4_f__-_Crystal_Reservoir_Dust_Hazard_Analysis_1.pdf

Fugitive Dust Hazard Analysis

Crystal Reservoir

Ash Meadows National Wildlife Refuge

The U.S. Fish and Wildlife Service (USFWS) is considering closing the Crystal Reservoir within Ash Meadows National Wildlife Refuge (AHME) (Figure 1). This would entail removing Crystal Dam and draining the reservoir. One concern is that allowing the reservoir area to dry could create a significant source of dust, and if that dust contains high metal concentrations, it could pose an inhalation hazard for Refuge workers and visitors. This memorandum evaluates that potential.

Ten sediment samples were collected from the 115-acre reservoir (Table 1) and analyzed for a suite of thirteen naturally occurring metals. In AHME, these metals originate from bedrock and are transported to the surface by groundwater. Saturated areas may have the greatest metal concentrations due to frequent transport of bedrock metals to the surface. The metals chosen for analysis are common target analytes for soil and sediment and have quantitative inhalation toxicity criteria appropriate for an environmental setting. Nine of the analytes were detected in at least one sample (Table 1): aluminum, arsenic, barium, beryllium, chromium, cobalt, manganese, nickel, and vanadium. Antimony, cadmium, lead, and mercury were not detected at levels that meet the Practical Quantitation Limit, which is the minimum concentration of an analyte that can be confidently identified using an EPA approved laboratory method, and are not further considered.

Three soil samples were collected from outside the reservoir. The purpose of these background samples was to identify whether metal concentrations in reservoir sediments had been enhanced over time (e.g., from evaporation causing metals to precipitate out of the water into sediments). These samples were collected in a transect extending out from the reservoir.

CSBG11 was collected in an area that is perennially moist, CSBG12 was collected from soil that is seasonally moist, and CSBG13 was collected from a location that is dry year-round. Data from these samples are summarized in Table 1.

Two lines of evidence were used to assess the potential dust hazard. The first line of evidence was to estimate the risk posed by metals in the reservoir dust. The second line of evidence was to compare concentrations detected in the sediment near Crystal Reservoir to those in the surrounding soil. Similar concentrations would indicate that, barring an increase in total dust levels, any hazard from reservoir dust would be comparable to the dust risk already posed by ambient surficial materials. We also compared these results with a recent study in which samples from Ash Meadows were analyzed for some of the same analyte concentrations (Goldstein et al. 2011).

Figure 1. Location of Crystal Reservoir within Ash Meadows National Wildlife Refuge.

Potential Toxic Hazards from Reservoir Dust

A risk-based framework was used to evaluate the potential for an inhalation hazard. This framework is utilized by the U.S. Environmental Protection Agency (EPA) in evaluating hazards where constituents have been released to the environment (EPA 1989). The framework has a four-step process:

● Hazard Assessment and Data Evaluation

● Exposure Assessment

● Toxicity Assessment

● Risk Characterization

The analysis below follows this framework.

Hazard Assessment and Data Evaluation: The hazard assessment and data evaluation consists of refining the data to only consider constituents likely to be responsible for the greatest risks and determining an exposure point concentration (EPC) for those constituents. Given the small number of samples collected and the limited number of analytes, all of the detected analytes were further evaluated in this risk memorandum. The EPC is the concentration used in the exposure and risk calculations.

The EPCs were calculated as the 95th percent upper confidence limit (95% UCL) on the mean.

The 95% UCL is a high-end estimate of the true mean concentration in the reservoir sediments.

The mean of the ten samples is a simple calculation, but if ten more samples were collected, it is highly unlikely that they would have exactly the same concentrations (i.e., the mean of the next ten samples might be higher or lower than that of the first samples). The 95% UCL is a high end estimate of the mean, such that there is only a 5% chance that the true mean of all of the sediment in the reservoir is higher than this value.

An EPA program, ProUCL Version 4.0, was used to calculate the UCLs (EPA 2007). ProUCL computes UCLs by comparing a data set to normal, lognormal, and gamma distributions. It recommends using the 95% UCL from the distribution that best fits the data as the EPC. If none of these distributions fit the data, ProUCL identifies one of several non-parametric methods (i.e., methods making no distributional assumptions) to determine an appropriate EPC. Outputs from the ProUCL analysis are included in Appendix A. The EPCs recommended by ProUCL for sediment are summarized in Table 2.

Exposure Assessment: The exposure route being evaluated in this memorandum is inhalation of dust that comes from the dried-up reservoir. Potential receptors are USFWS workers and Refuge visitors. USFWS workers will be exposed to dust more frequently and over a longer period of time than Refuge visitors, thus USFWS workers are considered to be the most sensitive (i.e., highest risk) receptor. Therefore, if there are no significant risks for workers, conditions will also be acceptable for Refuge visitors.

The amount of exposure a worker has depends on the number of hours per day that the person inhales dust, the number of days per year that they are exposed, the number of years they are exposed, and the average concentration of each inhalable constituent. The following equation was used to quantify exposure using those parameters:

Inhalation Concentration (mg/m3) =

AT

x ED x EF (ET/24)x Ca where:

Ca = Constituent Concentration in Air (mg/m3)

ET = Exposure Time (hrs/day)

EF = Exposure Frequency (days/year)

ED = Exposure Duration (years)

AT = Averaging Time (period over which exposure is averaged – days)

The values used for these parameters are summarized in Table 3.

The constituent concentration in air depends on the concentration of metals in the dust and the amount of dust generated. It is calculated as follows:

Ca = Cs ÷ PEF where:

Ca = Constituent Concentration in Air (mg/m3)

Cs = Constituent Concentration in Soil (mg/kg)

PEF = Particulate Emission Factor (kg/m3)

The constituent concentration in soil is the EPC. The PEF is the factor that relates the air calculation to the soil calculation and indicates expected dust generation. It is a modeled parameter that relates factors such as area, mean annual wind speed, and soil texture. Finely textured soils (i.e., more than 60% able to pass through a 1-mm sieve) are classified and modeled as an “unlimited reservoir.” To reach a conservative estimate, Crystal Reservoir sediments were modeled under this classification. The model also conservatively assumes that emission rates related to wind and mechanical entrainment will be continuous and steady (Cowherd et al. 1985). This model and its calculations are described in further detail in Appendix B.

Toxicity Assessment: Risks associated with exposure to the constituents detected at a site are a function of the inherent toxicity of each chemical, as well as the exposure concentration described above. Generally, there are two categories of information that are considered:

● Information on the non-cancer effects of chemicals

● Information on the potential for chemicals to initiate or promote cancers

Non-Cancer Effects: The assessment of non-cancer effects is complex, with a broad interaction of time scales (acute, sub-chronic, and chronic), routes of exposure (ingestion, inhalation, dermal), and varying kinds of effects. A single chemical may elicit several adverse non-carcinogenic effects depending on the dose, exposure route, and duration of exposure. This hazard assessment is focused on effects from chronic exposure, which is exposure occurring over several years.

For many non-carcinogenic effects, protective mechanisms must be overcome before the effect is manifested. Therefore, there is a finite dose (i.e., a threshold) below which adverse effects will not occur. Many toxicological studies focus on identifying where this threshold lies.

For a given chemical, the concentration that elicits no effect when evaluating the most sensitive response (i.e., the adverse effect that occurs at the lowest concentration) in the most sensitive species is used to establish an acceptable concentration (toxicity value) for non-cancer effects.

This dose can be related to an inhalation reference concentration (RfC). A chronic RfC is an estimate of a daily exposure level (with uncertainty spanning perhaps an order of magnitude) for which people, including sensitive individuals (e.g., children, elderly), do not have an appreciable risk of suffering significant adverse health effects. Exposure above an RfC could possibly cause health effects.

The RfC is calculated by starting with the concentration that elicited no adverse effects and applying uncertainty (safety) factors. Most uncertainty factors are standardized and include:

● 10-fold factor for extrapolation from animals to humans

● 10-fold factor for variability in the human population

● 10-fold factor for use of a less-than-chronic study

● 10-fold factor for extrapolation from a lowest observed adverse effect level to a no-observed adverse effect level

Other factors may further modify the RfC, such as epidemiological evidence or the absence of information on certain types of effects (e.g., reproductive effects).

Cancer Effects: Studies of carcinogenicity tend to focus on identifying the slope of the linear portion of a curve of concentration versus response. A plausible upper-bound value of the slope is called the cancer slope factor (SF). The product of the SF and the exposure concentration is an estimate of the risk of developing cancer. The unit risk factor (URF) is the upper-bound lifetime excess cancer risk estimated to result from continuous exposure to a concentration of 1 μg/m3.

Identification of chemicals as known, probable, or possible human carcinogens is based on an EPA weight-of-evidence classification scheme in which chemicals are systematically evaluated for their ability to cause cancer in mammalian species, and conclusions are reached about the potential to cause cancer in humans.

The EPA classification scheme (EPA 1989) contains six classes based on the weight of available evidence, as follows:

A Known human carcinogen

B1 Probable human carcinogen – limited evidence in humans

B2 Probable human carcinogen – sufficient evidence in animals and inadequate data in humans

C Possible human carcinogen – limited evidence in animals

D Inadequate evidence to classify

E Evidence of non-carcinogenicity

Some chemicals in Class D may have the potential to cause cancer, but adequate data are not currently available to change the classification. In this risk assessment, evaluations of the likelihood of a carcinogenic effect include chemicals in Classes A, B1, B2, and C. The EPA does not assign URFs to chemicals in Classes D and E.

Toxicity Values: Toxicity values used in this risk assessment are summarized in Table 4. The primary source of RfCs and URFs is the Integrated Risk Information System (IRIS), an electronic database for selected compounds, which was established and is maintained by EPA. This IRIS is based on a compendium of available toxicological data, containing both United States and international studies, and peer-reviewed and non-peer reviewed research. The database identifies a critical effect, which is the health endpoint upon which the RfC is based.

In accordance with EPA protocols, when toxicity values are not available on IRIS, EPA’s Provisional Peer-Reviewed Toxicity Values (PPRTV) database is to be used as a source of toxicity values (EPA 2003). When values are unavailable on IRIS or PPRTV, non peer-reviewed values from the National Center for Environmental Assessment (NCEA), minimal risk levels established by the Agency for Toxic Substances and Disease Registry (ATSDR), or the EPA Health Effects Assessment Summary Tables (HEAST) (EPA 1997) may be consulted. Direct access to the PPRTV database is not generally available, but values from this database are found on EPA’s Regional Screening Level Table (EPA 2009b).

Risk Characterization: Risks are characterized by the calculation of hazard quotients and cancer risks for non-cancer and cancer effects, respectively. These are described below.

Non-Cancer Effects: The potential for individuals to experience effects other than cancer is evaluated by comparing the exposure concentration during the period of exposure to the RfC.

This comparison takes the form of a ratio termed the hazard quotient (HQ):

Hazard Quotient =

EC

RfC where:

EC = Exposure Concentration (mg/m3)

RfC = Reference Concentration (mg/m3)

If the hazard quotient is less than or equal to 1, no adverse health effects are expected. If the hazard quotient is greater than 1, adverse health risks are considered possible. To estimate the overall non-carcinogenic potential for health effects, HQs are summed to give a hazard index

(HI).

Cancer Effects: For carcinogens, risk estimates represent the incremental probability that an individual will develop cancer over a lifetime as a result of exposure to a particular carcinogen or set of carcinogens (EPA 1989). The cancer risk is calculated by multiplying the exposure concentration, averaged over a lifetime, by the unit risk factor:

Risk = LDC x URF where:

LDC = Lifetime Daily Concentration (μg/m3)

URF = Unit Risk Factor (μg/m3)-1

To estimate the overall cancer risk, risks from each carcinogen are summed. Note that this is an estimate of the additional cancer risk posed by exposure to constituents in dust from the reservoir; it does not include the much larger probability of cancer from all other causes.

A carcinogenic risk is expressed as a probability, such as one additional cancer in an exposed population of one million (which is also expressed as 1 x 10-6). The lifetime daily intake is the exposure dose averaged over a 70-year lifetime. This is in keeping with the concept that there are no threshold doses for carcinogens. Risks below 1 x 10-6 are considered de minimis. Risks above 1 x 10-6 potentially require attention.

Site Risks: Table 5 presents the calculated hazard indices and cancer risks. The total hazard index was 0.008. This value is less than 1, meaning that non-cancer effects are not expected.

The cancer risk was 9 x 10-8. This value is less than 1 x 10-6, so cancer risks are considered de minimis. These results indicate that there will not be a dust inhalation hazard if the reservoir is drained.

An additional line of evidence is the comparison of the reservoir data to the background samples (Table 2). There was no apparent pattern within the background samples that would reflect the moisture of the sampling location (Table 1). Sometimes the samples taken from moist and seasonably moist locations were most similar (e.g., arsenic, barium, manganese, and nickel), but sometimes the moist and dry locations were the closest two results (e.g., aluminum, beryllium, and cobalt).

Six of the nine metals were detected at their highest concentrations in the reservoir sediments (barium, beryllium, chromium, cobalt, manganese, and nickel). The highest concentrations of aluminum, arsenic, and vanadium were in the background samples. Most of the time, the highest reservoir sediment concentration was higher than the highest background concentration by only a small margin. Only barium was higher than background by a factor of 2; the other metals exceeded background by between 8 and 33 percent. Nickel was the only metal that had a greater concentration near Crystal Reservoir than sediment samples collected in Ash Meadows by Goldstein et al. (2011). These results may be reflective of natural variability; however, to the extent that they show a real increase in concentration, the difference is relatively modest.

All detected metals were included in the risk calculations regardless of whether concentrations in the reservoir sediments were higher than those in the background samples. Aluminum, arsenic, and vanadium were not detected above background levels. If these metals were removed from the risk estimates (to establish a site-related risk), the hazard index and cancer risk from inhalation of dust derived from reservoir sediment would be cut approximately in half:

from 0.008 to 0.004 in the case of the hazard index, and from 9 x 10-8 to 4 x 10-8 in the case of the cancer risk.

Uncertainty Analysis: The risk estimates are subject to uncertainties contained in the calculations. For example, the EPC assumes that samples collected are representative of the reservoir as a whole. If an area where metals have precipitated out was not sampled, this area would not be reflected in the risk estimates. It should be noted that such an area would need to be both extremely large and have much higher concentrations than the reservoir as a whole to have a major effect. For example, if there was a 1-acre area with concentrations 10 times as high as the EPCs, the risk estimates would only be increased by 10 percent.

The analytes for this project were those trace metals with inhalation toxicity values. There are other metals that would be expected to be found in the sediment. However, the metals that were analyzed showed only a modest to moderate, rather than large, increase in concentration versus background. It is likely that other metals would behave similarly, indicating only a limited potential for risks increasing above background.

Exposure parameters are expected to be conservative (health-protective). The assumption that a worker will be exposed 8 hours per day, 250 days per year for 25 years is more reflective of the time that a long-term worker may be at the Refuge than the time that such a person would work near the reservoir. To the extent that the worker is away from the reservoir and does not inhale reservoir-related dust, risk calculations will be overestimates.

The model used to estimate dust assumes there are no rocks or shrubs that would limit dust generation. It also ignores the fact that snow cover will suppress dust for a portion of the year.

In these ways, the dust model is also conservative.

An RfC was not available for arsenic. Consequently, the potential for non-cancer effects from this constituent has not been quantified. However, the arsenic exposure concentration was 1.2 x 10-8 mg/m3 (Table 5). This is over 100 times less than the RfC of any of the other metals. It is unlikely that an RfC for arsenic would be so low (i.e., indicate the potential for a toxic effect at that concentration). In addition, the maximum reservoir concentration is less than that of the background samples, indicating that the reservoir would not significantly increase the arsenic-related risk, regardless of the RfC.

The toxicity of chromium is dependent upon its form; i.e., whether it is in the trivalent (Cr3+) or hexavalent (Cr6+) state. The form was not determined during the analysis. However, in the absence of a source of hexavalent chromium, most chromium is likely present as Cr3+.

Chromium was conservatively assumed to be present as Cr6+.

Reference concentrations incorporate safety factors of 100 to 1,000 when extrapolating from animal data to humans. These safety factors account for the potential for humans to be more sensitive to the chemical than laboratory animals, for humans that are more sensitive to the chemical than the main population, etc. This creates many conditions that can lead to an overestimate of risk; for example, humans can also be less sensitive to the chemical than laboratory animals.

The unit risk factors are upper bound values that fit carcinogenicity data to a specific mathematical function. The function selected is in itself generally conservative with respect to other mathematical functions that fit the data equally well.

In summary, there are many sources of uncertainty that may affect the estimation of risks.

However, assumptions incorporated into the calculations have a high degree of conservatism, and the actual risks are expected to be less than those calculated. Given that the hazard index was less than 1 and the cancer risk less than 1 x 10-6, and that risk calculations have likely overestimated the true risks, no adverse health effects are expected from dust that would result from draining the reservoir.

Once construction sequencing and methods for reservoir removal are determined, the Bureau of Air Pollution Control of the Nevada Department of Environmental Protection should be contacted to determine if permits are required for this project.

References

Cowherd, Chatten Jr., G. Muleski, P. Englehart, and D. Gillette. 1985. Rapid assessment of exposure to particulate emissions from surface contamination sites. EPA/600/8-85/002.

Goldstein, H.L., G.N. Breit, J.C. Yount, R.L. Reynolds, M.C. Reheis, G.L. Skipp, E.M. Fisher, and P.J.

Lamothe. 2011. Physical, chemical, and mineralogical data from surficial deposits, groundwater levels, and water composition in the area of Franklin Lake playa and Ash Meadows, California and Nevada: U.S. Geological Survey Data Series 607, 153 p.

U.S. Environmental Protection Agency (EPA). 1989. Risk Assessment Guidance for Superfund:

Volume 1 - Human Health Evaluation Manual, (Part A), Interim Final. EPA/540/1-89/002, December 1989.

U.S. Environmental Protection Agency (EPA). 1997. Annual Health Effects Assessment Summary Tables (HEAST) FY 1997. Office of Solid Waste and Emergency Response, Washington DC. EPA 540/R-94/036.

U.S. Environmental Protection Agency (EPA). 2002. Supplemental Guidance for Developing Soil Screening Levels for Superfund Sites. OSWER 9355.4-24. December 2002.

U.S. Environmental Protection Agency (EPA). 2003. Human Health Toxicity Values in Superfund Risk Assessments. OSWER Directive No. 9285.7-53, December 5, 2003.

U.S. Environmental Protection Agency (EPA). 2007. ProUCL Version 4.0 Users Guide.

EPA/600/R-07/038. April 2007.

U.S. Environmental Protection Agency (EPA). 2009a. Risk Assessment Guidance for Superfund Volume I: Human Health Evaluation Manual (Part F, Supplemental Guidance for Inhalation Risk Assessment). Final. EPA-540-R-070-002. January.

U.S. Environmental Protection Agency (EPA). 2009b. Regional Screening Levels (RSL) for Chemical Contaminants at Superfund Sites. Table Update. April 2009.

List of Abbreviations

AT: Averaging Time (period over which exposure is averaged – days)

ATSDR: Agency for Toxic Substances and Disease Registry

Ca: Constituent Concentration in Air (mg/m3)

Cs: Constituent Concentration in Soil (mg/kg)

EC: Exposure Concentration (mg/m3)

ED: Exposure Duration (years)

EF: Exposure Frequency (days/year)

EPA: U.S. Environmental Protection Agency

EPC: Exposure Point Concentration

ET: Exposure Time (hrs/day)

HEAST: EPA Health Effects Assessment Summary Tables

HI: Hazard Index

HQ: Hazard Quotient

IRIS: Integrated Risk Information System

LDC: Lifetime Daily Concentration (μg/m3)

NCEA: National Center for Environmental Assessment

PEF: Particulate Emission Factor (kg/m3)

PPRTV: Provisional Peer-Reviewed Toxicity Values

RfC: Reference Concentration (mg/m3)

SF: Slope Factor

UCL: Upper Confidence Limit

URF: Unit Risk Factor (μg/m3)-1

Table 1. Data Summary

BG = background sample location

Table 2. Sediment Data Statistics

Maximum Site Concentration 95% Upper Number Number Minimum Maximum Greater Than Confidence Exposure of of Detected Detected Maximum Average Standard Limit on Point Analyte Detections Samples Concentration Concentration Background? Concentration Deviation Mean Concentration(a)

(mg/kg) (mg/kg) (mg/kg) (mg/kg) (mg/kg) (mg/kg)

Antimony 0 10 NA NA No NA NA NA NA Arsenic 10 10 8.0 19 No 12 2.9 14 14 Barium 10 10 150 540 Yes 331 112 396 396 Beryllium 5 10 0.40 0.70 Yes 0.47 (b) 0.12 (b) 0.66 0.66 Cadmium 0 10 NA NA No NA NA NA NA Chromium 10 10 5.1 9.3 Yes 6.8 1.4 7.6 7.6 Cobalt 10 10 1.9 3.6 Yes 2.7 0.6 3.0 3.0 Lead 0 10 NA NA No NA NA NA NA Manganese 10 10 98 190 Yes 140 30 158 158 Mercury 0 10 NA NA No NA NA NA NA Nickel 10 10 15 21 Yes 19 2.0 20 20 Vanadium 10 10 10 16 No 13 1.8 14 14

Statistics are for samples from the reservoir. Background samples are not included in the statistics.

(a) Exposure point concentration is the 95% upper confidence on the mean, as calculated by ProUCL 4.0 (EPA, 2007). ProUCL outputs are provided in Appendix A.

(b) Mean and standard deviation calculated using half the detection limit for samples where analyte was not detected

NA = not applicable; constituent not detected

Table 3. Exposure Parameters

Parameter Symbol Value Rationale

Exposure Time (hrs/day)

ET

Typical length of a work day (EPA, 2009a)

Exposure Frequency (days/year) EF 250 Assumes a person works 5 days per week for 50 weeks per year (EPA 2002)

Exposure Duration (years) ED 25 95th percentile of job tenure (time spent with a single employer) (EPA 2002)

Averaging Time, non-cancer (days) ATnc 9,125 Number of days in the 25 year exposure duration

Averaging Time, cancer (days) ATc 25,550 Number of days in a 70-year lifetime

Particulate Emission Factor (m3/kg) PEF 259565552 Calculated value; see Appendix B

Table 4. Inhalation Toxicity Values

Chronic

Inhalation RfC

Target Organ/System or Critical Effect for Inhalation

Inhalation

URF

EPA Weight of

Evidence Classification for Inhalation

Constituent (mg/m3) (μg/m3)-1

Aluminum

5.0E-03

P

CNS

NA

Arsenic NA Pulmonary system 4.3E-03 I A

Barium 5.0E-04 H2 Respiratory, fetus (inhalation) NA D

Beryllium 2.0E-05 I Pulmonary system 2.4E-03 I B1

Chromium 1.0E-04 Id Respiratory system, nasal septum 1.2E-02 I A

Cobalt 6.0E-06 P Blood 9.0E-03 P NA

Manganese 5.0E-05 I CNS NA D

Nickel 9.0E-05 M NA NA NA

Vanadium 7.0E-06 P NA 8.3E-03 P NA

A Known human carcinogen

B1 Probable human carcinogen based on limited evidence of carcinogenicity in humans

CNS Central nervous system

D No evidence of carcinogenicity

H2 Health Effects Assessment Summary Tables (HEAST) Table 2: Alternate Methods (EPA 1997)

I Integrated Risk Information System (IRIS)

Id IRIS value for hexavalent chromium in the form of particulates

M Minimal risk level from the Agency for Toxic Substances and Disease Registry (as cited in: EPA 2009b)

NA Not available or not applicable

P Provisional Peer-Reviewed Toxicty value (either directly available or as cited in: EPA 2009b)

RfC Reference concentration

URF Unit risk factor

Table 5. Risk Calculation Results

Total: 0.008 9E-08

NA = Not available or not applicable

Metal

Sediment

Concentration

Chronic Daily Air

Reference

Lifetime Average

Air

Unit Risk

Factor

Hazard Quotient

Cancer Risk

(mg/kg) (mg/m3) (mg/m3) (mg/m3) (μg/m3)-1

Aluminum

9,563

8.4E-06

0.005

3.0E-06

0.0017

Arsenic 14 1.2E-08 NA 4.3E-09 0.0043 NC 1.9E-08

Barium 396 3.5E-07 0.0005 1.2E-07 NA 0.00070 NA

Beryllium 0.66 5.8E-10 0.00002 2.1E-10 0.0024 0.00003 5.0E-10

Chromium 7.6 6.7E-09 0.0001 2.4E-09 0.012 0.00007 2.9E-08

Cobalt 3.0 2.6E-09 0.000006 9.5E-10 0.009 0.00044 8.5E-09

Manganese 158 1.4E-07 0.00005 4.9E-08 NA 0.0028 NA

Nickel 20 1.7E-08 0.00009 6.2E-09 NA 0.00019 NA

Vanadium 14 1.2E-08 0.000007 4.3E-09 0.0083 0.0017 3.6E-08

APPENDIX A

PROUCL OUTPUTS

ProUCL Results

Crystal Reservoir Sediment Data

A B C D E F G H I J K L 1 General UCL Statistics for Data Sets with Non-Detects

2 User Selected Options 3 From File C:\Documents and Settings\Steven Glaser\My Documents\Consulting\Crystal Springs\Data_Pro

4 Full Precision OFF

5 Confidence Coefficient 95%

6 mber of Bootstrap Operations 2000

9 Aluminum

11 General Statistics

12 Number of Valid Samples 10 Number of Unique Samples 9

14 Raw Statistics Log-transformed Statistics

15 Minimum 5200 Minimum of Log Data 8.556

16 Maximum 11000 Maximum of Log Data 9.306

17 Mean 8450 Mean of log Data 9.017

18 Median 8400 SD of log Data 0.242

19 SD 1920

20 Coefficient of Variation 0.227 21 Skewness -0.201 23 Relevant UCL Statistics

24 Normal Distribution Test Lognormal Distribution Test

25 Shapiro Wilk Test Statistic 0.962 Shapiro Wilk Test Statistic 0.947

26 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

27 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

29 Assuming Normal Distribution Assuming Lognormal Distribution

30 95% Student's-t UCL 9563 95% H-UCL 9902

31 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 11289

32 95% Adjusted-CLT UCL 9407 97.5% Chebyshev (MVUE) UCL 12514

33 95% Modified-t UCL 9556 99% Chebyshev (MVUE) UCL 14920

35 Gamma Distribution Test Data Distribution

36 k star (bias corrected) 14.12 Data appear Normal at 5% Significance Level

37 Theta Star 598.6 38 nu star 282.3 39 Approximate Chi Square Value (.05) 244.4 Nonparametric Statistics

40 Adjusted Level of Significance 0.0267 95% CLT UCL 9448

41 Adjusted Chi Square Value 238.3 95% Jackknife UCL 9563

42 95% Standard Bootstrap UCL 9411

43 Anderson-Darling Test Statistic 0.212 95% Bootstrap-t UCL 9550

44 Anderson-Darling 5% Critical Value 0.725 95% Hall's Bootstrap UCL 9423

45 Kolmogorov-Smirnov Test Statistic 0.112 95% Percentile Bootstrap UCL 9380

46 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 9320

47 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 11096

48 97.5% Chebyshev(Mean, Sd) UCL 12241

49 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 14490

50 95% Approximate Gamma UCL 9761 51 95% Adjusted Gamma UCL 10012 53 Potential UCL to Use Use 95% Student's-t UCL 9563

U

56 Arsenic

58 General Statistics

59 Number of Valid Samples 10 Number of Unique Samples 6

61 Raw Statistics Log-transformed Statistics

62 Minimum 8 Minimum of Log Data 2.079

63 Maximum 19 Maximum of Log Data 2.944

64 Mean 11.91 Mean of log Data 2.453

65 Median 12 SD of log Data 0.227

66 SD 2.912

67 Coefficient of Variation 0.245 68 Skewness 1.542 70 Relevant UCL Statistics

71 Normal Distribution Test Lognormal Distribution Test

72 Shapiro Wilk Test Statistic 0.824 Shapiro Wilk Test Statistic 0.892

73 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

74 Data not Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

76 Assuming Normal Distribution Assuming Lognormal Distribution

77 95% Student's-t UCL 13.6 95% H-UCL 13.78

78 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 15.64

79 95% Adjusted-CLT UCL 13.9 97.5% Chebyshev (MVUE) UCL 17.25

80 95% Modified-t UCL 13.67 99% Chebyshev (MVUE) UCL 20.43

82 Gamma Distribution Test Data Distribution

83 k star (bias corrected) 14.69 Data appear Gamma Distributed at 5% Significance Level

84 Theta Star 0.811 85 nu star 293.8 86 Approximate Chi Square Value (.05) 255.1 Nonparametric Statistics

87 Adjusted Level of Significance 0.0267 95% CLT UCL 13.42

88 Adjusted Chi Square Value 248.8 95% Jackknife UCL 13.6

89 95% Standard Bootstrap UCL 13.35

90 Anderson-Darling Test Statistic 0.676 95% Bootstrap-t UCL 14.13

91 Anderson-Darling 5% Critical Value 0.725 95% Hall's Bootstrap UCL 21.86

92 Kolmogorov-Smirnov Test Statistic 0.257 95% Percentile Bootstrap UCL 13.42

93 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 13.7

94 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 15.92

95 97.5% Chebyshev(Mean, Sd) UCL 17.66

96 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 21.07

97 95% Approximate Gamma UCL 13.72 98 95% Adjusted Gamma UCL 14.06 100 Potential UCL to Use Use 95% Approximate Gamma UCL 13.72

103 Barium

105 General Statistics

106 Number of Valid Samples 10 Number of Unique Samples 9

108 Raw Statistics Log-transformed Statistics

109 Minimum 150 Minimum of Log Data 5.011

110 Maximum 540 Maximum of Log Data 6.292

111 Mean 330.5 Mean of log Data 5.745

112 Median 315 SD of log Data 0.36

113 SD 112.5

114 Coefficient of Variation 0.34 115 Skewness 0.471 117 Relevant UCL Statistics

118 Normal Distribution Test Lognormal Distribution Test

119 Shapiro Wilk Test Statistic 0.966 Shapiro Wilk Test Statistic 0.963

120 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

121 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

123 Assuming Normal Distribution Assuming Lognormal Distribution

124 95% Student's-t UCL 395.7 95% H-UCL 426.4

125 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 497.6

126 95% Adjusted-CLT UCL 394.7 97.5% Chebyshev (MVUE) UCL 569.5

127 95% Modified-t UCL 396.6 99% Chebyshev (MVUE) UCL 710.8

129 Gamma Distribution Test Data Distribution

130 k star (bias corrected) 6.511 Data appear Normal at 5% Significance Level

131 Theta Star 50.76 132 nu star 130.2 133 Approximate Chi Square Value (.05) 104.9 Nonparametric Statistics

134 Adjusted Level of Significance 0.0267 95% CLT UCL 389

135 Adjusted Chi Square Value 100.9 95% Jackknife UCL 395.7

136 95% Standard Bootstrap UCL 384.1

137 Anderson-Darling Test Statistic 0.212 95% Bootstrap-t UCL 407.7

138 Anderson-Darling 5% Critical Value 0.726 95% Hall's Bootstrap UCL 431.8

139 Kolmogorov-Smirnov Test Statistic 0.153 95% Percentile Bootstrap UCL 386

140 Kolmogorov-Smirnov 5% Critical Value 0.267 95% BCA Bootstrap UCL 389.5

141 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 485.5

142 97.5% Chebyshev(Mean, Sd) UCL 552.6

143 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 684.3

144 95% Approximate Gamma UCL 410.4 145 95% Adjusted Gamma UCL 426.5 147 Potential UCL to Use Use 95% Student's-t UCL 395.7

150 Beryllium

152 General Statistics

153 Number of Valid Samples 5 Number of Detected Data 5

154 Number of Unique Samples 5 Number of Non-Detect Data 0

155 Number of Missing Values 5 Percent Non-Detects 0.00%

157 Raw Statistics Log-transformed Statistics

158 Minimum Detected 0.4 Minimum Detected -0.916

159 Maximum Detected 0.7 Maximum Detected -0.357

160 Mean of Detected 0.54 Mean of Detected -0.639

161 SD of Detected 0.128 SD of Detected 0.238

162 Minimum Non-Detect N/A Minimum Non-Detect N/A

163 Maximum Non-Detect N/A Maximum Non-Detect N/A

166 UCL Statistics

167 Normal Distribution Test with Detected Values Only Lognormal Distribution Test with Detected Values Only

168 Shapiro Wilk Test Statistic 0.935 Shapiro Wilk Test Statistic 0.941

169 5% Shapiro Wilk Critical Value 0.762 5% Shapiro Wilk Critical Value 0.762

170 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

172 Assuming Normal Distribution Assuming Lognormal Distribution

173 DL/2 Substitution Method DL/2 Substitution Method 174 Mean 0.54 Mean -0.639

175 SD 0.128 SD 0.238

176 95% DL/2 (t) UCL 0.662 95% H-Stat (DL/2) UCL 0.712

178 Maximum Likelihood Estimate(MLE) Method N/A Log ROS Method 179 MLE method failed to converge properly Mean in Log Scale N/A

180 SD in Log Scale N/A

181 Mean in Original Scale N/A

182 SD in Original Scale N/A

183 95% Percentile Bootstrap UCL N/A

184 95% BCA Bootstrap UCL N/A

186 Gamma Distribution Test with Detected Values Only Data Distribution Test with Detected Values Only

187 k star (bias corrected) 9.031 Data appear Normal at 5% Significance Level

188 Theta Star 0.0598 189 nu star 90.31 191 A-D Test Statistic 0.279 Nonparametric Statistics

192 5% A-D Critical Value 0.679 Kaplan-Meier (KM) Method 193 K-S Test Statistic 0.679 Mean 0.54

194 5% K-S Critical Value 0.357 SD 0.115

195 Data appear Gamma Distributed at 5% Significance Level SE of Mean 0.0573

196 95% KM (t) UCL 0.662

197 Assuming Gamma Distribution 95% KM (z) UCL 0.634

198 Gamma ROS Statistics using Extrapolated Data 95% KM (jackknife) UCL 0.662

199 Minimum 0.4 95% KM (bootstrap t) UCL 0.719

200 Maximum 0.7 95% KM (BCA) UCL 0.624

201 Mean 0.54 95% KM (Percentile Bootstrap) UCL 0.624

202 Median 0.52 95% KM (Chebyshev) UCL 0.79

203 SD 0.128 97.5% KM (Chebyshev) UCL 0.898

204 k star 9.031 99% KM (Chebyshev) UCL 1.11

205 Theta star 0.0598 206 Nu star 90.31 Potential UCLs to Use

207 AppChi2 69.4 95% KM (t) UCL 0.662

208 95% Gamma Approximate UCL 0.703 95% KM (Percentile Bootstrap) UCL 0.624

209 95% Adjusted Gamma UCL 0.794 210 Note: DL/2 is not a recommended method.

213 Chromium

215 General Statistics

216 Number of Valid Samples 10 Number of Unique Samples 8

218 Raw Statistics Log-transformed Statistics

219 Minimum 5.1 Minimum of Log Data 1.629

220 Maximum 9.3 Maximum of Log Data 2.23

221 Mean 6.805 Mean of log Data 1.9

222 Median 6.625 SD of log Data 0.2

223 SD 1.374

224 Coefficient of Variation 0.202 225 Skewness 0.47 227 Relevant UCL Statistics

228 Normal Distribution Test Lognormal Distribution Test

229 Shapiro Wilk Test Statistic 0.948 Shapiro Wilk Test Statistic 0.956

230 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

231 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

233 Assuming Normal Distribution Assuming Lognormal Distribution

234 95% Student's-t UCL 7.602 95% H-UCL 7.729

235 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 8.687

236 95% Adjusted-CLT UCL 7.589 97.5% Chebyshev (MVUE) UCL 9.501

237 95% Modified-t UCL 7.612 99% Chebyshev (MVUE) UCL 11.1

239 Gamma Distribution Test Data Distribution

240 k star (bias corrected) 19.53 Data appear Normal at 5% Significance Level

241 Theta Star 0.348 242 nu star 390.6 243 Approximate Chi Square Value (.05) 345.8 Nonparametric Statistics

244 Adjusted Level of Significance 0.0267 95% CLT UCL 7.52

245 Adjusted Chi Square Value 338.5 95% Jackknife UCL 7.602

246 95% Standard Bootstrap UCL 7.488

247 Anderson-Darling Test Statistic 0.237 95% Bootstrap-t UCL 7.726

248 Anderson-Darling 5% Critical Value 0.725 95% Hall's Bootstrap UCL 7.644

249 Kolmogorov-Smirnov Test Statistic 0.132 95% Percentile Bootstrap UCL 7.52

250 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 7.57

251 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 8.699

252 97.5% Chebyshev(Mean, Sd) UCL 9.519

253 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 11.13

254 95% Approximate Gamma UCL 7.687 255 95% Adjusted Gamma UCL 7.853 257 Potential UCL to Use Use 95% Student's-t UCL 7.602

260 Cobalt

262 General Statistics

263 Number of Valid Samples 10 Number of Unique Samples 8

265 Raw Statistics Log-transformed Statistics

266 Minimum 1.9 Minimum of Log Data 0.642

267 Maximum 3.6 Maximum of Log Data 1.281

268 Mean 2.655 Mean of log Data 0.953

269 Median 2.575 SD of log Data 0.23

270 SD 0.611

271 Coefficient of Variation 0.23 272 Skewness 0.279 274 Relevant UCL Statistics

275 Normal Distribution Test Lognormal Distribution Test

276 Shapiro Wilk Test Statistic 0.907 Shapiro Wilk Test Statistic 0.914

277 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

278 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

280 Assuming Normal Distribution Assuming Lognormal Distribution

281 95% Student's-t UCL 3.009 95% H-UCL 3.082

282 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 3.502

283 95% Adjusted-CLT UCL 2.991 97.5% Chebyshev (MVUE) UCL 3.868

284 95% Modified-t UCL 3.012 99% Chebyshev (MVUE) UCL 4.588

286 Gamma Distribution Test Data Distribution

287 k star (bias corrected) 14.84 Data appear Normal at 5% Significance Level

288 Theta Star 0.179 289 nu star 296.8 290 Approximate Chi Square Value (.05) 257.9 Nonparametric Statistics

291 Adjusted Level of Significance 0.0267 95% CLT UCL 2.973

292 Adjusted Chi Square Value 251.6 95% Jackknife UCL 3.009

293 95% Standard Bootstrap UCL 2.965

294 Anderson-Darling Test Statistic 0.467 95% Bootstrap-t UCL 3.083

295 Anderson-Darling 5% Critical Value 0.725 95% Hall's Bootstrap UCL 2.951

296 Kolmogorov-Smirnov Test Statistic 0.216 95% Percentile Bootstrap UCL 2.965

297 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 2.965

298 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 3.498

299 97.5% Chebyshev(Mean, Sd) UCL 3.862

300 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 4.578

301 95% Approximate Gamma UCL 3.056 302 95% Adjusted Gamma UCL 3.132 304 Potential UCL to Use Use 95% Student's-t UCL 3.009

307 Manganese

309 General Statistics

310 Number of Valid Samples 10 Number of Unique Samples 8

312 Raw Statistics Log-transformed Statistics

313 Minimum 98 Minimum of Log Data 4.585

314 Maximum 190 Maximum of Log Data 5.247

315 Mean 139.8 Mean of log Data 4.919

316 Median 135 SD of log Data 0.218

317 SD 30.48

318 Coefficient of Variation 0.218 319 Skewness 0.358 321 Relevant UCL Statistics

322 Normal Distribution Test Lognormal Distribution Test

323 Shapiro Wilk Test Statistic 0.955 Shapiro Wilk Test Statistic 0.964

324 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

325 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

327 Assuming Normal Distribution Assuming Lognormal Distribution

328 95% Student's-t UCL 157.5 95% H-UCL 160.9

329 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 182

330 95% Adjusted-CLT UCL 156.8 97.5% Chebyshev (MVUE) UCL 200.3

331 95% Modified-t UCL 157.7 99% Chebyshev (MVUE) UCL 236.1

333 Gamma Distribution Test Data Distribution

334 k star (bias corrected) 16.56 Data appear Normal at 5% Significance Level

335 Theta Star 8.442 336 nu star 331.2 337 Approximate Chi Square Value (.05) 290 Nonparametric Statistics

338 Adjusted Level of Significance 0.0267 95% CLT UCL 155.7

339 Adjusted Chi Square Value 283.3 95% Jackknife UCL 157.5

340 95% Standard Bootstrap UCL 155.4

341 Anderson-Darling Test Statistic 0.216 95% Bootstrap-t UCL 161.1

342 Anderson-Darling 5% Critical Value 0.725 95% Hall's Bootstrap UCL 159.6

343 Kolmogorov-Smirnov Test Statistic 0.153 95% Percentile Bootstrap UCL 155

344 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 155.8

345 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 181.8

346 97.5% Chebyshev(Mean, Sd) UCL 200

347 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 235.7

348 95% Approximate Gamma UCL 159.6 349 95% Adjusted Gamma UCL 163.4 351 Potential UCL to Use Use 95% Student's-t UCL 157.5

354 Nickel

356 General Statistics

357 Number of Valid Samples 10 Number of Unique Samples 6

359 Raw Statistics Log-transformed Statistics

360 Minimum 15 Minimum of Log Data 2.708

361 Maximum 21 Maximum of Log Data 3.045

362 Mean 18.7 Mean of log Data 2.923

363 Median 19 SD of log Data 0.112

364 SD 2.003

365 Coefficient of Variation 0.107 366 Skewness -0.73 368 Relevant UCL Statistics

369 Normal Distribution Test Lognormal Distribution Test

370 Shapiro Wilk Test Statistic 0.92 Shapiro Wilk Test Statistic 0.903

371 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

372 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

374 Assuming Normal Distribution Assuming Lognormal Distribution

375 95% Student's-t UCL 19.86 95% H-UCL 20.02

376 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 21.58

377 95% Adjusted-CLT UCL 19.59 97.5% Chebyshev (MVUE) UCL 22.83

378 95% Modified-t UCL 19.84 99% Chebyshev (MVUE) UCL 25.27

380 Gamma Distribution Test Data Distribution

381 k star (bias corrected) 64.44 Data appear Normal at 5% Significance Level

382 Theta Star 0.29 383 nu star 1289 384 Approximate Chi Square Value (.05) 1207 Nonparametric Statistics

385 Adjusted Level of Significance 0.0267 95% CLT UCL 19.74

386 Adjusted Chi Square Value 1193 95% Jackknife UCL 19.86

387 95% Standard Bootstrap UCL 19.71

388 Anderson-Darling Test Statistic 0.415 95% Bootstrap-t UCL 19.76

389 Anderson-Darling 5% Critical Value 0.724 95% Hall's Bootstrap UCL 19.62

390 Kolmogorov-Smirnov Test Statistic 0.174 95% Percentile Bootstrap UCL 19.7

391 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 19.5

392 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 21.46

393 97.5% Chebyshev(Mean, Sd) UCL 22.66

394 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 25

395 95% Approximate Gamma UCL 19.98 396 95% Adjusted Gamma UCL 20.21 398 Potential UCL to Use Use 95% Student's-t UCL 19.86

401 Vanadium

403 General Statistics

404 Number of Valid Samples 10 Number of Unique Samples 8

406 Raw Statistics Log-transformed Statistics

407 Minimum 10 Minimum of Log Data 2.303

408 Maximum 16 Maximum of Log Data 2.773

409 Mean 12.75 Mean of log Data 2.537

410 Median 12.25 SD of log Data 0.141

411 SD 1.814

412 Coefficient of Variation 0.142 413 Skewness 0.471 415 Relevant UCL Statistics

416 Normal Distribution Test Lognormal Distribution Test

417 Shapiro Wilk Test Statistic 0.959 Shapiro Wilk Test Statistic 0.97

418 Shapiro Wilk Critical Value 0.842 Shapiro Wilk Critical Value 0.842

419 Data appear Normal at 5% Significance Level Data appear Lognormal at 5% Significance Level

421 Assuming Normal Distribution Assuming Lognormal Distribution

422 95% Student's-t UCL 13.8 95% H-UCL 13.91

423 95% UCLs (Adjusted for Skewness) 95% Chebyshev (MVUE) UCL 15.23

424 95% Adjusted-CLT UCL 13.78 97.5% Chebyshev (MVUE) UCL 16.3

425 95% Modified-t UCL 13.82 99% Chebyshev (MVUE) UCL 18.41

427 Gamma Distribution Test Data Distribution

428 k star (bias corrected) 39.19 Data appear Normal at 5% Significance Level

429 Theta Star 0.325 430 nu star 783.7 431 Approximate Chi Square Value (.05) 719.7 Nonparametric Statistics

432 Adjusted Level of Significance 0.0267 95% CLT UCL 13.69

433 Adjusted Chi Square Value 709.1 95% Jackknife UCL 13.8

434 95% Standard Bootstrap UCL 13.65

435 Anderson-Darling Test Statistic 0.259 95% Bootstrap-t UCL 14.02

436 Anderson-Darling 5% Critical Value 0.724 95% Hall's Bootstrap UCL 14.12

437 Kolmogorov-Smirnov Test Statistic 0.157 95% Percentile Bootstrap UCL 13.65

438 Kolmogorov-Smirnov 5% Critical Value 0.266 95% BCA Bootstrap UCL 13.65

439 Data appear Gamma Distributed at 5% Significance Level 95% Chebyshev(Mean, Sd) UCL 15.25

440 97.5% Chebyshev(Mean, Sd) UCL 16.33

441 Assuming Gamma Distribution 99% Chebyshev(Mean, Sd) UCL 18.46

442 95% Approximate Gamma UCL 13.88 443 95% Adjusted Gamma UCL 14.09 445 Potential UCL to Use Use 95% Student's-t UCL 13.8

APPENDIX B

FUGITIVE DUST MODELING

B-1

Appendix B: Fugitive Dust Modeling

Dust generated by wind was estimated using a model by Cowherd et al. (1985) that is recommended by U.S. Environmental Protection Agency (EPA), most recently by the Soil Screening Guidance Supplement (EPA, 2002). This model estimates the airborne particulate concentration from a site using the following equation:

PEF = Q/C x 3,600 sec/hr

0.036 x (1 - V) x (Um / Ut ) x F(x) where:

PEF = Particulate emission factor (m3/kg; see above)

V = Fraction of vegetative cover (unitless)

Um = Mean annual wind speed (m/s)

Ut = Equivalent threshold value of wind speed at 7 meters above ground surface (m/s)

F(x) = Function dependent upon Um and Ut

Q/C = Emission flux per unit concentration in air (g/m2-s per kg/m3)

B-2

The parameter Q/C is area dependent. It was calculated by the following equation:

where:

Q/C = Emission flux per unit concentration in air (g/m2-s per kg/m3)

A, B, C = Constants based on air dispersion modeling for climate-specific zones

Asite = Area of contamination (acres)

The concentration of a chemical in air is derived by dividing the sediment concentration

(mg/kg) by the PEF.

B-2

Parameter values for calculating the PEF are summarized in Table B-1. It was conservatively assumed that there was no vegetative cover. Values of Um, Ut, and F(x) were EPA default values (EPA, 2002). It should be noted that the default value of the mean annual wind speed (Um) of 4.69 m/s is significantly higher than the 2.5 m/s that has been recorded at a nearby wind station in Pahrump, Nevada (http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=KNVPAHRU5 &day=15&year=2009&month=12&graphspan=year). It is also significantly higher than any monthly average wind speed (mph) recorded at the Rogers Spring, AHME, weather station (http://www.wrcc.dri.edu/weather/nrog.html). This is therefore also a conservative value estimate, as higher wind speeds generate more dust.

The parameter Q/C relates the dust emission rate (which is calculated by using everything in the equation except for Q/C) to the resulting dust concentration in air. This parameter is dependent on the size of the dust source and on the meteorology of the area. The reservoir was estimated to be 115 acres in size. The meteorological fitting parameters A, B, and C were taken for Las Vegas.

Appendix B References

Cowherd, Chatten Jr., G. Muleski, P. Englehart, and D. Gillette. 1985. Rapid assessment of exposure to particulate emissions from surface contamination sites. EPA/600/8-85/002.

U.S. Environmental Protection Agency (EPA). 2002. Supplemental guidance for developing soil screening levels for Superfund sites. OSWER 9355.4-24. December 2002.

TABLE B-1.

PARAMETER VALUES FOR CALCULATION OF THE PARTICULATE EMISSION FACTOR FOR WINDBLOWN DUST

Parameter Units Value Rationale

Fraction of vegetative cover (V) unitless

Conservative assumption

Mean annual windspeed (Um) m/s 4.69 Default value (EPA 2002)

Equivalent threshold value of windspeed at 7 m (Ut) m/s 11.32 Default value (EPA 2002)

Function dependent on Um/Ut (F(x)) unitless 0.194 Default value (EPA 2002)

Site Area (Asite) acres 115 Site-specific

Constant A not specified 13.3093 Value for Las Vegas (EPA 2002) 2002) Constant B not specified 19.8387 Value for Las Vegas (EPA 2002) Constant C not specified 230.1652 Value for Las Vegas (EPA 2002)

Emission flux per unit concentration in air (Q/C) g/m2-s per kg/m3

35.8 Calculated value

Particulate emission factor (PEF) m3/kg 2.60E+08 Calculated value

Cancer Effects: Studies of carcinogenicity tend to focus on identifying the slope of the linear portion of a curve of concentration versus response. A plausible upper-bound value of the slope is called the cancer slope factor (SF). The product of t...