboratory. Chlorine was also considered for this sampling program, because different chlorinated compounds may be released during rocket motor burning, and these compounds may reside in the particulate phase. Table C-1 lists the contaminants selected for this sampling program, along with their detection limits.

    • Two particulate sampling methods were used in this program. First, every sampling location was equipped with an AIRmetrics MiniVol PM10 sampler. These devices operate from battery power, and are therefore well suited for collecting PM10 samples in remote areas without electrical power. In fact, given the proposed scope of work, CARB reportedly recommended that the Army contractors use battery-operated sampling devices for this program, provided at least one station include a co-located device (TetraTech NUS 2000). Based on this recommendation, one downwind station was equipped with two sampling devices, for purposes of evaluating measurement precision. All of the MiniVol samples collected air at a rate of 5 liters per minute.

    In addition to measuring PM10 with the MiniVol samplers, PM10 concentrations were also measured using a Graseby Anderson High Volume PM10 sampler. This sampler is a reference method device, which means that EPA has specifically approved use of this device (among several others) for evaluating whether ambient air concentrations of PM10 meet federal air quality standards. PM10 samples were collected with the reference method device only at the Patton Village sampling station.

    All PM10 filters were weighed in an analytical laboratory, and filters collected by the MiniVol samplers were analyzed for elemental content (i.e., concentrations of metals and chlorine). Following specifications in EPA's reference methods for PM10 measurements, all sampling filters were weighed under controlled conditions before being sent to the field, and after being returned from the field. Metals content in particulate matter was measured using x-ray fluorescence, following an EPA-published method titled "Determination of Metals in Ambient Particulate Matter Using X-Ray Fluorescence (XRF) Spectroscopy" (EPA 1999). The sampling program's summary report presents the estimated detection limits for this method in units of mass per filter; ATSDR calculated the ambient air concentrations that correspond to these detection limits, as described later in this section.

  • Data Quality. Overall, 308 sampling events were scheduled, and 277 valid samples were collected, resulting in a completeness fraction of 91%. Several measures were taken to reduce random and systematic errors in the ambient air measurements. These include having field personnel adhere to written standard operating procedures, calibrating all sampling equipment before the program began, verifying that sample flow rates fall in acceptable ranges, and conducting a thorough data quality review of all results. Moreover, several quantitative measures of data quality were reported, as summarized below. ATSDR's interpretation of the data quality measures are presented later in this section.

    • First, precision of the MiniVol sampler measurements was quantified by comparing concentrations measured simultaneously by the two co-located devices. The authors reported that these co-located devices had relatively poor precision. Based on the 19 times the devices simultaneously collected valid samples, the average relative percent difference (RPD) in the measured concentrations was 64%. Only 8 of the 19 co-located measurements had RPDs lower than 30%–a level of precision that many sampling methods specify as an acceptable level of precision. Many of the instances with poor repeatability occurred in samples with low PM10 concentrations.

    Second, the summary report compares 24-hour average PM10 concentrations that were measured by the co-located MiniVol sampler and reference method sampler at the Patton Village station. In this comparison, the average RPD was 46%, and the MiniVol sampler consistently measured lower concentrations than the corresponding FRM. For at least half of the sample pairs considered, the PM10 concentrations were 20 g/m3 or lower.

    Finally, the precision of the XRF analyses was examined by having an EPA laboratory re-analyze ten particulate filters collected during the program. The results from both laboratories agreed well: with only one exception, the ratio of the concentrations originally reported for this sampling program to the corresponding concentrations measured by the EPA laboratory was less than two.

  • Results Presented in the Final Report. The summary report for the air sampling program (TetraTech NUS 2001) presents the measured air concentrations, the amounts of waste material treated on sampling days, plume observations, and interpretations of these data. ATSDR summarizes many of the measurements later in this section. However, some key conclusions presented in the summary report follow:


    • The authors concluded that "statistical evaluations of PM10 measurements showed that the downwind monitors might have captured the plume from OD emissions on two occasions." This finding suggests that monitors did not capture the plume on the overwhelming majority of sampling events. The statistical evaluations were based on comparisons between upwind and downwind sampling stations using T-tests conducted at the 95% confidence level. ATSDR does not agree with this conclusion, as noted below.


    • The authors concluded that "OB/OD treatment emissions did not result in exceedances of state or federal health-based PM10 standards." They also noted that emissions from wildfires led to elevated PM10 concentrations in locations throughout the area.


    • The authors indicated that measurements of 3-hour average PM10 concentrations were not as precise as desired, largely because the ambient air concentrations of particulate matter were much lower than expected. Consequently, this low precision was observed primarily on days when PM10 concentrations were lowest. Metals measurements, on the other hand, were found to be highly precise, based on comparisons between concentrations reported by two different laboratories that analyzed the same subset of filters.


    • The authors concluded that "future ambient air monitoring of PM10 and metals utilizing currently available sample collection and analytical techniques is not recommended." This recommendation is apparently based on a judgment that available sampling and analytical methods cannot precisely quantify impacts from OB/OD emissions. The authors acknowledged, however, that future ambient air sampling may need to be considered if SIAD starts to treat larger quantities of rocket motors. The authors supported future studies of soil contamination, which they presume would reflect the impacts of atmospheric deposition of OB/OD emissions.

ATSDR's Interpretation of the Sampling Results. ATSDR's review of the 2000 air sampling study at SIAD follows. This review is organized by the same topics presented previously:

  • Level of OB/OD Activity. ATSDR notes that the average level of waste treatment activity during the sampling program–127 tons (gross weight) per day–is reasonably representative of SIAD's long-term average waste treatment quantities. If one assumes that SIAD treats waste materials for 150 days per year, for instance, then the average daily treatment rate during the sampling program would amount to an annual treatment rate of 19,050 tons per year. As Section III.E of this health consultation noted, SIAD's annual average waste treatment quantity between 1990 and 2000 was 19,000 tons per year. Thus, the sampling program was conducted during typical OB/OD operations in terms of total waste treated, though sampling occurred on only one day when rocket motors were burned.


  • Sampling Schedule. Samples were collected on 22 days between August and October, 2000, and sampling was intentionally scheduled for days when different types of OB/OD activities occurred. Sampling therefore characterized typical levels of air pollution following various waste treatment activities during the late summer and early fall. Though it is possible that levels of air pollution vary with season, ATSDR believes the sampling schedule considered for this program allows for a reasonable analysis of typical air quality impacts from SIAD's emissions. Recognizing that reviewing additional sampling data would give greater confidence in our overall conclusions, as is true for most any environmental health evaluations, ATSDR has recommended that further sampling occur near SIAD if OB/OD waste treatment operations resume (see Section X).


  • Sampling Locations. This program considered two types of sampling locations, which were selected for different reasons. The three sampling locations in populated areas were selected primarily to address community concerns. These were placed at locations in California and Nevada, including upwind and downwind locations. ATSDR believes the placement of these sampling devices was appropriate and consistent with the sampling program's objectives.

    • Additionally, nine sampling locations in unpopulated areas were selected to characterize maximum air quality impacts and to conduct model performance evaluations for the atmospheric dispersion analysis. These stations were placed primarily downwind from the OB/OD area, up to 8 miles from the installation boundary. The ultimate placement of devices was determined from reviews of on-site meteorological monitoring data and dispersion modeling analysis. ATSDR has confidence that the nine sampling locations in unpopulated areas were adequately placed, given the measured data and modeling results upon which this decision was made. The fact that CARB reviewed and approved SIAD's meteorological monitoring plan and that DTSC reviewed and approved this program's sampling plan (which detailed the findings of the dispersion modeling analysis) gives greater reassurance that proposed sampling locations were selected appropriately.

    ATSDR notes that equipment siting is also an important consideration when interpreting measured concentrations. Siting refers to the placement of sampling equipment and sampling probes at a given location. Though ATSDR scientists did not witness the 2000 air sampling study at SIAD, we note that representatives from DTSC were present when all sampling equipment was installed. We assume this oversight helped ensure that siting was performed adequately.

  • Contaminants Measured, Sampling Methods, and Analytical Methods. The sampling program measured ambient air concentrations of PM10, metals, and chlorine. ATSDR believes this contaminant selection was appropriate, and supported by the findings of the human health and ecological risk assessments. Although SIAD clearly emits many additional contaminants that were not measured, the sampling program focused on the pollutants that accounted for the majority of the emissions and estimated human health and ecological risks. Sampling for every contaminant identified in the various emission inventories likely would have been prohibitively expensive.

    • The overwhelming majority of samples were collected using battery-operated AIRmetrics MiniVol PM10 samplers. Although these are not EPA-designated reference or equivalent methods (EPA 2002c), the sampling devices have been widely used to measure PM10 concentrations, particularly in remote areas where electricity is not available, as was the case for most sampling stations in this program. Accordingly, ATSDR believes the MiniVol PM10 samplers were appropriate and consistent with this program's objectives. Use of the reference method device at Patton Village was also appropriate.

    Filters were analyzed using an XRF method that EPA published. The sampling report documented detection limits, expressed in terms of mass per filter, which ATSDR has converted to their corresponding ambient air concentrations. Table C-1 presents these detection limits, both for the 3-hour and 24-hour sampling durations, and compares them to health-based comparison values. The table shows that detection limits were lower for the sampling locations in populated areas, which was expected because larger volumes of air were sampled at these locations. ATSDR believes the analytical method was sensitive enough to measure concentrations at levels of interest for all contaminants, with the possible exception of cadmium. Average air concentrations of cadmium in remote locations are typically 0.001 g/m3 (ATSDR 1999), and the lowest detection limit in this program was 0.014 g/m3. Therefore, this study's methods are sufficient only for determining whether unusually high levels of cadmium are found in SIAD's air emissions.

  • Data Quality. The 2000 air sampling study at SIAD was conducted according to the specifications of a detailed plan, which state environmental regulators reviewed and approved. The high completeness fraction (91%) for this program suggests that sampling devices did not malfunction frequently and that sampling filters were not routinely lost or destroyed. Use of standard operating procedures likely reduced the possibility of field sampling personnel introducing random errors into the measurements. Although these observations are all encouraging, none confirm that the measured concentrations are of a known and high quality. ATSDR more closely examined three quantitative measures of data precision:


    • The RPD for the co-located PM10 3-hour average measurements with MiniVol samplers was 64%, which is quite high in comparison to precision observed in programs that use reference method devices to measure 24-hour average concentrations. To assess how the low measurement precision affects the reported concentrations, ATSDR examined the magnitude of the differences between the co-located measurements. For 18 co-located measurements,(9) the average difference in measured concentrations at the two sampling devices was 14 g/m3. This concentration difference is actually near the measurement sensitivity for the methods used in this program: given that the devices collected 0.9 m3 of air per sample and that the measurement scale had a sensitivity of 10 g, one would expect that replicate measurements of the same filter might exhibit concentration differences up to 11 g/m3. Therefore, the actual concentration difference observed (14 g/m3) is reasonably close to the measurement sensitivity.

      • Though ATSDR would certainly prefer to work with a more precise data set, we note that the relatively low measurement precision was reported for a sampling station that collected 3-hour average samples. We believe (though cannot prove) that higher measurement precision would have occurred for 24-hour average sampling durations, because the particulate mass loadings on the filters would be much higher than the measurement sensitivity of the scale.

    • ATSDR also examined the precision of measurements made by the MiniVol sampler and reference method sampler that were co-located at the Patton Village station. These measurements had an average RPD of 46%. Although this agreement is again relatively low, most of the paired observations had PM10 concentrations in a range (0 to 20 g/m3) where the MiniVol samplers have greater measurement variability. The average PM10 concentration difference between the MiniVol sampler and reference method sampler was 10 g/m3.


    • Metals concentrations measured during the air sampling program compared very well to those reported by the EPA laboratory that re-analyzed the filters. The ten filters that were re-analyzed had detections for aluminum, barium, chlorine, copper, iron, lead, manganese, and zinc. The agreement between these independent measurements indicate that laboratory analyses were not biased.


    After carefully evaluating these observations, ATSDR concluded that the PM10 and metals measurements are of sufficient quality to use in this health consultation. Two factors weighed heavily in this decision. First, although the measurement precision of the MiniVol samplers was relatively low, the measured concentrations in populated areas were well below health-based standards, as confirmed by a reference method sampling device. The low measurement precision, though undesirable, is not as critical when measured concentrations are far below levels of health concern(10). Second, ATSDR is comforted knowing that the low precision resulted largely from the fact that PM10 concentrations were much lower than had been anticipated when the sampling program was planned.

  • Results. This program's summary report presents raw data for 22 sampling events, but does not present descriptive summary statistics for the sampling stations. To interpret this program's results, ATSDR first loaded the raw data into an electronic database to examine how the measured PM10 concentrations varied from one location to the next. Figure C-1 illustrates the observed spatial variations. For ease of interpretation, different shading schemes are used to differentiate the sampling locations in populated areas, the upwind sampling locations in unpopulated areas, and the downwind sampling locations in unpopulated areas. The figure clearly shows that average PM10 concentrations were highest in the unpopulated areas downwind from the OB/OD area. Moreover, sampling location DW6 had the highest PM10 levels, and the difference between these levels and those observed at the upwind stations were significant at p-values of 0.06 (for comparison to station BK1) and 0.09 (for comparison to station WB1). It is interesting to note that the two sampling stations with the highest PM10 concentrations (DW5 and DW6) were located within the area where a modeling analysis predicted the maximum impacts would occur (TetraTech NUS 2000). Based on these observations, ATSDR believes that the sampling data are reasonably consistent with the dispersion modeling predictions.

    • In addition to examining PM10 levels, ATSDR evaluated spatial trends among the sampling results for metals and chlorine. Table C-2 lists the frequency with which the various elements were detected, but only among the sampling locations in unpopulated areas. The presentation focuses on frequency of detection because many elements were detected in few samples, thus complicating efforts to calculate average concentrations. The presentation is limited to only sampling locations in unpopulated areas, because this subset has a common set of detection limits. A direct comparison cannot be made to the samples collected in the populated areas, since they had lower detection limits due to the longer sampling times. In any case, the data in Table C-2 show that several elements were detected more frequently at the downwind sampling stations than they were at the upwind stations. This trend was particularly pronounced for aluminum and zinc, which the OB/OD operations emit in large quantities. Though the frequencies of detection also support our belief that the downwind sampling stations detected effects of OB/OD operations, we caution about inferring too much from the data in Table C-2, because variability in measurements tend to be greatest near detection limits.

    Given that the downwind stations had higher levels of PM10 and that metals were detected more frequently in their filters, ATSDR was surprised to read the conclusion that "the downwind monitors might have captured the plume from OD emissions on two occasions" (TetraTech NUS 2001). This conclusion was based on a statistical analysis of the data, in which differences between upwind and downwind concentrations on "background" days were compared to the differences observed on days when OB/OD waste treatment occurred. This comparison was made as follows:

    • For each downwind station, the authors computed three upwind/downwind concentration differences based on the results from the three "background" days.(11)


    • The authors then computed, for each station, the mean concentration difference from the three upwind/downwind concentration differences, and the 95% upper confidence limit (UCL) about these means.


    • Whether or not a particular station was viewed as being impacted by the OB/OD plume on a given day was then determined by comparing that day's upwind/downwind concentration difference to the 95% UCL concentration determined from the background sampling days. The authors concluded a station was impacted by the plume only if the upwind/downwind concentration difference on a given day was greater than that station's 95% UCL concentration.


    After carefully reviewing the data, ATSDR believes this statistical approach does not provide a meaningful interpretation of the sampling results. Our specific concern is with the limited sample size of measurements collected on "background" days: with only three days of measurements when OB/OD was not occurring (and considerable variability in PM10 levels on these three days), the 95% UCL concentration for almost every station was so high that only extremely large concentration differences would be considered "impacted by the plume," by the authors' approach. As an example of our concern, the summary report indicates that upwind/downwind concentration differences on a given day would have to reach the following levels for the authors to have concluded that a station was "impacted by the plume":

    DW3–53 g/m3
    DW4–19 g/m3    		 DW5–199 g/m3
    DW6–163 g/m3    		 DW8–169 g/m3

    In other words, for any given sampling event, station DW5 would be considered to be "impacted by the plume" only if the PM10 concentration at this station was 199 g/m3 higher than the PM10 concentration measured at the upwind station (BK1). Therefore, PM10 concentrations at this station could increase five times over background levels and still not be considered to be affected by a source. Not surprisingly, the only stations found to be "impacted by the plume" in this study were stations DW3 and DW4–the stations with the lowest UCL concentrations.

    Overall, ATSDR does not agree with the conclusion that measurements collected during this sampling program captured the effects of the plume on two events, because the statistical analyses were based on a very limited sample size. ATSDR's three major conclusions on this sampling program follow:

    • The average PM10 concentrations measured in some unpopulated areas downwind from the OB/OD area were higher than those measured in upwind locations, by as much as a factor of two. Several metals were detected more frequently in these downwind areas, when compared to the upwind areas. The most logical explanation for these trends is that the downwind stations were impacted by emissions from the OB/OD area.


    • The highest PM10 concentrations were observed in the same general area where dispersion modeling analyses predicted they would occur, but the magnitude of the predicted concentrations differed from that of the measured concentrations. This observation suggests that the meteorological data considered in the modeling analysis was representative of site conditions (because the location of maximum impact was predicted), but the emissions data in the modeling analysis may not be accurate (because the magnitude of the concentrations were not predicted well).


    • The sampling data are reasonably representative of actual exposures because samples were collected where people live and on days with typical OB/OD operations (in terms of the quantity of waste material treated). Two of the downwind sampling locations (DW1 and DW2) were located between the OB/OD area and one of the two ranches nearest the site.

C.2 Ambient Air Monitoring Data Collected in Susanville, California

To locate ambient air monitoring data for Lassen County, California, ATSDR consulted two resources: the state of California's "Ambient Air Quality Data" electronic database (CARB 2000) and EPA's Aerometric Information Retrieval System (AIRS) (EPA 2002b). These are both electronic databases of ambient air monitoring data maintained by state and federal environmental agencies, respectively. Both data sets include air quality measurements for just one ambient air monitoring station in Lassen County. This station is located in Susanville, which is approximately 45 miles west-northwest (and upwind) from SIAD.

The Susanville air monitoring station is located at 175 Russell Avenue, and is operated by the California Air Resources Board. This air monitoring station has measured ambient air concentrations of both TSP and PM10. The TSP measurements were made between 1974 and 1977, and again in 1987, during which time the station collected 208 valid measurements. However, the station has several months of missing observations. The TSP data are not summarized here because, since the 1970s, researchers have shown that exposures to finer particle sizes are more closely associated with adverse health effects. ATSDR notes, however, that the annual geometric mean concentrations of TSP at Susanville during this time frame were all lower than EPA's former National Ambient Air Quality Standard (NAAQS) for this pollutant.

In October, 1996, particulate matter monitoring resumed in Susanville, but PM10 concentrations were measured instead of TSP. Specifically, 24-hour average PM10 samples were collected once every 6 days, which is a typical particulate sampling schedule for assessing attainment status with EPA's NAAQS. During this time, 178 valid PM10 samples were collected and reported to AIRS. Though valid samples were routinely collected, a large data gap occurs from October 1997 through September 1998, presumably because no samples were collected during this fiscal year. Further, the AIRS database does not include any observations from the summer of 2000, when wildfires burned out of control in many parts of Lassen County. The PM10 samples at Susanville were collected using EPA-approved devices (i.e., reference methods) following strict data quality procedures, and the California Air Resources Board reviews all data for accuracy before submitting them to EPA's AIRS database. ATSDR therefore believes the PM10 monitoring results are of a known and high quality.

Table C-3 summarizes the PM10 concentrations that were measured in Susanville between 1996 and the present. Annual average concentrations and maximum 24-hour average concentrations are shown and compared to EPA's corresponding health-based NAAQS. As the summary table shows, annual average PM10 levels in Susanville have been lower than EPA's health-based standards since 1996, though this observation is based on a limited number of samples for 1996 and 1998. Further, all 178 24-hour average concentrations submitted to AIRS are lower than EPA's corresponding health-based standard. Based on these data, Lassen County is designated as being in attainment with EPA's PM10 air quality standards.

C.3 Ambient Air Monitoring Data Collected in Washoe County, Nevada

To evaluate air quality in areas downwind of SIAD in Nevada, ATSDR acquired data from the Nevada Division of Environmental Protection (NDEP 2001) and from EPA's AIRS database (EPA 2002b). When accessing these data, ATSDR learned that Washoe County, Nevada, is currently a non-attainment area for carbon monoxide, ozone, and PM10. This means that ambient air concentrations of these pollutants somewhere in Washoe County have exceeded EPA's corresponding health-based standards in recent years. Data collected most recently, including up through 2001 (WCDHD 2002), indicate that Washoe County's air quality in the last few years has not exceeded the federal air quality standards. Consequently, Nevada has asked EPA to designate Washoe County as being in attainment with the ozone standards, and that request is still being reviewed (NDEP 2001).

Though all of Washoe County may have a non-attainment designation for certain pollutants, the poor air quality that triggered these designations was measured strictly in the Reno area (NDEP 2001), which is more than 50 miles south-southwest from SIAD and not downwind from the OB/OD operations. The data ATSDR obtained show that the Nevada state and local agencies measured ambient air concentrations of criteria pollutants at several locations in the cities of Reno, Sparks, and in Reno's outlying areas. Elevated levels of carbon monoxide and particulate matter are believed to come predominantly from sources local to that area, such as motor vehicle exhaust, road sanding, and wind-blown dust. High ozone concentrations, on the other hand, result from the complex mixture of pollutants from many different sources.

As Section IV.A of this health consultation describes, ATSDR did not consider ambient air monitoring data from the Reno area when evaluating air quality impacts from SIAD. This decision was made due to the distance that separates SIAD and the Reno area, the fact that Reno is not downwind from SIAD, and the belief that the air quality problems in the Reno area are caused predominantly by local sources of air pollution.

Table C-1: Detection Limits for the 2000 Air Sampling Study at SIAD

Analyte Detection Limits (g/m3) Most Conservative Health-Based Comparison Value (g/m3) Type of Comparison Value (see Appendix A)
3-Hour Sampling Period 24-Hour Sampling Period
Aluminum 0.11 0.014 3.7 RBC-N
Barium 1.18 0.15 0.51 RBC-N
Cadmium 0.11 0.014 0.0006 CREG
Chlorine 0.068 0.0085 0.21 RBC-N
Copper 0.012 0.0015 150 RBC-N
Iron 0.018 0.0022 1,100 RBC-N
Lead 0.056 0.0069 1.5 NAAQS
Manganese 0.022 0.0028 0.04 EMEG (chronic)
Nickel 0.012 0.0015 0.2 EMEG (chronic)
Zinc 0.013 0.0017 1,100 RBC-N

Note: Detection limits were calculated from those reported in the sampling study (in terms of mass per filter), the reported sample flow rate (5 liters/minute), and the sample durations.


Table C-2: Frequencies with which Elements Were Detected during the 2000 Air Sampling Study at SIAD

Element Upwind Sampling Locations Downwind Sampling Locations
BK1 WB1 DW1 DW2 DW3 DW4 DW5 DW6 DW8
Aluminum 5% 6% 11% 11% 15% 40% 30% 20% 0%
Barium 0% 12% 0% 0% 0% 0% 0% 5% 10%
Cadmium 0% 0% 0% 0% 0% 0% 0% 0% 0%
Chlorine 0% 6% 5% 20% 0% 25% 10% 10% 0%
Copper 15% 19% 5% 11% 15% 20% 15% 5% 0%
Iron 100% 100% 89% 89% 80% 85% 95% 95% 81%
Lead 5% 19% 16% 16% 0% 25% 10% 20% 10%
Manganese 0% 12% 5% 5% 0% 15% 5% 0% 5%
Nickel 0% 0% 0% 0% 5% 5% 0% 0% 0%
Zinc 5% 19% 21% 16% 20% 50% 35% 65% 14%

Notes:

Data source: TetraTech NUS 2001.
For purposes of this table, measured concentrations and J-qualified results were considered to be detections.


Table C-3: PM10 Concentrations Measured in Susanville, California

Year Number of Samples Long-Term Exposure Levels Short-Term Exposure Levels
Annual Average Concentration
(g/m3)		 EPA's Health-Based National Ambient Air Quality Standard (g/m3) Highest 24-Hour Average Concentration (g/m3) EPA's Health-Based National Ambient Air Quality Standard (g/m3)
1996 10 22.9 50 42 150
1997 38 31.7 50 84 150
1998 14 26.8 50 52 150
1999 52 36.1 50 100 150
2000 32 37.2 50 80 150
2001 32 32.0 50 87 150

Notes:

Data source: EPA 2002b.
EPA's health-based standards for exposure to PM10 are shown. Some states (including California) have developed more stringent standards.


APPENDIX D: REVIEW OF AIR MODELING STUDIES

ATSDR views environmental sampling data as critical inputs to the public health assessment process and strongly recommends the use of validated sampling data as the basis for public health decisions. In some circumstances, however, sampling data are not sufficient to characterize all site-specific exposures. At the SIAD site, for example, only one site-specific air sampling project has been conducted. Though this project provided critical insights into local air quality (see Appendix C.1), the sampling data have limitations in that they characterize air quality over a relatively short time frame and only for certain air pollutants.

Given the limitations of the air sampling data, ATSDR considered findings from modeling studies when evaluating the public health issues for the SIAD site. Modeling studies are only capable of estimating exposure concentrations, based on a scientific understanding of how chemicals move in the environment. Because models have limitations and uncertainties, and may not accurately represent actual environmental conditions, ATSDR carefully reviews all modeling applications to determine whether they can be used in the public health assessment process.

Although several site documents include air dispersion modeling analyses (e.g., Brown and Root Environmental 1996b, TetraTech NUS 2000, TetraTech NUS 2001), the modeling that was performed specifically to evaluate human exposures is reported in the site's human health risk assessment (Brown and Root Environmental 1996a). The remainder of this appendix presents ATSDR's critical review of the modeling study documented in the human health risk assessment, and Section V describes how ATSDR used these modeling analyses to reach public health conclusions.

D.1 Dispersion Modeling Documented in the Human Health Risk Assessment

The human health risk assessment prepared for SIAD presents an atmospheric dispersion modeling analysis of emissions from the site's waste treatment operations (Brown and Root Environmental 1996a). The risk assessment was conducted as part of the permit application process for the OB, OD, and incineration waste treatment activities.

D.1.1 Approach Used to Estimate Emissions

The dispersion modeling analysis in the human health risk assessment considered air emissions of nearly 100 pollutants from four sourcesOD of high explosives, OB of propellants in burn pans, OB of rocket motors, and the incinerator exhaust. Emissions were estimated from these sources as follows:

  • OD of high explosives. The human health risk assessment includes emissions estimates for 85 pollutants released to the air following OD events at SIAD. Emissions were estimated assuming SIAD detonates the maximum amount of waste material that would have been allowed had the installation's permit application been approved. The treatment amounts considered in the emissions calculations were 18,200,000 pounds NEW per year and a short-term peak treatment rate of 140,000 pounds NEW per hour. This maximum hourly treatment rate assumes that detonations would simultaneously occur in all 14 OD pits, with each pit being filled with the maximum amount of waste material allowed. Because SIAD previously did not operate at its permitted limit, use of the maximum allowed treatment rates will lead to overestimated emission rates.

    • Emissions were estimated using multiple approaches. The approach taken varied by the class of pollutant, as summarized below:

    • Explosives and chemical by-products of explosions. Emission rates for these pollutants were calculated primarily from the results of source tests conducted in controlled field experiments at other military installations. These source tests, known as "Bangbox" studies, measured the amounts of chemical by-products of explosions that result from OB/OD activities. The Bangbox is a flexible structure in which ordnance is detonated or burned. Because the Bangbox is completely enclosed, pollutants released during the detonation or burn do not escape the structure and can be measured by air sampling equipment. The Bangbox has thus allowed scientists to estimate emission factors for OB/OD of various types of ordnance and propellants, many of which have similar composition to those the Army treats at SIAD. The emission factors estimate the amounts of chemicals released to the air per weight of NEW detonated or burned.

      • In the SIAD human health risk assessment, the authors reviewed the results of the available Bangbox studies (and other plume sampling studies) to select chemical-specific emission factors. For every chemical considered, the Bangbox emission factor used in the human health risk assessment is either the highest emission factor from all relevant Bangbox tests or the 95% upper confidence limit of the mean emission factor. The higher of these two values was selected for every chemical.

      ATSDR believes use of the Bangbox test results is appropriate for evaluating air quality impacts from SIAD, because these emission factors have been widely used to assess environmental impacts from OB/OD activities. For instance, the Open Burn/Open Detonation Model (OBODM), available from EPA's clearinghouse of dispersion models on the agency's technology transfer network, also uses the Bangbox emission factors to estimate air emissions from OB/OD. Further, the Bangbox studies examined air emissions following detonation of high explosive mixtures similar to those that SIAD commonly treats. ATSDR acknowledges that the emission factors from the Bangbox studies have inherent uncertainties, and may underestimate or overestimate air emission rates in actual OB/OD events. However, the fact that the emissions estimates in the human health risk assessment are based on the highest Bangbox emission factor for all chemicals considered provides some comfort that the approach taken does not underestimate emissions and human health risks.

    • Particulate matter. Particulate matter can be formed in many ways following an OD event. Large volumes of particulate matter emissions, for example, are soils that are ejected into the air following a detonation. The Army Research Laboratory has developed a model that predicts emissions of particulate matter from various battlefield activities, including detonation of high explosives (Army Research Laboratory 2000). This model predicts that particulate matter emissions are sensitive to several site-specific parameters, including soil type and where the detonation occurs with respect to local terrain features (e.g., at the surface, above the surface, in a pit, and so on).

      • The human health risk assessment used the Bangbox emission factors to estimate the amounts of particulate matter that OD activities released to the air. ATSDR is not convinced that these emission factors are representative of actual conditions at SIAD, because the detonations in the Bangbox tests were not conducted in pits, like the detonations at SIAD. Therefore, it is unclear whether the estimated emission rates for particulate matter in the human health risk assessment adequately characterize actual exposures. ATSDR does not view this as a critical data gap for this health consultation, because valid, representative ambient air monitoring data are available for evaluating air quality impacts of particulate matter emissions.

    • Metals. During OD activities, metals can be emitted by multiple mechanisms. The casings of the high explosives being destroyed, for example, contain metals, which can be released to the air during a detonation. Some ordnance contains metals (e.g., aluminum dust) in the high explosive charge, which also can become airborne during the waste treatment activities. Finally, detonations eject soils into the air, and the soils contain both naturally-occurring metals and metals contamination from past waste treatment activities.

      • The human health risk assessment estimated air emissions of metals by assuming that the entire weight of the casings is released to the air during a detonation. This assumption clearly overstates metals emissions from casings, because considerable amounts of scrap materials remain on the ground after detonation events. Although this source of metals emissions is over-represented, ATSDR notes that potential air emissions of metals in contaminated soils were not considered. Nonetheless, the assumption that the entire weight of bomb casings is completely released to the air most likely caused the human health risk assessment to overstate metals emission rates, possibly by a large margin. Fortunately, ambient air monitoring data are available for many of the metals considered, which allowed ATSDR to examine measured air quality impacts, rather than impacts estimated by modeling analyses.

    Overall, ATSDR found the approaches used to estimate air emissions from OD events to be technically sound and useful for evaluating inhalation exposures to SIAD's emissions. Emissions were estimated for waste treatment quantities larger than the amounts of waste SIAD typically treated. The contaminants with the highest estimated emission rates were PM10, copper, iron, and aluminumall of which were measured during the recent ambient air monitoring program.

  • OB of propellants in burn pans. The human health risk assessment estimated air emissions for 37 pollutants released during OB of propellants in the burn pans. Emissions were estimated based on the maximum treatment quantities being proposed in SIAD's permit application, which were 13,000,000 pounds of propellants being treated per year and 150,000 pounds of propellants being treated per hour. Because SIAD typically did not operate at its permitted capacity, use of the maximum allowable waste treatment quantities likely overstated the actual emissions. ATSDR's comments on the specific emissions calculations approaches are similar to those for the OD waste treatment events:


    • Propellants and chemical by-products of OB. The human health risk assessment used two approaches to estimate emissions of propellants and their chemical by-products. First, 0.01% of all propellants treated by OB in burn pans were assumed to be emitted into the air. This fraction is consistent with treatment efficiencies reported in the Bangbox studies. Second, emission rates of chemical by-products of OB events were estimated using the Bangbox emission factors and results from other field tests. ATSDR believes these emission factors are an appropriate basis for characterizing releases from OB events, as summarized in our review of OD emissions.


    • Particulate matter. Air emissions of particulate matter were calculated from the observations in the Bangbox studies. Although ATSDR was concerned that these studies might not characterize crater ejecta emissions for OD events, this concern does not apply to the OB waste treatment operations that occur in burn pans, because these treatments occur entirely within large steel pans and not on an exposed soil surface. Therefore, ATSDR believes using the Bangbox emission factors to estimate particulate matter emissions from OB waste treatment in burn pans is appropriate.


    • Metals. Several propellant mixtures treated in the burn pans contain various metallic constituents, including barium, copper, and lead. The human health risk assessment assumes that all metals in the propellant wastes were vaporized during an OB event. This assumption likely overstates actual emissions, because the ash that remains in burn pans after an OB event contains metals. Therefore, the modeling analysis likely overstates air quality impacts of metals resulting from OB events. However, actual air quality impacts of metals have been characterized in a recent air monitoring study (see Appendix C.1).


    ATSDR found the approaches used to estimate air emissions from OB waste treatments in burn pans to be technically sound and useful for evaluating inhalation exposures to SIAD's emissions. Emissions were estimated for waste treatment quantities larger than the amounts of waste SIAD typically treated. The contaminants with the highest estimated emission rates were nitrogen oxides, PM10, methane, and leadtwo of which were measured during the recent ambient air monitoring program.

  • OB of rocket motors. The propellants and fuel in rocket motors differ considerably from the materials treated by other OB/OD operations at SIAD. Specifically, rocket motors contain relatively large amounts of chlorinated compounds (e.g., ammonium perchlorate), while the high explosives and other propellants treated at SIAD contain relatively small amounts of these chemicals. Consequently, air emissions from OB of rocket motors are expected to differ from those generated by OB/OD of other wastes. For this reason, the human health risk assessment based air emission estimates largely on source testing results conducted at United Technology Corporation, a company that treated rocket motors identical to many that SIAD treated in the 1990s. These source test results were scaled up to the maximum treatment quantities allowed at SIAD: 10,400,000 pounds NEW per year, with a short-term peak treatment rate of 160,000 pounds NEW per hour. These treatment rates exceed actual treatment rates by a considerable margin. Emissions for the different classes of chemicals released were calculated as follows:


    • Propellants and chemical by-products of OB. The human health risk assessment assumed that 99.99% of the organic propellants in the rocket motors was destroyed during a typical OB event. This value is consistent with data observed in the Bangbox studies for OB of propellant mixtures. Chemical by-products of the OB operations were calculated using the source test results from United Technology Corporation. ATSDR finds this approach to be appropriate.


    • Particulate matter. Air emissions of particulate matter also were based on the source test data from United Technology Corporation, but critical information is missing on how the treatment environments differ between United Technology Corporation and SIAD. At SIAD, rocket motor treatments occur in the OD pits, and the energy generated from the waste treatment clearly can cause soil particles to eject into the air. No information is provided in the human health risk assessment on the treatment environment at United Technology Corporation. As a result, ATSDR cannot assess whether the estimated particulate matter emissions are representative of actual emissions at SIAD.


    • Metals and other inorganics. The human health risk assessment assumes that all metals in the rocket fuel are emitted to the air during OB events, which accounted for the majority of aluminum emissions from OB events. Further, the chlorine atoms in all chlorinated compounds (e.g., ammonium perchlorate) were assumed to be emitted as hydrogen chloride. ATSDR believes the assumption made to estimate emissions of metals is appropriate, but the assumption regarding the fate of chlorine atoms is not necessarily conservative and would understate risks if these atoms actually become part of toxic organic compounds. ATSDR notes, however, that chlorides were among the pollutants measured in SIAD's recent ambient air monitoring program (see Appendix C.1), which provides some insights on the levels of chlorinated compounds that may be found in particulate matter following a rocket motor treatment.


    In summary, ATSDR found the approach to estimating air emissions from OB of rocket motors to be generally appropriate for assessing whether maximum treatment rates present human health risks. ATSDR has some concern about assumptions made to estimate air emissions of particulate matter and chlorinated compounds, but results from the recent ambient air monitoring project provide further insights on air quality impacts of these pollutants.

  • Incineration exhaust. The human health risk assessment included emissions estimates for 34 pollutants emitted from SIAD's incinerator. Estimates were based largely on incinerator stack test data collected at two other military installations (Tooele Army Depot and Lake City Army Ammunition Plant) that destroy very similar munitions wastes using virtually identical treatment technologies. For the organic constituents of the wastes treated, SIAD assumed that the incinerator had a destruction efficiency of 99.99%. Emissions estimates were based on a maximum treatment quantity of 3,170 tons of waste material per year. This amount is considerably higher than the actual waste treatment quantities observed in recent years, which have been as low as 6 tons per year. The pollutants accounting for the greatest fraction of estimated emissions from the incinerator were PM10, carbon monoxide, and sulfur dioxide. However, total emissions from the incinerator were far lower than those from SIAD's other waste treatment operations.

In review, SIAD's human health risk assessment includes emissions estimates for numerous pollutants from the four different waste treatment operations conducted at the installation. These emissions estimates are based on source tests conducted in the field and in experimental settings. Further, emissions were estimated assuming SIAD treats the maximum amount of waste material that would have been allowed had its air permit been approved, and this amount was considerably higher than the actual treatment rates. With few exceptions, ATSDR found the emissions estimation approach to be technically sound and scientifically defensible. This section lists several concerns ATSDR has regarding the emissions estimates for particulate matter, metals, and chlorides. However, ambient air monitoring data are available for these contaminants, which provides additional insights on their air quality impacts. ATSDR found no evidence that the human health risk assessment grossly understates actual emission rates.

D.1.2 Approach Used to Model Atmospheric Dispersion

The human health risk assessment used two models to evaluate atmospheric dispersion of contaminants released from SIAD. The INPUFF model predicted air quality impacts from all OB/OD operations, and the ISCST model predicted impacts from incineration. Additional detail on these modeling approaches follow:

  • INPUFF modeling. Emissions from OB/OD operations are instantaneous in nature, meaning emissions occur over a very short time frame, with no emissions occurring at other times. The INPUFF model was originally designed to simulate atmospheric transport for these "puff-like" releases in areas of simple terrain (EPA 1986). The model was not designed to represent photochemical reactions or dispersion in areas of complex terrain. INPUFF can be used to predict deposition of contaminants, but it does not deplete the levels of contamination in the plume when doing so. As a result, the model will tend to overestimate air concentrations of particulate matter, especially at receptors furthest from the source.

    • To account for plume impacts in complex terrain, the risk assessment authors used the "half-height methodology," which is documented in the Rough Terrain Diffusion Model (EPA 1987a). According to this methodology, plume heights are adjusted to lower values based on whether the elevation at a given receptor is higher or lower than the plume heights observed during the field tests (as described in the next paragraph). Although more sophisticated complex terrain modeling algorithms have been published, ATSDR believes the approach used in the human health risk assessment is sufficient for assessing the public health implications of inhalation exposures. We note that most populated areas in the vicinity of SIAD are not located in areas with complex terrain.

    The INPUFF model typically requires inputs that characterize the temperature and velocity with which a source releases contaminants. Such information is not well documented for OB/OD operations of munitions waste, largely due to the dangers associated with placing sensing devices near these events. To quantify these inputs, the authors relied on the findings of a plume height study, which measured the heights of plumes at SIAD following OB/OD events. For all cases, the temperature of the OB/OD plumes at the point of release was assumed to be 1,727 oC (or 2,000 K). INPUFF was run using multiple exit velocities to determine which input value would replicate the observed plume height.

    In the human health risk assessment, concentrations were modeled at numerous locations, or receptors, up to 50 kilometers away from SIAD. The modeling included evaluations to estimate the locations in California and Nevada believed to receive the greatest air quality impacts. Additionally, extensive data were presented on 33 selected receptor locations. These included 16 locations in California, 13 locations in Nevada, and 4 locations along the border between these states. Of these 33 locations, 11 were in areas where people live (either cities or individual residences) and 22 were in uninhabited areas in the mountains near the OB/OD area. The receptor locations reportedly included the areas in California and Nevada closest to where waste treatment operations occurred. ATSDR believes the placement of receptors was appropriate for the analyses documented in the human health risk assessment.

    The human health risk assessment describes the dispersion modeling evaluation as a "screening analysis." In refined analyses, site-specific meteorological data are typically used to characterize actual dispersion patterns. In screening analyses, wind direction generally does not factor into the dispersion modeling, and all receptors are assumed to be downwind from the source of concern. In this case, dispersion was evaluated considering 47 combinations of wind speed and atmospheric stability. These combinations include all reasonably anticipated conditions for the OB/OD waste treatment activities. Specifically, the evaluation was limited to neutral and unstable atmospheres, excluding periods of calm winds or strong, sustained winds. The meteorological condition found to be associated with the highest air quality impacts was used to estimate maximum 1-hour average concentrations. These, in turn, were multiplied by a factor of 0.08 to estimate annual average impacts, consistent with EPA guidance on screening assessments of atmospheric dispersion (EPA 1992). According to EPA's guidance on this matter, "...a degree of conservatism is incorporated in the factors [0.08] to provide reasonable assurance that maximum concentrations...will not be underestimated" (EPA 1992).

  • ISCST modeling. Emissions from the incinerator, which tends to operate in a continuous fashion, were evaluated using the ISCST dispersion model. EPA recommends use of ISCST for modeling continuous releases of air contaminants from stacks in areas with simple terrain, much like the conditions at SIAD. Version 3 of the ISCST model also includes algorithms to evaluate dispersion in complex terrain. Source release parameters (i.e., stack height, diameter, exit temperature, and exit velocity) were based on actual observations at SIAD. Using a set of 65 predefined meteorological conditions, ISCST was used to estimate the highest 1-hour average air quality impact from incinerator emissions at the same set of receptors considered for the INPUFF modeling. Annual average concentrations were estimated by multiplying the 1-hour maximum levels by a factor of 0.08.

Overall, ATSDR finds the dispersion modeling analysis documented in the human health risk assessment reasonably portrays the air quality impacts from SIAD's waste treatment operations. The models selected and approaches used are consistent with the current state of the science. ATSDR believes the modeling analysis greatly overstates short-term peak concentrations for most pollutants, because the analysis assumes that SIAD would conduct its maximum allowed OB and OD waste treatment operations during a single hour with the least favorable meteorological conditions. It is unlikely that the maximum treatment rates ever occurred due to practical limitations (e.g., OB of rocket motors takes place in the OD pits, and such OB events would likely limit the extent of a OD operations on a given day), and it is improbable that the single hour with the highest treatment quantity would also be the single hour with the least favorable meteorological conditions.

D.2 Utility of the Dispersion Modeling Analysis in ATSDR's Health Consultation

ATSDR concludes that the dispersion modeling analysis documented in SIAD's human health risk assessment is a useful consideration in this health consultation. The approaches taken are generally consistent with other published methodologies for assessing air quality impacts from OB/OD waste treatment operations (Bjorklund et al. 1998, EPA 1998). As with any modeling analysis, many decisions were made to calculate emission rates, assign model inputs, and to run the models.

ATSDR finds that several assumptions made caused the modeling analysis to overstate actual air concentrations. Representatives from the California Environmental Protection Agency's Office of Environmental Health Hazard Assessment reached a similar conclusion when commenting on the human health risk assessment: "The results of the exposure analysis are very conservative and overestimate potential risks from carcinogenic chemicals and health hazards from non-carcinogenic chemicals. Current risks and hazards are significantly less than those calculated in the HRA [health risk assessment] (Brown and Root Environmental 1996a)." ATSDR believes the estimated maximum hourly concentrations grossly overstate actual air quality impacts, because the concentrations are estimated for a highly unrealistic scenario: SIAD simultaneously treating the maximum amount of wastes allowed in all 14 OD pits and 30 OB burn stations, and this waste treatment occurring during an hour with the least favorable meteorological condition.

Overall, the modeling analysis provides useful insights on the upper bound exposure levels, and ATSDR emphasizes that actual exposure levels were likely considerably lower than those predicted by the modeling analysis. ATSDR also notes that valid, representative monitoring data are available for several contaminants that are emitted in largest quantities (e.g., particulate matter, metals, and hydrogen chloride). We considered both the measured and modeled data for these chemicals. Section V describes how we factored the dispersion modeling results into our technical analyses.

8 Ten sampling locations were originally proposed for the unpopulated area, but one of the proposed locations (DW7) was not installed due to difficulties field sampling personnel encountered accessing the site.
9 This analysis excludes an outlier observation from September 14 when the concentrations differed by more than 100 g/m3. This occurred on the windiest day of the sampling program, which might explain the large difference.
10 At the three stations in populated areas, program-average and the highest 24-hour average PM10 concentrations were more than 20 g/m3 and 60 g/m3 lower than EPA's corresponding health-based standards.
11 Samples were scheduled to be collected on four "background" days during this program, but valid "background"samples were collected at the upwind stations on only three of these four days.

Table of Contents


Error processing SSI file