V. Evaluation of Air Quality Issues

This section of the PHA presents ATSDR's analyses of the four inhalation exposure scenarios defined in Section IV. Each scenario is addressed in a separate subsection. These subsections start by presenting the key question and ATSDR's response, followed by a review of the sampling and modeling data ATSDR considered to reach the conclusion. More detailed reviews of the environmental contamination data for this site can be found in Appendix C (sampling data) and Appendix D (modeling results). Readers interested in only a brief summary of ATSDR's technical analyses should refer to those provided in Sections I and VIII of this PHA.

A. Exposures to Wind-Blown Dust

Key Question:

On days when bombing did not occur, did wind-blown dust from the LIA pose a health hazard?

ATSDR's Response:

On days without bombing exercises, wind-blown dust from the LIA did not cause air concentrations of particulate matter, metals, or explosives to reach levels that could potentially present a public health hazard levels in the residential areas of Vieques. In fact, the air sampling data suggest that wind-blown dust from the LIA accounts for an extremely small portion of the levels of air pollution currently measured in the residential areas. ATSDR concludes that wind-blown dust from the LIA on days when bombing did not take place is not a health hazard.

ATSDR used the following information to reach this conclusion: 443 air samples that PREQB collected in Esperanza and Isabel Segunda, and levels of contamination measured in 43 soil samples from the LIA. ATSDR believes these sampling data are of a known and high quality. The remainder of this section provides more detail on the data that support this conclusion.

Analysis:

Residents of Vieques have expressed concern to ATSDR about dusts from the LIA blowing potentially unhealthy levels of contamination into their neighborhoods, including on days when military training exercises did not take place. Concerns have been specific to dust (or particulate matter) and the possibility that this dust contained high levels of metals and explosives. ATSDR's evaluations of this issue are presented below, organized by the different classes of contaminants.

ATSDR notes that wind-blown dust is a natural phenomenon, and the amount of dusts blown into the air is determined both by soil properties and local weather conditions. An EPA model of this phenomenon, for example, suggests that the amounts of dust generated by winds depend on the wind speed, the fraction of soil covered by vegetation, the relative size of soil particles, and other factors (EPA 1985). Because these parameters do not change considerably from one year to the next, the amount of wind-blown dust is not expected to exhibit considerable annual variations.

ATSDR notes that the LIA soils clearly release dust into the air as a result of steady winds blowing over this land and much of the area not being covered with dense vegetation. This dust may contain contaminants that are in the LIA soils. Some of the dust that blows into the air settles back to the ground, some deposits in the ocean, and a small fraction may remain airborne for longer time frames. To assess whether the dust releases present public health hazards, ATSDR had to evaluate whether dusts blow into the residential areas in appreciable quantities.

ATSDR believes the best approach to evaluating this scenario is to examine the air sampling results that PREQB has collected in the neighborhoods where people live. At the time we completed this report, ATSDR had access to 443 valid air sampling results for "particulate matter" (see text box on the following page) that were collected by PREQB. The samples were collected between July 2000 and December 2002. As Appendix C.1 states, ATSDR believes PREQB's data are of a known and high quality and sufficient for use in the public health assessment process.

Following are ATSDR's specific interpretations of the available sampling data that pertain to the issue of wind-blown dust.

• Total suspended particulates (TSP). Wind-blown dust includes different size fractions of particulate matter (see text box on the following page). However, larger particles (e.g., TSP) are more likely to settle back to the ground surface near their source than are smaller particles (e.g., PM10, particulate matter having aerodynamic diameters less than or equal to 10 microns). In other words, TSP is less likely than PM10 to transport from the LIA to the residential areas. Nonetheless, ATSDR evaluated the levels of both TSP and PM10 that might be blown into the air.

  ATSDR identified four air sampling studies that measured ambient air concentrations of TSP on Vieques (see Appendices C.1, C.4, C.5, and C.6). Due to data quality concerns regarding three of these studies, ATSDR bases its conclusion for wind-blown dust entirely on the data recently collected by PREQB. Not only are PREQB's data well-documented and collected using rigorous methods, but they are the only extensive account of TSP levels in locations near where people live: one sampling station is in Esperanza, and the other is in Isabel Segunda (see Figure 6). PREQB started collecting 24-hour average air samples at these stations every sixth day in July 2000, and sampling data are currently available through December 2002. Thus, samples have been collected during all seasons of the year. At the time this document was prepared, 222 valid TSP measurements were available to ATSDR for PREQB's two sampling stations on Vieques. As Appendix C.1 describes, the average levels of TSP measured at Esperanza and Isabel Segunda are 40.4 µg/m³ and 34.0 µg/m³, respectively, both of which are considerably lower than EPA's former annual health-based air quality standard for TSP (75 µg/m³). Similarly, the highest 24-hour average TSP concentrations observed in Esperanza and Isabel Segunda (163 µg/m³ and 177 µg/m³, respectively) are considerably lower than EPA's former 24-hour average health-based air quality standard (260 µg/m³). These comparisons indicate that wind-blown dust from the LIA between July 2000 and December 2002 did not cause levels of air pollution that could present a public health hazard. Further, because the total emissions of wind-blown dust from the LIA are not expected to vary considerably from year to year, it is reasonable to assume that wind-blown dust from the LIA has not caused levels of air pollution, that could present a public health hazard, in years when sampling did not take place. Therefore, wind-blown dust from the LIA did not cause ambient air concentrations of TSP in Esperanza and Isabel Segunda to reach levels that could present a public health hazard.

  To analyze this issue further, ATSDR examined whether ambient air concentrations of TSP were higher on days with strong winds, as one would expect if wind-blown dust truly accounted for a large portion of TSP in the residential areas of Vieques. Based on average wind speed data observed at US Naval Station Roosevelt Roads (NCDC 2001)⁽⁴⁾, ATSDR found the ambient air concentrations of TSP in both Esperanza and Isabel Segunda to be essentially uncorrelated with wind speed (R² = 0.028 for Esperanza, R² = 0.030 for Isabel Segunda). This observation suggests that the TSP levels measured in Esperanza and Isabel Segunda are not strongly affected by wind-blown dust from the LIA, but rather are more likely affected by local sources.

• PM10. The PREQB 2000-2002 air sampling data are the best indicators of potential inhalation exposures to PM10 in the residential areas of Vieques. When ATSDR completed this PHA, 221 valid 24-hour average ambient air concentrations of PM10 were available for the sampling stations in Esperanza and in Isabel Segunda. Figure 6 shows where these samples were collected.

  As Appendix C.1 notes, the average PM10 concentrations observed in Esperanza and Isabel Segunda (34.1 µg/m³ and 23.5 µg/m³, respectively) are lower than EPA's current annual average health-based standard for PM10 (50 µg/m³). Further, the highest 24-hour average PM10 concentrations observed in Esperanza and Isabel Segunda (79 µg/m³ and 94 µg/m³, respectively) are lower than EPA's corresponding 24-hour average health-based standard (150 µg/m³). ATSDR concludes from these observations that wind-blown dust from the LIA does not cause PM10 to reach levels that could present a public health hazard in the residential areas of Vieques.

  ATSDR also examined correlations between measured PM10 concentrations and daily average wind speed, but found that these observations also were virtually uncorrelated (R² = 0.037 for Esperanza, and R² = 0.045 for Isabel Segunda). The lack of correlation suggests that wind speed has essentially no effect on PM10 concentrations measured in the residential areas of Vieques–a trend that implies that wind-blown dust from the LIA accounts for a small portion of the PM10 that residents are breathing.

• Metals. Airborne particulate matter in all parts of the country contains trace levels of metals. The amounts of metals within these particles is one of the factors that may be used to determine whether people will get sick from breathing the air. To evaluate potential exposures to metals at Vieques, ATSDR first tried to access all valid air sampling results from existing studies. The only study that has collected such results is PREQB's ongoing air sampling in Vieques (see Appendix C.1). ATSDR has requested access to PREQB's sampling results (ATSDR 2001d), but has not yet received copies of the metals sampling data. Until these data are provided, ATSDR can only estimate the ambient air concentrations of metals on days when military training exercises did not occur. The rest of this section presents these estimates.

  Wind-blown dust causes surface soils, and metals within or attached to these soils, to become airborne. Therefore, if wind-blown dust were the only source of particulate air pollution, a reasonable assumption would be that the concentrations of metals within the airborne dust are the same as the concentrations of metals within the surface soil from which the dust originated. ATSDR used this approach to estimate ambient air concentrations of metals on days when bombing did not occur⁽⁵⁾. Specifically, ATSDR estimated the air concentrations by multiplying the average concentration of PM10 (34.1 µg/m³ in Esperanza) by the average metals concentrations in surface soils in the LIA (ATSDR 2001b).

  Table 4 compares the estimated ambient air concentrations using this approach to corresponding health-based comparison values. With one exception, the estimated annual average air concentrations of all metals considered are lower than their corresponding health-based comparison values. As the exception, the estimated ambient air concentration of arsenic (0.0003 µg/m³) is slightly higher than the lowest health-based comparison value (0.0002 µg/m³). Examining potential exposures further, ATSDR notes that the estimated concentration (0.0003 µg/m³) is within the range of ambient air levels of arsenic reported for remote areas in the United States and is lower than the ranges reported for rural and urban settings (ATSDR 2000a). Moreover, the estimated air concentration is considerably lower than the range of exposure concentrations (0.7-613 µg/m³) that have been shown to cause harmful health effects in humans (ATSDR 2000a). Because the estimated average ambient air concentrations for nearly every metal considered is lower than their corresponding health-based comparison values, and because the levels of arsenic are not of health concern, ATSDR concludes that the metals in wind-blown dust on Vieques did not present a public health hazard on days when military training exercises do not take place.

  Two assumptions made when evaluating exposures to metals in wind-blown dust deserve further attention. First, ATSDR used the comparison value for trivalent chromium to screen concentrations of "chromium" listed in Table 4. The available sampling data do not indicate whether the chromium detected is in the trivalent or the potentially more harmful hexavalent state. Knowing that chromium in soils tend to be in the trivalent state and that chromium air emissions from most combustion-related sources (to which explosions are similar) are believed to contain less than 1% hexavalent chromium (ATSDR 2000b), the use of the trivalent comparison value is an appropriate selection.

  Second, the data in Table 4 can be compiled in different fashions. For instance, one can attempt to construct maximum concentrations (rather than average concentrations) by multiplying the highest PM10 concentration by the highest metal content observed in surface soils. ATSDR performed such calculations, which did not result in any metals concentrations significantly higher than comparison values and levels of significant exposure appropriate for acute exposure scenarios. Therefore, metals in wind-blown dust on Vieques are not a public health hazard, both for short-term and long-term exposures.

• Explosives. According to the documents ATSDR has reviewed, no agencies or researchers have attempted to measure ambient air concentrations of explosives in the residential areas of Vieques. ATSDR estimated concentrations for this group of contaminants using the same approach we used to estimate concentrations of metals. Specifically, ATSDR multiplied annual average PM10 concentrations in Esperanza by the average concentration of explosives measured in the soils of the LIA (ATSDR 2001c). This approach almost certainly overestimates the actual concentrations of explosives by assuming that 100 % of the PM10 originates from the LIA. However, airborne particles in the residential areas clearly do not originate only from the LIA and many of the local sources of particulate matter (e.g., mobile sources) release particles that do not contain explosives. Nonetheless, ATSDR proceeded with this approach for a reasonable upper-bound estimate of actual exposures.

  Table 5 presents the estimated ambient air concentrations, which show that the levels of explosives are considerably lower than their corresponding health-based comparison values. In fact, the estimated ambient air concentrations of explosives are so low that they would not be detected by routine explosive sampling procedures. Based on this analysis, ATSDR concludes that ambient air concentrations of explosives, as with particulate matter and metals, did not reach levels that could present a public health hazard on days when military training exercises do not occur.

The previous analyses indicate that, on days without military training exercises, the levels of air pollution at Vieques did not present a public health hazard. In fact, the concentrations of most pollutants are orders of magnitude lower than levels believed to cause adverse health effects. This conclusion is based on a large set of sampling data, including 443 air samples collected in Esperanza and Isabel Segunda by PREQB and levels of contamination measured in the soils of the LIA. Though ATSDR believes these sampling results form an adequate basis for reaching this conclusion, the Agency is committed to reviewing the ambient air concentrations of metals that PREQB has been measuring in Esperanza and Isabel Segunda, once these data become available.

B. Exposures to Releases from Military Training Exercises Using "Practice" Bombs

Key Question:

Did contaminants released when the Navy uses "practice" bombs pose a health hazard?

ATSDR's Response:

From April 1999 to May 1, 2003, all bombing activities on Vieques were limited to use of practice bombs, or bombs that have almost all their explosive content replaced with an inert material, like sand or concrete. Exercises involving practice bombs released contaminants into the air, primarily dusts and chemicals that were previously found in the LIA soils.

The available sampling data indicate that ambient air concentrations of particulate matter in the residential areas of Vieques were higher on days with military training exercises involving practice bombs than they are on days when no exercises occur, though most of the differences were not statistically significant. Additionally, the concentrations of particulate matter were virtually uncorrelated with the weight of practice bombs that were dropped, meaning that levels of air pollution are not consistently worse on days with the most intense exercises. These observations indicate that no clear relationship exists between military training exercises using practice bombs and ambient air concentrations of particulate matter in the residential areas of Vieques.

Regardless of the results of the statistical comparisons, PREQB's sampling data clearly indicate that ambient air concentrations of particulate matter have not reached levels that could present a public health hazard in the residential areas of Vieques on days of military training exercises involving practice bombs. This finding is based on 51 valid ambient air samples that PREQB collected on 16 days when the Navy conducted air-to-ground and ship-to-shore training exercises between August 2000 and October 2001. ATSDR believes these sampling data are of a known and high quality. Furthermore, ATSDR estimated ambient air concentrations of metals and explosives for days when the Navy dropped practice bombs on Vieques, and these estimated concentrations were all lower than levels known to cause adverse health effects. ATSDR concludes, therefore, that levels of air pollution on days with military training exercises involving only practice bombs presented no health hazard to the residents of Vieques.

As Section III.D describes, the nature and extent of military training activities at Vieques changed after April 19, 1999, when a bombing accident killed a civilian guard. From that date through May 1, 2003, a Presidential executive order required that only practice bombs be used during these activities. Practice bombs have their entire explosive charge replaced by a non-explosive material, usually sand or concrete. Some of the practice bombs have very small quantities of explosives that are used for spotting purposes.

Figure 7 depicts the emissions that were typically associated with military training exercises using practice bombs. As the picture shows, emissions were generated when practice bombs impacted the ground. The force of this impact could create a small crater, and the soil ejected from this crater typically became airborne. Small pieces of the practice bomb might also have become airborne. After impact, however, most soil and bomb particles fell to the ground, often within a short distance of the crater. A portion of the soils that the practice bombs eject into the air remained airborne and traveled downwind. These emissions not only included soils, but any contaminants that were previously in the soils, including metals and explosives. Though emissions clearly occurred, the amounts of exposure are determined by where these contaminants went, at what levels, and for how long. The following paragraphs address these factors.

ATSDR believes an adequate set of sampling data are currently available to evaluate potential inhalation exposures during the military training exercises involving practice bombs, without the need for air quality modeling for this scenario. Specifically, as of the writing of the public comment release PHA, range utilization statistics indicate that the Navy dropped practice bombs on the LIA on nearly 80 days since April 19, 1999,⁽⁶⁾ and valid ambient air samples for particulate matter were collected in the residential areas of Vieques on 16 of these days. In other words, valid air samples were collected approximately one out of every five days when the Navy conducted military training exercises using practice bombs.

Though the sampling data did not capture every single practice bombing event, they provide useful perspective on the extent to which these activities contributed to exposures. Following is ATSDR's interpretation of potential inhalation exposures to airborne contaminants generated by use of practice bombs. These analyses are presented for four different groups of compounds: two forms of particulate matter (TSP and PM10), metals, and explosives.

• TSP. Table 6 summarizes PREQB's 24-hour average TSP sampling results collected in Esperanza and Isabel Segunda, both on days with no military training exercises and on days when exercises took place using practice bombs (see also Appendix C.1). These data indicate three important trends. First, the highest level of TSP measured on days when practice bombs were used (124 µg/m³) was considerably lower than EPA's former health-based standard for 24-hour average concentrations (260 µg/m³). Additionally, the average TSP concentrations in the residential areas on days with exercises involving practice bombs (53.3 µg/m³ in Esperanza and 43.8 µg/m³ in Isabel Segunda) were lower than EPA's former health-based standard for annual average concentrations for this pollutant (75 µg/m³). Thus, ATSDR concludes that the ambient air concentrations of TSP on days with military training exercises using only practice bombs did not present a likely public health hazard.

  Second, the data trends indicate that average concentrations of TSP on days with exercises using practice bombs were higher than the average concentrations on days without this activity, but these differences were not statistically significant. The lack of statistically significant differences results largely from the fact that only a limited number of TSP samples have been collected on days when exercises involving practice bombs have taken place.

  Third, for days with military training exercises involving practice bombs, ATSDR compared the concentrations of TSP measured in Esperanza and Isabel Segunda to the total weight of the bombs that were dropped. ATSDR conducted this analysis to test a hypothesis: if emissions from practice bombs truly accounted for a very large fraction of particulate matter measured in the residential areas of Vieques, then concentrations of TSP would likely be positively correlated with the weight of the bombs dropped. ATSDR found, however, that the ambient air concentrations of TSP in the residential areas of Vieques were essentially uncorrelated with the weight of practice bombs dropped (see Table 7). To a first approximation, the lack of correlations suggests that emissions from practice bombs was not the dominant factor affecting air quality on days when military training exercises take place.

  Without statistically significant differences in concentrations between days with and without practice bombing, and without correlations between the concentrations and the measured TSP concentrations, ATSDR concludes that no clear relationship existed between the military training exercises conducted with practice bombs and air quality in the residential areas of Vieques. More importantly, none of the 222 TSP concentrations measured on Vieques to date, including the 25 TSP concentrations measured during military training exercises with practice bombs, have exceeded levels of health concern.

• PM10. Table 6 presents a similar summary for PREQB's 24-hour average PM10 sampling data collected on Vieques on days when military training exercises have taken place using practice bombs (see also Appendix C.1). The conclusions from this table are also similar. First, none of the measured PM10 concentrations on days with training exercises using practice bombs exceeded EPA's 24-hour average health-based standard (150 µg/m³) and the average concentrations did not exceed EPA's annual average health-based standard (50 µg/m³). Therefore, ATSDR concludes that on days when military training exercises take place with practice bombs, no exposures occurred that presented a public health hazard.

  As Table 6 shows, average concentrations of PM10 on days when practice bombs were used are higher than the average levels on days without military training exercises at both Esperanza and Isabel Segunda; the difference is not statistically significant at Esperanza, and is statistically significant at Isabel Segunda. Even though a statistically significant increase was observed at Isabel Segunda, ATSDR emphasizes that the sampling data are not sufficient for drawing conclusions on what source or sources most likely account for this difference. As evidence of this, Table 7 illustrates that PM10 concentrations in the residential areas of Vieques were essentially uncorrelated with the weight of practice bombs that were dropped.

  In summary, ATSDR concludes that the available sampling records, which have been collected on days with military training exercises of varying intensity, indicate that ambient air concentrations of PM10 on Vieques did not present a public health hazard, even on days when military training exercises using practice bombs took place.

• Metals. As Figure 7 illustrates, practice bombs displaced soils at the LIA when the bombs hit the ground. The soils that were ejected into the air contain metals, which included both naturally occurring metals and metals that may have accumulated soils over the years that the Navy conducted military training exercises at Vieques. To assess potential exposures to these metals, ATSDR requested access to the air sampling data that PREQB collected on these contaminants (ATSDR 2001d), but did not receive those data. Without access to the measured air concentrations of metals in the residential areas, ATSDR estimated potential inhalation exposures using an understanding of how emissions are generated.

  Upon impact with the ground, practice bombs tended to break into fragments and smaller pieces. Because practice bombs do not contain large explosive charges, the impacts were not accompanied by high-temperature explosions that have the potential to vaporize bomb casings. The main contaminants released by the impacts, therefore, are the soils that are displaced when the practice bombs hit the ground surface. These soils undoubtedly contained some level of metals, both naturally occurring minerals and contaminants that resulted from the Navy's history of conducting military training exercises on Vieques. Because the practice bombs impacted various locations on the LIA, the concentration of metals in the soils that become airborne was likely comparable to the average concentration of metals in soils throughout the LIA.

  To evaluate potential exposures to metals, ATSDR estimated exposure concentrations following the approach used to evaluate exposures to metals in wind-blown dust. Specifically, ATSDR assumed that the ambient air concentrations of particulate matter in the residential area of Vieques were composed entirely of soils ejected from the LIA by practice bombs. By this approach, the exposure concentrations for metals were calculated by multiplying the measured ambient air concentrations of particulate matter and the average soil concentrations from the LIA. ATSDR found that the estimated ambient air concentrations of all metals considered were lower than health-based comparison values, except for arsenic⁽⁷⁾. Estimated ambient air concentrations for arsenic were within the range of those reported for remote areas of the United States and are not of health concern. ATSDR concludes, therefore, that military training exercises involving practice bombs did not cause ambient air concentrations of metals to reach levels that could present a public health hazard in the residential areas of Vieques.

  Some additional observations deserve further attention. First, the assumption that soil ejected from the LIA accounted for all of the particulate matter in the residential area of Vieques does not account for potential contributions from local sources (e.g., motor vehicles, construction activities, outdoor fires). It is likely, therefore, that sources other than those related to Navy training exercises contributed to actual ambient air concentrations of metals during military training exercises. ATSDR will consider this scenario when reviewing metals sampling data collected by PREQB, once they are provided. Second, while researchers may debate the exact quantity of metals emitted when practice bombs impact the ground surface, ATSDR believes metals emissions from practice bombing events are unquestionably less than the emissions that occur when live bombs (of the same weight) impact the ground surface. Because ATSDR's air quality modeling analysis for live bombing scenarios (see Section V.C) suggests that ambient air concentrations of metals did not exceed levels of health concern when the Navy used live bombs, one can reasonably infer that ambient air concentrations of metals during practice bombing exercises also are safely below levels of health concern.

  In summary, ATSDR's analyses indicate that the amounts of soil on the LIA that became airborne during practice bombing exercises did not carry levels of metals that could have presented a public health hazard to the residential areas of Vieques. This conclusion is based on PREQB's ambient air sampling data for PM10 and TSP and reasonable assumptions regarding the composition of these pollutants. PREQB has already collected additional data that likely provide additional perspective on exposures to metals. As Section IX notes, ATSDR remains committed to evaluating the public health implications of these data, once PREQB releases them to ATSDR.

• Explosives. During military training exercises using practice bombs, explosives might have been released to the air in two ways. First, spotting charges in these bombs might have released trace amounts of explosives, though amounts of explosives in these charges are far less than the high explosive charge in most live bombs. Further, range utilization statistics indicate that the total amount of explosives used in an entire day of practice bombs never exceeded the amount of explosives found in a single 1,000-pound live bomb. Second, falling practice bombs might have released soils to the air that were contaminated with explosives during the time when the Navy conducted military training exercises using live bombs. However, the soil sampling data ATSDR previously reviewed suggest that the LIA soils contain only trace amounts of explosives (at the part per million (ppm) level, see Table 5).

  ATSDR used two approaches to evaluate whether practice bombs caused explosives to be released to the air in levels that could have presented a public health hazard. First, according to the analysis of wind-blown dusts (see Table 5), the estimated ambient air concentrations of explosives were more than 1,000 times lower than health-based comparison values. Given that ambient air concentrations of particulate matter on days when practice bombs were used were not considerably different from those on days when no bombs were dropped, it is highly unlikely that emissions caused by practice bombs could increase the estimated levels of explosives by a factor of 1,000.

  Second, ATSDR notes that its air quality modeling analysis indicates that estimated ambient air concentrations of explosives did not reach levels that could present a public health hazard in the residential areas of Vieques, even when the Navy was using live bombs (see Section V.C). Because the amounts of explosives in practice bombs are substantially lower than the amounts in live bombs, one can reasonably infer that explosives released from practice bombs also did not cause ambient air concentrations of explosives that could have presented a public health hazard in the residential areas of Vieques. Thus, ATSDR's air quality modeling results indicate that emissions of explosives during military training exercises using practice bombs did not lead to ambient air concentrations of explosives of health concern.

The previous analyses indicate that, on days with military training exercises using practice bombs, the levels of air pollution at Vieques do not present a public health hazard. Both measured air concentrations and estimated air pollution levels are considerably lower than levels believed to cause adverse health effects. This conclusion is based largely on routine air sampling conducted by PREQB.

C. Exposures to Releases from Military Training Exercises Using "Live" Bombs

Key Question:

Did the contaminants released when the Navy used "live" bombs pose a health hazard?

ATSDR's Response:

ATSDR thoroughly evaluated the public health implications of contaminants released to the air during the time when the Navy used live bombs. Because no sampling programs extensively characterized air quality on Vieques during live bombing exercises, ATSDR relied entirely on a modeling study to evaluate this exposure scenario. To do so, ATSDR estimated the amount of chemicals that would be released to the air during bombing exercises, and then the agency evaluated how those chemicals would move through the air to where people might inhale them.

ATSDR's conclusions on this question depend on the type of contaminant. ATSDR estimated ambient air concentrations for more than 80 different explosives, metals, and organic by-products of explosions. For all contaminants considered, the estimated ambient air concentrations were considerably lower than levels of potential health concern. Though the modeling analysis involves some uncertainty, the estimated concentrations for most contaminants were orders of magnitude lower than relevant health-based comparison values. As a result, ATSDR is confident that airborne levels of explosives, metals, and organic by-products of explosions were not at levels that could present a public health hazard during the time when the Navy used live bombs.

For particulate matter, ATSDR evaluated two scenarios: annual average exposures and short-term (or maximum 24-hour) exposures. Over the long term, particulate matter emissions from the LIA had relatively little impact on air quality in the residential areas of Vieques. In fact, ATSDR's best estimates suggest that, when averaged over the year, emissions from the LIA accounted for less than 1% of the particulate matter found in the air in Esperanza and Isabel Segunda.

When evaluating acute exposure durations, on the other hand, ATSDR found that short-term increases (e.g., over the course of a day) in particulate matter did occur during military training exercises. For a given day, the amount of the increase depended on local weather conditions and the amounts and types of ordnance the Navy used. Based on detailed scientific analyses of the best available information, ATSDR found that the short-term increases in particulate matter in the residential areas were not at levels of health concern, even during the most intense exercises. These analyses are based on calculations and air quality modeling studies that have inherent uncertainties and the actual air concentrations of particulate matter might be slightly higher or lower than the levels ATSDR predicted. However, ATSDR's modeling approach is based on several assumptions that likely overstate actual exposure concentrations. Overall, ATSDR's detailed modeling analysis indicate that no exposures to particulate matter occurred that could present a public health hazard as a result of the Navy's past training exercises using live bombs.

The following discussion presents a general overview of ATSDR's analysis of the public health implications of live bombing exercises on Vieques. Refer to Appendix D.3 for a technical description of the air quality modeling analysis used to evaluate this issue.

Military training exercises involving live bombs were part of the Navy's operations at Vieques for many years. As Section III.D explains, the most intense activity at Vieques started in the early 1970s, when the Navy gradually stopped conducting exercises on Culebra, and continued through April 19, 1999, when a bombing accident killed a civilian guard. Between the early 1970s and 1999, the Navy's use of live bombs greatly varied from month to month, and even from day to day. However, relatively small variations in bombing activity occurred from one year to the next (see Figures 4 and 5).

Because they contain high explosive charges, live bombs release more contaminants to the air than practice bombs. Figure 8 identifies the types of contaminants emitted and how they are formed. When live bombs impact the surface, an explosion almost always follows. These explosions are a series of chemical reactions that consume the high explosive charge and release large amounts of energy. For instance, some live bombs used at Vieques contained 2,4,6-trinitrotoluene, or TNT. During explosions, chemical reactions rapidly break TNT down into smaller molecules. These reactions release energy previously stored in the chemical bonds of TNT. The energy released causes the bomb casings to fragment, a crater to form, and dust to be ejected into the air.

Explosions from live bombs release many different contaminants to the air, which fall into four general categories: particulate matter, chemical by-products of explosions, metals, and the explosives themselves (e.g., TNT). Analyses later in this section describe how each type of contaminant is formed, the amounts that are released, and the amounts that might have been found in the air in the residential areas of the island.

The primary focus of this analysis is to characterize potential exposures that occurred during the time when the Navy used live bombs. Because the center of the LIA is located 7.9 miles away from the nearest residential areas of Vieques, all contaminants released from live bombs dispersed greatly in the air before reaching locations where they might have been inhaled. Nonetheless, as this section shows, residents of Vieques were likely exposed to trace levels of various contaminants on days when live bombing exercises took place. The fact that exposure occurred does not mean that adverse health effects resulted. After all, residents of Vieques, like residents throughout the United States, are exposed to air contaminants from many sources of air pollution on a daily basis. The key question is not simply whether exposure occurred, but rather whether exposures occurred at levels that might be harmful to human health.

To quantify exposures to chemicals released by explosions, ATSDR first examined the available air sampling data, or measurements of what residents of Vieques might have actually breathed. Unfortunately, very few air samples were collected during the time when the Navy used live bombs, and documentation of these sampling studies is either incomplete or missing (see Appendix C.4, C.5, and C.6). As a result, ATSDR had to use air quality models to evaluate exposures to chemicals released from live bombing activities. ATSDR emphasizes that air quality modeling results only estimate air pollution levels and the model output may be higher or lower than actual levels. This is not to say, however, that models are not useful in the public health assessment process, because rigorous modeling studies can generate convincing, scientifically defensible conclusions. The utility of a given study depends on the limitations and uncertainties of the model selected and the assumptions made when running the model Thus, ATSDR carefully reviews these factors before making any conclusions based on modeling results.

ATSDR identified two existing air quality modeling studies that estimated air quality impacts from live bombing activities at Vieques. One was conducted by a contractor to the Navy (IT 2000, 2001), and the other by a local professional engineer (Cruz Pérez 2000). ATSDR critically reviewed these studies and identified strengths and weaknesses in both of them (see Appendix D.1 and D.2). To have the best information available for this PHA, ATSDR eventually decided to conduct its own air quality modeling study of how military training exercises using live bombs might have affected air quality at Vieques (see Appendix D.3). The following discussion summarizes ATSDR's findings, organized by four groups of contaminants:

• Particulate Matter. When live bombs explode at the ground surface, the energy released forms craters and ejects soil particles into the air. The amount of particles released depends on many factors, such as the total weight of high explosives in the bomb, whether the bomb explodes at or below the surface, and properties of the soil where the bomb is detonated. The particles released to the air vary in size, which causes them to move in the air differently. Much of the soil ejected from craters, for example, immediately returns to the ground in large clumps and does not blow to downwind locations. Other soil particles are ejected high into the air during an explosion and settle to the ground in the immediate vicinity of the crater. Finally, a small fraction of the particles are small enough that they can remain airborne for extended periods of time and thus blow with the wind toward the residential areas of Vieques.

  To evaluate the public health implications of the particulate matter that live bombs released to the air, ATSDR first reviewed available air sampling data as documented in three air sampling studies that measured levels of airborne particles during the 1970s. As Appendix C.4., C.5, and C.6 indicate, none of these studies is well documented and the quality of the sampling results is not known. With no information on data quality, ATSDR decided not to base its conclusions on the limited sampling results.

  Without sufficient sampling data to reach a conclusion, ATSDR decided to use modeling analyses to put potential exposures to particulate matter into perspective. Appendix D.3 describes ATSDR's modeling approach in detail. This modeling involved two steps: first estimating the amount of particulate matter released to the air and then predicting ambient air concentrations in the residential areas of Vieques. In its analysis, ATSDR used a model that the Army Research Laboratory has developed and enhanced over the last 15 years to estimate the amount of soil particles an explosion releases to the air (Army Research Laboratory 2000). This model has many desirable features, including the ability to estimate (although roughly) the size distribution of particles released to the air. ATSDR specifically used the model to estimate emissions of PM10, the particles most likely to transport longer distances⁽⁸⁾. ATSDR's estimated PM10 emission rate (280 tons per year) is considerably higher than that documented in a dispersion modeling analysis performed by a Navy contractor (80 tons per year).

  Estimated emissions of PM10 are not a direct measure of exposure, but they can be used with air quality models to generate reasonable estimates of ambient air concentrations. ATSDR used its emissions estimates as an input to the CalPUFF air quality model to predict what levels of exposure might take place in the residential areas of Vieques, both over the short term and the long term. The following paragraphs summarize the modeling results:

  • Annual average concentrations. ATSDR's air quality model simulations indicate that the Navy's live bombing exercises at the LIA would have caused annual average PM10 concentrations in the residential areas of Vieques to increase by 0.04 µg/m³. Recent air samples collected when no military training exercises occurred, however, indicate that annual average PM10 concentrations in Esperanza and Isabel Segunda are 34.1 µg/m³ and 23.5 µg/m³, respectively. Therefore, PM10 emissions from the past live bombing exercises at Vieques probably accounted for less than 1% of the total PM10 to which residents were typically exposed. Figure 9 illustrates this further in two pie charts. In short, the models suggest live bombing exercises at Vieques had little impact on long-term average PM10 exposures in the residential areas. More importantly, reasonable estimates of annual average PM10 concentrations in both Esperanza and Isabel Segunda are lower than 50 µg/m³, EPA's health-based standard for annual average concentrations of particulate matter.
  • Maximum 24-hour average concentrations. Recognizing that the nature and extent of military training exercise vary from day to day, ATSDR conducted additional modeling to determine whether increased PM10 emissions over the short term caused acute exposures that could present a public health hazard. To do so, ATSDR reviewed nearly 7 years of range utilization statistics to identify the day on which the largest amount (by weight) of explosive ordnance were used during a military training exercise. This search identified a day of operation on which the Navy dropped 38.9 tons of high explosives on the LIA. ATSDR used this level of bombing activity to evaluate the maximum 24-hour average air concentrations of particulate matter that may have occurred when the Navy used live bombs.

    On this date, ATSDR constructed an upper-bound exposure scenario to evaluate the highest exposures that may have occurred. Appendix D.3 lists the assumptions made in this evaluation. In short, ATSDR derived a reasonable upper-bound estimate of emissions, or the amount of PM10 released to the air. ATSDR also reviewed 5 years of meteorological data to identify worst-case atmospheric dispersion conditions. Combined, both upper-bound assumptions suggested that this intense military training exercises using live bombs caused the 24-hour average PM10 concentration in residential areas to increase by 10.2 µg/m³. This increase in PM10 concentrations, even when added to the highest PM10 concentration measured at Vieques to date (94 µg/m³), suggests that maximum 24-hour PM10 concentrations in the residential areas likely did not exceed 104 µg/m³–a level lower than EPA's 24-hour health-based standard (150 µg/m³). ATSDR acknowledges that the uncertainty associated with predicting a maximum 24-hour concentration is typically greater than the uncertainty associated with predicting annual average concentrations. However, ATSDR notes that its estimated ambient air concentration is based on a series of events that occur infrequently (e.g., the highest level of bombing activity occurring on the day with both the least favorable meteorological conditions and the highest "background" concentration of PM10). The likelihood that these events truly coincide seems remote.

  In summary, ATSDR thoroughly reviewed potential exposures to PM10, drawing from the best information readily available. ATSDR's modeling suggests that, during the time the Navy conducted military training exercises using live bombs, residents of Vieques were not exposed to levels of particulate matter that could present a public health hazard, either over the long-term and the short-term. In fact, on the majority of days bombing exercises took place, ATSDR estimates that emissions from the explosions at the LIA account for a very small fraction of the PM10 in the air 7.9 miles downwind in the residential areas of the island. Appendix D.3 presents extensive details on ATSDR's dispersion modeling analysis on which this conclusion rests. It should be noted that ATSDR also considered the possibility of military training exercises releasing fine particulate matter to the air, and our modeling data indicate that air emissions of particulate matter, whether coarse or fine, were not at levels of health concern for the residential areas of Vieques. Our response to Comment #8 (see Appendix E) presents further information on fine particulates.

• Chemical By-products of Explosions. During an explosion, chemical reactions not only consume high explosives in bombs, but they also form a variety of explosion by-products, both organic and inorganic chemicals. The overwhelming majority of the explosion by-products are generally benign from a public health perspective. Examples include water vapor, nitrogen, solid carbon, and carbon dioxide–all of which are relatively abundant in the atmosphere. Several researchers have estimated that these by-products tend to account for a very large proportion of the overall amounts of chemicals that explosions release (e.g., Bjorklund et al. 1998; Cooper 1996; Defense Nuclear Agency 1981).

  To evaluate potential exposures to chemical by-products of explosions, ATSDR conducted a modeling analysis, because air quality measurements for almost all known explosion by-products are not available. As Appendix D.3 describes in detail, ATSDR's modeling analysis is based largely on studies that measured air emissions of explosion by-products for various types of high explosives. In these studies, called "Bangbox studies," explosives are detonated in an enclosed structure, after which the air within the structure is sampled for chemical by-products of the explosions. The sampling results can then be used to estimate emissions for explosions in similar scenarios.

  Table 8 presents ATSDR's estimated annual average concentrations for explosive by-products for the residential areas of Vieques that result from live bombing exercises. Thus, the concentrations in the table are estimates of how much levels of air contamination would increase in the residential areas as a result of the live bombing exercises. As the table shows, every estimated ambient air concentration is substantially lower than the corresponding health-based comparison values. In fact, for almost every contaminant considered, the estimated concentrations are much lower than levels most air sampling methods can reliably detect. ATSDR also estimated maximum 24-hour average concentrations, but these too were all lower than the corresponding health-based comparison values shown in Table 8. Based on these evaluations, ATSDR concludes that residents of Vieques were not exposed to chemicals formed as by-products of explosions in amounts that could present a public heath hazard.

  ATSDR recognizes the uncertainties associated with assuming that emission factors determined during static detonations in the relatively controlled setting of the "Bangbox" apply to the field setting in Vieques, especially considering that bombs at Vieques are fired from remote locations. However, two factors give ATSDR confidence that its use of the "Bangbox" emission factors did not lead to erroneous conclusions. First, the Bangbox studies tested the same type of explosive material that account for the majority of high explosives used at Vieques (e.g., various mixtures of TNT, RDX, and aluminum powder). Second, ATSDR's estimated ambient air concentrations are all several orders of magnitude lower than levels that might warrant more detailed evaluations. Thus, even if the Bangbox studies underestimate actual emissions by an extremely large factor, perhaps even 1,000, estimated ambient air concentrations for almost every contaminant considered would still be lower than the most conservative health-based comparison values.

• Metals. Metals are pervasive in the environment. At Vieques, metals are naturally found in soils and airborne dusts, but they are also found in the high explosives and bomb casings that the Navy fired at the LIA. Unfortunately, no ambient air samples were collected and analyzed for concentrations of metals during the time when the Navy fired live bombs at Vieques. Therefore, the only basis ATSDR has for reaching conclusions is using air quality models to estimate potential exposure concentrations.

  ATSDR identified and modeled the following sources of metals emissions associated with the Navy's former live bombing exercises: metals in the bomb casings that vaporize during an explosion; metals in the explosive charge that vaporize; and metals in the soil that is ejected into the air. Appendix D.3 describes the various assumptions that ATSDR made to estimate emission rates. For instance, to estimate the amount of metals released from casings, ATSDR assumed that all metals present vaporize during an explosion. This assumption clearly overstates emissions, because large pieces of bomb casings fragment in explosions and the metals in these fragments do not become airborne. Using this and other assumptions, ATSDR compiled reasonable emissions estimates for 28 different elements, with iron, aluminum, calcium, copper, and zinc having the highest emission rates.

  Table 9 lists ATSDR's estimates of how much annual average air concentrations of metals in the residential areas of Vieques increased as a result of the live bombing exercises. These increases are all lower than the metals' corresponding health-based comparison values. ATSDR also evaluated short-term increases in air pollution, and none of the estimated metals concentrations exceeded concentrations of potential concern for acute exposure scenarios (see Appendix D.3). Based on these observations, ATSDR concludes that any increase in ambient air concentrations of metals that resulted from live bombing exercises are of no public health significance. Appendix D.3 presents a detailed account of the data ATSDR considered to reach this conclusion.

• Explosives. The live bombs used at Vieques contained high explosive charges of varying quantities. Once initiated, an explosion is a series of chemical reactions that rapidly consume the high explosive charge and release large amounts of energy. Explosives within the charge are chemicals with a structure and composition that greatly facilitates the chemical reactions (i.e., oxidation reactions) that occur during an explosion.

  To estimate emissions and ambient air concentrations of explosives at Vieques, ATSDR first evaluated the proportion of explosive chemicals that are not consumed during an explosion and thus are available for downwind transport. In other words, ATSDR considered how efficient explosions are in destroying their high explosive charges. This efficiency has not been measured specifically for the bombing exercises at Vieques. However, researchers have reported that open burning and open detonation of explosives are much more than 99% efficient at destroying explosive chemicals (Radian 1996; Halliburton NUS 1995).

  Appendix D.3 describes how ATSDR evaluated releases and atmospheric transport of the explosives the Navy has used at Vieques. Based on an assumed 90% destruction efficiency and the maximum explosive content of ordnance used in 1998, ATSDR estimated the following exposure point concentrations:

  Estimated exposure point concentrations

  Explosive	Highest Estimated Air Concentration	Comparison Value
  RDX	0.002 µg/m3	0.057 µg/m3 (RBC-c)
  >TNT	0.003 µg/m3	0.21 µg/m3 (RBC-c)
  All others	< 0.0003 µg/m3	NA

  Notes:

  RDX = hexahydro-1,3,5-trinitro-1,3,5-triazine (CAS #121-82-4)

  The highest estimated air concentrations are the highest annual average concentrations estimated for locations in the residential areas of the island.

  The comparison values used in this table are both Risk-Based Concentrations for carcinogenic effects developed by EPA Region 3. See Appendix A for more information on these comparison values.

  These data show that the estimated ambient air concentrations for the explosives used in highest quantities are considerably lower than health-based comparison values, or levels that would require more detailed evaluations. Comparison of estimated annual average concentrations to the comparison values is appropriate, given that the comparison values are derived for long-term average exposure scenarios. In the table, "all other" explosives refer to various high explosive materials that comprise relatively small portions of high explosive charges. These include lead azide, HMX, and other impurities. The highest estimated ambient air concentrations for these compounds appear to be lower than highly sensitive sampling methods would be able to detect.

  ATSDR recognizes that the ambient air concentrations listed above are estimates and some uncertainty was involved in deriving them. Arguably the most critical assumption was assigning a destruction efficiency of 90% to the live bombing activities. ATSDR notes, however, that estimated ambient air concentrations of explosives would still be lower than health-based comparison values when considering a very wide range of destruction efficiencies. For instance, even if the destruction efficiencies were 10% (an unrealistically low value), the estimated ambient air concentrations would still be lower than health-based comparison values. Thus, all reasonable estimates of destruction efficiencies would lead to the same conclusion: Explosives in live bombs are chemicals that are largely destroyed during explosions. Reasonable modeling studies show that live bombing exercises did not release explosive chemicals at levels of health concern.

The previous analyses suggest that air pollution on Vieques did not reach levels that could present a public health hazard during the time when the Navy used live bombs. This conclusion is based entirely on ATSDR's air quality modeling study, which estimated ambient air concentrations that would result from live bombing exercises. Key assumptions, limitations, and uncertainties associated with the model are document throughout the previous paragraphs and, in far greater detail, in Appendix D.3. Though live bombing exercises release many contaminants, these contaminants disperse greatly in the air over the 7.9 miles that separates the center of the LIA from the nearest residential areas of the island. Contaminants disperse to even lower levels before they reach the more populated areas of Isabel Segunda and Esperanza, both located at further downwind distances.

Reasonable emissions estimates show that annual average concentrations of all contaminants considered were lower than corresponding health-based comparison values, often by very large margins. Increases of air pollution over the short term (i.e., on days with live bombing exercises) also were not at levels of health concern, even when considering releases from the most intense military training exercises.

Throughout this section, ATSDR has noted that air quality modeling studies can predict or estimate levels of air pollution, and modeling results should not be viewed as actual measurements of environmental contamination. Recognizing the limitations of environmental models, ATSDR usually recommends actions to reduce uncertainties in its public health evaluations based primarily on modeling results. We make no recommendations in this case, because past levels of air pollution obviously cannot be measured today.

D. Exposures to Releases Associated with Other Activities

Key Question:

Did open burning and open detonation or the Navy's past use of other chemicals (e.g., depleted uranium, chaff) pose a health hazard?

ATSDR's Response:

The following paragraphs present ATSDR's analyses of open burning and open detonation activities and the Navy's past use of chemicals and materials other than those released by bombs. These latter analyses focus specifically on depleted uranium and chaff. The best available information suggests that past open burning and open detonation activities and the previous usage of depleted uranium and chaff did not cause adverse health effects among residents of Vieques. In fact, estimated exposures to these materials are at levels considerably lower than levels believed to be harmful to human health.

• Open burning and open detonation (OB/OD). As Section III.D indicates, the Navy conducted OB/OD operations on Vieques to treat both unused waste munitions (i.e., munitions that were never dropped on the LIA) and unexploded ordnance collected during range clearance activities (i.e., munitions that were dropped on the LIA but did not detonate). The data available on the extent of the OB/OD operations are limited. Based on queries of EPA's Biennial Reporting System and data documented in the Navy's dispersion modeling analysis (IT Corporation 2001), ATSDR found waste management statistics for the OB/OD operations for the years 1993, 1995, 1998, and 1999. Data from these years indicate that the highest annual amount of wastes treated in OB/OD operations was 30.945 tons.

  ATSDR's evaluation of the OB/OD operations examined whether treating 30.945 tons of waste (whether waste munitions or unexploded ordnance) was expected to cause levels of air pollution to reach levels that could present a public health hazard, both over the short term and the long term. To evaluate short-term or acute exposures, ATSDR considered the possibility that the Navy used OB/OD to treat 30.945 tons of waste munitions on a single day. Recognizing that emissions from OB/OD treatment of waste munitions are likely not considerably different from emissions from munitions detonated in military training exercises, ATSDR used its conclusion for live bombing exercises to evaluate how OB/OD treatment may have affected air quality. Specifically, because a single day of live bombing exercises involving 38.93 tons of high explosives did not appear to cause ambient air concentrations to reach levels that could present a public health hazard (see Secti on V.C), it is reasonable to assume that OB/OD treatments involving 30.945 tons of waste munitions annually also do not cause exposures that could present a public health hazard in the residential areas of Vieques. ATSDR believes this assumption is justified because the composition of waste material treated in OB/OD operations is similar to the composition of material in live bombs.

  Regarding long-term or chronic exposures, ATSDR considered whether treating 30.945 tons of waste munitions over the course of a calendar year would contribute to levels of contamination that could present a public health hazard. To assess the impacts of these operations, ATSDR reflected on its findings for military training exercises involving live bombs, for which range utilization statistics indicate that the Navy detonated, on average, 353 tons of high explosives per year. In other words, the amount of high explosives treated in OB/OD operations at Vieques accounted for less than 10% of the amount of high explosives that were detonated during exercises involving live bombs. Based on these relative quantities, the OB/OD operations likely accounted for only small increases (less than 10%) in the estimated ambient air concentrations shown in Tables 8 and 9 and Figure 9 ⁽⁹⁾. Such increases would not have caused any of the estimated ambient air concentrations to exceed their corresponding health-based comparison values.

  Overall, these analyses indicate that OB/OD operations at Vieques, whether conducted to treat waste munitions or unexploded ordnance collected during range clearance activities, did not cause levels of air pollution that could present a public health hazard in the residential area of Vieques.

• Depleted uranium. Over the last 2 years, ATSDR has received several inquiries about the public health implications of the use of depleted uranium (DU) penetrators on the LIA during a February 1999 military training exercise. Specifically, residents have expressed concern that ongoing exercises at Vieques might cause soils potentially contaminated with DU to become airborne and blow downwind to the residential areas of the island. The following paragraphs address these concerns, first by summarizing past DU usage at Vieques and then by evaluating potential exposures. Based on ATSDR's analyses, as well as analyses conducted by the U.S. Nuclear Regulatory Commission (NRC), the amount of DU previously used at Vieques does not pose a public health hazard.

  Background information on DU. Uranium occurs in various chemical forms in nature. Naturally occurring uranium is actually a mixture of three different types (or isotopes) of uranium. All uranium isotopes are radioactive, meaning they are unstable and gradually decay through a series of transformations to form stable elements. Naturally occurring uranium is found at trace levels in rocks and soils throughout the world, including the rocks and soils on Vieques.

  Many industries process uranium to create materials for various products and purposes. A by-product from some of these industrial processes is depleted uranium (DU). Like naturally occurring uranium, DU is a mixture of isotopes. However, it is mostly depleted of certain radioactive uranium isotopes. As a result, DU is considerably less radioactive than the uranium typically found in nature. DU has been used to make a variety of products, including some aircraft, certain types of sailboats, and protective shielding for industrial applications.

  Because DU is a very dense material, the military uses DU in some types of ammunition, known as penetrators, which can travel through certain materials that other types of ammunition cannot. When fired upon tanks, rocks, or other hard objects, DU penetrators typically are crushed into fragments and dust and some of the DU may vaporize and ignite and eventually enter the air as aerosols (UNEP 1999). Because DU is dense, almost twice as dense as lead, it does not travel far in air and often deposits near its release point.

  When fired upon dirt and sandy surfaces, however, DU penetrators generally are not destroyed. Rather, they remain largely intact and penetrate as far as 1 meter beneath the soil surface (UNEP 1999). The DU in these penetrators will remain in the soil for extended periods of time. Eventually, the DU in the soils will either transport to other locations by various natural environmental processes, be removed from the soils by some type of man-made intervention (e.g., a clean-up activity or a military training exercise), or remain in place and gradually decay to form more stable elements.

  Usage of DU at Vieques. As Section III.D indicates, 263 DU penetrators were fired on the LIA during a military training exercise on February 19, 1999. These penetrators each contained 148 grams (about 0.33 pounds) of DU (Navy 1994). Overall, therefore, roughly 86 pounds of DU landed on the LIA. On March 5, 1999, the Naval Radiation Safety Committee notified the NRC of this unauthorized use of DU. Shortly thereafter, the Navy began an effort to remove all DU penetrators that could be identified in the LIA soils.

  To date, the Navy has removed the equivalent of 116 DU penetrators from the LIA soils, leaving the equivalent of 147 DU penetrators not accounted for. Accordingly, 38 pounds of DU have been removed from the LIA, and 48 pounds of equivalent penetrators have not been recovered. The fate of the unrecovered penetrators is uncertain: they might have fragmented and become airborne shortly after their use, they might have been buried in soils and become airborne during later military training exercises, or they might still be buried in the LIA soils at depths beyond the range of equipment used to detect the penetrators. ATSDR scientists who toured the field where the DU penetrators were recovered noted that the area is covered with soils without large rocks or boulders–a surface that DU ammunition is known to penetrate without significant fragmenting.

  Evaluation of potential non-radiological hazards. Studies of uranium toxicity have generally focused on two issues: whether uranium exposures present chemical hazards (to the kidney) and whether exposures presents radiological hazards. ATSDR considered both types of hazards when evaluating the public health implications of DU usage at Vieques. Findings specific to potential chemical hazards are presented first, followed by those specific to potential radiological hazards.

  To evaluate the chemical hazards associated with potential exposures to DU, one must first know where the DU transports in the environment, and at what levels. In June 2000, the NRC evaluated this issue by collecting 114 environmental samples for analysis of uranium content. These samples were collected from soils, sediments, surface water, and vegetation in the LIA, on other Navy property, and on the residential areas of Vieques. All environmental samples were analyzed in a laboratory, using methods known to generate high quality observations of uranium concentrations. Representatives from the Puerto Rico Department of Health witnessed, and assisted with, the NRC involvement at Vieques. Based on its sampling results, NRC concluded that ". . . there was no spread of DU contamination to areas outside of the LIA and that contamination from the DU inside the LIA was limited to the soil immediately surrounding the DU penetrators" (NRC 2000).

  ATSDR notes that NRC's findings are consistent with conclusions reached by the United Nations Environment Programme (UNEP) regarding the potential use of a similar quantity of DU penetrators in Kosovo in 1999 (UNEP 1999). Specifically, UNEP assembled a panel of international experts to examine the public health implications of the localized use of 22 pounds of DU–a scenario quite similar to the usage of DU at Vieques, where 48 pounds of DU have not been recovered. The UNEP analyses, which were based on modeling evaluations and not on sampling data, concluded that firing of 22 pounds of DU would cause no chemical toxic effects among people who did not visit the specific areas where DU penetrators were fired. UNEP evaluated whether people who inhale dusts when walking around a target area (after the DU penetrators had been fired) could breathe amounts of DU that might present a public health hazard. The UNEP conclusion was that the amounts of uranium inhaled in such circumstances, even by an individual who spent an entire year in the affected area, would not exceed levels known to cause chemical toxicity (UNEP 1999).

  In addition to the NRC and UNEP analyses, ATSDR conducted its own evaluation of the specific community concern (i.e., whether ongoing military training exercises are causing harmful air releases of the unrecovered DU). To conduct this evaluation accurately, one would need to know how much DU is released to the air, but such information is not available. As a defensible estimation, ATSDR assumed that the entire mass of unrecovered DU at Vieques has been released to the air by the various military training exercises that have taken place since February 1999. In other words, ATSDR assumed that the entire 48 pounds of unrecovered DU has been released between the time the DU was fired and today. Based on this and other assumptions,⁽¹⁰⁾ ATSDR estimated the DU emission rate to be no more than 0.017 pounds per hour. ATSDR emphasizes that this is an upper-bound estimate of the actual emission rate over the long term, because some of the unrecovered DU may still remain buried at depth.

  Combining this estimated emission rate with the findings of ATSDR's air quality modeling analysis (see Appendix D.3), ATSDR estimates that the long-term average ambient air concentration of uranium in the residential areas of Vieques attributed specifically to the DU usage at the LIA is likely not greater than 0.000008 µg/m³. This ambient air concentration is nearly 40,000 times lower than ATSDR's chronic inhalation minimal risk level (0.3 µg/m³). In other words, the estimated amounts of uranium that people of Vieques might breathe do not present a public health hazard, with a very large margin of safety.⁽¹¹⁾

  To put this estimated concentration into perspective, ATSDR calculated that a resident of Vieques might inhale a total of 56 nanograms (ng) of uranium per year from the past DU usage at the LIA–a finding based on the conservative assumption that all unrecovered DU from the LIA soils have been released by military training activities since February 1999. This estimated intake was calculated by multiplying the estimated air concentration by the average inhalation rate of an adult. As the table below shows, the estimated intake is considerably lower than the amounts of uranium that some people encounter in their daily lives:

  Estimated Uranium Intake

  Scenario	Estimated Uranium Intake
  Estimated amount of uranium inhaled from releases of unrecovered DU from the LIA at Vieques	56 ng/year
  Estimated amount of uranium inhaled from smoking two packages of cigarettes per week for a year	1,125 ng/year (a)
  Estimated amount of naturally occurring uranium ingested in normal dietary intake	328,500 ng/year (b)

  Notes:

  (a) Source of information: UNEP 1999.

  (b) Source of information: ATSDR 1999a. Intake of naturally occurring uranium selected is the lowest estimate of average daily intake in Chapter 5.5 of the toxicological profile. Naturally occurring uranium is found at trace levels in a variety of food products throughout the United States and the world.

  As the information above shows, the amounts of uranium that might be released from unrecovered DU penetrators and transported to the residential areas of Vieques are very low in comparison to the amounts of naturally occurring uranium that residents may encounter normally in their daily lives. Moreover, the estimated ambient air concentrations of uranium associated with past usage of DU penetrators are well below levels believed to cause adverse health effects in humans. Therefore, the DU penetrators that were fired at Vieques do not pose a health hazard in terms of their chemical toxicity.

  Potential exposure to radiation as a result of DU usage. Because uranium is radioactive, ATSDR evaluates potential exposures to radiation at most sites where uranium contamination has been documented. ATSDR notes, however, that both naturally occurring uranium and DU are weakly radioactive, and people exposed to large amounts of these types of uranium typically experience chemical toxicity effects before they experience effects of radiation (ATSDR 1999a). To be thorough, ATSDR evaluated potential exposures to radiation as a result of the DU usage on Vieques.

  Knowledge of the radiation that uranium emits is critical in evaluating the potential for adverse health effects to occur. When undergoing radioactive decay, all three isotopes of uranium that comprise DU release alpha particles (or alpha radiation) (ATSDR 1999a), and subsequent steps in the uranium decay series release other types of radiation. Alpha radiation has relatively low penetrating power and typically does not travel long distances in the environment. In fact, alpha particles typically travel less than 10 centimeters in air before they reach their resting point (ATSDR 1999b). Because the uranium isotopes in DU primarily emit alpha particles, decaying uranium at the LIA is expected to affect radiation levels only in very localized areas.

  Analyses by NRC and UNEP confirm that DU penetrators tend to affect radiation levels only in their immediate proximity, with virtually no impacts observed even short distances away. For instance, based on its extensive sampling project witnessed by the Puerto Rico Department of Health, NRC concludes that ". . . members of the public [on Vieques] could only have received a measurable dose from the DU penetrator event if they directly accessed a DU penetrator for extended periods of time" (NRC 2000). Similarly, UNEP concluded that radiation hazards among the population in Kosovo may exist for very limited scenarios, such as placing a DU penetrator in one's pocket and carrying it continuously for several weeks (UNEP 1999). Clearly, both exposure scenarios are not realistic for the population at Vieques and ATSDR concludes that no residents of the island are exposed to levels of radiation that could present a public health hazard as a result of past usage of DU penetrators during military training exercises.

  ATSDR is aware that a recent press release from the Committee for the Rescue and Development of Vieques (CRDV) reports that levels of radiation on Vieques increased during certain military training exercises–an increase the authors seem to attribute to the past use of DU penetrators at the LIA (CRDV 2001). Specifically, this press release suggests that levels of radiation at certain parts of Vieques increased by as much as 248% during military training exercises that occurred between July and October, 2001. However, the press release does not indicate how levels of radiation were measured, what types of radiation were measured, and the actual amounts of radiation detected, all of which are critical considerations when evaluating data on radiation. ATSDR has contacted CRDV to learn more about this sampling effort (ATSDR 2001e), but did not receive a response in time to address specific information in this release of the PHA. ATSDR will evaluate CRDV's data in greater detail in future rel eases of this PHA if data are received in a timely fashion.

  Although ATSDR cannot confirm that radiation levels increased on Vieques during recent military training exercises, ATSDR must emphasize that 248% increases in levels of radiation do not necessarily indicate that public health hazards are occurring. The more important indicator of exposure is the actual level of radiation, not the relative increase. However, ATSDR evaluated the public health implications of the reported increases in radiation nonetheless.

  In June 2000, NRC made dose rate measurements of radiation at 29 locations of the residential areas of Vieques using a Ludlum Model 19 microR meter (NRC 2000). These observations were collected at a distance of 1 meter above the ground surface and the average exposure rate of the 29 measurements was 4 microroentgens per hour (µR/hour), which is approximately equal to 4 microrem per hour (µrem/hour). ATSDR will assume for the following analysis that this dose rate represents background levels of external radiation in the residential areas of Vieques.

  If the CRDV data are based on similar dose rate observations, a 248% increase in radiation would imply that radiation levels increased from 4 µrem/hour to 14 µrem/hour, or a net increase above background of 10 µrem/hour. Even if ATSDR assumes this increase above background occurs 24 hours per day for 90 days per year (i.e., the maximum amount of time the Navy is currently allowed to conduct military training exercises on Vieques), the overall increase in radiation dose for the year would be 22 mrem–a level well below ATSDR's chronic MRL for ionizing radiation. This MRL is an increase in ionizing radiation dose of 100 mrem above background per year. Based on this analysis, ATSDR does not believe that CRDV's press release necessarily indicates radiation exposures at levels of concern. However, to be certain of this finding, ATSDR would like to review the original data compiled by CRDV before issuing the final release of this PHA.

  For perspective on the reported increases in radiation at Vieques, ATSDR notes that many activities that people undertake lead to increased exposures to radiation. Such increases are generally not viewed as unhealthy, but simply occur as people come closer to sources of radiation, such as the sun or certain medical equipment. For example, individuals who take round-trip flights across country typically receive increased radiation doses of 10 mrem during their air travel, by virtue of being closer to sources of cosmic radiation; and individuals who receive chest x-rays typically receive increased radiation doses of 14 mrem per procedure (ATSDR 1999b). Such increases due to single events are comparable to the increase in radiation that ATSDR calculated from the CRDV data collected on Vieques (see the previous paragraph). This comparison shows that periodic increases in exposures to radiation are not an adequate basis for judging whether adverse health effects might occur.

  Finally, ATSDR notes that the levels of radiation measured in the study cited by CRDV appear to be well within background levels observed throughout the United States. Specifically, a press release other than CRDV's announced that the highest level of radiation measured during the recent survey on Vieques was 18 µR/hour (Fellowship of Reconciliation 2001), which is approximately equal to 18 µrem/hour. Not only are these levels comparable to survey readings collected elsewhere in the United States, but they are actually considerably lower than background measurements from many areas at elevations of several thousand feet, such as Denver, Colorado (ATSDR 1999b). Although this information suggests that the levels of radiation measured at Vieques do not appear to be notably elevated, ATSDR hopes to review data provided by CRDV for a more complete analysis of the matter.

  Conclusion. Overall, ATSDR concludes that any exposures to uranium as a result of past usage of DU penetrators at Vieques are trivial in comparison to the daily exposures to naturally occurring uranium that residents experience through their diets and other activities. Further, ATSDR's conservative modeling analysis predicts that any exposures to uranium are considerably lower than levels believed to cause either chemical or radiological health effects in humans–a finding that is consistent with studies published by NRC and UNEP. Though the best information currently available indicates that levels of radiation on Vieques are not at levels of concern, ATSDR will review this finding further upon receipt of data collected by CRDV.

• Chaff. Both community members and various media reports have voiced concern about the public health implications of the Navy's past use of chaff at Vieques. As Section III.D indicates, chaff is a material that the military uses to confuse radar signals, which allows aircraft to operate without being easily detected. Chaff is aluminum-coated glass fibers. Therefore, the main metallic elements in chaff are aluminum and silicon–two of the most abundant elements naturally occurring in the Earth's crust. Chaff fibers typically are 25 microns (µm) thick and between 1 and 2 centimeters long (Naval Research Laboratory 1999). In other words, chaff fibers are visible to the human eye and have the appearance of short, very fine, hair-like fibers.

  ATSDR searched various records on the Navy's usage of chaff both at Vieques and across the country. Nationwide statistics indicate that the Navy's annual average usage of chaff between 1991 and 1997 at all domestic installations was 133 tons of chaff per year (GAO 1998). This includes amounts of chaff that were released from aircraft (69 tons per year, on average) and from ships (64 tons per year, on average). Chaff usage statistics specific to Vieques are not documented in any of the reports that ATSDR was provided and reviewed. Thus, ATSDR can only conclude that the previous chaff usage at Vieques was not greater than 133 tons per year and was probably considerably lower than this amount, since several Navy installations other than Vieques also used chaff.

  At Vieques, the Navy used chaff to conduct realistic military training exercises, in which hiding aircraft from radar sources was desired. During these exercises, chaff was intentionally released into the air in the offshore training area near the LIA. As Section III.D noted, the Navy prohibited chaff from being released directly over the island of Vieques and over the warning and restricted areas that extend several miles from the Vieques shoreline. Since the chaff was released at elevations where airplanes fly (i.e., several thousand feet above the ground), the fibers drifted in the wind and remained airborne over long distances. In fact, a recent study has suggested that chaff fibers can transport aloft for hundreds of miles before depositing on the ground (GAO 1998). This observation is consistent with data from weather radar signals, which have detected chaff particles floating in the air at locations several hundred miles from where they were rel eased. In short, when released high in the air, chaff fibers can drift over extremely large areas and are greatly dispersed before ever reaching the Earth's surface.

  ATSDR notes that no researchers quantified the fate of chaff used at Vieques (e.g., what amounts deposited on the island, and what amounts deposited in the ocean waters around the island). A general understanding of where chaff transports can be derived from some basic observations of the local geography. Specifically, the Earth's surface in the vicinity of Vieques (see Figure 1) is primarily covered with water. This observation, combined with knowledge that chaff released aloft can transport for hundreds of miles, suggests that much of the chaff used at Vieques probably deposited in waters surrounding the island, and a very small portion of the chaff that was released settles in the residential areas.

  To address health concerns related to chaff, ATSDR conducted two evaluations of potential exposure scenarios. ATSDR's first evaluation interpreted the existing air sampling data for particulate matter to assess potential impacts of chaff fibers on air quality. Given the shape and composition of chaff, one would expect that its greatest impacts on air quality, if any, would be observed in the measured concentrations of particulate matter (and possibly aluminum and silicon). Though ATSDR would prefer to base toxicological evaluations of chaff on published data documenting responses to actual exposures, no extensive data exist on exactly how chaff affects people who come into contact it. In the absence of such data, ATSDR evaluated the public health implications of the Navy's usage of chaff by characterizing exposures to the overall material (as PM10) as well as to chaff's principal components (aluminum and silicon).

  As Sections V.A and V.B explain, 443 air samples have been collected on Vieques and analyzed for particulate matter and not a single measurement has been at levels of potential health concern. The available sampling data, therefore, show no evidence of chaff significantly affecting concentrations of particulate matter on Vieques. ATSDR acknowledges, however, that only a small subset of the air sampling results were collected during military training exercises and ATSDR has no data on the corresponding amounts of chaff used during these exercises, if any.

  ATSDR's second evaluation considered chaff usage statistics and reasonable assumptions about how chaff moves through the air to estimate potential ambient air concentrations of the material. In this evaluation, ATSDR assumed a variety of daily chaff usage rates and general transport behavior, namely the area and depth over which chaff might evenly disperse. Moreover, ATSDR assumed that all airborne chaff fibers break up into particles small enough to be considered PM10–or particle sizes small enough to be considered respirable. This assumption almost certainly leads to an overestimate of potential exposures, because chaff fibers probably do not break into hundreds of pieces when settling in the atmosphere. (Chaff fibers, which are typically between 1 and 2 centimeters long, would have to break into hundreds of pieces in order to be measured as PM10.)

  Nonetheless, even under these assumptions that likely overstate potential exposures, ATSDR estimated that PM10 concentrations at Vieques would not increase by more than 4 µg/m³ by virtue of chaff usage⁽¹²⁾. This figure should be viewed as an upper bound of the actual air quality impacts of chaff: the increase in PM10 levels, if any, was probably much lower because chaff fibers almost certainly did not uniformly degrade into respirable particles. Regardless, an increase in PM10 of 4 µg/m³ over the background levels observed at Vieques, even if such an increase occurred, would not lead to a public health hazard, either in terms of particulate matter or in terms of the metallic components of the fibers.

  The outcome of ATSDR's evaluations is consistent with the general scientific understanding of chaff and how people might be exposed to it. For instance, a panel of independent experts from various universities and research institutes concluded that chaff fibers are too large to be inhaled into the lungs and are therefore not of health concern for inhalation exposure (Naval Research Laboratory 1999). The large particles would instead be collected in the mouth or nasal tract, and presumably may be ingested (or swallowed).

  ATSDR notes that the amount of aluminum that might be swallowed by chaff depositing in the mouth or nose is trivial in comparison to the quantities of aluminum that people consume in food products and medicines. Specifically, given that chaff is 40% aluminum by weight, ATSDR's previous analyses suggest that ambient air concentrations of aluminum resulting from chaff usage are likely no higher than 2 µg/m³ in the residential areas of Vieques (or half of the calculated increase in particulate concentrations, using conservative assumptions). This concentration should be viewed strictly as the highest estimated aluminum levels that might result from chaff usage.

  ATSDR evaluated this potential exposure by considering people who breathe air containing large particles with 2 µg/m³ of aluminum. Assuming all of these large particles deposit in the mouth and are swallowed, and assuming an average inhalation rate of 20 m³/day, an individual in this scenario will ingest 15 milligrams of aluminum from this source in a given year. This annual ingestion intake is the same amount of aluminum that people ingest from a single tablet of buffered aspirin or antacid (ATSDR 1999c). Thus, the usage of chaff does not cause people to ingest amounts of aluminum that could present a public health hazard.

  The previous review of sampling data and reasonable exposure scenarios all suggest that the usage of chaff at Vieques does not pose a public health hazard, whether the chaff particles are inhaled or deposited in the mouth and swallowed. This conclusion is based on sampling data of limited duration and realistic calculations of potential exposures.

---

⁴ When the public comment release of this report was prepared, ATSDR had obtained meteorological data through April 30, 2001, and ambient air monitoring data through March 30, 2002. ATSDR computed data correlations from this subset of data.
⁵ ATSDR acknowledges that source of air pollution in the residential areas of Vieques (such as mobile sources) undoubtedly release metals into the air. It is possible that emissions of metals from these local sources cause actual ambient air concentrations of metals to be higher than those listed in Table 4. This possibility can only be verified by reviewing the concentrations of metals in the PM10 filters collected in Esperanza and Isabel Segunda. As Section IX of this PHA indicates, ATSDR will review PREQB's sampling results as soon as they are released.
⁶ This figure accounts for all military training exercises that have occurred in calendar years 2000 and 2001. Further, the figure indicates the number of days on which the Navy actually dropped practice bombs or fired non-explosive ordnance from ships, not the number of days the Navy had scheduled to do so. In many cases, practice bombs are dropped on only a small subset of the days within a given military training exercise. ATSDR based this number of days on range utilization statistics that the Navy routinely compiles.
⁷ Table 4 presents ATSDR's estimates of ambient air concentrations of metals for exposures to wind-blown dust. These were calculated based on an average PM10 concentration of 34.1 µg/m³. The highest average PM10 concentration on days with military training exercises using practice bombs (40.1 µg/m³) was only marginally higher. Therefore, the estimated ambient air concentrations of metals during the practice bombing exercises are only marginally higher (roughly 18% higher, not enough of a difference to represent a public health concern) than those shown in Table 4. This PHA does not include a separate table to document these marginally higher levels.
⁸ Particles larger than PM10 are more likely to deposit on the ground than blow several miles down wind. As a result, ATSDR did not model TSP emissions.
⁹ As Appendix D.3 indicates, when estimating ambient air concentrations resulting from live bombing exercises, ATSDR assumed that every bomb used in an exercise detonated on the LIA. In reality, a small fraction of the bombs dropped do not detonate when dropped and remain on the LIA until range clearance operations collect these unexploded ordnance for waste treatment. As a result, ATSDR's dispersion modeling analysis for live bombing exercises actually accounts both for emissions during these exercises and emissions that result from treatment of unexploded ordnance.
¹⁰ For an upper-bound estimate, ATSDR assumed that military training exercises using practice bombs caused all of the unrecovered DU to be emitted to the air. To calculate an emission rate, ATSDR assumed that these exercises took place 16 hours per day on 90 days per year for 2 years of duration.
¹¹ The exposure scenario considered above–releases of uranium over a 2-year time frame–was used to address the specific community concerns that ATSDR received. In addition to this scenario, ATSDR evaluated other scenarios, such as the entire unrecovered amounts being released on a single day or during a 2-week military training exercise. Those evaluations also found estimated ambient air concentrations of uranium considerably lower than their appropriate health-based comparison values (i.e., acute-duration and intermediate-duration comparison values). The assumptions made in these evaluations are very conservative, since some DU penetrators will likely remain buried and not be entirely released over durations considered in this evaluation.
¹² In this particular evaluation, ATSDR assumed that the Navy uses 1 ton of chaff per day of military training exercises. ATSDR notes that this daily usage rate, if it were to occur on the maximum number of days that the Navy is authorized to conduct exercises on Vieques (i.e., 90 days), would account for nearly 90% of the Navy's annual chaff usage across the nation. In short, the assumed usage rate is an overestimate of the actual chaff usage. Next, ATSDR assumed that the chaff disperses evenly over an area of 150 square miles–an area approximately three times as large as Vieques. ATSDR also assumed that the chaff disperses evenly in the lowest 2,000 feet of the atmosphere. These assumptions likely overstates exposures, since radar images and engineering analyses have demonstrated that chaff released from planes can remain aloft for extended periods of time and transport over much larger distances (GAO 1998).