Peer Review of the Surface Impoundment Study Technical Plan for Human Health and Ecological Risk Assessment

General Comments and Recommendations by Curtis Travis

The objective of the Surface Impoundment Study is to determine, with an acceptable degree of certainty, the risks to human health and the environment posed by industrial wastewaters managed in surface impoundments. EPA estimates that there are 19,000 impoundments, located at 8,500 facilities in the United States. The EPA will use a sample of 215 facilities as the basis for the Surface Impoundment Study. The Study will be conducted in two phases: Phase 1, a screening approach to reduce the number of constituents, impoundments, and facilities that require fate and transport modeling, and Phase 2, a more detailed multimedia modeling study on some units and constituents.

During Phase 1, cumulative risks will be calculated for each impoundment and each facility and for each constituent. The cumulative risk estimates will be used to build initial risk distributions for surface impoundments within the scope of the study. The risk distributions will be used to exclude constituents, impoundment types or facilities for further consideration.

I first want to say that the document is well written and clearly organized. It is easy to follow the major ideas and the authors have gone the great length to break the presentation into bite-sized chunks that are easy to follow and understand. There is also ample use for figures and diagrams to enhance communication. All in all, the document is an above-average demonstration of how to communicated complicated technical ideas.

I now answer the questions posed to the reviewers.

1) Derivation of human health screening factors: a) is the methodology for calculating screening factors a suitable methodology? and b) are other data sources available that would provide credible benchmarks for the cases where no benchmarks are available?

Human health screening factors are used to convert reported concentration data (for impoundment sludge/soil, water, and air) into risk numbers (the process is defined on page 2-4). The screening factors (there is one each for sludge/soil, water, and air) represent the concentrations in sludge/soil, water, or air that would result in a 10-5 cancer risk or 1 HQ for non-carcinogens if receptors are exposed under the assumed exposure scenarios.

In a sense, the question of the suitability of the methodology for calculating screening factors is not the most important question. Human health screening factors arise only in Phase 1A. Given that practically no constituents, impoundments or facilities will be eliminated from further consideration in Phase 1A (see below), the assumptions and procedures developed in Phase 1B will have more impact on the overall validity of the Surface Impound Study.

The methodology for defining the human health screening factors is presented in Table 2-1 and Table 2-2. This methodology is protective of human health. The risk assessment procedures employed in their derivation are acceptable and the equations for development of human health screening factors (Table 2-1) are standard. The exposure parameter values (Table 2-2) represent typical or central tendency values, but this is acceptable given the highly conservative nature of the rest of the methodology (the assumption of direct exposure to impoundment constituents). >p>The question of the adequacy of human health benchmarks is more important than the adequacy of the procedure for deriving human health screening factors. The methodology states that the screening factors are developed using EPA health benchmarks. However, EPA has established benchmarks for relatively few compounds. The document states (page 2-10) that when EPA approved benchmarks are not available, interim benchmarks will be developed. However, the document does not state how this will be done in Phase 1 (the document does address this issue for Phase 11).

With regards to question b), I know of no other data sources that would provide credible benchmarks for the cases where benchmarks are unavailable? However, I believe the procedure outlined in Phase 11 (Section 3.2.3.4) is adequate and should be employed in Phase 1.

2) Derivation of ecological screening factors: a) is the list of representative species suitable for a screening level assessment to identify the highest ecological risks associated with surface impoundment? and b) are there gaps in the proposed consideration of ecological risks?

The document separates ecological assessment endpoints from ecological measurement endpoints. Assessment endpoints specify environmental values to be protected (like protection of a population or community), while measurement endpoints identify measurements of effect (reproductive or developmental toxicity) that have a direct bearing on the values to be protected. This seems to be a very positive direction, but as stated, it has a few limitations:

The document states (page 2-47) that the screening factors are based on "the theory that protection of 95 percent of the species in the community will provide a sufficient level of protection for the community". It would be better to just assume that protection of 95% of the species will be protective of the community (rather than refer to a theory). A weakness in the methodology is that the document does not state how one determines that 95% of the species are being protected. It would be better to state as an assumption that protection of the representative species (because of their representative nature) is assumed to result in protection of 95% of community species.
The document notes that toxicological data for mammals, birds and reptiles is limited. The use of surrogate chemicals as indicators of the toxicity of impoundment constituents introduces some uncertainty. It is not clear from the document how many constituents have sufficient eco-toxicity data to carry out a Phase 1 assessment. Thus, the degree of uncertainty introduced by the use of surrogates is unclear.
The list of representative habitats, receptor groups, and representative species seem very suitable for a screening level assessment to identify the highest ecological risks associated with surface impoundments. However, it is not clear that sufficient ecological toxicity data exist to implement the ecological methodology outlined in Phase 1.

3) Level of protectiveness: does the proposed screening methodology include sufficient protective assumptions as to have a negligible possibility of overlooking potential risks?

The purpose of the human health screening methodology is to (1) reduce the number of constituents, impoundments, and facilities that require detailed fate and transport modeling, and (2) develop cumulative probability distributions of the risks posed by surface impoundments, that serves both as a decision tool and to provides a preliminary overview of the risks of surface impoundments nationwide.

The basic approach to this problem is to compare the reported concentration data (in surface impoundment water, sludge and emissions) collected from a facility survey with screening factors developed to be protective of human health and ecological receptors. Thus, the acceptability or non-acceptability of this approach depends on the methodology for deriving the human health and ecological screening factors.

Phase 1A

This is a very conservative methodology. There is a very small probability of overlooking potential risks. The primary and largest introduction of conservatism in this methodology is the assumption that exposure will be based on direct ingestion of the surface impoundment water, ingestion of impoundment sludge/soil, and direct inhalation of the air emissions. This is equivalent to assuming that an adult resident and one child are living next to the surface impoundment and obtaining all of their drinking water as impoundment water, directly ingesting impoundment sludge/soil, and continuously breathing volatile emissions coming off of the impoundment 7 days a week, 350 days per year for a period of 30 years. In addition, to pass Phase 1A, the cumulative cancer and noncancer risk from all constituents must pass the screening criteria, which is 10 times less than the risk criteria. Thus, any constituent, impoundment, or facility that is eliminated from further consideration during Phase 1A has a negligible potential to cause harm to human health.

EPA provides additional protection by recognizing the potential for indirect exposures and giving special consideration to constituents with high potential to bioaccumulate or persist in the environment. This is a very conservative methodology. In fact, one has to wonder if any facilities or impoundments will be eliminated from further consideration. It would mean that every constituent in the facility or impoundment must be sufficiently non-toxic to be acceptable for direct ingestion or inhalation without harm. Thus, it would appear that the primary result of the Phase 1A screening will be to eliminate some constituents, but no facilities, from further consideration.

Phase 1B

The purpose of the Phase 1B assessment is to evaluate those constituents, impoundments, and facilities that were not eliminated from further consideration in Phase 1A. Again, the major purpose of this phase is to limit the number of constituents, impoundments, and facilities that will require detailed analysis in Phase 11. My comments follow:

Conservatism. The real answer to the question of the level of protectiveness of the screening methodology lies in Phase 1B. The vast majority of constituents, impoundments, and facilities will make it to the Phase 1B screening.
Description. The description in the document of the Phase 1B methodology is not adequate.
The document is not clear as to the importance of the Phase 1B screening, given that most impoundments and facilities will require further analysis in Phase 1B. The amount of space given in the document to a discussion of the Phase 1B screening is small in comparison to the Phase1A screening.
The document is not clear as to the criteria for passage from Phase 1A to Phase 1B. On page 1-9, the document states, "EPA expects to use screening-level fate and transport models to supplement the initial screening under Phase 1A. In general these screening models will be appropriate where the major routes of exposure are expected to be direct ingestion of drinking water or direct inhalation". This sounds like there is a choice as to when to use the Phase 1B models.
The exposure scenarios to be used in Phase 1B are not explicitly defined (especially for groundwater exposures). Are they the same as in Phase 1A?
Why is soil ingestion considered in Phase 1A, but not Phase 1B? Suppose a constituent passes from Phase 1A to Phase 1B based on a high risk from soil ingestion. Then in Phase 1B it has low risk from groundwater ingestion and inhalation, and is eliminated from further consideration. This would not be protective.
It should be pointed out that Phase 1B screening does not use risk-screening factors or the risk screening formula on page 2-4.Instead, it directly computes risk using IWAIR and IWEM.
Inhalation Assessment. All constituents, impoundments, and facilities that are not eliminated by Phase 1A will be analyzed under Phase 1B. Here the CHEMDAT8 model will estimate volatilization from the impoundment and the IWAIR model will be used to calculate inhalation risk. Exposure will be to a close in resident. CHEMDAT8 has undergone extensive peer review and appears to be an adequately protective model for use in this screening assessment. However, the document gives no description of how CHEMDAT8 will be run. Will it predict mean expected volatilization or 90th percentile volatilization? How will nationwide variability in volatilization modeling parameters be handled?
ISCST3 is a respected model. The only question is its applicability for a close in resident. EPA has guidance for the use of dispersion factors for multiple source area sizes, receptor distances, and meteorological conditions. Is a close in resident one of the acceptable receptor distances for use of ISCST3? Given resolution of these issues, I find the inhalation assessment component of Phase 1B to be very conservative.
Groundwater Assessment. The IWEM model will be used to calculate the risks due to exposure to groundwater containing constituents released from surface impoundments. The model relies on a dilution attenuation factor (DAF). The DAFs were generated using the EPACMTP groundwater model, which has undergone extensive peer review and is an expectable model for this type of assessment. The DAF selected represents the amount of dilution and attenuation that would occur in at least 90 percent of the cases in the United States. As the document states, "The DAF is protective in at least 90 percent of the modeled cases". Thus, we finally come to the point where we can answer the question "Does the proposed screening methodology include sufficient protective assumptions as to have a negligible possibility of overlooking potential risks? The answer depends on who you define negligible probability. The groundwater pathway is probably the limiting pathway and the current methodology may under predict groundwater concentrations in 10 % of the cases. However, the exposure assumptions are very conservative. They assume that a nearby resident will drink groundwater 350- days a year and this is unlikely to be the case for most of these sites. In addition, the use of the 90th percentile DAR is consistent with other EPA analyses and its use has been reviewed and approved for national level analyses by the EPA Science Advisory Board and the public. Thus, I conclude that the screening methodology provides an adequate margin of protection to human health and the environment given the degree of uncertainty inherent in the data and this type of analysis.

Two possible limitations of this approach are as follows:

Representative Sample. It is assumed that the sample of 215 selected facilities used, as the basis for this study is representative of the 8,500 facilities, 19,000 surface impoundments, and hundreds of constituents in these surface impoundments. If this is not the case, the screening analysis will not give a true picture of the risk posed by nonhazardous waste surface impoundments in the United States. The justification of the representatives of this sample needs further justification on page 1-3. This is an important point that should not be relegated to one sentence.
Percentage of Constituents for Which There Are Health Data. The current methodology relies on carcinogenic and noncarcinogenic risk screening factors derived from cancer slope factors and RfDs. However, EPA has developed slope factors and RfDs for only a limited number of compounds. For constituents that lack established health benchmarks, EPA is proposing to develop provisional benchmarks. This proposal is acceptable, as long as the provisional benchmarks have an adequate regulatory or toxicological basis, as it is desirable to expand the screening analysis to the largest possible class of constituents. The document states that of the 256 study constituents, at least one benchmark is available from EPA sources for all but 25 constituents. For the few remaining chemicals for which benchmarks cannot be developed, a quantitative risk assessment will not be possible (page 3-25). This is a limitation of the current study, but not a serious one.

Approach for dealing with lack of information on chemical composition of wastewater in the impoundments or emissions: Are the proposed methods for dealing with situations where facilities do not report concentration or emissions data sufficient?

If concentration data are not available for a media at an impoundment, then one determines if data are available for that media for other impoundments at other facilities in the same industrial category. This is an entirely acceptable procedure. Phase 1B and Phase 11 use computer models to estimate the unknown concentrations of constituents in one media from the known concentrations in another media. This is also acceptable in that all of the models have received peer review.

4) Approach for representing cumulative risks: Is the proposed methodology for accumulating and representing risks adequate?

To answer this question, one must know the intended purpose of the cumulative risk distributions. Page 1-9 of the document states, "The purpose of the cumulative risk distributions is two fold: (1) screening out particular constituent/unit/pathway combinations of negligible concern from further analysis; and (2) developing an initial overall relative risk ranking of the surface impoundment universe". On page 1-12 the document further states, "EPA anticipates that the risk estimates generated during Phase 11 will provide a comprehensive National profile of potential risk posed by the universe of surface impoundments.". On page 1-13, the document states, "The study will provide a profile of the surface impoundment universe by unit type, industry type, and constituent; provide a descriptive profile of the subset of the universe that is of negligible concern and requires no further risk analysis; and provide a relative-risk profile for the entire universe of surface impoundments for these 256 constituents".

The development of the cumulative risk distributions (as proposed by the document) is a three-step process. First, during Phase 1A, an initial risk distribution is generated to represent the risk of direct exposure (ingestion and inhalation) to impoundment constituents. Those constituents with risk estimates falling above the Screening Criteria will have improved risk estimates calculated using procedures defined in Phase 1B, and all Phase 1A risk estimates falling above the Screening Criteria will be replaced by the corresponding estimates from Phase 1B. Constituents from this second-tier analysis with risk estimated falling about the Screening Criteria will have new risk estimates calculated using the multimedia modeling approach from Phase 11 and all Phase 1B risk estimates falling above the Screening Criteria will be replaced by the corresponding estimates from Phase 11.

Thus, the final cumulative risk distribution will be a compilation of three risk distributions. It will consist of those Phase 1A risk estimates that were below the Screening Criteria; together with those Phase 1B risk estimates that were below the Screening Criteria, together with all risk estimates calculated during Phase 11.

This procedure for constructing a cumulative risk distribution is adequate for eliminating constituents, impoundments, and facilitates from the detailed Phases 11 analysis. Only constituents that are not harmful even under the very stringent conditions of direct exposure are eliminated during the Phase 1A analysis. This is a very conservative process that is highly unlikely to eliminate any constituents or impoundments that should undergo further, more detailed, analysis. The elimination process under Phase 1B is not quite a stringent. However, the methodology is consistent with other EPA analyses and its use has been reviewed and approved for national level analyses by the EPA Science Advisory Board and the public. Thus, I conclude that the use of the cumulative risk distributions for the purpose of eliminating constituents, impoundments, and facilities from the more detailed Phase 11 analysis is adequate and protective of human health and the environment.

Now for the question of whether the final cumulative risk distributions obtained under the above defined procedure will "provide a comprehensive National profile of potential risk posed by the universe of surface impoundments". As stated above, the final cumulative risk distribution will be a compilation of three different risk distributions (with some numbers from the Phase 1A analysis, some numbers from the Phase 1B analysis, and some numbers from the Phase 11 analysis), each with different levels of conservatism. Thus the numbers (risk estimates) are not directly comparable. The final cumulative risk distribution obtained under the above procedure will not provide a comprehensive and realistic National profile of potential risk posed by the universe of surface impoundments. The upper end of the cumulative risk distribution will be realistic, as it will be composed of Phase 11 risk estimated. However, no more than 25% of the risk estimates in the cumulative risk distribution will be Phase 11 estimates. The rest of the distribution will be composed of Phase 1A and 1B risk estimates, which by their nature were designed to be highly conservative. Thus, the overall cumulative risk distribution will provide overestimates of the risk posed by a least 75% of surface impoundments and will not provide a realistic national profile of potential risk posed by the universe of surface impoundments. It will however provide an upper bound estimate of impoundment risks and thus will be useful in deciding issues that need further analysis.

5) Modeling Approaches: a) is the overall methodology for conducting screening-level modeling during Phase 1 sufficient? b) is the overall methodology for conducting modeling during Phase 11 sufficient? c) if a large number of sites require Phase 11 modeling, are pre-defined representative scenarios acceptable?

The overall Phase 1 methodology for conducting screening-level modeling is very conservative and hence is protective of human health and the environment. There is a very small probability of overlooking potential risks.

The Phase 11 methodology (fate and transport modeling) is state of the art. The exposure methodology includes all relevant exposure pathways and follows the standard EPA approach to exposure assessment. The approach to development of health effect benchmarks if complete. The various components of the Phase 11 methodology have previously undergone peer review by the EPA Science Advisory Board and been found to be acceptable of a national-level analysis.

The use of pre-defined representative scenarios is acceptable.

Specific Comments

2.0 Phase 1 Screening Assessment

2.2 Human Health Risk Screening

Page 2-4, paragraph 3. The screening risk formula is not clear. The terms "Risk Screening Factor" and "Risk Criteria" are not defined. This section should begin with a definition of Risk Screening Factor and Risk Criteria. What does the statement mean that the screening risk calculation is "equivalent to a standard forward calculation of risk, where exposure concentration is converted to a dose and multiplied by the cancer slope factor"? The formula has the same form, but none of the expressions have the same meaning. There is no calculation of dose.

What is the meaning of the sentence "The above calculation is easier to use for this SI Study screening phase"? What above calculation? Is easier to use than what?

Page 2-5, paragraph 2. There are several problems with the first sentence. First, the use of the word "concerns" makes it sound like it is not known whether this issue is a problem or not. Second, it is definitely true that a child would have a higher level of exposure if he/she were exposed to the same level of intake as an adult. However, this will not be the case. A child does not drink the same amount of water or eat the same amount of food. This point needs rewording. Finally, ending the sentence with "as an adult" leads to a awkward construction. This entire sentence needs to be rewritten. Actually the best reason for using a time-weighted child/adult is that cancer is a disease that results from a lifetime of exposure. Thus, it makes sense to use time-weighted body weights and intakes. There aren't used for noncarcinogenic risks because noncarcinogenic risks result from acute exposures.

It is not clear why the child is exposed to the soil and groundwater ingestion pathways and the adult is only exposed to the air inhalation pathway. The document states that the RfC for inhalation is protective of both child and adult receptors. This explains why the inhalation pathway is not considered for children. Why aren't the soil and groundwater ingestion pathways considered for adults? (I now see on page 2-8 that for noncarcinogenic risks, the age group that produces the lowest screening factor is used in the exposure assessment. This is the 1 to 5 age group of children for soil and groundwater ingestion. This point needs to be explained here).

Page 2-5, paragraph 3. The description of the human exposure pathways is clear. However, they differ from the ones listed under Human Receptor Types.

Page 2-9, paragraph 1. I agree that exposure factors should be selected to represent typical or central tendency values, not high-end values.

Page 2-9, paragraph 2. The exposure duration of 30 years seems adequate. The exposure frequency of 350 days per year seems overly conservative. This would assume, for inhalation for example, that an adult never leaves the site at all (except for two weeks vacation) and is exposed to volatile emissions all day long, an unlikely event.

Page 2-9, paragraph 3. The inhalation rates seem reasonable.

Page 2-9, paragraph 4. The soil ingestion rate of 200 mg/d is conservative and, when used in conjunction with an assumed exposure frequency of 350 days per year, results in a high intake from soil ingestion. It is unlikely that a child will consume 200 mg/d of soil for 350 d/yr.

Page 2-9,paragraph 5. The drinking water intake rates seem reasonable.

Page 2-10, paragraph 1. The proposed method of obtaining body weights seems reasonable.

Page 2-10, paragraph 3. The consideration of cumulative effects is reasonable. However, this section does not say how this will be done.

Page 2-10, paragraph 6.The document states that "surface impoundment wastewater concentrations will be used in Phase 1A". Figure 2-4 states that a step in the process is to "estimate leachate concentration". Which is used, surface impoundment wastewater concentrations or leachate concentrations? Or are they the same thing? This needs clarification.

Page 2-13, paragraph 1. The constituent risk and HI will be calculated by summing the risks and hazard quotients for all pathways for a particular constituent. This procedure is reasonable except that many constituents may only have wastewater concentration data. You might state here that if air concentration data and sludge concentration data are not available, they will be estimated from the wastewater concentration data.

Page 2-13, paragraph 4. I agree that the use of average concentrations of constituents in impoundments is a reasonable approach to evaluating long-term chronic exposures by receptors.

Page 2-17, paragraph 1. The methods for calculating impoundment risks, constituent risk, and facility risks appear to be reasonable.

Page 2-17, paragraph 5. The use of cumulative frequency histograms to represent risks is appropriate.

Page 2-18, paragraph 2. Example. The example is somewhat confusing. The document says that the cancer risk number is multiplied by the sample weight for the facility (38.7) and added to the risk bin. For the noncancer risk bin, the document says that 38.7 is assed to the 0.01 to 0.1 bin. Why isn't 38.7 multiplied times the HI and then added to the noncancer risk bin?

Page 2-18, paragraph 3. Risk Screening. The risk screening procedure outlined in Figure 2-9 seems reasonable. It is sufficient to identify facilities, impoundments, and constituents that should proceed to the next phase.

Page 2-20, paragraph 2. This paragraph concerning the use of a margin of protection is confusing. First, it says that this procedure "may" be used. On page 1-8, the document says that the procedure will be used. The rest of the paragraph is difficult to understand. It would be better if you explained what each step is trying to accomplish.

Page 2-20,paragraph 2.Example. The example is clear and well written. However, the risk screening results in Table 2-3 are based on decision criteria rather than screening criteria as mentioned in paragraph 1, page 2-20. Also, the decision criteria at the bottom of Table 2-3 are not clear. The 10-5 for cancer risk is the same as the risk criteria for carcinogens given on page 1-8. However, the0.1 for noncancer risk is 0.1 lower than the risk criteria for noncarcinogens given of page 1-8. Is there a mistake here? If not, this needs an explanation.

Page 2-20, paragraph 3. The documents states that "This example illustrates that the decision criteria incorporate a margin of protection in determining which units and constituents proceed to Phase 1B and 11". It is not clear how this example illustrates a margin of protection. Is it because the decision criteria at the bottom of Table 2-3 incorporate the margin of protection? As I pointed out above, there seems to be some discrepancy in these numbers.

Page 2-22, paragraph 1.This paragraph clarifies the use of the margin of protection.

Page 2-22, paragraph 2. Two points. First, the documents states ".screening based on direct ingestion of the surface impoundment influent." This is a point that I have mentioned before. Is the direct ingestion exposure assumed to be to the impoundment influent or the impoundment leachate? Second, the document mentions indirect exposure as the justification for the margin of protection. More needs to be said about this. What is the assumed scenario for the indirect exposures? Is it irrigation of crops with impoundment influents? Is it growing crops on impoundment sludge? What is the justification for the value of 0.1 for the margin of protection? Will it be sufficient to screen for the food chain pathways?

Page 2-22, paragraph 6. The document states ".certain industries were given higher priority in the SI Study" What is the SI Study?

Page 2-23. In the discussion of Table 2-4, indicate that the 215 facilities listed in the Table are the facilities selected to participate in the detailed survey to determine the characteristics of surface impoundments.

Page 2-23. Risk Screening. It is not clear if the constituent distribution is composed of the constitutions across all facilities or just the facilities that fail the facility screening and must proceed to Phase 1B.

Page 2-25. The Example. This example is clear except for a couple of points. The document states that "The third distribution is on a constituent basis and shows the constituents that will be assessed in Phase 1B. This constituent distribution applies to the subset of impoundments and facilities that have proceeded to Phase 1B". First, make it clear that there are multiple constituent distributions, one for each constituent. Second, there is an inconsistency here. Page 2-17 says that the constituent risk is the maximum risk for the constituent across all impoundments for the particular facility. The constituent risk is thus a number associated with a particular facility. The constituent distribution will therefore show the risk of a particular constituent across all facilities, both those that require further analysis and those that don't. The sentence above (taken from the Example) says that the constituent distribution represents only those facilities that have proceeded to Phase 1B. Thus, there is a contradiction between pages 2-17 and the Example on page 2-25. Further, if Figure 2-11 only represents facilities that have proceeded to Phase 1B, why are there risk values less than the Screening Criteria in Figure 2-11? This may be correct, but it needs explanation. In addition, the whole concept of the constituent distribution needs clarification.

Page 2-27, paragraph 1. The use of screening models to evaluate the fate and transport of constituents that have proceeded to Phase 1B is reasonable.

Page 2-27, paragraph 4. The document states that "these risk estimates will replace or supplement the corresponding Phase 1A screening risk estimates and therefore will refine and improve the overall Phase 1A risk distributions". This statement is only partially true. Since there are different levels of conservatism in the Phase 1A risk estimates and the risk estimates produced by the screening models, to two are not exactly comparable. The new risk distribution will give a better estimate of the facilities, impoundments, and constituents that can be eliminated from further consideration. It will also give a truer picture of the risk of the facilities, impoundments, and constituents that need further analysis. However, the risk distributions for the facilities, impoundments, and constituents that have been eliminated from further consideration will not be consistent. Despite this inconstancy, the updated risk distributions are the best way to display the risk data. The best way to handle this situation is to discuss the limitations of these figures in the document.

Page 2-27. IWAIR. The IWAR model is a reasonable way to estimate the inhalation risks from airborne volatile constituents released from surface impoundments. The component submodels of IWAIR have undergone extensive peer review. The IWIAR model is fast and easy to use and requires minimal input data.

Page 2-28. IWEM. The IWEM model is acceptable for estimating risks due to exposure to groundwater containing constituents released from surface impoundments. The model uses a dilution attenuation factor (DAF) to account for the dilution and attenuation that occur as the constituents move through soil to reach groundwater. The DAF is conservative in that it is generated using the EPA Composite Module for Leachate Migration with Transformation Products (EPACMTP), a model that has received extensive peer review, and uses a DAF that is generated through a Monte Carlo procedure that accounts for variability in groundwater parameters nationwide.

Page 2-29, paragraph 1. The computation of the DAF is appropriate for the present application. It represents a dilution and attenuation factor that is conservative in 90 percent of the groundwater situations nationwide. The Monte Carlo procedure and its applicability to national analyses has been extensively reviewed external reviewers and by the EPA Science Advisory Board and found to be acceptable.

Page 2-29, paragraph 3. The documents states "To maintain consistency with Phase 1A in the risk calculation, only DAFs from IWEM will be used". Since no DAFs were used in Phase 1A, it is not clear what this sentence means. I think that the consistency refers to the rest of this paragraph and not to the use of the DAF. This needs clarification.

Page 2-29, paragraph 4. This procedure for calculating the DAF of chemicals not included in the IWEM Tier 1 table seems reasonable.

Page 2-30, paragraph 1. The first bullet states, "Risks will be calculated for each constituent, impoundment, facility." However, the sentence following these bullets states, "The analysis will be performed only for a subset of constitutes, impoundments, and facilities defined by the Phase 1A screening results". This contradicts the first bullet. In addition, how will Phase 1A define those units requiring Phase 1B analysis?

Page 2-30, paragraph 1. The document states "The primary difference between Phase 1A and 1B is the procedures used for calculating risks". Actually, the primary difference is the procedures used for calculating exposure concentrations. The exposure factors (e.g., amount ingested and inhaled) and toxicity data are the same.

Page 2-30. Figure 2-12 is clear.

Page 2-35, paragraph 2. This paragraph is well written and easy to understand. Figure 2-16 makes sense.

Page 2-36. Specials Cases. The approach to evaluating constituents that may be persistent or bioaccumulative seems reasonable. The revised WMPT appears to be appropriate for this type of prioritization. The text states "Persistence is based on a steady-state, nonequilibrium multimedia partitioning model to estimate constituent half-life". What are the compartments of the model? The half-life of the constituent in what environmental media is used to determine persistence? (Air, water, soil?).

2.3 Ecological Screening Assessment

Page 2-38, paragraph 2. The last sentence on this paragraph is confusing. It says that "results from the ecological screening will be used to inform the selection ..most likely to pose significant risks to both human and ecological receptors." Does this mean that to be considered for Phase 11 evaluation, a constituent, impoundment, or facility must score above the screening criteria for both human health risk and ecological risk? Or does it mean that they make it to the Phase 11 evaluation if they score high for either the human health risk or ecological risk?

Page 2-38. Design Goals and Overview. This section is well written and clear. I agree that the use of screening factors will be protective of ecological receptors.

Page 2-42, paragraph 2. The approach to ecological risk assessment based on organizing receptors into feeding guilds and selecting a species to represent each guild seems reasonable for a screening level assessmenet.

Page 2-42, paragraph 3. The ecological exposure pathways listed for use in development of ecological screening factors appear reasonable and complete. The exclusion of dermal absorption from surface waters and inhalation pathways also appears reasonable.

Page 2-45, paragraph 1. In defense of not including inhalation of volatile compounds in the development of ecological screening factors, the document states "Concentrations of volatile chemicals released from soil to aboveground air are drastically reduced, even near the soil surface". If this is true, why is the inhalation pathway included in the human health risk assessment?

This section also states "Significant concentrations of VOCs would be required to induce noncarcinogenic effects in wildlife based on inhalation toxicity data for laboratory rats and mice". Why would this not also be true for humans? Again, given these two points, why is the inhalation pathway included in the human health risk assessment?

Page 2-45, paragraph 2. The document states, "As suggested in Table 2-6, risk to four groups of receptors (mammals, birds, amphibians, and reptiles) will be estimated based on endpoints relevant to population sustainability". First, Table 2-6 also lists plants and invertebrates. Why aren't these also being used? Second, the first footnote in Table 2-6 says that reptiles will not be assessed in Phase 1 due to lack of toxicity data. Why do you say in this paragraph that reptiles will be assessed? Finally, you say that risk will be estimated based on endpoints relevant to population sustainability. What endpoints relevant to population sustainability? In the very next two sentences you say you are not using population studies, but rather physiological endpoints like reproductive or developmental effects.

Page 2-46. Duration of Exposure. The definition of chronic exposure as greater than 50 percent seems reasonable for the present analysis. However, for long-lived mammals, this may be an overly long time period. It is possible that a shorter time period could be used. This is an issue that could stand further study.

Page 2-46. Timing of Exposure. The focus on exposures occurring during reproductive and/or developmental stages is appropriate in the absence of chronic studies.

Page 2-46. Endpoint of Interest. The hierarchy of endpoints of interest presented (reproductive or developmental, physiological, tumorigenic) is appropriate. I agree that tumorigenic and carcinogenic toxicity studies should not be considered ecologically relevant.

Page 2-47. Data Gaps. The use of surrogate chemicals to derivate ecological screening factors when other data are not available seems appropriate. It is not clear how the use of uncertainty factors can over come the existence of data gaps. The text should be clearer in the statement of this alternative.

Page 2-47, paragraph 3. The text states "Screening factors for communities generally are not based on community-level studies in the sense that they do not reflect endpoints relevant to community dynamics". What does the "they" refer to? This sentence does not make sense to me. If I understand it correctly it says that community-level studies do not reflect endpoints relevant to communities. How can this be? The endpoint of community-level studies is the community by definition.

The next sentence states "They are based on the theory that protection of 95% of the species in the community will provide a sufficient level of protection for the community". Again, what does the "they" refer to? I assume that it means community-based screening factors. But if that is the case, the sentence says that community-based screening will be based on protection of 95% of the species in the community. Are you proposing this as the definition for developing community-based screening factors? If so, you will have to identify all of the species in the community so you know if you are protecting 95% of them. I don't think you want to do this. In addition, you would have to assess the impact on all species to see if you are protecting them. I don't think you want to do this.

Finally, in the paragraph, the document states, "As with the wildlife populations, ecotoxicological data on individual species will be used to infer risks to the community". What does "As with wildlife populations" mean? I thought we were developing screening factors for wildlife populations.

Page 2-47, paragraph 4. The reliance on existing compendia to derive screening factors seems reasonable.

Page 2-49, paragraph 4. The uncertainty factors proposed for use in extrapolating from acute exposures to chronic exposures seem reasonable.

Page 2-50, paragraph 2. The scaling of toxicological effect across species using the body weight formula 2-2 is appropriate. This is the default methodology EPA uses for carcinogenicity assessments.

Page 2-50, paragraph 3. Based on the Mineau et. al. study, the scaling formula 2-3 appears reasonable.

Page 2-50, paragraph 4. The proposed uptake/accumulation factors appear reasonable.

Page 2-51, paragraph 4. The proposed method for calculating screening factors appears reasonable.

Page 2-55, paragraph 4. The inclusion of terrestrial receptors and exclusion of aquatic and sediment-associated biota for impoundments seems appropriate.

Page 2-56. The HQ methodology for estimating risks to receptors is appropriate. However, formula 2-9 refers to receptor i associated with that impoundment and facility. This raises the question as to whether the screening factors are for individual receptors or for communities as suggested on page 2-54. Since the toxicity data being used are for individual species (page 2-47, paragraph 4), how are community screening factors derived?

Page 2-59. The document proposes that a surface impoundment or facility will proceed to Phase 1C if (1) the hazard quotient for plants exceeds 10 and (2) the hazard quotient for at least one other receptor group exceeds the risk criterion of 1. I agree with the reasons given to support this procedure.

Page 2-60, paragraph 1. It is reasonable that the ecological risk screening be used in support of the overall prioritization, not as am independent screen to identify facilities.

Page 2-60, paragraph 1. The document states, "The ecological risk characterization will characterize facility-level risks to assess cumulative effect of multiple surface impoundments and constituents". How will the cumulative effect of multiple surface impoundments be assessed? The document has not discussed this issue. I assume that the facility HI is the summation of constituent His for each ecological receptor. This needs to be stated.

Page 2-60, paragraph 4. The document proposes to give a high priority of facilities within 2 kilometers of the listed managed areas. This seems reasonable, although the methodology assumes that each receptor species forages only within the contaminated area (the impoundment) and an impoundment 2 kilometers from a managed area must only represent a very small fraction of a species home range. Thus, this assumption represents a very conservative approach to protecting managed areas.

Page 2-62, paragraph 2. The designation of impoundments within 1 kilometer of a wetland as high priority seems reasonable. In fact, this assumption is more reasonable than that for managed areas because of the potential for surface runoff and the fact the birds frequent wetlands and can easily travel the 1-kilometer to the surface impoundment.

2.4 Phase 1C Initial Prioritization

The purpose of the Phase 1C prioritization is to insure that the number of constituents, units and facilities that move into Phase 11 is not greater than 25 percent of the study sample. This seems reasonable given that Phase 11 analysis is resource intensive and recognizing that the whole purpose of the Screening Study was to limit the number of Phase 11 analyses that had to be done.

Page 2-62, paragraph 4. The first sentence states " The goal of Phase 1C is to prioritize the Phase 11 analysis of these constituents, impoundments, and facilities" What does the "these" refer to?

Page 2-63, paragraph 3. The document states, "For ecological risk, a single facility risk based on the maximally impacted receptor type is provided". This provides clarification of how the ecological risk of a facility is determined. This point should be made earlier in the document.

Page 2-64, paragraph 2. Either of the two options seems reasonable. However, option 2 is not clear. The option says, "Some constituents (and impoundments) may constitute more of the facility risk than others". This is true, but the facility risk is the highest of the constituent risk, so how does this fact affect facility prioritization? The next sentence says, "The constituent score will be used to prioritize the rank". The rank of what? The facility? And how will it be used?

Page 2-64, paragraph 3. The proposed method of combining of human health and ecological scores seems reasonable.

3.0 Phase 11 Risk Assessment

Page 3-1, paragraph 2. The document states, "Because the high-end and central tendency scenarios will be based on real receptors, they will provide a realistic span of potential risks". It is not clear at this point that the Phase 11 assessment if using high-end and central tendency scenarios. What does it mean that they will be based on real receptors? Very few of these sites will have real receptors.

Page 3-4, paragraph1. The document states, "Alternatives also exist in terms of necessary modifications to the 3MRA modules and modeling system" What does this sentence mean?

Page 3-6. The conceptual exposure model for human receptors discussed in section 3.2 and figure 3-3 seems reasonable. This conceptual model is similar to those used in most health risk assessments. It considers all reasonable exposure pathways and receptor groups.

Page 3-8. The special scale considered seems appropriate.

Page 3-12. The temporal scale considered seems appropriate. However, 5,000 years seems like too long of a time period over which to run the model. I understand that the intent is to capture the potential impact of slow moving contaminates. However, the longer the time period that a model tries to simulate, the larger the errors in the final estimates. When a model is run over a 5,000 year time period, the confidence bound on the best estimate is larger than the estimated itself. That is, the best estimate is meaningless. I also understand that it is standard EPA policy to run models over these time periods. However, you can tell from my comment that I believe this policy is incorrect.

Page 3-15. Temporal Integration. The proposed procedure for aggregating risk over a given time period is correct.

Page 3-17, paragraph 2. The document states, "Types of outputs at Tcrit include population-weighted risk or hazard quotient". The document has not described how it will compute population-weighted risk. The process outlined in steps 1 and 2 page 3-17 does not mention population size.

I can see obtaining population risks for inhalation exposures, but how is the methodology going to determine population risks for groundwater? Will it assume that all people living within the 2-km radius obtain all of their drinking water from wells contaminated by the surface impoundment? What about the fishing pathway? All people living within the 2-km radius obtain their fish from streams near the impoundment?

3.2.3 Human Health

Page 3-18. Human Receptor Types. The matrix of receptor types and age cohorts is appropriate.

Page 3-19. Human Exposure Pathways These are the standard exposure pathways and represents the standard EPA approach to exposure assessment. This exposure approach has received extensive peer review. This section does point out that ingestion of contaminated groundwater is for private groundwater wells only and that fish ingestion is for all recreational fishers.

The decision to exclude the dermal routes of exposure is justified given the lack of health benchmarks for dermal toxicity.

Table 3-4 is clear and represents the standard EPA approach. This approach is highly protective and provides an over estimate of risk in all but a very small number of conceivable cases.

Page 3-21 Human health Effect Benchmarks. The proposed health effect benchmarks are standard. Appendix A provides a list of EPA and Non-EPA available benchmarks. At least one benchmark exists for all but 25 constituents. EPA proposes to other data sources to develop draft provisional benchmarks where necessary. This approach is reasonable given that this is a screening study whose purpose is to identify possible health problem resulting from surface impoundments. If any constituents with provisional benchmarks are identified as potential problems, the validity of the approach can be investigated with greater detail.

The data for lead due not fit the standard RfD methodology in that health impacts from lead may not have a threshold. Based on a biokinetic model relating lead exposure to blood lead levels in children, EPA has developed a soil screening level of 400mg/kg for lead. In the present study, EPA is considering use of a screening value of 400mg/kg in sludge and 10 to 15 mg/L for water. These values seem appropriated given the extensive peer review that the EPA approach to lead has undergone.

EPA is in the process of developing a revised RfD for perchlorate. The proposed value is 0.0009 mg/kg-d. A provisional RfD of 0.002 mg/m3 was derived for the listing rule for solvents. Either of these two numbers seems reasonable to use as a provisional benchmark. However, since EPA, after extensive review, believes that 0.0009 mg/kg-d is the best health benchmark available, it would seem to be best to use this number even though it is still undergoing review.

The last sentence on page 3-25, paragraph 4 seems to be out of place. It is a repeat of the last sentence in the previous paragraph on the screening value for lead. Or is EPA really proposing a screening value of 400 mg/kg for perchlorate in sludge? If so, which of the provisional health benchmarks was used to derive this number?

Page 3-25, paragraph 5. The document states, "In summary, EPA benchmarks exist for 240 SI constituents. Of the 24 with benchmarks,." This adds up to 264 constituents. On page 3-23, the document states that there are 256 constituents.

Page 3-25 Human Health Risk Measures. The procedures for calculating cancer and noncancer risk follow standard EPA practice. The document states, "Cancer risks will not be summed if inhalation or ingestion of a chemical results in health effects at the point of exposure or has other immediate effects". The meaning of this sentence is not clear. I assume that it means that if a chemical has acute effects, it is treated as an acute toxin and it's cancer risk is not added to the other cancer risks. What the sentence actually says is that if one chemical has acute effects, none of the cancer risks will be summed.

Page 3-26 Lead Evaluations. The proposed methodology for calculating blood-levels appears sound, even if it may be complex to the models. The document does not say what pathways will be considered in evaluating lead exposure to young children. If only soil ingestion is considered, then it may not be necessary to run the IEUBK model.

Page 3-27 Infant Risks. The document says, "The risk will be characterized as a margin of error (MOE) using a measure of background for comparison purposes". Explain what a margin of error approach is.

Page 3-27 Individual vs. Population-Weighted Risks. The document should mention in the beginning that both individual and population-weighted risks will be estimated.

The document states, "Normalized population risk estimates will be provided for each receptor type to provide the distribution of risk across all receptors in the study area". This sentence is unclear. How will normalized population risk estimates provide the distribution of risk across all receptors? I guess each receptor type will have a single normalized population risk number assigned to it and the collection of these single estimates is the distribution.

The definition given by formula 3-1 is clear.

Page 3-27 Uncertainty Analysis. This section needs a little amplification. How will a semi quantitative uncertainty analysis be done? There are multiple models and multiple data sets involved. Will uncertainty analysis be done on all of the environmental transport models? How will it be done? Will uncertainty is the exposure scenarios also be evaluated? How will uncertainty is the health benchmarks be addressed? As pointed out on page 3-18, an analysis of sources of uncertainty in the predicted risk estimates is an important part of a complete human health risk assessment. This section needs a more complete treatment.

3.3.4 Ecological Health

Page 3-28 Representative Ecological Receptors. The ecological receptors to be evaluated appear appropriate. The list of habitats on page 3-29 seems representative of habitats across the United States that may be found near surface impoundments. It is appropriate to exclude estuarine and marine ecosystems from the analysis, since there currently does not exist an automated system for such an analysis. The representative ecological receptors that will be used to populate habitats seem appropriate.

The ecological exposure pathways presented in Table 3-5 are appropriate. It is appropriate to exclude dermal exposure and inhalation as pathways of exposure for the reasons given.

While I admit that the concepts of assessment and measurement endpoints are new to me, I am a little confused by the write up on page 3-31. Assessment endpoints are defined as the ecological values to be protected, while measurement endpoints are the eco-toxicological effects used to support benchmarks. Thus, as the text describes, measures of effect reflect toxicity to the individual while assessment endpoints represent ecological values that go beyond the individual receptor. I therefore find confusing the statements on page 3-32 that say, " Assessment endpoints are defined based on the identification of potentially exposed ecological receptors and potentially complete exposure pathways. Development of assessment endpoints is based on the assumption that protection of a population or community can be inferred from protection of developmental and reproductive functions in individuals". How can the ecological values to be protected (assessment endpoints) be based on the identification of potentially exposed ecological receptors? One would first chose the ecological value to be protected (in this case, ecological communities) and then determine receptors for which to develop benchmarks. Second, how can the ecological values to be protected be inferred from protection of developmental and reproductive functions in individuals? First, one chooses the ecological values to be protected (in this case, ecological communities) and then one assumes that protection of the developmental and reproductive functions of individuals will provide protection to the desired ecological value. These descriptions need clarification.

Page 3-33. Wildlife Exposure Factors. The section on exposure factors seems reasonable. The use of mean wildlife exposure factors to evaluate the impact on receptors at large is reasonable. The last sentence in this section needs attention. It states," Ingestion rates are a function of body weight and, thus, will also, reflect central tendency values". Just because ingestion rates are a junction of body weight does not mean that they reflect central tendency values. If you used maximum ingestion rates divided by body weight, they would reflect maximum values.

Page 3-35. The use of the ¼ scaling factor seems appropriate, as does the use of a scaling factor of 1 for birds. Formula 3-2 has a "0" as the exponent of body weight. This is incorrect. It should be "1". Also the definition of MATCt is given as the maximum acceptable toxicant concentration, where as it should be the maximum acceptable toxicant intake (the units are mg/kg-d).

Page 3-38. The methodology for converting total metal concentrations in the water column to total dissolved concentrations seems appropriate. It is based on a series of filtration experiments conducted by EPA.

Page 3-42 Uncertainty Analysis. The documents states, "A methodology for a qualitative analysis of uncertainty was developed for the HWIR analysis and will be applied to the surface impoundments ecological risk assessment". Some description of this methodology should be given.

The document speaks of confidence indicators on pages 3-42 and 3-43. Are the confidence indicators in addition to the uncertainly analysis mentioned above?

3.3 Overview of Simulation Models

Page 3-51. The air model ISCST3 is an accepted air transport code. Its use as a legacy code within the 3MRA framework seems reasonable. ISCST is a regulatory approved model that has had extensive use in the past. The modification to ignore plum depletion is reasonable given that less than 1 percent of the plum is deposited within a 2-km radius of the source.

The proposed groundwater model seems reasonable. It is based on EPACMPT, an accepted model for estimating the migration of compounds in the subsurface and predicting potential exposure at a downstream receptor will.

Top of Page

Peer Review of the Surface Impoundment Study Technical Plan for Human Health and Ecological Risk Assessment

Local Navigation