Nez Perce Air Toxics Data Analysis

Mike McCarthy, Theresa O’Brien, Jordan Stone, Hilary HafnerSonoma Technology, Inc.

Petaluma, CA

Nez Perce Tribe’s Environmental Restoration and Waste Management Division (ERWM), Air Quality Program

Lapwai, ID

June 26, 2009

909027-3646

2

Outline

• Introduction• Study Objectives• Starting Conceptual Model• Analyses

– Spatial– Temporal– Comparative

• Conclusions and Recommendations

3

Introduction to STI

STI 25th anniversary in 2007

STI data analysis division

4

Study Objectives

• Characterize air toxics concentrations in Lewiston and nearby Nez Perce Reservation– Assess spatial pattern and gradient in toxics

concentrations (especially around the pulp and paper mill)

– Assess the mill’s contribution to air toxics concentrations in the valley

– Develop a picture of spatial patterns of air toxics concentrations

– Predict concentrations in areas without monitors• Dispersion modeling (sub-objectives not listed)

• Characterize air toxics risk – Focus on the risks from pulp and paper mill

Note: Objectives in red were not a focus of STI’s analysis.

5

Starting Conceptual Model

• We can identify the pulp and paper mill influence on air toxics concentrations through a combination of:– Spatial analysis–a clear spatial gradient in air toxics

concentrations– Temporal analysis–mill emissions will result in higher

concentrations in winter (lower mixing heights and wind speeds).– Chemical analysis–distinct “fingerprints” of air toxics will reveal

emissions sources (e.g., the mill vs. motor vehicles)• Concentrations of most toxics will be low compared to

other monitored areas in the United States (low population).

• A large fraction of toxics risk will originate from background air.

6

Nez Perce County Demographic Comparison

Expect lower concentrations than in other U.S. data.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

100 1,000 10,000 100,000 1,000,000 10,000,000

Population

Frac

tion

of c

ount

ies

US countiesCounties with metals measurementsCounties with VOC measurementsNez Perce County

Median county population The subsets of counties with metals or VOC measurements have median populations that are at the upper end of the distribution compared to all US counties.

37786

6th percentile

26th percentile

7

Chemical Fingerprints

Emissions from different sources can have identifying signatures of chemicals in characteristic ratios.

• Motor vehicles have characteristic emissions of hydrocarbons such as benzene, 1,3-butadiene, toluene, and xylenes.

• The Clearwater pulp and paper mill emits acetaldehyde, formaldehyde, lead, manganese, and chlorinated hydrocarbon byproducts.

• Other sources may also be important. Using fingerprints with spatial gradients and temporal patterns may provide consensus for identifying emissions sources for some pollutants.

8

Tetrachloroethylene and Formaldehyde Collocated Comparisons

Collocated data should have a slope of 1.0 +/- 0.1 and R2 should be >= 0.90.These data meet our expectations for data quality.

y = 1.09x - 0.00R2 = 0.93

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 0.05 0.1 0.15 0.2 0.25 0.3ITD 1 (ppb)

ITD

2 (p

pb)

Tetrachloroethylene

y = 1.00x - 0.14R2 = 0.91

0

5

10

15

20

25

0 5 10 15 20 25ITD 1 (ppb)

ITD

2 (p

pb)

Formaldehyde

9

Collocated Benzene Data Comparison

y = 0.41x + 0.33R2 = 0.17

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

ITD 1 (ppb)

ITD

2 (p

pb)

Benzene

Misidentified peak?Co-eluting interferent?

10

How Reliable is the Hydrocarbon Data?

Since benzene and 1,3-butadiene are not reliably and accurately measured at the same location, our confidence in the accuracy of annual average concentrations used for spatial analysis and risk screening is reduced for these compounds.

y = 0.83xR2 = 0.51

0

0.5

1

1.5

2

2.5

0 0.5 1 1.5 2 2.5ITD1 (ppb)

ITD

2 (p

pb)

Tolueney = 0.66x + 0.02

R2 = 0.40

0

0.05

0.1

0.15

0.2

0.25

0.3

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

ITD1 (ppb)

ITD

2 (p

pb)

1,3-Butadiene

11

Data Quality• Data quality for formaldehyde, acetaldehyde,

and chlorinated hydrocarbons is within acceptable parameters.

• Many metals are below MDLs, but this is typical for these trace species.

• Data for hydrocarbons, especially benzene and 1,3-butadiene, are very imprecise and may also have systematic accuracy problems. – Interpret results for these pollutants with a large

degree of uncertainty.

12

Wind Roses

Hatwai wind rose for toxics sampling days.Wind speeds and frequencies are shown in the direction from which they originate.

Winds are from the east-northeast about 14% of the time. In the 45 degree arc, it is 27% of the time.

13

24-hr Wind Roses Map

Winds are aligned along valleys. Topography is channeling the winds up and down the river valleys.

Most winds are low (<10 mph).

14

Daytime Wind Roses (9 a.m. to 5 p.m.)

Daytime winds are predominantly from the west along the Snake River and from the north in Lapwai.

Wind speeds are more likely to be high in the daytime.

15

Nighttime Wind Roses (9 p.m. to 5 a.m.)

Nighttime winds are from the east (south for Lapwai).

Wind speeds are low.

Overall, winds flow upslope during the day and downslope during the night. This is classic drainage flow.

16

Mixing Heights and Seasonal Analysis

Taken from C. Ramos et al., Modeling of Air Toxics from Urban and Industrial Sources within Complex Terrain, Final Report, 2009

Higher mixing heights lead to increased dilution of emissions from local sources. Conversely, lower mixing heights reduce dilution and enhance concentrations.

Therefore, summer concentrations of pollutants should be lower if emissions throughout the year are invariant.

Formaldehyde and acetaldehyde have much higher summer concentrations.

17

Implications of Meteorology

Assuming emissions from the mill and/or Lewiston area• Pollutants with higher nighttime concentrations are

likely to be higher at the Western sites. Benzene, toluene, and xylenes would fit this model.

• Pollutants with higher daytime concentrations might be higher at the Eastern sites. Formaldehyde, acetaldehyde, and ozone would fit this model.

• Higher wind speeds reduce concentrations (they blow away the pollution). If emissions from the mill are the same over a 24-hr period, we expect sites to the east of the mill to be impacted the most because of the lower wind speeds and mixing heights.

18

Chemical Fingerprints

The following maps show which pollutants are high or low at each of the monitoring sites over the entire year, comparing each site to the overall average. • Bars going down indicate lower than average

concentrations of a pollutant. • Bars going up indicate higher than average

concentrations of a pollutant.

19

Hydrocarbons

The spatial gradient in concentrations goes from west (highest) to east (lowest) for most hydrocarbons.

20

Chlorinated HydrocarbonsNo clear pattern. Some pollutants are high at individual sites.

21

Metals

High concentrations at the Idaho Transportation Department and downtown Lewiston sites.

22

Fingerprints Implications

• Hydrocarbons were highest at Lewiston State Office Building. Motor vehicles, trucks, boats, and the port may be likely sources.

• Chlorinated hydrocarbons were variable.– Dichlorobenzenes and dichloromethane were highest at LSOB,

possibly solvent use?– Chloroform, tetrachloroethylene, and trichloroethylene were

highest at ITD, likely from the mill as chlorination biproducts.– Chlorobenzene was high at Lapwai. Unknown local source?

• Most metals were high at ITD and some were high at LSOB. Sources could include the port, brake linings, road dust, and the mill.

23

National Comparison

• Compare Lewiston area concentrations to national averages at all monitoring sites (mostly urban) from 2003 through 2005.

• Selected pollutants only. – See also

• http://www.epa.gov/ttnamti1/files/ambient/airtox/2007-workshop/03_100407_hafner_mccarthy.pdf

• McCarthy M.C., O’Brien T.E., Charrier J.G., and Hafner H.R. (2009) Characterization of the chronic risk and hazard of hazardous air pollutants in the United States using ambient monitoring data. Environmental Health Perspectives (in press). Available on the Internet at <http://www.ehponline.org/docs/2009/11861/abstract.html>.

24

25

Hatwai:

>75th: Acetaldehyde, formaldehyde, chromium VI (est’d).>50th: Benzene, toluene

Remember, Nez Perce County has much lower population than most places being monitored.

Note: Est’d Chromium VI is 34% of total chromium

26

Lapwai:

>95th: Chlorobenzene>75th: Acetaldehyde, formaldehyde, benzene, chromium VI (est’d)>50th: 1,4-dichlorobenzene

27

ITD:

>75th: Acetaldehyde, formaldehyde, benzene, chromium VI (est’d), chloroform>50th: Toluene, tetrachloroethylene,1,4-dichlorobenzene

We suspect the high chloroform and tetrachloroethylene levels are a result of emissions from the Mill.

28

LSOB: High 1,3-butadiene

Sunset: Not highest for any pollutant.

All have high formaldehyde, acetaldehyde, benzene, and estimated chromium VI.

29

National Comparison Implications

Concentrations of most pollutants are higher than expected based on population comparisons.• Unfavorable topography may be a major

reason for high concentrations. The river valley may trap emissions locally, especially under low-wind nighttime conditions.

• Pollutants with the highest relative concentrations include formaldehyde, acetaldehyde, chromium VI, benzene, and toluene. Other pollutants are high at certain sites.

30

Regional Comparison

Downloaded air toxics data from eastern Oregon, eastern Washington, and Boise sites from 2006 and 2007. Goals

• Assess whether the Lewiston sites have higher concentrations than other sites in the inland Northwest.

• Determine if concentrations arise from transport or from local emissions.

31

Box Notch Whisker Plots

HATWAI 1

HATWAI 2ITD 1

ITD 2

LAPWAILSOB

SUNSET

SITEID

0.00

0.05

0.10

0.15

0.20

0.25

CH

LOR

OFO

RM

• Display distribution of concentrations monitored at each site.

• Boxes show 25th and 75th

percentile points in the distribution.

• Notches show median concentration at a site.

• Whiskers indicate 1.5 times interquartile range.

• X’s and O’s indicate outliers in the distribution.

We use these plots to show statistically significant differences between sites. In this case, Lapwai and Hatwai are significantly lower than ITD, LSOB and Sunset

32

Regional Comparisons: Benzene

B1_IDB2_ID

M1_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

LG_OR

SITEID

0.0

0.5

1.0

1.5

2.0

BE

NZE

NE

(ppb

v )

B1_ID = E. Linden St. Boise, ID LG_OR = La Grande, ORB2_ID = Cabarton Ln. Boise, IDM1_ID = S. Worth Way, Meridian, ID

Lewiston sites are not significantly different from Boise.

La Grande and Meridian are significantly lower.

Ada County (Boise) has a 2008 population of >380,000.

33

Regional Comparisons: Tetrachloroethylene

B1_IDB2_ID

M1_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

LG_OR

SITEID

0

5.00E-2

1.00E-1

1.50E-1

2.00E-1

TETR

AC

HLO

RO

ETH

YLE

NE

( ppb

v)

B1_ID = E. Linden St. Boise, IDB2_ID = Cabarton Ln. Boise, ID LG_OR = La Grande, ORM1_ID = S. Worth Way, Meridian, ID

Measurements of tetrachloroethyleneare not sensitive enough to allow comparisons to other regional sites.

Other sites’are too high! This is also true for 1,3-butadiene, chloroform, carbon tetrachloride, styrene, dichlorobenzene, chlorobenzene

MDLs

34

Regional Comparisons: Formaldehyde

B1_IDB2_ID

M1_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

LG_OR

SITEID

0

5

10

15

20

FOR

MA

LDE

HY

DE

(ug/

m3) Median

concentrations at these sites are statistically indistinguishable.

However, Lewiston has much higher episodic concentrations of formaldehyde than other areas.

B1_ID = E. Linden St. Boise, IDB2_ID = Cabarton Ln. Boise, ID LG_OR = La Grande, ORM1_ID = S. Worth Way, Meridian, ID

35

Regional Comparisons: Acetaldehyde

B1_IDB2_ID

M1_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

LG_OR

SITEID

0123456789

10

AC

ETA

L DE

HY

DE

( ug/

m3) Median

concentrations at these sites are statistically indistinguishable.

Again, Lewiston has higher episodic concentrations.

B1_ID = E. Linden St. Boise, IDB2_ID = Cabarton Ln. Boise, ID LG_OR = La Grande, ORM1_ID = S. Worth Way, Meridian, ID

36

Regional Comparisons: Chromium

M2_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

SP_WALG_OR

ST_OR

SITEID

0.000

0.001

0.002

0.003

CH

RO

MIU

M (u

g /m

3 )

LG_OR = La Grande, ORST_OR = Starkey, ORM2_ID = S. Eagle Rd., Meridian, IDSP_WA = Spokane, WA

Meridian, Spokane, and Starkey show median concentrations of 0 (reporting issues?). La Grande, Oregon, shows concentrations that are indistinguishable from those in Lewiston.

Outlier concentrations at all sites cover the same range.

37

Regional Comparisons: Lead

M2_ID

LAPWAI

HATWAI

SUNSET ITDLSOB

SP_WALG_OR

ST_OR

SITEID

0.000

0.005

0.010

0.015

LEA

D (u

g/m

3)

LG_OR = La Grande, ORST_OR = Starkey, ORM2_ID = S. Eagle Rd., Meridian, IDSP_WA = Spokane, WA

Lead concentrations significantly higher at LSOB and Spokane sites. Rural Oregon is significantly lower than all other sites.

High concentrations of lead are likely from local emissions sources. Leaded fuel from river vessels or airplanes and local incinerators may be potential contributors.

38

Regional Comparison Implications

• High concentrations are likely locally derived and do not appear to be transported in.

• More than half of the concentrations for key risk drivers such as formaldehyde, acetaldehyde, and potentially benzene and 1,3-butadiene, are likely to be locally derived.

39

Other Information

• Health risk data and key pollutants– ATSDR and IDOH report

• Seasonal patterns can help indicate the source of emissions– Use time series plots from analytical

laboratory to examine seasonal variability.

ATSDR = Agency for Toxic Substances and Disease RegistryIDOH = Idaho Department of Public health

40

Key Pollutants from Risk Screening

• Comparison values based on chronic multiyear exposures in order of importance– Formaldehyde – Benzene– Chromium (est’d hexavalent values)– 1,3-Butadiene– Carbon tetrachloride– Acetaldehyde– Chloroform– 1,4-Dichlorobenzene

Pollutants listed in blue are emitted by the mill, according to the Toxics Release Inventory (TRI)

Note: Using screening values from other agencies or using more conservative assumptions may cause further pollutants to be added to the bottom of this list (e.g., arsenic, tetrachloroethylene).

41

Formaldehyde Time Series

0

5

10

15

20

25

01-May-06 01-Jul-06 31-Aug-06 31-Oct-06 31-Dec-06 02-Mar-07 02-May-07

conc

entr

atio

n (p

pb)

HATWAI 1 ITD average LAPWAI LSOB SUNSET 5-site average

42

Possible Sources of Formaldehyde and Acetaldehyde in Lewiston

• Local emissions sources– Boats and pleasure craft– Fires– Motor vehicles (unlikely)– Mill (?)

• Indirect sources (formed in the atmosphere or regional)– Any of the above– Biogenic (higher in Lewiston than La Grande?)– Transport (unlikely)

Of these sources, only those that have high activity in the summer are likely contributors.

43

Benzene Time Series

0

0.5

1

1.5

2

2.5

5/1/2006 7/1/2006 8/31/2006 10/31/2006 12/31/2006 3/2/2007 5/2/2007

Con

cent

ratio

n (p

pb)

HATWAI 1ITD 1ITD 2LAPWAILSOBSUNSET5-site average

44

0

369

121518

21242730

3336

Formald

ehyd

e

Chromium

Benze

ne1,3

-Buta

diene

Carbon

tet

Acetal

dehy

deChlo

roform

1,4-di

chlor

oben

zene

Tetrac

hloroe

thene

Chr

onic

can

cer r

isk

(per

mill

ion) 30-year chronic cancer risk screening

values from EPA Region 3 and highest annual mean concentrations from any site in the Lewiston study were used to generate estimated risk numbers.

Chronic Inhalation Cancer Risk Comparison

45

Risk Picture

• Formaldehyde – largest cancer risk – Significant risk from “background” (>20%)– Large local and seasonal component

• Summer source of formaldehyde is larger than other nearby areas.Spatial and temporal pattern of concentrations is inconclusive

– Acetaldehyde – same diagnosis as formaldehyde but a smaller contributor to risk

• Mobile sources – Benzene and 1,3-butadiene• Chromium – Dust?• Mill contributions – Almost certainly chloroform and

tetrachloroethylene. Potentially contributes to formaldehyde, acetaldehyde, and the metals, but these data are not convincing or conclusive.

46

Key Conclusions

• Concentrations of formaldehyde and acetaldehyde are much higher than expected. Formaldehyde is the biggest contributor to estimated cancer risk and acetaldehyde is a significant contributor.

• Concentrations of benzene and 1,3-butadiene are also higher than expected, although the monitoring data may not be reliable.

• The mill is likely responsible for local concentrations of chloroform and tetrachloroethylene. The mill may also contribute to high concentrations of other pollutants (e.g., formaldehyde), but the available data is not conclusive.

47

Objectives – Spatial Gradients

• Hydrocarbons were highest in the west and decreased as one went further east.– Benzene, 1,3-butadiene, formaldehyde, acetaldehyde, and

other BTEX species– Note that the collocated analysis indicates the

concentrations are unreliable for the benzene and 1,3-butadiene species.

• Carbon tetrachloride was invariant, as expected for a background species.

• Chloroform, tetrachloroethylene, and trichloroethylene were highest at the ITD site and are likely from the mill.

• 1,4-Dichlorobenzene was highest at the Lewiston site.

48

Contributions of Mill to Key Air Toxics Concentrations (1 of 2)

• Chloroform, tetrachloroethylene, and trichloroethylene (high certainty, >50% contributor)– Concentrations were highest near the mill at the ITD site and were lower

farther away.– These pollutants are emitted from the mill (chlorine bleaching process).– The sites with lowest concentrations were at or below the 25th percentile

nationally, consistent with expected concentrations for a community of this size.

• Formaldehyde and acetaldehyde (low certainty, potentially large contributor)– These pollutants are emitted from the mill as fugitive emissions.– Concentrations were highest at LSOB and Sunset, not ITD, and there

was almost no gradient.– Average concentrations are dominated by very high episode days in the

summer. Could be the mill (although seasonality is not consistent) or some local summer source.

49

Contributions of Mill to Key Air Toxics Concentrations (2 of 2)

• Benzene and 1,3-butadiene (no contribution, low certainty)– Spatial gradient is not consistent with emissions from the mill.– Mill doesn’t emit these pollutants in large quantities.

• Carbon tetrachloride (no contribution, high certainty)– Carbon tetrachloride is a background pollutant in the

atmosphere. It has the same concentration in Lewiston as everywhere else in North America.

• Chromium VI and other metals (potential contribution, low certainty)– Metals concentrations are high at the ITD site.– Mill emits manganese and lead in small quantities.– Could also be road dust (high manganese, aluminum, and iron

may indicate a soil signature).

50

Recommendations

• Examine QC data from laboratory to quantify accuracy and precision of all measurements for each species.

• Monitoring– Can analytical issue with benzene and 1,3-butadiene be

resolved without remonitoring? Reassess chromatographs from GC-MS to determine source of imprecision.

– Set up one monitoring site in Lewiston area for long-term toxics measurements.

– Consider a special study with time-resolved measurements to help identify sources of formaldehyde and acetaldehyde during the peak summer season.

• Multiple 3-hr canisters/cartridges during peak season like PAMS network measurements

• Aethalometer• CO monitor• Auto-GC