Assessing Transboundary Influences in the Lower Rio
Grande Valley
99-197
Shaibal Mukerjee
U.S. Environmental Protection Agency, Research Triangle Park, North Carolina
Douglas S. Shadwick
Mantech Environmental Technology, Inc., Research Triangle Park, North Carolina
Kirk E. Dean
Texas Natural Resource Conservation Commission, Austin, Texas
Linda Y. Carmiehael
QST Environmental, Gainesville, Florida
ABSTRACT
The Lower Rio Grande Valley Transboundary Air Pollution Project (TAPP) was a U.S.-Mexico
Border XXI Program project to assess transboundary air pollution in and near Brownsville,
Texas. The study used a three-site air monitoring network very close to the border to capture the
direct impact of local sources and transboundary transport. Ambient data included particulate
mass and elemental composition, VOCs, PAHs, pesticides, and meteorology. Also, near real-
time, PM2 5 mass measurements captured potential pollutant plume events occurring over 1-h
periods. Data collected were compared to screening levels and other monitoring data to assess
general air pollution impacts on nearby border communities. Wind sector analyses, chemical
tracer analyses, principal component analyses, and other techniques were used to assess the extent
of transboundary transport of air pollutants and identify possible transboundary air pollution
sources.
Overall, ambient levels were comparable to or lower than other urban and rural areas in Texas and
elsewhere. Movement of air pollution across the border did not appear to cause noticeable
deterioration of air quality on the U.S. side of the Lower Rio Grande Valley. Dominant
southeasterly winds from the Gulf of Mexico were largely responsible for the clean air conditions
in the Brownsville airshed. Few observations of pollutants exceeded effects screening levels,
almost all being VOCs; these appeared to be due to local events and immediate influences, not
regional phenomena or persistent transboundary plumes.
INTRODUCTION
The Lower Rio Grande Valley Transboundary Air Pollution Project (TAPP) was a U.S.-Mexico
Border XXI Program project to assess transboundary air pollution conditions in and near the
border city of Brownsville, Texas. The TAPP was developed by the U.S. EPA, in cooperation
with the Texas Natural Resource Conservation Commission (TNRCC), as a follow-up
1
-------
investigation to air pollution findings from a previous, multimedia monitoring effort in 1993
known as the Lower Rio Grande Valley Environmental Scoping Study (LRGVESS)1'2. Both the
LRGVESS and TAPP were initiated in response to increased concerns about a possible increase
in air pollution from industrial and waste emissions in Mexico as a result of the North American
Free Trade Agreement (NAFTA). Although the LRGVESS was of short duration, it was
determined that more information was needed concerning overall contact with pollutants from
cross-border transport and pesticide applications in the Valley. With this perspective, the TAPP
study objectives were to: 1) characterize air quality within the Valley and establish a baseline of
air quality data for future reference, 2) characterize transport of air pollutants across the U.S.-
Mexican border, and 3) define what, if any, additional studies might be needed.
Ambient air monitoring in the TAPP was conducted in the immediate vicinity of the border for a
full year. These measurements were compared with similar pollutant species collected in urban
and rural areas in Texas and elsewhere. Wind sector analyses, principal component analyses
(PCA), and non-parametric statistical methods were also performed to assess the extent of
transboundary transport of air pollutants and to identify possible transboundary air pollution
sources.
MATERIALS & METHODS
In the TAPP, a three-site air monitoring network was established very close to the border to
capture the direct impact of local sources and transboundary transport. Figure 1 shows their
locations; all three sites were approximately one km from the Rio Grande River, which forms the
boundary between Texas, U.S.A. and Tamaulipas, Mexico. Major sources close to the sites are
noted in Figure 1. Air monitoring locations were selected by U.S. EPA personnel, with input
from TNRCC and local community leaders. Site 1 was adjacent to downtown Brownsville and
just southeast of the Gateway International Bridge, the main traffic route between Brownsville
and Matamoros, Mexico. It was located near automotive, diesel truck, and industrial emissions.
It was also near the central site where air monitoring was done during the LRGVESS1. Site 2
was approximately 5 km (3 miles) northwest of Site 1 and was ideally suited to measure impacts
of emissions from Brownsville to the east and southeast, emissions from Matamoros to the south
and southeast, and agricultural activities to the north. It was presumed that anthropogenic
transboundary emissions could be assessed at this site due to its location, which was usually
downwind from those sources under predominant wind flows (from the southeast). Finally, Site 3
was approximately 35 km (22 miles) northwest of Site 2 Site 3 was in a rural community, near
agricultural fields and relatively free from local industrial- emissions. The new Free Trade Bridge
at Los Indios, Texas was approximately 2-3 km (1-2 miles) southeast This site was selected
primarily to provide information about air pollution from agricultural activities.
Air quality and meteorological data similar to those collected in the LRGVESS were acquired at
each site for a full year (March 1996-March 1997). Ambient data collection included fine
particulate mass (PM2 5) and elemental composition, particulate carbon, coarse particulate mass
(PM2 S.10) and elemental composition, volatile organic compounds (VOCs), polycyciic aromatic
hydrocarbons (PAHs), pesticides, and meteorology. With the exception of VOCs (which were
collected every sixth day at Sites 1 and 2 with 6-L evacuated canisters), most inorganic PM and
semivolatile measurements were performed with an experimental air monitoring device known as
2
-------
the dual fine particle sequential sampler (DFPSS, Model URG-2000-01K) (URG Corp., Chapel
Hill, NC). Details of the DFPSS used in this study are discussed elsewhere3. Dichotomous
samplers collected PM2S (analyzed for elemental and volatile carbon species), and PM2 5_10 every
third day at the three sites. Near real-time, PM25 mass measurements were conducted using a
tapered element oscillating microbalance (TEOM® Series 1400a) (Rupprecht & Patashnick Co.,
Inc., Albany, NY) with a cyclone inlet3. TEOMs were at each site to capture potential pollutant
plume events occurring over 1-h periods. Finally 10-m meteorological towers were used at the
three sites to continuously record wind speed, direction, temperature and relative humidity. Siting
requirements outlined in the U.S. EPA Ambient Air Quality Surveillance regulations were
followed.
OVERALL RESULTS
Overall, air pollution levels measured were not unusually high nor of a persistent nature (Table 1).
The majority of air pollutants were similar or lower than data from samples collected in other
areas. Air pollutants having higher concentrations than regional background levels were primarily
VOCs from automobile and gasoline evaporative emissions.
Interpretation of the results in Table 1 is based upon how the chemical information analyzed in
this study compared with information on data collected in other studies. As done in the
LRGVESS, the TAPP results were compared with Effects Screening Levels (ESLs)2 developed
by the TNRCC4. ESLs are based on health effects data, odor nuisance potential, vegetation
effects, or corrosion effects. ESLs are used for screening purposes and are not regulatory
standards4. As in the LRGVESS, the vast majority of air pollutants measured during the TAPP
did not exceed the ESL; hence, adverse effects were not expected. Besides ESLs, other
comparative data from exposure monitoring studies were used and are presented in Table 1. Of
the more than 250 pollutants measured, only seven air pollutants had levels above ESLs, usually
occurring in maximum values (not shown in Table 1), Of the approximately 2600 particle,
semivolatile and VOC samples taken in this study, silver, 2-nitropropane, benzene, methylene
chloride, and vinyl acetate exceeded the ESL only once as maximum values and were not
expected to result in long-term adverse effects. No pesticide or PAH values exceeded ESLs or
comparison values; particle carbon results were also insignificant (Table 1). While acrolein and
methanol concentrations exceeded the ESL more than once, collection of these pollutants using
devices in this study is difficult; caution should be exercised when comparing these findings to
ESLs. If an air pollutant exceeded the ESL, it does not necessarily mean there is a problem, but
rather is an indication that further review is warranted. Further review may include additional
sampling or consideration of ambient levels in the environment. As with all comparison values,
ESLs undergo periodic review and revision to insure that they are based on current scientific
literature4.
TRACER & WIND SECTOR ANALYSES
Based on previous air pollution research, certain air pollutants encountered can be considered
tracers-of-opportunity for specific sources. Total particle (PM2,5+PM2 5_10) chlorine levels were
probably associated with sea salts. A dominant Gulf of Mexico influence was seen in much of the
3
-------
data, both in the TAPP and LRGVESS. Many of the elevated VOCs (benzene, methanol, 2-nitr-
opropane, and methylene chloride) are present in solvents or are emitted into the air by many
sources (for example, benzene is also found in automobile emissions)5. Since measurements of air
emission sources on both sides of the border were not done, it was not possible to identify a
specific source for VOCs (or for the other pollutant species, as well).
The highest levels of silver occurred at Site 1; the highest levels of methylene chloride, and vinyl
acetate were at Site 2. Maximum values for these pollutants came from the southeast direction;
as presented elsewhere3, most of the high hourly PM25 data showed a similar distribution. It is
possible that these maximum levels came from man-made sources in Mexico or more immediate
sources in the U.S.; it is also possible that higher wind velocities from this direction may have
affected air pollutant levels. An additional reason was due to a greater chance of detecting a
random event (i.e., an emission) from the southeast since winds predominate from that direction;
wind sector plots and wind sector-based bar charts bore this out. Daily PM2 5 data (calculated
from the hourly results) were highest from the southeast since approximately 50 percent of daily
winds came from that direction (Figure 2). This wind direction pattern was typical for almost all
of the fine and coarse particle elements and most of the VOC data. In terms of interpreting bar
charts and plots, air pollution data are displayed in a format similar to wind rose plots with the
direction from where the pollutant was coming from being plotted on a 360° circle with 0° (or
360°) indicating North, 90° indicating East, 180° indicating South, and 270° indicating West.
The magnitude of each pollutant concentration is proportional to its distance from the origin of
the plot. Overall average values of the plotted data, shown as lines in the bar charts, are
represented by circles. The dotted lines are the boundaries for the eight wind sectors divided in
the same way as explained for the bar charts.
Many of the detected chemicals could have come from either side of the border. For example, of
the five highest methanol values, three came from the south and two came from the north. The
three high levels from the south could have come from Mexico; the two high levels from the north
could have come from the U.S. One exceedance of the ESL for methylene chloride occurred
from the southeast at Site 2. While transboundary influences occurring at Site 2 from either side
of the border may be initially deduced from these observations, it should be noted that this site
was in close proximity to a propane/butane filling station and this may have influenced some of
the VOC data. The maximum values for benzene (at Site 1) and 2-nitropropane (at Site 2) came
from the north.
PRINCIPAL COMPONENT ANALYSIS (PCA)
PCA was applied to PM2 5 element and VOC data to identify possible local and regional influences
that affect air quality in the Valley. PCA is a factor analysis procedure that has been used as a
screening tool to identify air emission source categories and to select independent emission
tracers. In PCA, the purpose is to group air pollutant tracers according to their common
correlations; groups of high correlation are considered factors or sources. The factors are
uncorrelated with each other and are ordered in terms of the amount of variability they explain.
The first factor represents the maximum amount of variance in the data set with the maximum
amount of remaining variance being incorporated into the second factor, and so on. A large
proportion of the total variance can be explained by the first few principal factors. As is
4
-------
commonly performed in ambient air studies, a transformation technique known as orthogonal
varimax rotation6 was used to aid in interpretation of the factors. Factor loadings represent the
association of each pollutant to a principal factor (or component) and are used to help identify
that component. High factor loadings on a component (greater than 0.5) were used to identify the
component. Both element and VOC species used were tracers-of-opportunity for different
sources in order to prevent biasing the PCA results in favor of a given source6. Since a large data
set is needed in PCA, PM2 5 element data were used in the PCA regardless of analytical
uncertainty; for VOCs, tracer species that were below detection were given a value of half the
detection limit for PCA Factors with eigenvalues greater than one were retained for varimax
rotation.
Table 2 presents the varimax-rotated factor loadings of the fine fraction element species at Site 1.
The first principal factor accounted for 38% of the variance in the data set (proportion of
variation for each factor [prior to varimax rotation] x 100); Factor 1 was dominated by species
that were primarily of crustal origin (Al, Ca, Fe, K, Mn, Si, and Ti with high factor loadings >0.5)
and sea salt origin (CI with a high factor loading > 0.5). Principal component analyses of PM2.5
near the northern coastal city ofBoston have also revealed sources associated with soil dusts and
sea salts7. Previous analyses of fine particles from resuspended road dusts and yard soils in the
LRGVESS confirmed that these elements (in Factor 1) were abundant in crustal/salt sources1.
Factor 2, accounting for 15% of data variance, was dominated by species of possible marine
influences (Br, and S) and/or crustal origin (intermediate K loading). Factor 3 (accounting for
only 9% of variance) had high loadings for Pb and Zn; this may be associated with incineration
emissions8. Factor 4 (accounting for 8% of variance) had high loadings for Ni and V which was
associated with oil burning from boilers or power plants6,7. Factor 5 was an unknown source of
variation (high loading for arsenic [As]; this may have merely been an artifact of the PCA
technique.
Although not shown for brevity, a similar factor solution as seen at Site 1 was also revealed at
Sites 2 and 3, At Site 2, the five factors were interpreted as representing crustal (Factor 1, 24%
of data variance), incineration (Factor 2, 18% data variance), marine/crustal influence (Factor 3,
10% data variance), boiler or electric utility oil combustion (Factor 4, 9% data variance), and
unknown As and Se (Factor 5; again, possible PCA artifact, 7% data variance). At Site 3, the five
factors were interpreted as representing a crustal influence (Factor 1, 27% data variance),
marine/crustal influence (Factor 2, 18% data variance), boiler or electric utility oil burning (Factor
3, 10% data variance), incineration (Factor 4, 9% data variance), and unknown As and Se (Factor
5, possible PCA artifact, 7% data variance).
As seen in Table 3a for Site 1 and Table 3b for Site 2, the dominant source for VOCs was a
transportation-related source, this would include sources as vehicle exhaust, gasoline, and
gasoline evaporation. Dominance of a transportation source category in the Brownsville air shed
for VOCs was expected since this was similar to findings in other VOC investigations in urban
areas Both Sites 1 and 2 were near areas where traffic influences occurred. Biogenic emissions
in Brownsville were a distant second in overall contribution, this was also expected since
agriculture is an important economic factor in the Valley. As shown in Table 3b, some of the
principal factors may have been the result of the nearby filling station and possible industrial
5
-------
sources. As with the PM2 5 results, actual VOC source profiles would be required before
definitive conclusions can be made.
NON-PARAMETRIC STATISTICAL ANALYSES
Two types of non-parametric statistical analyses that relate levels of pollutant tracers with average
wind direction were carried out. The basic premise of the statistical analyses employed were to
associate a pollutant tracer with wind direction and statistically determine which wind sectors
were significantly impacting a receptor. In this way, tracers-of-opportunity associated with an
emission source category can be presumed to influence the monitoring site.
The first analysis was used to calculate the linear-angular correlation between the level of
pollutant tracer and average wind direction. Linear-angular correlation was calculated in
accordance with a statistic proposed by Mardia9; details in application of this and the next
technique to assess local air emission source impacts are discussed elsewhere10. The chi-squared
approximation of the level of significance, designated as Un by Mardia9, was computed to judge
the significance of the computed linear-angular correlation and, thus, accept or reject the null
hypothesis (Hq) that pollutant concentration is independent of wind direction. The text statistic
Un ~ X22 denotes "chi-square distribution with 2 degrees of freedom". The second test, known as
the Kruskal-Wallace test1011, determines whether the median ranked scores for the concentration
of a given tracer are equivalent across separate wind sectors, this being Hq. The Kruskal-Wallace
test is essentially a non-parametric one-way analysis of variance (ANOVA). Overall differences
between wind sector classes in the ANOVA were judged from the Kruskal-Wallace test on
summed ranks of pollutant concentration. The eight wind sector classes in the ANOVA were
determined by average wind direction; like the plots, each wind sector was 45° wide, extending
22 .5° on either side of the eight cardinal wind directions. To interpret levels of significance in the
Kruskal-Wallis statistical analyses, a level of significance of 0.05 was used to indicate that results
departed from the expected behavior under that the distribution of concentrations from the
eight wind sectors are not different.
The techniques were used on the fine particle CI and Si concentration data for each site. These
pollutant tracers were used since tracer analyses and PCA determined sea breezes from the Gulf
and aeolian generated soil dusts were dominant influences in particulate matter measurements.
Table 4 shows linear-angular correlation results for fine particle CI and Si at the three sites.
Examining the p-values (probability{ X22 > Un}), evidence of association with wind direction was
found for CI at all three sites while Si showed an association with wind direction at Sites 1 and 2.
Based on the linear-angular correlations for these pollutants, the Kruskal-Wallis test was applied
to provide an indication of which wind direction may have influenced the correlations. Based on
the mean rank concentrations for CI, the results for Sites 1 and 2 indicated CI impacts due to Gulf
sea breeze influences from the southeast (Table 5). Although association of CI with wind
direction was found at Site 3, impacts were determined to come from the northeast and southwest
(Table 5); however, this conclusion was based on only five observations from the northeast and
one observation from the southwest. As determined from visual examination of wind sector plots
at the three sites presented elsewhere3, this may suggest diminished sea breeze influences further
inland; Site 3 was much further inland than either Sites 1 and 2 (Figure 1). Aeolian dusts from
these strong sea breeze influences from the Gulf were also confirmed in the Si results. Since
6
-------
strong sea breeze influences occurred at Sites 1 and 2, association with wind direction for Si was
found at these sites (Table 4) and these impacts principally came from the southeast in which
winds from the Gulf were expected to come (Table 5); although not shown, this result was
generally the same for other crustal tracers like Al, Ca, Fe, and K.
CONCLUSIONS
Overall, ambient levels of air pollution were comparable to or lower than other urban and rural
areas in Texas and elsewhere. In addition, transboundary transport of air pollution plumes
originating in Mexico did not appear to cause noticeable deterioration of air quality on the U.S.
side of the Lower Rio Grande Valley border. Dominant southeasterly winds from the Gulf of
Mexico were largely responsible for the clean air conditions in the Brownsville airshed. Few
observations of pollutants exceeded effects screening levels, almost all being VOCs; these
appeared to be due to local events and immediate influences, not regional phenomena or
persistent transboundary plumes. Experimental error in the data cannot be ruled out, as well.
Some elevated pollutant levels may have been influenced by transboundary transport of air
pollution from the south and southeast. However, predominant wind flows came from the
southeast and may have resulted in a greater opportunity for pollutants to come from these
directions. Emission sources in the immediate location of the sites could have also influenced
short-term observations. While the TAPP data do not support the immediate need for
monitoring, it nevertheless provides a baseline to assess future trends in air quality as industrial
and population growth continues in the Valley as a result ofNAFTA. As shown in Figure 1,
TNRCC continues to monitor air quality at Site 1 and other sites in the Valley to determine such
trends.
ACKNOWLEDGEMENTS & DISCLAIMER
We thank Mike and Karen Stowe and Ron Moser of Brownsville who performed site operations.
We also thank Jon Bowser (formerly of QST) for initiating the field monitoring program, Larry
Purdue (QST) for help on project report development and the teams of Steve Hern/Joe Behar
(U.S. EPA) and Carl Snow/Janet Pichette (TNRCC) for reviewing the manuscript. The
information in this document has been funded wholly or in part by the United States
Environmental Protection Agency under Contracts 68-D2-0134 to QST Environmental, Inc and
68-D5-0049 to ManTech Environmental, Inc. It has been subjected to agency review and
approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
REFERENCES
1. Special Issue. Environmental Aspects of the Lower Rio Grande Valley; Mukeijee, S. Ed.;
Environ. Int. 1997, 23, 593-744.
2. U.S. Environmental Protection Agency. Lower Rio Grande Valley Environmental
Monitoring Study: Report to the Community on the Pilot Project, U.S. EPA: Research
Triangle Park, NC 1994, pp 3, 13, 25.
7
-------
3. Mukerjee, S., Shadwick, D.S.; Bowser, J.J.; etal. Field Anal. Chem. Technol., submitted.
4. Texas Natural Resource Conservation Commission. Effects Screening Levels (ESLs) List,
available on the Internet at http://www.tnrcc.state.tx.us/exec/chiefeng/tara/, September 5,
1997.
5. Graedel, T.E. Chemical Compounds in the Atmosphere; Academic Press. New York,
1978; p 107.
6. Lioy, P.J.; Kniep, T.J.; Daisey, J.M. Receptor Model Technical Series VI: A Guide to The
Use of Factor Analysis and Multiple Regression (FA/MR) Techniques in Source
Apportionment, U.S. Environmental Protection Agency: Research Triangle Park, NC
1985; EPA-450/4-85-007; pp 1-23, 26, 34.
7. Thurston, G.D.; Spengler, J.D. J. Applied Meteor. 1985, 24, 1245-1256.
8. Olmez, I., Sheffield, A.E.; Gordon, G.E.; et al. JAPCA 1988, 38, 1392-1402.
9. Mardia, K.V. Biometrika 1976, 63, 403-405.
10. Somerville, M.C.; Mukeijee, S.; Fox, D.L.; et al. Atmos. Environ. 1994, 28, 3483-
3493.
11. Hollander, M.; Wolfe, D.A. Nonparametric Statistical Methods; Wiley: New York,
1973; pp 114-119.
12. Ozkaynak, H.; Xue, J.; Weker, R , et al. The Particle TEAM (PTEAM) Study: Analysis of
the Data. Final Report, Volume III. U.S. Environmental Protection Agency: Washington,
DC 1996; EPA/600/R-95/098; p 5-38.
13. Davis, B.L.; Johnson, L R ; Stevens, R.K.; et al. Atmos. Environ. 1984,18, 771-782.
14. Johnson, D.L.; Davis, B.L.; Dzubay, T.G.; et al. Atmos. Environ. 1984, 18, 1539-
1553.
15. Shah, J.J.; Singh, H.B. Environ. Sci. Technol. 1988, 22, 1381-1388.
16. Chuang, J.C.; Mack, G.A.; Kuhlman, M.R. etal Atmos. Environ. 1991, 25B, 369-380.
17. Whitmore, R W ; Immerman, F.W.; Camann, D.E.; et al. Arch. Environ. Contam. Toxicol.
1994, 26, 47-59.
8
-------
Table 1. Selected air pollutant concentrations (regardless of uncertainty for elements; if organics
below detection, assigned Vi detection limit). ESLs and other comparative data shown.
Site 1
Site 2
Site 3
ESL*
(annual)
Other
Comparative
Data
n
Mean
n
Mean
n
Mean
Fine particle data (concentrations in nanograms per cubic meter, ng/m3)
pm25
Silver
Arsenic
Chromium
Iron
Potassium
Lead
Sulfur
cE
Cv
262
256
256
256
256
256
239
256
101
98
10370
3.4
0.8
0.6
83.6
86.6
2.2
844.1
60
360
295
287
287
287
287
287
212
287
99
99
7550
4.9
0.7
0.8
30.5
60.4
2.3
823.3
60
350
299
295
295
295
295
295
212
295
95
95
6720
4.8
0.7
0.7
23.4
59.4
1.9
752.1
50
310
10
10
100
5000
2000
5000
50000'2
2513
l13
324.612
181.212
1500**, 24.212
1555.5'2
142014
5680'4
Coarse particle data (concentrations in nanograms per cubic meter, ng/m3)
PM25.10
Chlorine (as chloride)
102
100
18450
2161.8
106
100
17020
2130.8
100
97
10410
946.5
VOCs (concentrations in parts per billion by volume, ppbV)
1,1,1-
trichloroethane
2-nitropropane
2,2,4-
trimethylpentane
Acrolein
Benzene
Carbon
tetrachloride
Methanol
Methylene
chloride
Toluene
Vinyl acetate
m,p-Xylene
58
NM
58
NM
58
58
NM
58
58
NM
58
0.14
NM
0.08
NM
0.90
0.11
NM
0.05
0.65
NM
0.22
59
59
59
59
59
59
59
59
59
59
59
0.11
0.45
0.17
0.60
0.47
0.10
169.23
1.11
1.20
0.73
1.12
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
NM
200
1.40
75
0.10
1
2
200
7.5
50
4
50
0.91115
2.800"
0.168"
1.619"
7.775"
PAHs, fine fraction (concentrations in ng/m3)
Naphthalene
Phenanthrene
Pyrene
19
19
19
2.1
11.8
1.5
19
19
19
2.2
13.3
1.6
21
21
21
1.9
13.0
2.3
50000
50
50
17016
3116
4.416
Pesticides, fine fraction (concentrations in ng/m3)
Chlorpyrifos
Malathion
Methyl parathion
Trifluralin
19
19
19
19
1.0
1.0
2.1
1.0
19
19
19
19
2.1
1.1
2.5
1.2
21
21
21
21
1.1
1.1
2.1
1.2
200
5000
200
10000
16.717
0.317
n = number of samples measured
NM = not measured;
9
-------
CE= Elemental carbon
Cv = Volatilizable carbon
~ESL (annual) = TNRCC Effects Screening Levels (based on a yearly time frame). Developed by the Texas Natural
Resource Conservation Commission (TNRCC). The TNRCC ESLs are screening levels and are not ambient air
standards. If measured airborne levels of a certain chemical do not exceed the screening level, it is interpreted to
mean that adverse health or welfare effects are not expect©!. If the measured level exceeds the screening level, it does
not necessarily mean there is a health or welfare problem, but rather is an indication that further review is warranted.
Note: mean and median values need to be compared with annual ESLs since they are both based on data for the
full year.
~*U.S. EPA National Ambient Air Quality Standard for Pb is based on a three-month average of 24 h samples.
"Consecutive, 12 h day-night PM25 mean values converted to 24-h mean values from ref. 12.
"Mean PM2S values measured in El Paso, Texas from ref. 13.
I412 h mean PM25 values from Houston, Texas in ref 14.
"Mean outdoor air values in Columbus, OH from ref. 16.
17Data from Jacksonville, Florida during Summer 1986 from ref 17.
10
-------
Table 2. Varimax-rotated factor pattern for fine particulate element data at Site 1. Similar factor
pattern seen at Sites 2 and 3.
Element
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Al
0.95
-0.14
-0.05
0.13
0.05
As
0.02
-0.01
0.04
-0.01
0.91
Br
-0.14
0.86
0.10
-0.10
-0.10
Ca
0.65
0.08
0.18
-0.17
-0.09
CI
0.64
0.09
-0.34
-0.17
0.20
Fe
0.97
-0.12
-0.01
0.11
0.04
K
0.72
0.50
0.06
0.02
-0.06
Mn
0.94
-0.05
0.08
0.17
0.09
Ni
0.05
-0.11
0.05
0.85
-0.13
Pb
0.02
0.12
0.73
0.04
-0.40
S
-0.06
0.80
0.09
0.30
0.03
Si
0.98
-0.11
-0.03
0.09
0.04
Se
0.01
0.42
0.52
-0.11
0.17
V
0.13
0.30
-0.06
0.76
0.13
Ti
0.96
-0.12
-0.08
0.11
0.05
Zn
-0.02
0.02
0.86
0.02
0.16
Eigenvalue
6.09
2.40
1.49
1.34
1.12
Proportion
of variation
0.38
0.15
0.09
0.08
0.07
Factor 1: crustal/sea salt source (high loadings of Al, Ca, CI, Fe, K, Mn, Si, and Ti)
Factor 2: marine/crustal mix source (high loadings of Br and S; intermediate loading of K)
Factor 3: possible incineration source (high loadings of Pb and Zn with unknown Se)
Factor 4: oil combustion (high loadings of Ni and V); could be boiler/utility oil combustion
Factor 5: unknown; possible artifact (high loading of As)
11
-------
Table 3a. Varimax-rotated factor pattern for VQC data at Site 1.
voc
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Factor 6
1,2,4-Trimethylbenzene
0.63
0.08
-0.06
0.48
0.39
0.13
2,3,4-Trimethylpentane
0.76
0.57
-0.01
-0.10
0.12
0.11
2-Methylpentane
0.77
0.61
-0.05
0.005
0.09
0.03
3-Methylhexane
0.52
0.80
-0.05
0.16
0.02
-0.06
3-Methylpentane
0.80
0.52
-0.07
0.17
0.13
0.03
alpha-Pinene
0.09
0.06
-0.06
0.96
-0.01
-0.02
Benzene
0.14
0.06
-0.09
0.004
-0.02
0.97
beta-Pinene
0.03
-0.05
-0.08
-0.01
0.87
-0.06
Cyclohexane
-0.06
0.98
-0.02
-0.06
-0.01
0.02
Heptane
0.11
0.97
-0.03
0.13
-0.01
0.04
Hexane
0.35
0.91
-0.09
0.07
0.06
0.02
Nitropentane
0.74
0.62
-0.04
-0.11
0.02
0.08
Isopentane
0.94
0.17
-0.05
0.01
0.09
-0.02
Isoprene
0.59
-0.13
-0.11
0.11
-0.35
-0.03
m,p-Xylene
0.73
0.50
-0.10
0.19
0.29
0.10
Methylcyclopentane
0.30
0.93
-0.03
0.01
0.02
0.08
Methylenechloride
0.86
-0.05
-0.14
0.17
-0.01
0.20
o-Xylene
0.43
0.18
0.05
0.38
0.49
0.13
Pentane
0.90
0.31
-0.06
0.02
0.03
-0.01
Propylene
-0.07
-0.06
0.83
-0.02
-0.10
-0.05
Toluene
0.71
0.50
-0.02
0.21
0.14
0.28
Acetylene
-0.11
-0.05
0.86
-0.03
0.04
-0.04
Eigenvalue
11.39
2.88
1.48
1.44
1.05
1.01
Proportion of variation
0.52
0.13
0.07
0.07
0.05
0.05
Factor 1: transportation-related sources (vehicle exhaust [high loadings of 1,2,4-
Trimethylbenzene, m,p-Xylene, and Toluene], gasoline [high loadings of 2-Methylpentane, 3-
Methylhexane, 3-Methylpentane, m,p-Xylene, and Toluene], gasoline evaporation [high loadings of
2-Methylpentane, 3-Methylpentane, Isopentane and Pentane]), some possible biogenic emissions
(Isoprene).
Factor 2; transportation-related sources (vehicle exhaust [intermediate loading of m,p-Xylene],
gasoline [high loadings of 2-Methylpentane, 3-Methylhexane, 3-Methylpentane, Cyclohexane,
Hexane, m,p-Xylene, Methylcyclopentane, and intermediate loading of Toluene], gasoline
evaporation [high loadings of 2-Methylpentane, 3-Methylpentane, and Hexane]), some possible
architectural coatings (Heptane)
Factor 3: transportation-related source (vehicle exhaust [high loadings of propylene and acetylene]
Factors 4 and 5: biogenic source (high loadings of alpha- and beta-pinene)
Factor 6: benzene artifact
12
-------
Table 3b. Varimax-rotated factor pattern for VOC da
ta at Site 2.
voc
Factor 1
Factor 2
Factor 3
Factor 4
1,2,4-Trimethylbenzene
0.96
0.25
0.07
-0.03
2,3,4-Trimethylpentane
0.98
-0.007
0.10
-0.02
2-Methylpentane
0.995
-0.03
-0.07
-0.02
3-Methylhexane
0.99
-0.01
-0.10
-0.004
3-Methylpentane
0.99
-0.03
-0.07
-0.01
Acrolein
-0.01
0.74
0.08
0.32
alpha-Pinene
0.55
0.80
-0.06
-0.05
Benzene
0.95
0.19
0.18
0.005
beta-Pinene
0.004
0.99
0.06
-0.06
Cyclohexane
0.99
-0.02
-0.12
-0.005
Heptane
0.99
-0.02
-0.12
-0.01
Hexane
0.99
-0.02
-0.10
-0.01
Nitropentane
-0.02
0.48
0.76
-0.06
Isopentane
0.03
0.99
0.07
-0.06
Isoprene
0.0005
0.99
0.06
-0.06
m,p-Xylene
0.99
0.09
-0.04
-0.01
Methanol
-0.03
0.95
0.03
0.12
Methylcyclopentane
0.99
-0.02
-0.10
-0.01
Methylenechloride
0.005
0.99
0.06
-0.06
n-Butane
0.67
0.20
0.53
-0.06
o-Xylene
0.98
0.16
-0.02
-0.02
Pentane
0.25
0.96
0.05
-0.06
Propylene
-0.18
-0.06
0.72
0.10
Toluene
0.97
0.23
0.01
-0.005
Vinyl acetate
-0.02
0.01
0.05
0.98
Eigenvalue
13.8
7.17
1.42
1.09
Proportion of variation
0.55
0.29
0.06
0.04
Factor 1: transportation-related sources (vehicle exhaust [high loadings of 1,2,4-
Trimethylbenzene, Benzene, m,p-Xylene, o-Xylene and Toluene], gasoline [high loadings of 2-
Methylpentane, 3-Methylhexane, 3-Methylpentane, Cyclohexane, Hexane, Methylcyclopentane, m,p-
Xylene, o-Xylene and Toluene], gasoline evaporation [high loadings of 2-Methylpentane, 3-
Methylpentane, and Hexane])
Factor 2: biogenic source (high loadings of alpha- and beta-Pinene, and Isoprene), gasoline
evaporation (high loadings of Isopentane, Pentane, and Methanol), vehicle exhaust (high loading
of acrolein), Methylene chloride source?
Factor 3: propane/butane liquified petroleum filling station? (high loading of n-Butane) with
unknown Nitropentane and vehicle exhaust [Propylene]
Factor 4: Vinyl acetate artifact or industrial source (Vinyl acetate)
13
-------
Table 4. Linear-angular rank correlation coefficients, test statistics, and p-values for fine particle CI
and Si at Sites 1 to 3.
Site
Pollutant
n
Dn
un
p-value
1
CI
201
0.18
36.35
>0.0001
Si
229
0.10
22.39
>0.0001
2
CI
151
0.15
22.66
>0.0001
Si
218
0.08
16.14
0.0003
3
CI
154
0.05
7.6
0.0223
Si
240
0.001
0.17
0.9181
Table 5. Kruskal-Wallis test for fine particle CI and Si at Sites 1 to 3.
Site 1
Wind Sector
CI
Si
n
Mean Rank
n
Mean Rank
N
18
56.3
29
74.12
NE
15
61.47
17
106.53
E
20
88.32
22
118.32
SE
124
119.7
127
133.29
S
13
89.77
16
77.62
SW
1
44.00
1
106.0
w
3
79.0
4
116.75
NW
7
43.21
13
79.15
X2
40.17
29.97
p-value
0.0001
0.0001
14
-------
Site 2
Wind Sector
CI
Si
n
Mean Rank
n
Mean Rank
N
5
35.3
32
84.98
NE
6
60.33
12
99.67
E
12
65.58
17
95.59
SE
87
89.39
106
126.22
S
32
57.53
38
96.14
sw
1
69.00
1
150.0
w
1
65.0
1
174.0
NW
7
56.92
11
88.5
X2
21.06
17.78
p-value
0.0037
0.0130
Site 3
Wind Sector
CI
Si
n
Mean Rank
n
Mean Rank
N
10
38.2
34
131.57
NE
5
102.5
10
112.8
E
13
49.77
24
107.38
SE
96
80.6
120
123.93
S
21
96.1
37
113.4
SW
1
104.0
1
78.0
W
1
12.5
1
91.0
NW
7
74.43
11
115.77
X2
20.99
3.14
p-value
0.0038
0.8717
15
-------
5
San Ferlita
ape woolley
Bcrgill
Lyfora
Monte Alto
J
IJ^lDrth Floodv*a
Santa Rosa
dinbur
Lagun
Atascosa
Naftcnat.
Witctife Refu
oouch
La Joya
f^nMcAiienPharr
Ro Hondo
~Ftimera
La l,n9® n
in Flaodway
San Benito
Fresnos
Santa Ami
National
\Wdffe Refuge
Sft&S
Fima Industrial Park
A.Y.U.S.A. (Los Darddos)
C.Y.L.S.A.
A. Y. U.S.A. (Db! Logo)
Ciudad Industrial
EPA Trare boundary Air
Pollution Project Sit©
own will©
m TNRCC site
Figure 1. Air monitoring Sites 1, 2, and 3 used in the TAPP and other sites established by TNRCC in the Lower Rio Grande Valley.
Maquiladora industrial parks listed and identified by arrows to show location relative to monitoring sites.
16
-------
PIG J. Daily PM2.5
UG/CU. M
fPAM
SITE 1
VI 99 115 SS X 4 1*
KM
J A J
M-
ua/cu, n
KM -
N HB * as « BW W NW
49 SS at m 70 5 I 16
SITE 2
J
lAjj-
N NB B
B aw W NW
SITE 3
m
SLI
ni
^y.
«u
Ml
tLS
1 T 1 r^-f—
a «j v-
t-*^—s r~
SU
' * '
¦ -iWm
—n 1 r®*
tu a
¦it
K US
* 81
a sw w sw
-------
0600
TECHNICAL REPORT DATA
1. REPORT NO,
EPA/600/A-02/063
2.
3.RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
5.REPORT DATE
Assessing Transboundary Influences in the Lower Rio
Grande Valley
6.PERFORMING ORGANIZATION CODE
7. AUTHORSSi
8.PERFORMING ORGANIZATION REPORT MO.
Shaibal Mukerjee1, Douglas S. Shadwick2, Kirk E.
Dean3, Linda Y. Carmichael4
Research
9. PERFORMING ORGANIZATION NAME AND ADDRESS
1US Environmental Protection Agency,
Triangle Park, NC 27711
2ManTech Environmental Technology, Inc., Research
Triangle Park, NC 27709
3Texas Natural Resource Conservation Commission,
Austin, TX 78711-3087
4QST Environmental, Durham, NC 27713
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-D2-0134, 68-DS-0049
12. SPONSORING AGENCY NAME AND ADDRESS
National Exposure Research Laboratory,
Office of Research and Development,
US Environmental Protection Agency,
Research Triangle Park, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
Peer Reviewed Conference
Paper, March, 1996 - March,
1997
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The Lower Rio Grande Valley Transboundary Air Pollution Project (TAPP) was a U.S.-Mexico Border XXI
Program project to assess transboundary air pollution in and near Brownsville, Texas. The study used a
ths:2e-site air monitoring network very close to the border to capture the direct impact of local sources
anu transboundary transport. Ambient data included particulate mass and elemental composition, VOCs,
FAHs, pesticides, and meteorology. Also, near real-time, PM;.S mass measurements captured potential
pollutant plume events occurring over 1-h periods. Data collected were compared to screening levels and
other monitoring data to assess general air pollution impacts on nearby border communities. Wind sector
analyses, chemical tracer analyses, principal component analyses, and other techniques were used to
assess the extent of transboundary transport of air pollutants and identify possible transboundary air
pollution sources.
Overall, ambient levels were comparable to or lower than other urban and rural areas in Texas and
elsewhere. Movement of air pollution across the border did not appear to cause noticeable deterioration
of air quality on the U.S. side of the Lower Rio Grande Valley. Dominant southeasterly winds from the
Gulf of Mexico were largely responsible for the clean air conditions in the Brownsville airshed. Few
observations of pollutants exceeded effects screening levels, almost all being VOCs; these appeared to be
due to local events and immediate influences, not regional phenomena or persistent transboundary plumes.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED TERMS
C.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21.NO. OF PAGES
20. SECURITY CLASS (This Page)
UNCLASSIFIED
22. PRICE
------- |