Click for next page ( 36


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 35
Exposure Assessment BACKGROUND The National Research Council (NRC) defines exposure to a contaminant as "an event that occurs when there is contact at a boundary between a human and the environment with a contaminant of a specific concentration for an in- terval of time; the units of exposure are concentration multiplied by time" (NRC 1991~. To reliably estimate exposure, environmental monitoring should be conducted to determine contaminant levels, modeling can be used to sup- plement the monitoring, and potentially exposed populations must be identi- fied and enumerated. Conducting a good exposure assessment requires char- acterizing the real variability in exposures that are experienced by different groups of people, and different individuals within those groups. In addition, a good exposure assessment integrates an analysis of the uncertainty of the ex- posure data. The California Department of Pesticide Regulation's (DPR's) risk charac- terization document provides exposure estimates for a wide variety of worker and resident exposure scenarios in Sections IV.B, Risk Assessment, and V.C, Risk Characterization, as well as in Appendices F-K. Summary estimates of exposure to methyl bromide, listed in Tables ~ 6-20 of the DPR report, corre- spond to occupational (Section {V.B.~) and residential (Section TV.B.2) expo- sure scenarios. The estimates of exposure presented in Tables ~ 6-20 are based on exposure data contained in a report "Estimation of exposure of persons to methyl bromide during and/or after agricultural and non-agr~cultural uses" by Thongsinthusak et al. (1999) (HS-1659), which is included as Appendix F of the main document. 35

OCR for page 35
3 6 METHYL BROMIDE RISK CHARY CTERIZA TION IN CALIFORNIA The exposure information collected in the DPR report came from numerous studies that were conducted for a variety of purposes by several registrants, and therefore were not conducted in a consistent manner nor were they part of a comprehensive and systematic monitoring plan. For instance, DPR points out that many of these studies were not conducted in compliance with Good Laboratory Practices as described in 40 CFR 160 (EPA 1997~. Although a variety of analytical techniques were used to determine methyl bromide con- centrations in the air samples, these were not reliably tested. In addition, data were collected under different sampling protocols and field conditions (e.g., temperature, relative humidity). For some exposure scenarios, DPR used "default" values due to the lack of specific data on the specific exposure sce- narios. In addition to the limitations described above, DPR acknowledges that the exposure data set is incomplete, as not all potential exposure scenarios are discussed. As stated in the Thongsinthusak et al. (1999) report, "The Depart- ment of Pesticide Regulation does not have data to assess all worker exposure scenarios or potential exposure to the public from all methyl bromide appli- cations." However, DPR fails to enumerate what these data gaps are. The lack of a discussion in the DPR report of the limitations of the exposure data set, including the data gaps, undermines the subcommittee's confidence in the data presented by DPR. The remainder of this chapter addresses the following three aspects of an exposure assessment; ( 1 ) the scenarios used to characterize different exposure groups; (2) the quality of data available for characterizing exposures, includ- ing the analytical methods used to quantify the air concentrations, and the rep- resentativeness of the available air sampling that was conducted; and (3) the modeling used to estimate exposures that were not directly measured. For each of these items, the subcommittee assesses DPR's treatment of the data and its methodology for estimating exposure. LIKELY EXPOSURE SCENARIOS The DPR document describes a wide variety of occupational and some resi- dential exposure scenarios. DPR presents valuable information on the uses of methyl bromide in Tables 2 through 5 of Appendix E (pp. 248-250), which provide an understanding of where the most likely exposures might occur. Approximately 95% of the methyl bromide consumed in California is used in soil fumigation, so this mode of use is necessarily a major focus of the analy- sis. Structural fumigation comprises about 3/O to 4% of the methyl bromide use, and commodity fumigation comprises a relatively minor proportion,

OCR for page 35
EXPOSURE ASSESSMENT 3 7 about ~ TO to 2%. Based on these use data for methyl bromide, the committee believes that it is important to describe the exposure scenarios within the fol- lowing categories: (~) occupational; (2) residential, school, and other; and (3) residents returning to fumigated houses. Each of these categories and DPR's coverage of these exposure groups is addressed below. "Occupational" refers to people who work directly in or around fumigation operations. These individuals are likely to have the most intense exposures and include such labor categories as field applicators (soil fumigators- including pilots, copilots, shovelmen, and workers who remove tarps), struc- tural applicators, and commodity fumigators and aerators. The occupational exposure estimates presented in the DPR report are based on measurements conducted in soil fumigation and commodity fumigation scenarios. The jobs evaluated for exposure and the corresponding estimates of exposure are listed in Tables 16-20 (DPR 1999, pp. 96-106), which include estimates for acute (daily), short-term (7-day), seasonal (90-day), and chronic exposures (annual). A total of ~ 60 exposure categories are listed. Most of the exposure data were measured with personal monitoring devices. The exposure estimates are re- ported in parts per billion (ppb) and the acute exposure category includes both high and mean values. All other exposure categories are listed as mean val- ues. The 24-hr time-weighted maximal exposures range from a high of 8,458 ppb for sea-container aerators to a Tow of 0.6 ppb for shallow-shank nontarped bed shovelmen. Numerous job exposures are listed as "n/a," which the table footnote explains as either "not applicable" or "no exposure information avail- able." Unfortunately, it is not clear to the subcommittee which situation ap- plies for a given job category and there is no explanation as to why certain categories of exposure are not applicable to certain jobs. "Residential, school, and other exposures" refers to people who are ex- posed to methyl bromide due to its atmospheric transport from the site of di- rect application. This category specifically includes residents in houses, stu- dents in schools, and occupants of buildings near fumigated fields, structures, or near fixed commodity fumigation facilities. This category is expected to contain the most sensitive groups of potentially exposed persons, because it is a cross section of the entire population, and therefore would include the very young and old, as well as other persons that might have heightened sensitivity. DPR provides no data on exposures to individuals in homes or other build- ings near fumigated fields; however, it does provide exposure data on struc- tural fumigations. Gibbons et al. (1996a,b, as cited in DPR 1999) measured methyl bromide concentrations for 24-hr periods in houses located within 50 to 100 feet of fumigated houses. Air sampling in the nonfumigated houses was conducted in rooms closest to the fumigated houses. The measured con- centrations range from 0.024 parts per million (ppm) (the limit of detection) to

OCR for page 35
35 METHYL BROMIDE RISK CHARACTERIZATIONIN CALIFORNIA 0.406 ppm. It is unclear from the DPR report how many samples were non- detects. Mean concentration values were 0.024 ppm for nearby houses and 0.060 ppm for "downwind" houses. Downwind is not defined in the DPR re- port. Information on exposures to people in residences, schools, and unrelated workplaces near commodity fumigation facilities is based on exposure esti- mates for workers in those facilities (Haskel 1 998a,b). No actual air sampling was conducted to evaluate this nonworker scenario. The assumptions used in this scenario (DPR 1999, Appendix H. page 343) specify that residents are exposed to methyl bromide concentrations at 210 parts per billion (ppb) (24- hr time-weighted average), the maximum permissible exposure level specified in the permit. The subcommittee considers the information provided by DPR insufficient for evaluating the quality of the data used for this assumption and for evaluating the validity of extrapolating from worker exposures to expo- sures of nearby residents. "Residents returning to fumigated houses" can be subjected to a wide vari- ety of concentrations, depending on the characteristics of the house and the retention of methyl bromide in spaces in the houses, such as wall voids. This exposure group includes highly-susceptible individuals such as children (NRC ~ 993), the ill, the elderly, and those with genetic polymorphisms (see Chapter 2~. DPR presents exposure measurements from five houses in southern Cali- fornia that were fumigated on a single day followed by 24-hr of active aera- tion, such as with a fan. (These data are discussed in greater detail below in the section entitled "Exposure of Residents in a Fumigated House".) In addition to DPR's coverage of the three exposure scenarios above, there are other population groups and exposure scenarios that are never addressed by DPR. For example, DPR never describes or provides data on exposures to children and the elderly, who might be more sensitive to methyl bromide than the worker or general adult populations. Furthermore, DPR never addresses exposure scenarios for residents living near fumigated fields, because these homes were considered to be outside of the permit buffer zone. Therefore, DPR assumed that the maximum concentration to which these individuals could be exposed was 210 ppb. However, the DPR report fails to provide any monitoring data that supports this assumption. Other exposure scenarios not covered by DPR include elevated exposures that occur when multiple agricultural fields are treated during the same time period in one area (e.g., many strawberry fields treated simultaneously or con- secutively in SaTinas county). it is possible that workers and individuals liv- ing near the treated fields could experience higher exposure levels than pre- dicted by the permit conditions. For example, there are no data on 6-week to 3-month exposures that individuals who live in agricultural areas might re-

OCR for page 35
EXPOSURE ASSESSMENT 39 ceive if multiple methyl bromide applications occur during a season. A rea- sonable worst-case scenario could be described as multiple nearby fields be- ing treated simultaneously with the air mass moving towards a residential neighborhood located in a lower area of a valley. The subcommittee is not confident that under these conditions, exposures of children and adults to methyl bromide concentrations above the 6-week reference concentrations of and 2 ppb, respectively, do not, or are unlikely to, occur. Finally, DPR does not address less common exposure scenarios that might occur under unique weather and terrain conditions, such as when a low-level temperature inversion or other similar low-wind condition prevents the dilu- tion of methyl bromide that would normally be expected to occur. Workers and residents living in such an area could be exposed to high methyl bromide concentrations. DPR describes such an exposure scenario in Appendix F. page 253, where 35 bystanders experienced methyl bromide poisoning as a result of low winds and a temperature inversion during and following the ap- plications. The subcommittee recognizes the difficulty DPR would have in consider- ing all these potential exposure scenarios. However, the subcommittee be- lieves that these likely scenarios need to be evaluated, either by collection of additional monitoring data or by appropriate modeling. Only by doing so can the public have confidence in DPR's assertion that the concentrations to which they are exposed are consistently below regulatory levels. QUALITY OF DATA AVAILABLE FOR C1 IARACTERIZING EXPOSURES This section addresses issues relating to the quality of data available for characterizing exposures, including (1) the analytical methods used in quanti- rtying methyl bromide concentrations; (2) the representativeness of available exposure measurements; (3) the appropriateness of normalizing assumptions used by DPR for different application rates; and (4) the appropriateness of exposure-duration assumptions used in the risk characterization document. Analytical Methods The subcommittee has serious concerns about the analytical methods used by DPR and others to determine atmospheric concentrations of methyl bro- mide. For the most part, these concerns focus on the fact that the initial field- sampling studies were conducted prior to the development of standardized

OCR for page 35
40 METHYL BROMIDE RISK CHARA CTERIZA TION IN CALIFORNIA analytical recovery methodologies. In addition, the lack of information on the atmospheric conditions under which the field samples were collected calls into question the recovery values that were used to calculate actual concentra- tions of methyl bromide in ambient air. The uncertainty in the recovery values is expressed in a report by Biermann and Barry (1999), which was written after collection of all of the f~eld-sam- pling data for methyl bromide between ~ 992 and ~ 998. Although the analyti- cal method for extracting methyl bromide from the samples had been used previously, it appears that a rigorous testing of the method has not been con- ducted. The primary uncertainty with the analytical method centers on the procedure for recovering methyl bromide from the charcoal tubes that are used to collect ambient air samples. Prior to the Biermann and Barry (1999) study, recovery values were determined by adding a known amount of methyl bromide in solution to the charcoal, followed by extraction of the charcoal with an organic solvent. It was assumed that addition of methyl bromide in solution to the charcoal was identical to collecting methyl bromide from the gas phase through the charcoal. The percent of methyl bromide recovered from the solution application was considered by DPR to be identical to the percent of methyl bromide recovered from the charcoal in the actual air sam- ples. Biermann and Barry (1999) demonstrated that recovery of air-trapped methyl bromide from the charcoal is only about 50/0, whereas the recovery of solution-added methyl bromide from charcoal was reported to be 69% in the field tests. Therefore, the field sample concentrations determined prior to the Bie~ann and Barry (1999) study were assumed by DPR to have been under- estimated by approximately 50%. In its report, DPR calculated the expected concentrations for all the sampling data using the 50/0 recovery value, based on the Biermann and Barry (1999) study. The subcommittee considers the 50% recovery estimate of Biermann and Barry (1999) to be questionable for many ofthe air samples. The 50% recov- ery estimate is based on samples collected under normal laboratory conditions with ambient air temperatures of between 20C and 25C and 20% and 80% relative humidity. However, when Biermann and Barry (1999) took the test system outdoors and did air sampling during the warm daytime temperatures, recoveries were as low as 2 ~ % to 26%. In contrast, when the same tests were conducted during the night, recovery estimates were 45% to 48%. Further- more, when air samples were taken at very Tow relative humidity (0/0), recov- eries of methyl bromide were only 0/0 to 3%. Because relative humidity and air temperature were not considered when the exposure- assessment data were compiled by DPR, and because the sampling data were primarily collected during the daytime, the actual recoveries might be Tower than the 50% used by DPR. In addition, the recovery of methyl bromide from the charcoal tubes

OCR for page 35
EXPOSURE ASSESSMENT 41 appears to be dependent on the initial methyl bromide concentration. For in- stance, in the storage-stability experiment described in Biermann and Barry (1999), recoveries of methyl bromide concentrations at 95 ppb were 5/O to 10% Tower than recoveries of concentrations at 710 ppb. At even Tower con- centrations (Biermann and Barry 1999, Table 11), charcoal spiked with 19 ppb methyl bromide yielded 0% recovery; however, only one sample at this Tow concentration was examined. This 19-ppb concentration was twofold higher than the reporting (detection) limit of the California Depa~l~ent of Food and Agriculture laboratory that did the analysis. The subcommittee is concerned about the lack of reliable recovery estimates at low methyl bromide concentrations, because the reference concentrations (RfCs) for subchronic and chronic exposures to children and adults are 1 and 2 ppb, respectively. The subcommittee believes that DPR and other analytical laboratories might not be able to adequately measure atmospheric concentrations of methyl bro- mide at or near these RfCs. The recovery study by Biermann and Barry (1999) provides quantitative information on several environmental factors (e.g., humidity, concentration, temperature) that appear to affect the reliability of ambient air-sampling re- sults of methyl bromide in the field. The field-sampling data presented in the DPR report were collected by at least six different groups, during several time periods (July 1992; October 1992; November 1992; February 1993; and March 1993) and at various locations in California (Santa Maria, Arvin, Chowchilla, SaTinas, Hayward, Watsonville, and Madera). Because of the different times and locations at which the air sampling was conducted, it is to be expected that the temperature and humidity levels for each study varied considerably. Daytime temperatures in July and August in the Central Valley of California are often above 100F, probably near the temperature at which the outdoor recovery study of Biermann and Barry (l 999) was conducted, for which reported methyl bromide recoveries were 2 ~ to 26%. Air samples ob- tained in the cooler months of the year (November-April) were probably col- lected at temperatures reflective of the 50% recovery of the Biermann and Barry (1999) laboratory samples. Several of the studies by Siemer and Associates, TriCal, Inc., and AG-Tn- dustrial reported that the sampling data was initially adjusted for a recovery of 69% (DPR 1999~. However, the DPR report presents no information on whether these 69% recoveries were based on actual samples taken at the time these studies were conducted, or were based on a standardized recovery value. The subcommittee believes that it is unlikely that the 69% recovery used by the several researchers was based on actual laboratory testing, given the uni- formity of the recovery estimates. Furthermore, DPR states that, "a field for- tification recovery study was not carried out in many of the exposure studies"

OCR for page 35
42 METHYL BROMIDE RISK CHARACTERIZATIONIN CALIFORNIA (DPR 1999, Appendix H. p. 274~. Radian Corporation conducted an addi- tional sampling study and used a slightly different analytical technique (head- space gas chromatography) to determine the methyl bromide air concentra- tions, but did not report a recovery value (DPR 1999~. Air Toxics Limited conducted yet another study using charcoal tubes and a limited number of stainless steel (SUMMA) canisters, which do not have the same recovery problems as charcoals. Recoveries were reported to be in the range of 74% to 125%. Finally, DPR itself conducted residential exposure studies in fumi- gated houses. Average recoveries were reported to be 7 ~ .4%, with a range of 49/0 to 102%. The location, temperature, and relative humidity for each house appears to be subject to the same variability and uncertainty as for the outdoor air-sampling studies discussed previously. The analytical data from these studies are clearly compromised by the lack of a robust analytical method for measuring methyl bromide concentrations in air. Because of the ease and lower cost of methyl bromide collection using charcoal as compared with stainless steed canisters, the charcoal method will probably continue to be the method of choice. Therefore, the subcommittee finds that (1) a systematic study should be conducted to assess the usability of the previous sampling data obtained with charcoal tubes and (2) a sampling method should be developed that will provide reliable air concentration data. To accomplish these goals, the following issues should be addressed: 1. Are there types of charcoal (e.g., coconut shell) that give more reliable recoveries than the petroleum-based charcoal used for many of the re- ported exposure studies? 2. What are the effects of temperature on recovery values? Should the charcoal tubes be maintained at some specific temperature (e.g., 15- 18C) during sampling to minimize degradation of methyl bromide dur- ing Tong (e.g.,12 fir) collection times? 3. Are there methods to minimize the effect of humidity on sample recov- ery? 4. For each sampling trip, what is the minimum number of samples that should be taken using an alternative method (e.g., stainless steel canis- ters) to compare recoveries? 5. How does the recovery vary with time of sampling and concentration? What is the limit of detection? Stainless steel canisters are generally evacuated and the sample is captured by allowing air to flow into the canisters. The only surfaces that the methyl bromide comes into contact with in these canisters is the relatively inert stainless steel surface, which is distinctly different from the very large and complex surface of charcoal.

OCR for page 35
EXPOSURE ASSESSMENT 43 6. A routine method for conducting f~eld-recovery studies should be devel- oped that permits direct air sampling, rather than solvent spiking. Con- ducting a recovery study using gaseous samples would reduce the uncer- tainty in the available exposure data. The subcommittee recognizes the difficulties faced by DPR in using the available sampling data for the exposure assessment. Because the initial field sampling was conducted prior to the critical recovery study, DPR was obliged to use a single recovery study to reevaluate a large number of sampling stud- ies. The air concentrations used in the exposure assessment include an unde- termined level of uncertainty due primarily to the uncertainty in the actual analytical recoveries obtained when the samples were collected under field conditions. Nevertheless, the subcommittee feels the data are still very useful and provide important intonation on methyl bromide emissions from treated areas. With the caveats mentioned previously about air temperature, humid- ity, and concentration effects on recovery, the 50/0 adjustment used by DPR appears to be reasonable for most of the samples collected in the cooler months and for concentrations that are greater than 50 ppb. For air samples taken at higher temperatures, the methyl bromide concentrations are probably underestimated, potentially by a factor of 2. If, for example, the outdoor re- covery values of 21% to 26% were to prove typical, then the average methyl bromide concentrations would be expected to be about double those estimated by DPR. Because this data set was the primary information used to develop the exposure assessment, and it appears to be the bulk of the information pres- ently available, it is important to place some level of uncertainty on the data. For these purposes, the subcommittee suggests that the actual exposures might be considerably higher than even the adjusted estimates presented by DPR. Representativeness of Available Exposure Measurements A representative sample of a diverse group of exposures is a sample that is constructed such that the central tendency (mean) and distribution (standard deviation) of exposure levels observed in the sample are likely to be free of systematic differences from actual exposures that are being assessed. The data presented by DPR reflect a wide variety of occupational exposure scenarios and explicitly represent differences in such factors as soil application meth- ods, depth of application, type of tarping, and soil characteristics. However, even within the occupational exposure groupings, the data indicate very large ranges in exposure concentrations, often of several orders of magnitude. For instance, 24-hr time-weighted average exposures varied widely: for preplan"

OCR for page 35
44 METHYL BROMIDE RISK CHARA CTERIZA TION IN CALIFORNIA soil injection of methyl bromide they ranged from 0.6 to 835 ppb, for fumiga- tion of grain products from 6 to 6,039 ppb, and for residents downwind of a fumigated house exposures were estimated to range from 40 to 296 ppb. The sources of these variability ranges have not been characterized in the DPR report. However, aside from these broad ranges in variability, the measurements made for individual scenarios frequently reflect only a single set of samples collected on a single day for one type of exposure. There is little or no discus- sion in the DPR analysis of how well factors affecting the air sampling, such as air temperature, soil type, wind conditions, and humidity, reflect the actual exposure level distributions in practice for the occupational groups studied. In general, there is an absence of information on the conditions (e.g., tempera- ture, wind conditions, humidity) under which air-concentration measurements were made. Therefore, the subcommittee believes that there is considerable uncertainty about how accurately the observed measurements represent the real distributions of exposure concentrations and durations in the occupational groups that were studied. Appropriateness of Normalizing Assumptions for Different Application Rates To estimate occupational exposure levels from soil fumigation, DPR made a simple linear adjustment from the application rates used to the maximum permitted application rates. For example, if the maximum permitted applica- tion rate was 400 Ib/acre, but only 200 Ib/acre was used on the field, where the air concentrations were measured, DPR adjusted the measured air concentra- tions upward by twofold. The subcommittee has two reservations about this procedure: the first per- tains to the physical transport and transformation of methyl bromide, and the second pertains to the stated goals of the exposure analysis. In the first case, a simple linear adjustment is reasonable if one assumes that the only impor- tant mechanisms involved in the transport of methyl bromide between the sites of soil injection and the workers' breathing zones are mixing and dilu- tion, which lead to simple first-order Toss independent of concentration. How- ever, if physical sorption to soil particles, and chemical reactions with soil constituents are important, then it is possible that there could be a distribution of sites of high affinity adsorption, or high rate reaction, and that these prefer- ential binding/reaction sites would not be available during methyl bromide soil applications. In this case, methyl bromide applied at higher rates could encounter less effective sorption or reaction in the soil than methyl bromide applied at lower rates, and relatively more methyl bromide could be expected

OCR for page 35
EXPOSURE ASSESSMENT 45 to be available for inhalation by workers. Therefore, there is some risk that the worker exposures at maximally permitted application rates could be some- what understated. In the second case, if the goal of the exposure analysis is to represent expo- sures under the worst-case conditions permitted by the pesticide labels, then the subcommittee agrees that some adjustment for application rates should certainly be made. However, if the goal of the exposure analysis is to repre- sent the distributions of exposure levels that actually exist for the workers, then DPR's goal should be to assure that the exposure data collected appropri- ately reflect the actual distribution of application rates that are used in prac- tice. If the collected data differ from the exposure distribution being studied, then adjustments should be made to reflect the actual distribution of applica- tion rates. Appropriateness of F`nn~llre nllr~tion ~c~llmution~ To calculate exposures for durations longer than a single day, DPR has made a large number of assumptions (some of which might be considered rel- atively conservative) about how many days workers might be exposed at mean levels observed in ~ -day studies (DPR ~ 999, Appendix F. pp. 284-289~. The explanation for these assumptions is contained in a single paragraph on page 261: Calculations of exposure rely on factors, including application rates, work periods specified in the current California permit conditions, fre- quency and (luration of exposure. Types of tarpaulins, application equip- ment, and injection depth are used in the permit conditions to determine the maximum daily work time for each type of soil injection fumigation. DPR has requested registrants to provide frequency and duration of ex- posure for acute and non-acute exposures (Donahue 1997, as cited in DPR 1999~. So far, registrants have provided some data as requested. Consequently, default frequency and duration of exposure for many ex- posure scenarios were generated from data obtained from various sources and the use of professional judgment (Haskell 199Sa,b, as cited in DPR 1999~. These default values are shown in Appendix A tof the DPR document]. Without more explicit documentation of the specific derivation of the num- bers in Appendix F. and the overall goals of this exposure analysis, the sub- committee cannot readily assess the appropriateness of the exposure duration assumptions used.

OCR for page 35
46 METHYL BROMIDE RISK CHARACTERIZATIONIN CALIFORNIA ACCURACY AND APPROPRIATENESS OF AVAILABLE MODELING TOOLS Exposure Estimates Based on Modeling Modeling is an essential too] of risk analysis. It allows us to use our mech- anistic understanding of a system to draw inferences about exposure levels and associated risks, even in cases in which we do not have an extensive set of direct observational data. As discussed in more detail below, the subcommit- tee concludes that in general the basic structure of the residential indoor air dilution and outdoor air dispersion models used in the DPR exposure assess- ment are appropriate. However, the subcommittee finds that in some cases important questions about the variability of modeled exposures have not been addressed in the DPR report. For example, the subcommittee questions whether DPR has made an appropriate effort to juxtapose model predictions with field observations to characterize the quantitative uncertainties in the model predictions. The subcommittee questions whether DPR has used its models to assess the relevant variability in exposures and risks to different individuals and populations. DPR presents exposure estimates for individuals in fumigated homes or living near commodity-fumigation facilities in Table 19 (DPR 1999, p.105) of the report. Several of these estimates are based either on regulatory permit levels that are apparently derived from modeling or on model projections themselves; these include exposures of (l ~ residents in a fumigated house (Ta- ble 19-c), (2) residents living near commodity fumigation facilities (Table 19- d), and (3) residents living near fumigated fields (Table ~ 9-e). The modeling approaches supporting each of these cases are addressed below, with (2) and (3) discussed concurrently. Exposure of Residents in a Fumigated House The data for the analysis of exposure of residents in a fumigated house were drawn from air concentrations measured in five houses in southern Cali- fomia fumigated on a single day (April 7, 1992) at I.5 Ib/1000 ft3, and active- ly aerated using fans for 24 hr before closing the windows. (Data for a sixth house were excluded, reportedly because of a relatively short sampling time.) The data, consisting of a total of 32 methyl bromide concentration measure- ments made at times ranging from 3 to 92 hr after the end of the initial 24-hr aeration period, are presented in Table 36 (Columns 1 and 2) (DPR 1999, Ap- pendix F). DPR used a single-compartment, simple-diTution model to esti-

OCR for page 35
EXPOSURE ASSESSMENT 4 7 mate methyl bromide concentrations after 72 hr of active aeration using the aggregate data from all five homes. This was done by fitting a simple linear regression line to a plot of the logarithm of the observed concentrations versus time. The fitted line (Equation 3-1) is Log(MB) = - O.OOS ~ 95 x (t) - 0.14SO86 r2= 0.34955, (3-1) where MB is concentration of methyl bromide (in ppm), t is number of hours after 24-hour aeration, and r2 is correlation coefficient. DPR used this fitted regression line to predict residential exposures for a 1- week period (168 fir) beginning at either 48 or 49 hr after the 24-hr active ven- tilation (72 hr after the fumigation) without apparent further adjustment for the possibly greater reduction in concentrations that might occur from the 48 additional hours of active ventilation. (DPR requires that active ventilation be carried out for 72 hr after the fumigation, although for these data active venti- lation was only done for 24 hr.) To estimate exposure concentrations in northern California, where the fumigation rate is twofold higher (3.0 Ib methyl bromide/l,OOO ft3) than in southern California, a simple linear twofold adjustment was made to the methyl bromide concentrations (DPR 1999, Table 36, Columns 4 and 54. The subcommittee reproduced the regression equation above and derived confidence limits on the rate of exponential decline in methyl bromide con- centration over time in Equation 3-2 below. Log (MB) = - O.OO8 l 97 ~ 0.00204 x (t ~ std error) - 0.148O (3-2) r2 = 0.3497, where MB is concentration of methyl bromide (in ppm), t is number of hours after 24-hr aeration, and r2 is correlation coefficient. This regression equation allowed the subcommittee to verify the stated 7- day mean concentrations and associated confidence limits in Table 37 (DPR lL999, Appendix F) of 86 ~ 73 ppb (15-229) and 172 +147 ppb (30-458) for southern and northern California, respectively. lit also permitted the subcom- mittee to determine 24-hr estimates of methyl bromide concentrations to com- pare with the regulatory target level of 210 ppb that is assumed to apply for the 24 hr immediately following the reentry of residents into their homes. These data are presented in Table 3-1, in which the estimated average methyl bromide concentrations for 1 day and 7 days after the 24-hr ventilation period are shown, along with the standard errors. A comparison of the subcommittee projections of the central tendency and

OCR for page 35
48 METHYL BROMIDE RISK CHARA CTERIZA TION IN CALIFORNIA upper 95% confidence limits for the 7-day average exposure levels (Table 3- T) with the data in Table 37 (DPR 1999, Appendix F) shows that the values correspond closely. However, DPR appears to have made a twofold error in transposing these 7-day results to Table 19c (DPR 1999, p. 105) where the values are given as 172 ~ 146 and 344 ~ 294 ppb for southern and northern California, respectively. Aside from the apparent transposition error for the 7-day results, there ap- pear to be deeper problems with DPR's analysis. The grouping of data from five different houses yields, at best, a central estimate of the concentration levels that is likely to be present for residents reentering an average house. This estimate does not reflect the variability among houses in air exchange rates between contaminated wall spaces and the main living areas, and be- tween the living areas and outdoor air. The subcommittee believes that sepa- rate analyses of data from each of the five houses would have allowed DPR to make a first-cut assessment of the differences among houses in both initial concentrations of methyl bromide (following 24 hr of active ventilation) and the rates of decline. Because the average methyl bromide concentrations are already relatively high in relation to the regulated target level of 2 ~ 0 ppb (Ta- ble 3-1), neglecting this variability raises some concern, although the concern is somewhat mitigated by the fact that DPR apparently made no adjustment for the increased active ventilation period that might occur in practice (i.e., 72 hr versus 24 hr of active ventilation). Finally, DPR's assumption that the acute 24-hr exposure limit of 210 ppb is achieved is not supported by even the central tendency (median) estimate from the modeled data for the northern California application rate (Table 3-1~. This 210-ppb level is based on a calculation that assumes that methyl bromide TABLE 3-l Projected 1- and 7-Day Average Methyl Bromide Concentrations (ppb) for Residents Reentering Fumigated Homes Median -2 SEa -1 SE Estimate +1 SE +2 SE Southern California 7-day mean 39 Southern California 1-day mean (48 to 72 hr following 24-hr ventilation) Northern California 7-day mean Northern California 1-day mean (48 to 72 hr following 24-hr ventilation) - al and 2 standard error (SE) departures from the central estimate of the regression slope. 57 133 77 175 114 266 351 87 231 174 463 138 305 275 611 226 404 452 808

OCR for page 35
EXPOSURE ASSESSMENT 49 concentrations measured at electrical outlets or other enclosed spaces within the wall of a home will be equal to or less than 3 ppm when reentry is permit- ted. The 210-ppb level also implies that these within-wall measurements ac- curately reflect the average concentration in a well-mixed wall volume that represents only about 5.6% of the volume of the house, and that the 24-hr av- erage concentrations for the residents reflects immediate mixing of the wall volume contents with those of the living areas of the house, and no loss of methyl bromide from the house during the first 24 hr. Several of these assumptions appear incompatible with the direct observations made from the analysis of the five houses modeled above. First, the slope ofthe exponential decline in methyl bromide concentrations (Equation 3-2) reflects a half-life of about 37 hr (with 95/O confidence limits of 24 to 73 fur). The average air exchange rate, a general method for express- ing ventilation, is 0.019 exchanges per hour (95% confidence limits of 0.009 to 0.02S air changes/hr) in these houses (see Appendix C of this report). This air exchange rate estimate is considerably lower than rates observed in the living areas of other homes. (For example, EPA (1996) reported 24-hr aver- age air exchange rates from approximately 0.33 to 2.2 air changes/hr (10% to 90/O range) for 175 houses in Riverside, California.) The Tow air exchange rates observed for these five homes indicate that the controlling factor for the overall decline of methyl bromide concentrations over time (as observed in DPR ~ 999, Table 36, Appendix F) cannot be attributed to general house venti- lation, but probably reflects slow transfer of methyl bromide between the wall spaces and the living areas. Given this, and the convoluted geometry of wall spaces, the subcommittee questions DPR's assumption that measurements made at one or a few enclosed spaces within the wall are representative of a well mixed space. The subcommittee also has concerns with the fact that all the data used in the analysis (DPR 1999, Table 36, Appendix F) come from fumigations made on the same day in a similar area of southern California. This means that the data do not account for differences in varying external temperatures, wind conditions, and humidity, and possibly, house structural characteristics in dif- ferent areas of California and at different times of the year. Because of the uncertainties surrounding the current data set on exposures of residents returning to fumigated homes, the subcommittee finds that DPR's conclusion that current fumigation practices result in methyl bromide concen- trations that do not exceed the regulatory exposure level of 210 ppb does not seem warranted. Further data collection and analysis of exposure concentra- tions in routinely fumigated homes at different seasons and for different types of homes in various areas of California seems necessary if methyl bromide use as a house fumigant is to be continued with confidence.

OCR for page 35
50 METHYL BROMIDE RISK CHARD CTERIZA TION IN CALIFORNIA Exposure of Residents Downwind from Soil Fumigations The other major modeling effort in the DPR exposure analysis examines whether residents living near fumigated fields and commodity fumigation fa- cilities are exposed to methyl bromide concentrations that exceed the acute (24-hr) regulatory limit of 210 ppb. These exposures are regulated by an ex- tensive set of permit requirements implemented at the county level and are based on assumptions about the rate of air emissions from soil fumigation op- erations of various types. A standard air dispersion modeling system, the In- dustrial Source Complex-Short Term computer model (EPA 1995), is used to calculate the size of buffer zones that are required to prevent methyl bromide concentrations at the boundary from exceeding 210 ppb. DPR usecl the 210- ppb value to represent the acute exposures of residents near fumigated fields and commodity fumigation facilities in Table 19-d (DPRI999,p. 105~. This 2 ~ 0 ppb value represents an assumption by DPR that the permitting system as currently implemented is working. However, DPR fails to enumerate any un- derlying conservative assumptions used in their modeling, and does not de- scribe the variability or uncertainty associated with the actual implementation of the permits. The subcommittee attempted to evaluate DPR's assumption that the 210- ppb exposure level is not being exceeded at the buffer zone boundary. To conduct this analysis, empirical data contained in Table Hi (DPRI999, Ap- pendix H) of the DPR report were compared with the 210-ppb limit. Table Hi lists 39 maximum methyl bromide concentrations measured between 1992 and 1998 at or near buffer-zone boundaries at field fumigation sites. DPR describes the sampling methodology used to generate these data as follows (DPR1999, Appendix H. p. 357~: In these studies, air monitoring was conducted using personal air sampling pumps equipped with activated charcoal tubes. The samplers were set up around the field at a distance of 30 feet from the edge of the field and at the permit condition buffer zone determined for the application. Sampling was initiated at the start of the application and continued for one to seven days, with each sampling interval 6-12 hr. The air flow rate for all samplers was calibrated to approximately 15 mL/min. Wind speed, wind direction, air temperature, and relative humidity were recorded every five minutes with a Met-1 meteorological station. In Table 3-2, some of the data from Table HI (DPR 1999, Appendix H) have been reproduced, showing the sampling year, sampling distance, permit condition buffer zone, and methyl bromide concentration (Columns b, c, 6,

OCR for page 35
EXPOSURE ASSESSMENT 51 TABLE 3-2 Maximum Methyl Bromide Air Concentrations Mom Different Application Methods (e) (f) (g) (a) (c) (~) Permit Sampling 24-hr Case Sampling Permit Buffer- Distanceas Max MB Numberin (b) Distance Condition Sampling Fraction of Concen. Table Hi Year (ft) Buffer (ft) Distance Permit Buffer (ppm) 1 92 300 390 90 0.77 0.042 2 92 300 330 30 0.91 0.260 3 92 50 330 280 0.15 0.550 4 98 200 200 0 1.00 0.150 5 92 600 1060 460 0.57 0.700 6 92 600 1170 570 0.51 0.610 7 98 510 510 0 1.00 0.110 8 93 200 2010 1810 0.10 0.560 9 93 200 940 740 0.21 0.340 10 95 80 780 700 0.10 0.110 ... ... ... ... ... ... ... 30 97 625 420 -205 1.49 0.590 31 92 300 300 0 1.00 0.060 32 96 330 550 220 0.60 1.700 33 97 360 360 0 1.00 0.160 34 97 360 360 0 1.00 0.550 35 98 60 200 140 0.30 0.160 36 98 30 100 70 0.30 0.066 37 98 30 100 70 0.30 0.072 38 98 30 100 70 0.30 0.065 39 98 30 100 70 0.30 0.042 Mean 0.260 Std. Deviation 0.332 Std. Error 0.052 Geom. Mean 0.145 Geom. Std. 2.882 Dev. aCase numbers 11-29 have been deleted. Source: Adapted from DPR 1999, Appendix H. Table H-1, pp. 358-360. and g, respectively). Additional calculations have been made by the subcom- mittee, including Column e, the absolute distance between sampling distance

OCR for page 35
52 METHYL BROMIDE RISK CHARA CTERIZA TION IN CALIFORNIA (Column c) and the buffer boundary (Column d), and Column f, the ratio of the sampling distance (Column c) to the buffer boundary (Column d). The discussion of these data in the DPR report (DPR ~ 999) notes that: of the 39 applications monitored, seven exceeded the 0.21 ppm target level at the buffer zone distance...TarpauTin-bedded applications and applications using "very high barrier" tarpaulins appeared to have higher air concentrations than originally assumed in the permit conditions. Of the seven tarpaulin-bedded applications monitored, four exceeded the 0.21 ppm target level at the original buffer zone distance. Of the five very high barrier tarpaulin applications monitored, three exceeded the target level at the original buffer zone distance. None of the other appli- cation methods exceeded the target level at the buffer zone distance. in addition, a footnote in Appendix H (DPR 1999, p. 361) notes that "DPR revised the buffer zones in ~ 997 and ~ 998 to provide a higher margin of safe- ty. Under the revised buffer zones, none of the 39 fields monitored exceed 0.21 ppm at the buffer zone distance." Unfortunately, aside from this footnote in the report, no details are provided on the nature and extent of the modifica- tions of buffer zones for the individual cases listed in Table 3-2, nor does DPR indicate how they adjusted the measured data to arrive at their conclusion that none of the cases would have exceeded 0.21 ppm had measurements been taken at the new buffer-zone boundaries. In Table 3-3, the subcommittee has summarized the data presented in Table 3-2. Methyl bromide concentrations are stratified by distances greater or less than 90% ofthe buffer zone forpre-1998 and 1998 periods. The data in Table 3-3 suggest that in 1998, methyl bromide concentrations at the prescribed buffer-zone boundaries were lower than those measured prior to 1998. Forty- three percent of the pre- 1998 concentrations at the buffer-zone boundary were expected to exceed the regulatory limit of 210 ppb, whereas only 7% of the 1998 concentrations were expected to be over this limit. Overall, if the ~ 998 data presented in Tables 3-2 and 3-3 are representative of current permit conditions, the percentage of soil fumigation operations that would result in methyl bromide concentrations at the buffer-zone boundary of greater than 2 ~ 0 ppb is expected to be relatively modest. To make such a con- clusion, the subcommittee finds that further data are needed. The 1998 data set of measurements at or near the buffer zones of 30 and 100 feet is very lim- ited. As indicated in Table 3-3, there are only four measurements taken in 1998 at distances greater than or equal to 90% of the buffer zone boundaries. Data collected prior to 1998 suggest that the modeling program estimated methyl bromide concentrations at the buffer-zone boundary that are at or near

OCR for page 35
53 JO Ct 7 CC CC Cd By .s o 50 - o o _ _ In _ CASIO,, U. .~ cn .~ A Ct U. .o Ct 1 ~ ~ O So In C) ~ .m O no of 00 0 at ~ A U. Ct t~ Cq ~ ON so o\ o`` A Cal U. ~ .~ o To oo^ o V U) an =~ .= ~ ~ A= ^ Cat at, o ~ \ I_ o 1 0 V o 'e v: ~ oo ~ ~ oo ~ ~ ~ oo ~ ~ o o o - - c~ - } o o ~ o ~ ~ o o o oo o . c<) ~ ~ - } o o ~ o ~ ~ o o - a~ ~ - ~ ~ ~ - - ~ - - v o o o c~ ~ ~ ~ C~] ~ ~ d- ~ ~ o o - o ~ ~ - ~ ~ .o ~ ~ ~= h ~ ~ ~ x t C - E ,, o o -~ ~ =

OCR for page 35
54 METHYL BROMIDE RISK CHAR24CTERI~TIONIN CALIFORLNIA the 210-ppb limit. This is supported by the arithmetic and geometric concen- tration means of 0.286 ppm and 0.177 ppm, respectively. However, the sub- committee notes that there is a certain proportion of the measurements that exceed 2 ~ 0 ppb at the buffer-zone boundary, occasionally by several-fold, as indicated by concentrations of 0.65 ppm, 0.94 ppm, and 1.4 ppm at the pro- jected 90th, 95th, and 9Sth percentiles, respectively. The subcommittee reviewed two DPR documents that update the material provided in Appendix H of the DPR report (Segawa et al. 2000a,b). These documents provide detailed directions for calculating flux rates and buffer- zone distances for the proposed regulations. Although it is not within the subcommittee's task to comment on the appropriateness ofthe proposedregu- lations, it is relevant to the foregoing analysis to note DPR's comparison of buffer-zone distance with monitoring data (Segawa et al. 2000a, p. 8~. The authors state that, based on new modeling for 34 applications examined, buffer zone table distance was greater than the distance to 0.21 ppm esti- mated by the ISC "Industrial Source Complex ~ model for 33 of the 34 applications. On average, the buffer zone table distance exceeded the distance to 0.21 ppm by 520%, with a median of 400/O (Table 3~. We made these calculations when the monitoring data were originally ana- lyzed using unadjusted air concentrations of the first version of the ISC model. DPR is updating these calculations using adjusted air concentra- tions and version 3 of the 1:SC model. Segawa et al. (2000b) contains a table similar to Table Hi in Appendix H of DPR's report showing maximal concentrations measured at 30 feet, appli- cation rates, and proportions calculated to be volatilized using both unadjusted and adjusted measurement recoveries. However, there is no direct presenta- tion analogous to Table Hi of methyl bromide concentrations expected at the revised buffer-zone distances. Therefore, the subcommittee cannot determine the frequency distribution for maximally observed concentrations at the re- vised buffer-zone distances based on the available information. Accordingly, the subcommittee is unable to fully evaluate the accuracy of the modeling used for estimating off-site residential exposures in the DPR report, nor can the subcommittee determine if the proposed, or even current, buffer zones actually protect nearby residents from exposures to methyl bromide concen- trations greater than 210 ppb. SUMMARY The DPR report contains a large compilation of exposure data, particularly on worker exposures. However, the subcommittee finds that DPR's exposure

OCR for page 35
EXPOSURE ASSESSMENT 55 analysis is lacking in several respects. Certain exposure scenarios are not dealt with at all in the report, including exposures to residents living near fu- migated fields and potentially elevated exposures to residents and workers resulting from methyl bromide application to several fields simultaneously. The subcommittee believes that it is extremely important for DPR to address such exposures, considering that 95% of methyl bromide is used in soil fumi- gation. Furthermore, there is considerable uncertainty surrounding the analyt- ical recovery methods used in the exposure-assessment studies. Much of the data presented by DPR is based on single air-concentration measurements. There is no discussion of the representativeness of these measurements to the actual exposures experienced by the potentially exposed populations. In addi- tion, DPR makes numerous assumptions regarding durations and levels of ex- posures, which the subcommittee believes are not explained in sufficient de- taiT to understand their appropriateness. The subcommittee believes that fur- ther data collection and analysis are necessary to accurately assess both work- er and residential exposures to methyl bromide.