- Email: [email protected]
Potential particulate impacts at the Grand Canyon from northwestern Mexico
The Science of the Total Environment 276 Ž2001. 69᎐82
Potential particulate impacts at the Grand Canyon from northwestern Mexico Delbert J. Eatougha,U , Mark Green b, Will Moran c , Rob Farber c a
Department of Chemistry and Biochemistry, Brigham Young Uni¨ ersity, Pro¨ o, UT, USA b Desert Research Institute, Las Vegas, NV, USA c Southern California Edison, Rosemead, CA, USA Received 17 March 2000; accepted 20 October 2000
Abstract Project MOHAVE was a major air quality and visibility research program conducted from 1990 to 1999 to investigate the causes of visibility impairment in the Grand Canyon National Park region. At Meadview, a remote monitoring site just west of the Grand Canyon National Park, on September 1 and 2, 1992, the concentrations of sulfate Ž3.1 and 4.3 g sulfaterm 3 . were the highest seen in 6 years of monitoring at this site. During this period, the concentrations of SO 2 at Meadview were also abnormally high and approximately three times the sulfate concentrations, on a nmolrm3 basis. High concentrations of sulfate and SO 2 extended south into southern Arizona and northwestern Mexico. Based on ambient atmospheric conditions, emissions from the Mohave Power Project ŽMPP. 110 km upwind of Meadview could not have been responsible for the majority of the regionally observed sulfur oxides. The geographical distribution of SO 2 and sulfate, and available source information suggest that northwestern Mexico was a significant source of the unusually high observed sulfur oxides. A CMB model developed during Project MOHAVE was used to apportion sulfur oxides at Meadview and other sampling sites throughout the study region for August 31᎐September 2, 1992. The results indicate that the contribution of MPP to sulfate at Meadview was typical. However, the transport of SO x from northwestern Mexico was elevated throughout much of the region during this time period. This led to the large increase in sulfate concentrations at Meadview on September 1 and 2. These results indicate that emissions from Mexico can be a significant source of particulate material in the Grand Canyon. 䊚 2001 Elsevier Science B.V. All rights reserved. Keywords: Project MOHAVE; Mexico; Air quality; Visibility; Sulfate; Emissions
U
Corresponding author.
0048-9697r01r$ - see front matter 䊚 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 0 7 7 2 - 0
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
70
1. Introduction
ber 1 and 2, 1992 were the highest concentrations ever reported at this site in a total of 6 years of monitoring. During this period, the concentrations of SO 2 at Meadview were also high and approximately three times the sulfate concentrations, on a nmolrm3 basis. Meadview is a remote sampling site located approximately 20 km from the western boundary of the Grand Canyon National Park. Meadview was a monitoring site during the 5-year SCENES Ž1984᎐1989. as well as during Project MOHAVE Ž1991᎐1992.. Some Project MOHAVE investigators assumed the elevated SO 2 at Meadview during September 1 and 2, 1992 emanated from the Mohave Power Project, MPP, located 110 km ‘upwind’ from Meadview ŽPitchford et al., 1999.. However, the concentrations of sulfate and SO 2 were also com-
Project MOHAVE ŽMeasurement of Haze and Visual Effects, Pitchford et al., 1999. was a joint partnership research program between the US Environmental Protection Agency, the National Park Service and Southern California Edison. The primary goal of the program was to establish the relative contribution of emissions from the Mohave Power Project, MPP Ža 1580-MW lowsulfur coal burning power plant., to visibility degradation in the Grand Canyon region. The Project MOHAVE study area, location of sampling sites and location of the MPP relative to the Grand Canyon National Park are shown in Fig. 1. The concentrations of sulfate measured by Project MOHAVE at Meadview ŽFig. 1. on Septem-
Table 1 Concentrations of sulfate and SO x Žnmolrm3 . at various Project MOHAVE sampling sites during 24 August through to 2 September, 1992 a Month
Aug
Sep
Day
24 25 26 27 28 29 30 31 1 2
Wickenberg
San Gorgonio
Desert Center
Parker
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
4.9 5.9 NAb NA NA NA 17.1 17.0 16.2 NA
7.9 29.7 NA NA NA NA 36.7 55.3 57.1 NA
37.8 48.5 26.8 22.6 18.6 18.6 21.6 20.7 12.8 31.5
52.0 66.0 46.3 44.0 28.3 25.6 28.6 26.3 17.5 42.3
18.9 13.5 30.6 23.9 20.8 15.0 11.1 19.6 44.4 39.1
45.6 27.3 66.8 66.5 90.5 81.3 34.4 34.4 93.5 83.2
NA 13.8 15.5 23.4 18.3 19.0 13.7 32.6 53.5 44.0
NA 15.8 23.3 34.6 37.5 26.5 22.8 80.2 139.2 69.9
44.9 36.7 22.6 17.3 13.8 14.0 12.3 17.3 22.0 29.2
119.2 89.9 39.6 30.8 26.5 71.6 20.6 24.7 50.8 87.2
Essex Aug
Sep a
24 25 26 27 28 29 30 31 1 2
Sulfate 22.2 20.3 20.2 24.8 23.9 21.4 13.8 24.0 28.6 47.4
Kingman SOx 41.5 34.8 41.0 60.6 25.3 78.5 42.4 63.4 68.8 128.1
Sulfate 10.7 15.2 23.5 15.7 17.3 18.3 14.4 17.8 30.9 37.4
SOx 19.0 21.1 28.6 22.8 32.5 40.7 33.8 83.0 112.3 83.4
Joshua Tree
Dolan Springs
Las Vegas Wash
Meadview
Sulfate 13.2 16.8 27.1 11.1 16.7 16.6 10.1 18.2 32.7 47.3
Sulfate 20.5 26.4 40.8 13.3 23.3 24.3 14.0 19.1 35.7 54.5
Sulfate 15.6 20.1 24.7 9.8 19.1 16.8 9.0 15.2 32.2 45.2
SOx 18.4 25.9 37.8 21.0 47.4 41.3 50.5 66.9 98.3 116.4
SOx 52.6 64.0 59.6 37.9 59.0 93.1 35.9 51.7 94.8 143.8
SOx 22.9 29.3 36.3 14.3 48.2 31.7 10.5 37.4 85.5 98.4
Uncertainty in the data is approximately "1᎐2 nmolrm3 ŽPitchford et al., 1999.. Locations of the sampling sites are given in Figs. 1 and 2. b NAs Data not available.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
71
Fig. 1. Project MOHAVE study area. Geographical locations given in this paper are indicated.
parably high throughout the region including all stations south of Žand upwind of. MPP ŽTable 1 and Fig. 2.. Chemical mass balance ŽCMB. receptor based apportionment techniques have been used to estimate the point sources and regional contributors to SO x ŽSO 2 plus particulate sulfate . and fine particulate sulfate present in the study region during the 12 July᎐30 August Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a.. Source profiles of SO x from the four coal-fired generating stations closest to the Grand Canyon National Park ŽFig. 1. and from the various regional sources which can impact the Grand Canyon
have been developed using spherical aluminosilicate ŽSAS. particles present in coal-fired power plant emissions ŽEatough et al., 1996, 1997., Ftotal wHF Žg. plus fine particulate fluoridex, fine particulate Se, As, Pb and Br, and light absorption by fine particles ŽEatough et al., 1997.. These source profiles allow CMB discrimination among the various points and regional sources of SO x in the southwestern United States ŽEatough et al., 1999, 2000a.. Details of the identification of the various regional sources and the establishment of the source profiles used in the CMB analysis have been given ŽEatough et al., 1997.. The profiles of emis-
72
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Fig. 2. Concentrations of SO x Žsolid line. and sulfate Ždashed line. in nmolrm3 at various Project MOHAVE sampling sites during the period from 24 August through to 2 September, 1992, showing the regional character of the elevated concentrations of sulfate and SO x during the September 1 and 2 episode.
sions from the coal-fired generating stations nearest the Grand Canyon wMohave Power Project ŽMPP., Navajo Generating Station and Reid Gardner Generating Station, Fig. 1, were established from both ambient and stack data. The eight regional sources present in the Grand Canyon region during the July᎐August 1992 summer intensive ŽEatough et al., 1997; Eatough et al., 2000a. are summarized below. In addition, the profile of emissions from the coal-fired generating stations in Arizona to the southeast of the Grand Canyon was determined from the ambient data. The identified regional sources ŽEatough et al., 1997. and their probable geographical origins ŽEatough et al., 1997, 2000a. include Žsee Fig. 1.: SJ: Air masses from the San Joaquin Valley, CA area.
LA: Air masses transported from the South Coast Air Basin in southern CA. Emissions from this area include SO x from refineries, metal processing industries and mobile sources. BC: Air masses for the Baja California source region originate from north-central and northwestern Mexico, and the California Imperial Valley. Emissions from this portion of Mexico are not well characterized. However, there are no major point sources but numerous uncontrolled small industries which emit SO x ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001. exist in the region. The Imperial Valley has only minor sources of SO x ŽChow et al., 2000.. AZ: Characterized by air masses from the area generally south of the Grand Canyon. This includes emissions from smelters in both Arizona and Mexico.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
SE: Characterized by air masses from southeast of the Grand Canyon and Petrified Forest National Parks. This source region will also include smelter SO x emissions. NW: Days when the sampled air mass at a given site originated from the region north and west of the Grand Canyon established the NW region source profiles. The data were characterized by three distinctly different source regions. These included the profiles for NW1, a region which includes emission from coal-fired power plants Žprobably in northern Nevada. as evidenced by the elevated SASrSOx ratio, a NW2 local region characterized by a high FTotalrSOx ratio, probably originating from high fluoride in agricultural emissions ŽEatough et al., 1997., and a LV profile, characterized by no SAS particles or FTotal , identified from a synoptic meteorological airmass analysis between Las Vegas and the MV sampling site 100 km east of Las Vegas as originating from the urban Las Vegas area; see Fig. 1. A hybrid CMB model ŽEatough et al., 2000a. was applied to the Project MOHAVE summer intensive data at Meadview and Hopi Point in the Grand Canyon, Fig. 1. This model was developed and run using a combination of regional area and point source profiles which accounted for all of the source profile species concentrations and for all of the ambient SO x and sulfate. In this paper, we have used a combination of meteorological and CMB analyses to investigate the probable source of the elevated regionally distributed SO 2 and sulfate during September 1᎐2, 1992. The meteorological data indicate that MPP could not have been the dominant source of the regionally elevated SO 2 and sulfate during this episode. The CMB model used to apportion sulfur oxides at Meadview and Hopi Point during the Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a. was run for several sampling sites throughout the region for August 31᎐September 2, 1992. The results indicate that the contribution of MPP to sulfate at Meadview was no higher than, but comparable to, that modeled during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. The CMB results also
73
indicate that the excess regional SO x emanated from the Baja California source area during this time. This led to the large increase in sulfate concentrations at Meadview on September 1 and 2. These results are consistent with those obtained for the earlier Project MOHAVE summer intensive period where the Baja California CMB source region was found to be a major contributor to both sulfate ŽEatough et al., 2000a. and visibility reduction ŽEatough et al., 2000b. in the Grand Canyon region. The justification for these conclusions is presented in this paper.
2. Methods The data used in the CMB analysis for the episode period were obtained from Project MOHAVE IMPROVE results ŽPitchford et al., 1999.. Fine particulate sulfate was determined using the IMPROVE ŽEldred et al., 1990. protocols of collection of particles on a Teflon filter for the determination of sulfur by PIXE and ion chromatographic analysis of the carbonate filter following a Teflon filter for sulfate for the determination of SO 2 Žg.. Soot was determined on Teflon filters from the laser integrated plate analysis ŽEldred et al., 1990. measurement of fine particle absorption. The various trace elements were determined from PIXE analysis of Teflon filters.
3. Synoptic meteorological description of the August 30 through September 2, 1992 regional sulfate episode The period August 30 through September 2, 1992 was characterized by a weak split flow at 500 mb, Fig. 3. This split resulted in moderately low mixing heights Ž5850 m. at the 500-mb level for this time of year. As a result of this split, the flow was predominately southwesterly to westerly at 500 mb and fairly dry. Surface dew points were typically in approximately 5⬚C across the desert southwest. Satellite, radar and surface observations showed some scattered to broken layers of clouds associated with this split flow. On August
74
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
30, cumulonimbus clouds with bases at 2700 m above ground level Žagl. were reported at Las Vegas, Nevada. Both Las Vegas and Cedar City, Utah Ž200 km north of the Grand Canyon National Park. reported heavy cloud cover for most of the day on August 30 with some light precipitation. Potentially, these meteorological conditions could lead to substantial conversion of SO 2 to particulate sulfate. However, perfluorocarbon tracer concentrations ŽocPDCH. at Meadview released from the Mohave Power Project stack ŽPitchford et al., 1999; Watson et al., 1993. indicate that only a small portion of the high Meadview sulfate concentrations could have been due to MPP SO 2 . During the tracer release period an upper limit of MPP sulfate can be obtained by multiplying ambient tracer concentrations by the ratio of SO 2 to tracer in the MPP stack. The upper limit is obtained assuming 100% conversion of SO 2 to sulfate and no gaseous or particle
depositional losses. Using these assumptions, for August 30, the MPP emissions could have contributed, at most, 0.3 grm3 of sulfate Ž3 nmolrm3 of particulate S. at any of the Project MOHAVE monitoring sites. The CMB analyses results Ždescribed in the following section. indicate that MPP was responsible for less than 0.1 g of sulfaterm 3 at Meadview on August 30. Since MPP stack releases were terminated on August 31 at 07.00 h MST, the tracer data cannot be used to place upper limits on MPP sulfate much after this time. However, eliminating the possibility of substantial MPP sulfate on August 30 is significant because this day had the best potential for substantial wet processes sulfate formation due to cloud cover. After August 30, the atmosphere dried quickly. From August 31᎐September 2 only scattered clouds were observed throughout the region with bases above 4000 m agl.
Fig. 3. Surface pressure map for 12Z, August 30, 1992 for the southwestern United States. Map shows typical summer pattern with relatively high pressure over the western United States. The semi-permanent thermal flow is over the desert southwest interior. Resulting transport flow in the boundary layer is from the southwest through southeast into this thermal low centered over the Lower Colorado River Valley. All data in Figs. 3 and 4 are from the National Weather Service. Pressures are in millibars to the nearest tenth with the last three digits of the pressure reading at each circle. Wind speeds are in knots with a full barb denoting 10 knots.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
From the surface through over 1000 m agl, there was a well developed thermal low centered over the lower Colorado River Valley, the region to the south of Las Vegas along the Colorado River, Figs. 1 and 4. This is a semi-permanent feature in the desert southwest during the summer, resulting from differential heating between the Pacific Ocean, Gulf of California and the interior deserts. Surface pressure gradients between the coastal Los Angeles Basin and interior deserts were strongly to moderately onshore on August 30᎐31, decreasing to approximately average by September 1᎐2. This pattern resulted in a typical southwesterly flow in the transport layer across the desert southwest. On August 30, based on surface pressure gradient analyses, pollutants equally exited the Cajon Pass and Beaumont Passes, Fig. 1, from the Los Angeles Basin. Material exiting the Cajon Pass tends to travel through Essex ŽFig. 1. directly to the Colorado River while material exiting Beaumont
75
Pass takes a more circuitous route through San Gorgonio ŽFig. 1. to Desert Center ŽFig. 1., where the material catches the southerly channeled Colorado River Valley winds and is transported northward. By September 2, three times as much Los Angeles Basin pollutant material was transported along the Beaumont pass route through San Gorgonio compared to the Cajon Pass path through Essex. This synoptic analysis is collaborated by the increase in sulfur oxide concentrations at San Gorgonio on September 2, as compared to the previous 2 days ŽFig. 2. and by a sharp drop in the concentrations of methylchloroform Ža marker specific to emissions from the LA Basin, Bastable et al., 1990. at Spirit Mountain Žjust east of Essex, Fig. 1. on August 31 and September 1, as compared to August 30 ŽPitchford et al., 1999.. In response to these southerly transport winds across the region, the MPP plume was also transported northward throughout this entire pe-
Fig. 4. Analysis map for 500 mb, 00Z, August 31, 1992, for the western United States. The map shows a weak split flow over the southwestern US resulting in a weak trough with limited sub-tropical moisture over the desert southwest. For each observing station, the height of the atmosphere in decameters above sea level at which the 500-mb pressure occurs is given to the right of each circle.
76
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
riod except for approximately 5 h from 05.00 h until 10.00 h MST on the morning of August 31. This brief wind direction shift on the morning of August 31 could have been in response to the passage of the upper level split flow trough. Transport winds at effective plume height Ži.e. approx. 700 m agl. were approximately 10 mrs at the beginning of the period on August 30. However, there was a gradual decrease in the transport speeds as the pressure gradients relaxed to approximately 2᎐3 mrs by September 2. Transport speeds during the 5 h of ‘plume reversal’ were approximately 2 mrs, resulting in the plume traveling no further south than approximately 40 km. Furthermore, pressure gradients resulting in southerly transport winds were three times stronger during this period compared to the periods in mid-August when MPP emissions were still transported to the north, as indicated by the ocPDCH data. Thus, MPP emissions cannot account for the high sulfur oxide concentrations at stations south of MPP, Fig. 2. Back-trajectories from Meadview for the period of September 1᎐2 were computed using the ATAD model ŽHeffter, 1980.. The ATAD model does not take terrain into account. However, ATAD back trajectory results generally agreed with more sophisticated model results during the Project MOHAVE summer intensive ŽPitchford et al., 1999. when both were run. Only ATAD model results were available for the 1᎐2 September time period. The back-trajectories showed transport from the southwest and then west, with trajectories going south of MPP to just north of Mexicali near the Mexican border, Fig. 1, and then over the Los Angeles metropolitan area. While the trajectories may be expected to reveal the general transport direction, definitive conclusions cannot be drawn from them due to inaccuracies in wind fields due to the sparseness of data used ŽGreen et al., 2000.. In particular, flows channeled along the lower Colorado River valley are not represented due to the lack of wind data used in the model runs. Thus, the back-trajectories neither confirm nor refute the CMB results Ždescribed in the following section. that sources
in northwestern Mexico contributed significantly to the ambient sulfate levels. However, the Project MOHAVE sulfur oxide data ŽFig. 2. indicate that transport from northwestern Mexico and the California Imperial Valley into the study regions did occur. High concentrations of both SO 2 and sulfate at Desert Center and Parker were observed on September 1 and 2, with relatively lower concentrations coming from the LA Basin as indicated by the San Gorgonio data ŽFig. 2.. Similar high concentrations of SO x associated with transport from northwestern Mexico were seen at other times during the 50-day summer intensive at Parker, Desert Center and El Centro. However, these episodes did not result in the very elevated concentrations of SO 2 as far north as Parker. A more typical episode is illustrated by the August 27᎐29 data ŽFig. 2. where high concentrations occur at Desert Center Žconcentrations at El Centro were twice those at Desert Center., but the increase in SO x concentrations at Parker were much smaller. Data for El Centro are not shown in Fig. 2 because no samples were collected at this site after August 30. Pollutants coming from Baja California Žnorthwestern Mexico. and the California Imperial Valley followed a typical transport path during the summer, as illustrated by the general transport of the perfluorocarbon tracer ŽPitchford et al., 1997, 2000. release from El Centro during the Project MOHAVE summer intensive ŽFig. 5.. The great majority of these pollutants are expected to originate from Mexico and not from the Imperial Valley ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001.. Pollutants were transported across the border in a northwesterly direction to a location where eventually they would catch the southerly-channeled flow in the lower Colorado River Valley. At the beginning of the episode ŽAugust 30. average transport speeds were between 5 and 12 mrs with highest speeds in the afternoon and early evening. By September 2, these regional transport speeds had decreased to approximately 3᎐6 mrs with lowest speeds at night. The regional nature of SO x throughout the study area during September 1 and 2 is seen in Fig. 2.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
77
Fig. 5. Map of the frequency that tracer was detected above background for PFT releases from El Centro Žmarked [. Žfrom Pitchford et al., 1997.. The shaded areas indicate contours over which the average concentration of the tracer was the indicated value above background concentration Ždetails are given in Pitchford et al., 1997.. The data illustrate the expected transport of pollutants from Mexico and the California Imperial Valley up the Colorado River Valley during the summer.
4. CMB analysis and results during the September 1 and 2 episode The CMB model has been used to apportion sulfur oxides at Meadview and Hopi Point during the 50-day Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a.. For this 3-day August 31᎐September 2 episode, the model was run for additional receptor sites. It was applied to Project MOHAVE data from the Meadview ŽMV., Las Vegas Wash ŽLVW., Cottonwood Cove ŽCC., Dolan Springs ŽDS., Essex ŽES., Yucca ŽYC., Parker ŽPA., Wickenberg ŽWB. and Painted Desert ŽPD. sampling sites, Figs. 1 and 6. These regionally distributed sites were all included in
the CMB analysis because the Project MOHAVE data indicated they were all associated with a marked increase in SO 2 and sulfate on September 1 and 2, e.g. see Fig. 2. For most sites, the sulfate concentrations were higher than those measured during any other portion of the Project MOHAVE summer intensive. The source profiles previously developed as part of the CMB model ŽEatough et al., 1997. were used without change. The paradigms previously used for the estimation of SO 2 and sulfate deposition and conversion, and for the regional and Meadview area specific transport times were also used without change ŽDu et al., 1996; Eatough et al., 2000a.. Two of the markers, FTotal and spherical alumi-
78
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Fig. 6. Locations of the various sampling sites included in the CMB analysis for 31 August᎐2 September 1992, and the expected origins of emissions from the sources included in the analysis; MPP, Las Vegas, LA, BC, AZ and SE.
nosilicate ŽSAS. particles, used in the CMB analysis were not available for the 31 August᎐2 September event. SAS is a specific marker for emissions from a region which includes coal-fired power plant emissions. FTotal appears to originate from both industrial and agricultural regions and was particularly useful in identifying the SJ and NW2 source regions. The absence of these markers restricted the number of source regions which could be included in any single CMB estimation. Meteorological analyses were used to further guide the CMB model construction. It was assumed that emissions from MPP, Las Vegas ŽLV., Los Angeles ŽLA., Baja California ŽBC. and Arizona ŽAZ. source areas ŽFig. 6. were the only sources of sulfur oxides at all but one of the regionally analyzed sites. It was assumed that the southeast ŽSE. source region could also impact the PD site. It was further assumed that emissions from MPP were not present at sampling sites south of MPP or Dolan Springs Ži.e. at YC, ES, PR or WB, Fig. 6. and LV emissions could only be present at LVW and MV. The unavailability of the SAS particle source profile data restricted the robustness of the CMB analysis when emissions from certain point andror source region combinations were present. A sensitivity analysis was performed using the summer intensive data when SAS particle concen-
trations were available. The CMB results were adversely affected by a combination of emissions from the MPP point source, and LV and BC source regions. When emissions from these three sources were simultaneously present, the elimination of the SAS data frequently prevented CMB from separating LV and BC source signals. One or the other could be included in the fit with approximately equal ‘goodness of fit’ results, but both could not be included. This was true whether or not emissions from other source regions Že.g. LA. were present. Impacts from all other source regions were properly identified. This limitation of the CMB analysis without SAS particle data was evident in the analyses reported here for MV. No other receptor site was so adversely effected. At Meadview, this complication was due to the presence of comparable concentrations of SO x from LV and BC and high SO x from MPP. When BC emissions were present at CC, LVW and DS, the MV fit which included LV was adjusted for a BC impact consistent with the CC, LVW and DS results, i.e. it was assumed that the decrease in the BC source contribution seen at LVW as compared to CC also occurred at MV as compared to DS. As a result, the estimates of BC SO x at Meadview given in are based on an extrapolation of the CMB results at CC and DS and are not obtained directly from the Meadview data. While
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
the treatment cannot be directly validated from the data set, this approach was used because the assumption prevented an apparent overestimation of the BC source impact at Meadview. This adjustment did not significantly alter the predicted impact from MPP for any sampling period, e.g. see the sensitivity tests given in Table 3. The CMB analysis estimates of sulfur oxide emission sources present in the study region during September 2 are given in Fig. 7. Emissions from both LA and BC were generally present throughout September 1᎐2 at all sites ŽEatough et al., 1999.. The impact from BC on September 2 ŽFig. 7. was greater than from LA for both SO 2 and sulfate. However, the relative importance of emissions from LA increased from Essex northward, consistent with the expected transport path from the LA region. Similar trends were seen on September 1 ŽEatough et al., 1999. The highest
79
concentrations of impacts from BC were found at Parker, Essex and Yucca. The highest impacts from LA were observed at Essex and Dolan springs. High impacts for SO 2 Žbut not sulfate . from MPP were observed at Dolan Springs, Las Vegas Wash and Meadview. Significant impacts from LV were found at Las Vegas Wash and Meadview. The site-to-site and day-to-day changes ŽEatough et al., 1999. in sulfate and SO 2 were consistent with a southerly flow, resulting in generally decreasing BC and LA SO x concentrations as pollutants traveled north and east, but with an increasing sulfate to SO x ratio for the SO x from the BC and LA source areas throughout the region. The increase in the sulfate to SO x ratio results from both the conversion of SO 2 to sulfate and the faster deposition of SO 2 , as compared to sulfate during emission transport ŽDu et al., 1996; Eatough et al., 2000a..
Fig. 7. Results of the CMB analysis for SO x and sulfate Žnmolrm3 . on 2 September 1992 at the sampling sites identified in Figs. 1 and 6 for the sources identified in Fig. 6.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
80 Table 2 Meadview sulfate source September 1992,grm3 Date
MPP
LV
apportionment, 31
LA
BC
August᎐2
Sulfate Not Fit, % Žgrm3 .
31 Aug 1 Sep 2 Sep
0.3 0.3 0.6
0.3 1.4 0.8
0.2 0.4 0.6
1.1 0.9 1.8
4.2 Ž0.1. 8.4 Ž0.2. 14.8 Ž0.6.
Based on this analysis, the sulfate impacts at Meadview from all sources for each sampling day are given in Table 2. The results of sensitivity tests on the effect of changing the CMB MV assumptions are given in Table 3. The highest concentrations of sulfate from MPP were calculated to be 0.69 grm3 at Dolan Springs on 1 September and 0.55 grm3 at Meadview on 2 September. All other estimated MPP sulfate impacts were less than 0.4 grm3. In contrast, the highest estimated sulfate impact at Meadview from any source was 1.8 grm3 from BC on September 2.
5. Results from the Project MOHAVE summer intensive The results obtained during this episode were consistent with the importance of the BC source in the Grand Canyon region during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. Fig. 8 displays the total daily sulfate at the Hopi Point sampling site in the Grand Canyon and the daily sulfate attributed to the BC source Table 3 Sensitivity to changes in the assumption of the relative amounts of SO x at Meadview from LV and BC on MPP sulfate, grm3 Date
31 Aug 1 Sep 2 Sep a
MPP Sulfate SOx from both BC and LVa
SOx from LV but not BC
SOx from BC but not LV
0.3 0.3 0.6
0.5 0.4 0.6
0.3 0.3 0.9
Most probable scenario.
Fig. 8. Total and BC sulfate at GC. Northern Mexico is often a major contributor to sulfate at the Grand Canyon.
region during the Project MOHAVE summer intensive. Similar trends were seen at Meadview, but the relative impact of BC was less pronounced due to the importance of LV sulfate at MV. Sulfate from the BC region was frequently present at Hopi Point. When emissions from this source region were present, they often accounted for the majority of the sulfate. While concentrations as high as the 1.8-g sulfaterm 3 level at Meadview on September 2 were not seen at Hopi Point during the summer intensive, concentrations from 0.5 to 1.5 g sulfaterm 3 were frequently seen. The CMB prediction of BC SO x at Meadview and Hopi Point always coincided with the presence of the fluorocarbon tracer released from El Centro ŽEatough et al., 2000a.. The BC region accounted for an average of 42% of the sulfate present at Hopi Point during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. During the July 12 through August 30 summer intensive, the highest concentration of BC sulfate at Meadview was 0.8 grm3, with the average concentration of BC sulfate at Meadview being 0.3 grm3. In contrast, the highest concentration of BC sulfate at Meadview during the days just following the end of the summer intensive was 1.8 grm3 and averaged 1.3 grm3 , a factor four times higher. During the summer intensive, the average contribution of BC to SO x and sulfate at Meadview was 6% and 12% of the total, respectively. However, during the 3-day episode ŽAugust 31᎐September 2., BC contributed an average of
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
19% of the SO x and 44% of the sulfate. The LA source area contributed 18% of the sulfate during the summer intensive and a comparable 14% of the total during this 3-day episode. The relative ratios of SO 2 and sulfate present from the BC source is comparable to that seen during the summer intensive. Thus, the conversion of SO 2 to sulfate was normal during this episode period. This result is consistent with the dry condition of the lower boundary layer of the atmosphere where the pollutants were located during the August 31 through September 2 time period. Thus, the 6-year record high concentrations of sulfate observed at Meadview during August 31᎐September 2, 1992 can be attributed to a significant increase in transport of material from northwestern Mexico to Meadview. What remains an open question is whether there was an increase in SO x emissions from BC andror was the meteorological transport particularly conducive to transport of BC emissions during this period. However, since we do know that MPP emissions are ‘base loaded,’ that there was no stagnation, and that desert atmospheric conditions were conducive to only slow gas-phase conversion rates, MPP could only be contributing its ‘normal’ amount of sulfate, consistent with the CMB results. BC was probably also the largest contributor to extinction in the Grand Canyon during the summer intensive. Analysis performed elsewhere ŽEatough et al., 2000b. demonstrated that BC is a significant haze contributor in this region because of the frequency of the presence of pollutants from this source and overall magnitude of its pollutant components. Thus, combination of the CMB and light extinction data obtained during Project MOHAVE suggest that emissions from northwestern Mexico were the largest single urban area or point anthropogenic source of visibility impairing particulate material at Hopi Point in the Grand Canyon during Project MOHAVE.
6. Conclusions An analysis of available source information ŽChow et al., 2000; Chow and Watson, 2001;
81
Watson and Chow, 2001., meteorological patterns, CMB results and distributions of SO x and sulfate throughout the region are all consistent with the high concentrations of sulfate observed at Meadview on September 1 and 2, 1992 emanating from sources in northwestern Mexico. Similar results were obtained from the CMB analysis of all Project MOHAVE summer intensive data. The CMB results for the July and August Project MOHAVE summer intensive period showed that the BC source area was a frequent contributor to sulfate at Meadview and Hopi Point. During the Project MOHAVE summer intensive, the BC source was the main contributor to sulfate at Hopi Point in the Grand Canyon. Particles from the BC source area were also the main anthropogenic contributor to visibility reduction at Hopi Point during this period ŽEatough et al., 2000b.. Similar transport of SO x from Mexico into the Big Bend region in Texas has also been reported ŽGebhart et al., 2000a,b.. This transport, however, included significant point source emissions, as contrasted with the probable smaller, but greater in number, SO x sources in Northwestern Mexico ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001..
Acknowledgements Support for this work was provided by Southern California Edison to participate in Project MOHAVE. However, the results, findings, and conclusions expressed in this paper are solely those of its authors and are not necessarily endorsed by the management and sponsors of Project MOHAVE. References Bastable H, Rogers D, Schorran D. Tracers of opportunity and pollutant transport in southern California. Atmos Environ 1990;24:137᎐151. Du AY, Mangelson NF, Eatough DJ, Farber RJ, Watson JG. The deposition of SO 2 Žg. and particulate sulfate in the desert southwest. Proceedings Air & Waste Management Association Annual Meeting, 96-MP1A.04, 1996:1᎐14.
82
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Chow JC, Watson JG. Zones of representation for PM 10 measurements along the USrMexico borders. Sci Total Environ 2001;276:49᎐68. Chow JC, Watson JG, Green MC, Lowenthal DH, Bates B, Oslund W, Torres G. Cross-border transport and spatial variability of suspended particles in Mexicali and California’s Imperial Valley. Atmos Environ 2000;34:1833᎐1843. Eatough DJ, Eatough M, Eatough NL, Lewis EA. Apportionment of sulfur oxides at Canyonlands during the winter of 1990. III. Source apportionment of SO x and sulfate and the conversion of SO 2 to sulfate in the Green River Basin. Atmos Environ 1996;30:295᎐308. Eatough DJ, Du AY, Joseph JM, Caka FM, Sun B, Lewis L, Mangelson NF, Eatough M, Eatough NL, Farber RJ, Watson JG. Regional source profiles of sources of SO x at the Grand Canyon during Project MOHAVE. J Air Waste Manage Assoc 1997;47:101᎐118. Eatough DJ, Green M, Moran W, Farber R. A case study of particulate impacts on the Grand Canyon from northern Mexico. Proceedings, 92nd Annual A& WMA Meeting Paper No 99-497, 1999. Eatough DJ, Farber RJ, Watson JG. Second generation chemical mass balance source apportionment of sulfur oxides and sulfate at the Grand Canyon during the Project summer intensive. J Air Waste Manage Assoc 2000a; 50:759᎐774. Eatough DJ, Cui W, Hull J. Fine particulate chemical composition and light extinction at Meadview, Arizona Žin preparation. 2000b. Eldred RA, Cahill TA, Wilkinson LK, Feeney PJ, Chow JC, Malm WC. Measurement of fine particles and their chemical components in the IMPROVErNPS networks. In: Mathai CV, editor. Transactions: visibility and fine particles. Pittsburgh, PA: Air & Waste Management Association, 1990:187᎐196.
Gebhart KA, Malm WC, Flores M. A preliminary look at source᎐receptor relationships in the Texas᎐Mexico border area. J Air Waste Manage Assoc 2000a;50:759᎐774. Gebhart KA, Kriedenweis SM, Malm WC. Visibility at Big Bend National Park: a summary of source attribution analyses of the historical database and the 1996 scoping study. Sci Total Environ 2000b:276. Green, MC, Pai P, Ashbaugh L, Farber RJ. Evaluation of wind fields used in the Grand Canyon Visibility Transport Commission analyses. J Air Waste Manage Assoc 2000;50:809᎐817. Heffter JL. Air Resources Laboratories Atmospheric Transport and Dispersion Model ŽARL-ATAD., Technical Memorandum ERL ARL-81, NOAA, Rockville, 1980. Pitchford M, Green M, Kuhns H, Farber R. Characterization of regional transport dispersion using Project MOHAVE tracer data. Visual air quality: aerosols and global radiation balance. Air Waste Manage Assoc VIP 1997;76Ž1.:181᎐200. Pitchford M, et al. Ž1999.. Project MOHAVE Final Report. Internet version Žwww.epa.govrregion09rairrmohaverreport.html.. Pitchford M, Green M, Kuhns H, Farber R. Characterization of regional transport and dispersion using Project MOHAVE tracer data. J Air Waste Manage Assoc 2000;50:733᎐745. Watson JG, Green M, Hoffer TE, Lawson DR, Eatough DJ, Farber RJ, Malm WC, McDade CE, Pitchford M. Project MOHAVE data analysis plan, Proceedings, 86th Annual A& WMA Meeting 1993, Paper No. 93-MP-4.10, 1993. Watson JG, Chow JC. Source characterization of major emission sources in the Imperial and Mexicali Valleys along the U.S.rMexico border. Sci Total Environ 2001;276:33᎐47.
Potential particulate impacts at the Grand Canyon from northwestern Mexico Delbert J. Eatougha,U , Mark Green b, Will Moran c , Rob Farber c a
Department of Chemistry and Biochemistry, Brigham Young Uni¨ ersity, Pro¨ o, UT, USA b Desert Research Institute, Las Vegas, NV, USA c Southern California Edison, Rosemead, CA, USA Received 17 March 2000; accepted 20 October 2000
Abstract Project MOHAVE was a major air quality and visibility research program conducted from 1990 to 1999 to investigate the causes of visibility impairment in the Grand Canyon National Park region. At Meadview, a remote monitoring site just west of the Grand Canyon National Park, on September 1 and 2, 1992, the concentrations of sulfate Ž3.1 and 4.3 g sulfaterm 3 . were the highest seen in 6 years of monitoring at this site. During this period, the concentrations of SO 2 at Meadview were also abnormally high and approximately three times the sulfate concentrations, on a nmolrm3 basis. High concentrations of sulfate and SO 2 extended south into southern Arizona and northwestern Mexico. Based on ambient atmospheric conditions, emissions from the Mohave Power Project ŽMPP. 110 km upwind of Meadview could not have been responsible for the majority of the regionally observed sulfur oxides. The geographical distribution of SO 2 and sulfate, and available source information suggest that northwestern Mexico was a significant source of the unusually high observed sulfur oxides. A CMB model developed during Project MOHAVE was used to apportion sulfur oxides at Meadview and other sampling sites throughout the study region for August 31᎐September 2, 1992. The results indicate that the contribution of MPP to sulfate at Meadview was typical. However, the transport of SO x from northwestern Mexico was elevated throughout much of the region during this time period. This led to the large increase in sulfate concentrations at Meadview on September 1 and 2. These results indicate that emissions from Mexico can be a significant source of particulate material in the Grand Canyon. 䊚 2001 Elsevier Science B.V. All rights reserved. Keywords: Project MOHAVE; Mexico; Air quality; Visibility; Sulfate; Emissions
U
Corresponding author.
0048-9697r01r$ - see front matter 䊚 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 1 . 0 0 7 7 2 - 0
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
70
1. Introduction
ber 1 and 2, 1992 were the highest concentrations ever reported at this site in a total of 6 years of monitoring. During this period, the concentrations of SO 2 at Meadview were also high and approximately three times the sulfate concentrations, on a nmolrm3 basis. Meadview is a remote sampling site located approximately 20 km from the western boundary of the Grand Canyon National Park. Meadview was a monitoring site during the 5-year SCENES Ž1984᎐1989. as well as during Project MOHAVE Ž1991᎐1992.. Some Project MOHAVE investigators assumed the elevated SO 2 at Meadview during September 1 and 2, 1992 emanated from the Mohave Power Project, MPP, located 110 km ‘upwind’ from Meadview ŽPitchford et al., 1999.. However, the concentrations of sulfate and SO 2 were also com-
Project MOHAVE ŽMeasurement of Haze and Visual Effects, Pitchford et al., 1999. was a joint partnership research program between the US Environmental Protection Agency, the National Park Service and Southern California Edison. The primary goal of the program was to establish the relative contribution of emissions from the Mohave Power Project, MPP Ža 1580-MW lowsulfur coal burning power plant., to visibility degradation in the Grand Canyon region. The Project MOHAVE study area, location of sampling sites and location of the MPP relative to the Grand Canyon National Park are shown in Fig. 1. The concentrations of sulfate measured by Project MOHAVE at Meadview ŽFig. 1. on Septem-
Table 1 Concentrations of sulfate and SO x Žnmolrm3 . at various Project MOHAVE sampling sites during 24 August through to 2 September, 1992 a Month
Aug
Sep
Day
24 25 26 27 28 29 30 31 1 2
Wickenberg
San Gorgonio
Desert Center
Parker
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
Sulfate
SOx
4.9 5.9 NAb NA NA NA 17.1 17.0 16.2 NA
7.9 29.7 NA NA NA NA 36.7 55.3 57.1 NA
37.8 48.5 26.8 22.6 18.6 18.6 21.6 20.7 12.8 31.5
52.0 66.0 46.3 44.0 28.3 25.6 28.6 26.3 17.5 42.3
18.9 13.5 30.6 23.9 20.8 15.0 11.1 19.6 44.4 39.1
45.6 27.3 66.8 66.5 90.5 81.3 34.4 34.4 93.5 83.2
NA 13.8 15.5 23.4 18.3 19.0 13.7 32.6 53.5 44.0
NA 15.8 23.3 34.6 37.5 26.5 22.8 80.2 139.2 69.9
44.9 36.7 22.6 17.3 13.8 14.0 12.3 17.3 22.0 29.2
119.2 89.9 39.6 30.8 26.5 71.6 20.6 24.7 50.8 87.2
Essex Aug
Sep a
24 25 26 27 28 29 30 31 1 2
Sulfate 22.2 20.3 20.2 24.8 23.9 21.4 13.8 24.0 28.6 47.4
Kingman SOx 41.5 34.8 41.0 60.6 25.3 78.5 42.4 63.4 68.8 128.1
Sulfate 10.7 15.2 23.5 15.7 17.3 18.3 14.4 17.8 30.9 37.4
SOx 19.0 21.1 28.6 22.8 32.5 40.7 33.8 83.0 112.3 83.4
Joshua Tree
Dolan Springs
Las Vegas Wash
Meadview
Sulfate 13.2 16.8 27.1 11.1 16.7 16.6 10.1 18.2 32.7 47.3
Sulfate 20.5 26.4 40.8 13.3 23.3 24.3 14.0 19.1 35.7 54.5
Sulfate 15.6 20.1 24.7 9.8 19.1 16.8 9.0 15.2 32.2 45.2
SOx 18.4 25.9 37.8 21.0 47.4 41.3 50.5 66.9 98.3 116.4
SOx 52.6 64.0 59.6 37.9 59.0 93.1 35.9 51.7 94.8 143.8
SOx 22.9 29.3 36.3 14.3 48.2 31.7 10.5 37.4 85.5 98.4
Uncertainty in the data is approximately "1᎐2 nmolrm3 ŽPitchford et al., 1999.. Locations of the sampling sites are given in Figs. 1 and 2. b NAs Data not available.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
71
Fig. 1. Project MOHAVE study area. Geographical locations given in this paper are indicated.
parably high throughout the region including all stations south of Žand upwind of. MPP ŽTable 1 and Fig. 2.. Chemical mass balance ŽCMB. receptor based apportionment techniques have been used to estimate the point sources and regional contributors to SO x ŽSO 2 plus particulate sulfate . and fine particulate sulfate present in the study region during the 12 July᎐30 August Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a.. Source profiles of SO x from the four coal-fired generating stations closest to the Grand Canyon National Park ŽFig. 1. and from the various regional sources which can impact the Grand Canyon
have been developed using spherical aluminosilicate ŽSAS. particles present in coal-fired power plant emissions ŽEatough et al., 1996, 1997., Ftotal wHF Žg. plus fine particulate fluoridex, fine particulate Se, As, Pb and Br, and light absorption by fine particles ŽEatough et al., 1997.. These source profiles allow CMB discrimination among the various points and regional sources of SO x in the southwestern United States ŽEatough et al., 1999, 2000a.. Details of the identification of the various regional sources and the establishment of the source profiles used in the CMB analysis have been given ŽEatough et al., 1997.. The profiles of emis-
72
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Fig. 2. Concentrations of SO x Žsolid line. and sulfate Ždashed line. in nmolrm3 at various Project MOHAVE sampling sites during the period from 24 August through to 2 September, 1992, showing the regional character of the elevated concentrations of sulfate and SO x during the September 1 and 2 episode.
sions from the coal-fired generating stations nearest the Grand Canyon wMohave Power Project ŽMPP., Navajo Generating Station and Reid Gardner Generating Station, Fig. 1, were established from both ambient and stack data. The eight regional sources present in the Grand Canyon region during the July᎐August 1992 summer intensive ŽEatough et al., 1997; Eatough et al., 2000a. are summarized below. In addition, the profile of emissions from the coal-fired generating stations in Arizona to the southeast of the Grand Canyon was determined from the ambient data. The identified regional sources ŽEatough et al., 1997. and their probable geographical origins ŽEatough et al., 1997, 2000a. include Žsee Fig. 1.: SJ: Air masses from the San Joaquin Valley, CA area.
LA: Air masses transported from the South Coast Air Basin in southern CA. Emissions from this area include SO x from refineries, metal processing industries and mobile sources. BC: Air masses for the Baja California source region originate from north-central and northwestern Mexico, and the California Imperial Valley. Emissions from this portion of Mexico are not well characterized. However, there are no major point sources but numerous uncontrolled small industries which emit SO x ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001. exist in the region. The Imperial Valley has only minor sources of SO x ŽChow et al., 2000.. AZ: Characterized by air masses from the area generally south of the Grand Canyon. This includes emissions from smelters in both Arizona and Mexico.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
SE: Characterized by air masses from southeast of the Grand Canyon and Petrified Forest National Parks. This source region will also include smelter SO x emissions. NW: Days when the sampled air mass at a given site originated from the region north and west of the Grand Canyon established the NW region source profiles. The data were characterized by three distinctly different source regions. These included the profiles for NW1, a region which includes emission from coal-fired power plants Žprobably in northern Nevada. as evidenced by the elevated SASrSOx ratio, a NW2 local region characterized by a high FTotalrSOx ratio, probably originating from high fluoride in agricultural emissions ŽEatough et al., 1997., and a LV profile, characterized by no SAS particles or FTotal , identified from a synoptic meteorological airmass analysis between Las Vegas and the MV sampling site 100 km east of Las Vegas as originating from the urban Las Vegas area; see Fig. 1. A hybrid CMB model ŽEatough et al., 2000a. was applied to the Project MOHAVE summer intensive data at Meadview and Hopi Point in the Grand Canyon, Fig. 1. This model was developed and run using a combination of regional area and point source profiles which accounted for all of the source profile species concentrations and for all of the ambient SO x and sulfate. In this paper, we have used a combination of meteorological and CMB analyses to investigate the probable source of the elevated regionally distributed SO 2 and sulfate during September 1᎐2, 1992. The meteorological data indicate that MPP could not have been the dominant source of the regionally elevated SO 2 and sulfate during this episode. The CMB model used to apportion sulfur oxides at Meadview and Hopi Point during the Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a. was run for several sampling sites throughout the region for August 31᎐September 2, 1992. The results indicate that the contribution of MPP to sulfate at Meadview was no higher than, but comparable to, that modeled during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. The CMB results also
73
indicate that the excess regional SO x emanated from the Baja California source area during this time. This led to the large increase in sulfate concentrations at Meadview on September 1 and 2. These results are consistent with those obtained for the earlier Project MOHAVE summer intensive period where the Baja California CMB source region was found to be a major contributor to both sulfate ŽEatough et al., 2000a. and visibility reduction ŽEatough et al., 2000b. in the Grand Canyon region. The justification for these conclusions is presented in this paper.
2. Methods The data used in the CMB analysis for the episode period were obtained from Project MOHAVE IMPROVE results ŽPitchford et al., 1999.. Fine particulate sulfate was determined using the IMPROVE ŽEldred et al., 1990. protocols of collection of particles on a Teflon filter for the determination of sulfur by PIXE and ion chromatographic analysis of the carbonate filter following a Teflon filter for sulfate for the determination of SO 2 Žg.. Soot was determined on Teflon filters from the laser integrated plate analysis ŽEldred et al., 1990. measurement of fine particle absorption. The various trace elements were determined from PIXE analysis of Teflon filters.
3. Synoptic meteorological description of the August 30 through September 2, 1992 regional sulfate episode The period August 30 through September 2, 1992 was characterized by a weak split flow at 500 mb, Fig. 3. This split resulted in moderately low mixing heights Ž5850 m. at the 500-mb level for this time of year. As a result of this split, the flow was predominately southwesterly to westerly at 500 mb and fairly dry. Surface dew points were typically in approximately 5⬚C across the desert southwest. Satellite, radar and surface observations showed some scattered to broken layers of clouds associated with this split flow. On August
74
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
30, cumulonimbus clouds with bases at 2700 m above ground level Žagl. were reported at Las Vegas, Nevada. Both Las Vegas and Cedar City, Utah Ž200 km north of the Grand Canyon National Park. reported heavy cloud cover for most of the day on August 30 with some light precipitation. Potentially, these meteorological conditions could lead to substantial conversion of SO 2 to particulate sulfate. However, perfluorocarbon tracer concentrations ŽocPDCH. at Meadview released from the Mohave Power Project stack ŽPitchford et al., 1999; Watson et al., 1993. indicate that only a small portion of the high Meadview sulfate concentrations could have been due to MPP SO 2 . During the tracer release period an upper limit of MPP sulfate can be obtained by multiplying ambient tracer concentrations by the ratio of SO 2 to tracer in the MPP stack. The upper limit is obtained assuming 100% conversion of SO 2 to sulfate and no gaseous or particle
depositional losses. Using these assumptions, for August 30, the MPP emissions could have contributed, at most, 0.3 grm3 of sulfate Ž3 nmolrm3 of particulate S. at any of the Project MOHAVE monitoring sites. The CMB analyses results Ždescribed in the following section. indicate that MPP was responsible for less than 0.1 g of sulfaterm 3 at Meadview on August 30. Since MPP stack releases were terminated on August 31 at 07.00 h MST, the tracer data cannot be used to place upper limits on MPP sulfate much after this time. However, eliminating the possibility of substantial MPP sulfate on August 30 is significant because this day had the best potential for substantial wet processes sulfate formation due to cloud cover. After August 30, the atmosphere dried quickly. From August 31᎐September 2 only scattered clouds were observed throughout the region with bases above 4000 m agl.
Fig. 3. Surface pressure map for 12Z, August 30, 1992 for the southwestern United States. Map shows typical summer pattern with relatively high pressure over the western United States. The semi-permanent thermal flow is over the desert southwest interior. Resulting transport flow in the boundary layer is from the southwest through southeast into this thermal low centered over the Lower Colorado River Valley. All data in Figs. 3 and 4 are from the National Weather Service. Pressures are in millibars to the nearest tenth with the last three digits of the pressure reading at each circle. Wind speeds are in knots with a full barb denoting 10 knots.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
From the surface through over 1000 m agl, there was a well developed thermal low centered over the lower Colorado River Valley, the region to the south of Las Vegas along the Colorado River, Figs. 1 and 4. This is a semi-permanent feature in the desert southwest during the summer, resulting from differential heating between the Pacific Ocean, Gulf of California and the interior deserts. Surface pressure gradients between the coastal Los Angeles Basin and interior deserts were strongly to moderately onshore on August 30᎐31, decreasing to approximately average by September 1᎐2. This pattern resulted in a typical southwesterly flow in the transport layer across the desert southwest. On August 30, based on surface pressure gradient analyses, pollutants equally exited the Cajon Pass and Beaumont Passes, Fig. 1, from the Los Angeles Basin. Material exiting the Cajon Pass tends to travel through Essex ŽFig. 1. directly to the Colorado River while material exiting Beaumont
75
Pass takes a more circuitous route through San Gorgonio ŽFig. 1. to Desert Center ŽFig. 1., where the material catches the southerly channeled Colorado River Valley winds and is transported northward. By September 2, three times as much Los Angeles Basin pollutant material was transported along the Beaumont pass route through San Gorgonio compared to the Cajon Pass path through Essex. This synoptic analysis is collaborated by the increase in sulfur oxide concentrations at San Gorgonio on September 2, as compared to the previous 2 days ŽFig. 2. and by a sharp drop in the concentrations of methylchloroform Ža marker specific to emissions from the LA Basin, Bastable et al., 1990. at Spirit Mountain Žjust east of Essex, Fig. 1. on August 31 and September 1, as compared to August 30 ŽPitchford et al., 1999.. In response to these southerly transport winds across the region, the MPP plume was also transported northward throughout this entire pe-
Fig. 4. Analysis map for 500 mb, 00Z, August 31, 1992, for the western United States. The map shows a weak split flow over the southwestern US resulting in a weak trough with limited sub-tropical moisture over the desert southwest. For each observing station, the height of the atmosphere in decameters above sea level at which the 500-mb pressure occurs is given to the right of each circle.
76
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
riod except for approximately 5 h from 05.00 h until 10.00 h MST on the morning of August 31. This brief wind direction shift on the morning of August 31 could have been in response to the passage of the upper level split flow trough. Transport winds at effective plume height Ži.e. approx. 700 m agl. were approximately 10 mrs at the beginning of the period on August 30. However, there was a gradual decrease in the transport speeds as the pressure gradients relaxed to approximately 2᎐3 mrs by September 2. Transport speeds during the 5 h of ‘plume reversal’ were approximately 2 mrs, resulting in the plume traveling no further south than approximately 40 km. Furthermore, pressure gradients resulting in southerly transport winds were three times stronger during this period compared to the periods in mid-August when MPP emissions were still transported to the north, as indicated by the ocPDCH data. Thus, MPP emissions cannot account for the high sulfur oxide concentrations at stations south of MPP, Fig. 2. Back-trajectories from Meadview for the period of September 1᎐2 were computed using the ATAD model ŽHeffter, 1980.. The ATAD model does not take terrain into account. However, ATAD back trajectory results generally agreed with more sophisticated model results during the Project MOHAVE summer intensive ŽPitchford et al., 1999. when both were run. Only ATAD model results were available for the 1᎐2 September time period. The back-trajectories showed transport from the southwest and then west, with trajectories going south of MPP to just north of Mexicali near the Mexican border, Fig. 1, and then over the Los Angeles metropolitan area. While the trajectories may be expected to reveal the general transport direction, definitive conclusions cannot be drawn from them due to inaccuracies in wind fields due to the sparseness of data used ŽGreen et al., 2000.. In particular, flows channeled along the lower Colorado River valley are not represented due to the lack of wind data used in the model runs. Thus, the back-trajectories neither confirm nor refute the CMB results Ždescribed in the following section. that sources
in northwestern Mexico contributed significantly to the ambient sulfate levels. However, the Project MOHAVE sulfur oxide data ŽFig. 2. indicate that transport from northwestern Mexico and the California Imperial Valley into the study regions did occur. High concentrations of both SO 2 and sulfate at Desert Center and Parker were observed on September 1 and 2, with relatively lower concentrations coming from the LA Basin as indicated by the San Gorgonio data ŽFig. 2.. Similar high concentrations of SO x associated with transport from northwestern Mexico were seen at other times during the 50-day summer intensive at Parker, Desert Center and El Centro. However, these episodes did not result in the very elevated concentrations of SO 2 as far north as Parker. A more typical episode is illustrated by the August 27᎐29 data ŽFig. 2. where high concentrations occur at Desert Center Žconcentrations at El Centro were twice those at Desert Center., but the increase in SO x concentrations at Parker were much smaller. Data for El Centro are not shown in Fig. 2 because no samples were collected at this site after August 30. Pollutants coming from Baja California Žnorthwestern Mexico. and the California Imperial Valley followed a typical transport path during the summer, as illustrated by the general transport of the perfluorocarbon tracer ŽPitchford et al., 1997, 2000. release from El Centro during the Project MOHAVE summer intensive ŽFig. 5.. The great majority of these pollutants are expected to originate from Mexico and not from the Imperial Valley ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001.. Pollutants were transported across the border in a northwesterly direction to a location where eventually they would catch the southerly-channeled flow in the lower Colorado River Valley. At the beginning of the episode ŽAugust 30. average transport speeds were between 5 and 12 mrs with highest speeds in the afternoon and early evening. By September 2, these regional transport speeds had decreased to approximately 3᎐6 mrs with lowest speeds at night. The regional nature of SO x throughout the study area during September 1 and 2 is seen in Fig. 2.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
77
Fig. 5. Map of the frequency that tracer was detected above background for PFT releases from El Centro Žmarked [. Žfrom Pitchford et al., 1997.. The shaded areas indicate contours over which the average concentration of the tracer was the indicated value above background concentration Ždetails are given in Pitchford et al., 1997.. The data illustrate the expected transport of pollutants from Mexico and the California Imperial Valley up the Colorado River Valley during the summer.
4. CMB analysis and results during the September 1 and 2 episode The CMB model has been used to apportion sulfur oxides at Meadview and Hopi Point during the 50-day Project MOHAVE summer intensive ŽEatough et al., 1997, 2000a.. For this 3-day August 31᎐September 2 episode, the model was run for additional receptor sites. It was applied to Project MOHAVE data from the Meadview ŽMV., Las Vegas Wash ŽLVW., Cottonwood Cove ŽCC., Dolan Springs ŽDS., Essex ŽES., Yucca ŽYC., Parker ŽPA., Wickenberg ŽWB. and Painted Desert ŽPD. sampling sites, Figs. 1 and 6. These regionally distributed sites were all included in
the CMB analysis because the Project MOHAVE data indicated they were all associated with a marked increase in SO 2 and sulfate on September 1 and 2, e.g. see Fig. 2. For most sites, the sulfate concentrations were higher than those measured during any other portion of the Project MOHAVE summer intensive. The source profiles previously developed as part of the CMB model ŽEatough et al., 1997. were used without change. The paradigms previously used for the estimation of SO 2 and sulfate deposition and conversion, and for the regional and Meadview area specific transport times were also used without change ŽDu et al., 1996; Eatough et al., 2000a.. Two of the markers, FTotal and spherical alumi-
78
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Fig. 6. Locations of the various sampling sites included in the CMB analysis for 31 August᎐2 September 1992, and the expected origins of emissions from the sources included in the analysis; MPP, Las Vegas, LA, BC, AZ and SE.
nosilicate ŽSAS. particles, used in the CMB analysis were not available for the 31 August᎐2 September event. SAS is a specific marker for emissions from a region which includes coal-fired power plant emissions. FTotal appears to originate from both industrial and agricultural regions and was particularly useful in identifying the SJ and NW2 source regions. The absence of these markers restricted the number of source regions which could be included in any single CMB estimation. Meteorological analyses were used to further guide the CMB model construction. It was assumed that emissions from MPP, Las Vegas ŽLV., Los Angeles ŽLA., Baja California ŽBC. and Arizona ŽAZ. source areas ŽFig. 6. were the only sources of sulfur oxides at all but one of the regionally analyzed sites. It was assumed that the southeast ŽSE. source region could also impact the PD site. It was further assumed that emissions from MPP were not present at sampling sites south of MPP or Dolan Springs Ži.e. at YC, ES, PR or WB, Fig. 6. and LV emissions could only be present at LVW and MV. The unavailability of the SAS particle source profile data restricted the robustness of the CMB analysis when emissions from certain point andror source region combinations were present. A sensitivity analysis was performed using the summer intensive data when SAS particle concen-
trations were available. The CMB results were adversely affected by a combination of emissions from the MPP point source, and LV and BC source regions. When emissions from these three sources were simultaneously present, the elimination of the SAS data frequently prevented CMB from separating LV and BC source signals. One or the other could be included in the fit with approximately equal ‘goodness of fit’ results, but both could not be included. This was true whether or not emissions from other source regions Že.g. LA. were present. Impacts from all other source regions were properly identified. This limitation of the CMB analysis without SAS particle data was evident in the analyses reported here for MV. No other receptor site was so adversely effected. At Meadview, this complication was due to the presence of comparable concentrations of SO x from LV and BC and high SO x from MPP. When BC emissions were present at CC, LVW and DS, the MV fit which included LV was adjusted for a BC impact consistent with the CC, LVW and DS results, i.e. it was assumed that the decrease in the BC source contribution seen at LVW as compared to CC also occurred at MV as compared to DS. As a result, the estimates of BC SO x at Meadview given in are based on an extrapolation of the CMB results at CC and DS and are not obtained directly from the Meadview data. While
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
the treatment cannot be directly validated from the data set, this approach was used because the assumption prevented an apparent overestimation of the BC source impact at Meadview. This adjustment did not significantly alter the predicted impact from MPP for any sampling period, e.g. see the sensitivity tests given in Table 3. The CMB analysis estimates of sulfur oxide emission sources present in the study region during September 2 are given in Fig. 7. Emissions from both LA and BC were generally present throughout September 1᎐2 at all sites ŽEatough et al., 1999.. The impact from BC on September 2 ŽFig. 7. was greater than from LA for both SO 2 and sulfate. However, the relative importance of emissions from LA increased from Essex northward, consistent with the expected transport path from the LA region. Similar trends were seen on September 1 ŽEatough et al., 1999. The highest
79
concentrations of impacts from BC were found at Parker, Essex and Yucca. The highest impacts from LA were observed at Essex and Dolan springs. High impacts for SO 2 Žbut not sulfate . from MPP were observed at Dolan Springs, Las Vegas Wash and Meadview. Significant impacts from LV were found at Las Vegas Wash and Meadview. The site-to-site and day-to-day changes ŽEatough et al., 1999. in sulfate and SO 2 were consistent with a southerly flow, resulting in generally decreasing BC and LA SO x concentrations as pollutants traveled north and east, but with an increasing sulfate to SO x ratio for the SO x from the BC and LA source areas throughout the region. The increase in the sulfate to SO x ratio results from both the conversion of SO 2 to sulfate and the faster deposition of SO 2 , as compared to sulfate during emission transport ŽDu et al., 1996; Eatough et al., 2000a..
Fig. 7. Results of the CMB analysis for SO x and sulfate Žnmolrm3 . on 2 September 1992 at the sampling sites identified in Figs. 1 and 6 for the sources identified in Fig. 6.
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
80 Table 2 Meadview sulfate source September 1992,grm3 Date
MPP
LV
apportionment, 31
LA
BC
August᎐2
Sulfate Not Fit, % Žgrm3 .
31 Aug 1 Sep 2 Sep
0.3 0.3 0.6
0.3 1.4 0.8
0.2 0.4 0.6
1.1 0.9 1.8
4.2 Ž0.1. 8.4 Ž0.2. 14.8 Ž0.6.
Based on this analysis, the sulfate impacts at Meadview from all sources for each sampling day are given in Table 2. The results of sensitivity tests on the effect of changing the CMB MV assumptions are given in Table 3. The highest concentrations of sulfate from MPP were calculated to be 0.69 grm3 at Dolan Springs on 1 September and 0.55 grm3 at Meadview on 2 September. All other estimated MPP sulfate impacts were less than 0.4 grm3. In contrast, the highest estimated sulfate impact at Meadview from any source was 1.8 grm3 from BC on September 2.
5. Results from the Project MOHAVE summer intensive The results obtained during this episode were consistent with the importance of the BC source in the Grand Canyon region during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. Fig. 8 displays the total daily sulfate at the Hopi Point sampling site in the Grand Canyon and the daily sulfate attributed to the BC source Table 3 Sensitivity to changes in the assumption of the relative amounts of SO x at Meadview from LV and BC on MPP sulfate, grm3 Date
31 Aug 1 Sep 2 Sep a
MPP Sulfate SOx from both BC and LVa
SOx from LV but not BC
SOx from BC but not LV
0.3 0.3 0.6
0.5 0.4 0.6
0.3 0.3 0.9
Most probable scenario.
Fig. 8. Total and BC sulfate at GC. Northern Mexico is often a major contributor to sulfate at the Grand Canyon.
region during the Project MOHAVE summer intensive. Similar trends were seen at Meadview, but the relative impact of BC was less pronounced due to the importance of LV sulfate at MV. Sulfate from the BC region was frequently present at Hopi Point. When emissions from this source region were present, they often accounted for the majority of the sulfate. While concentrations as high as the 1.8-g sulfaterm 3 level at Meadview on September 2 were not seen at Hopi Point during the summer intensive, concentrations from 0.5 to 1.5 g sulfaterm 3 were frequently seen. The CMB prediction of BC SO x at Meadview and Hopi Point always coincided with the presence of the fluorocarbon tracer released from El Centro ŽEatough et al., 2000a.. The BC region accounted for an average of 42% of the sulfate present at Hopi Point during the Project MOHAVE summer intensive ŽEatough et al., 2000a.. During the July 12 through August 30 summer intensive, the highest concentration of BC sulfate at Meadview was 0.8 grm3, with the average concentration of BC sulfate at Meadview being 0.3 grm3. In contrast, the highest concentration of BC sulfate at Meadview during the days just following the end of the summer intensive was 1.8 grm3 and averaged 1.3 grm3 , a factor four times higher. During the summer intensive, the average contribution of BC to SO x and sulfate at Meadview was 6% and 12% of the total, respectively. However, during the 3-day episode ŽAugust 31᎐September 2., BC contributed an average of
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
19% of the SO x and 44% of the sulfate. The LA source area contributed 18% of the sulfate during the summer intensive and a comparable 14% of the total during this 3-day episode. The relative ratios of SO 2 and sulfate present from the BC source is comparable to that seen during the summer intensive. Thus, the conversion of SO 2 to sulfate was normal during this episode period. This result is consistent with the dry condition of the lower boundary layer of the atmosphere where the pollutants were located during the August 31 through September 2 time period. Thus, the 6-year record high concentrations of sulfate observed at Meadview during August 31᎐September 2, 1992 can be attributed to a significant increase in transport of material from northwestern Mexico to Meadview. What remains an open question is whether there was an increase in SO x emissions from BC andror was the meteorological transport particularly conducive to transport of BC emissions during this period. However, since we do know that MPP emissions are ‘base loaded,’ that there was no stagnation, and that desert atmospheric conditions were conducive to only slow gas-phase conversion rates, MPP could only be contributing its ‘normal’ amount of sulfate, consistent with the CMB results. BC was probably also the largest contributor to extinction in the Grand Canyon during the summer intensive. Analysis performed elsewhere ŽEatough et al., 2000b. demonstrated that BC is a significant haze contributor in this region because of the frequency of the presence of pollutants from this source and overall magnitude of its pollutant components. Thus, combination of the CMB and light extinction data obtained during Project MOHAVE suggest that emissions from northwestern Mexico were the largest single urban area or point anthropogenic source of visibility impairing particulate material at Hopi Point in the Grand Canyon during Project MOHAVE.
6. Conclusions An analysis of available source information ŽChow et al., 2000; Chow and Watson, 2001;
81
Watson and Chow, 2001., meteorological patterns, CMB results and distributions of SO x and sulfate throughout the region are all consistent with the high concentrations of sulfate observed at Meadview on September 1 and 2, 1992 emanating from sources in northwestern Mexico. Similar results were obtained from the CMB analysis of all Project MOHAVE summer intensive data. The CMB results for the July and August Project MOHAVE summer intensive period showed that the BC source area was a frequent contributor to sulfate at Meadview and Hopi Point. During the Project MOHAVE summer intensive, the BC source was the main contributor to sulfate at Hopi Point in the Grand Canyon. Particles from the BC source area were also the main anthropogenic contributor to visibility reduction at Hopi Point during this period ŽEatough et al., 2000b.. Similar transport of SO x from Mexico into the Big Bend region in Texas has also been reported ŽGebhart et al., 2000a,b.. This transport, however, included significant point source emissions, as contrasted with the probable smaller, but greater in number, SO x sources in Northwestern Mexico ŽChow et al., 2000; Chow and Watson, 2001; Watson and Chow, 2001..
Acknowledgements Support for this work was provided by Southern California Edison to participate in Project MOHAVE. However, the results, findings, and conclusions expressed in this paper are solely those of its authors and are not necessarily endorsed by the management and sponsors of Project MOHAVE. References Bastable H, Rogers D, Schorran D. Tracers of opportunity and pollutant transport in southern California. Atmos Environ 1990;24:137᎐151. Du AY, Mangelson NF, Eatough DJ, Farber RJ, Watson JG. The deposition of SO 2 Žg. and particulate sulfate in the desert southwest. Proceedings Air & Waste Management Association Annual Meeting, 96-MP1A.04, 1996:1᎐14.
82
D.J. Eatough et al. r The Science of the Total En¨ ironment 276 (2001) 69᎐82
Chow JC, Watson JG. Zones of representation for PM 10 measurements along the USrMexico borders. Sci Total Environ 2001;276:49᎐68. Chow JC, Watson JG, Green MC, Lowenthal DH, Bates B, Oslund W, Torres G. Cross-border transport and spatial variability of suspended particles in Mexicali and California’s Imperial Valley. Atmos Environ 2000;34:1833᎐1843. Eatough DJ, Eatough M, Eatough NL, Lewis EA. Apportionment of sulfur oxides at Canyonlands during the winter of 1990. III. Source apportionment of SO x and sulfate and the conversion of SO 2 to sulfate in the Green River Basin. Atmos Environ 1996;30:295᎐308. Eatough DJ, Du AY, Joseph JM, Caka FM, Sun B, Lewis L, Mangelson NF, Eatough M, Eatough NL, Farber RJ, Watson JG. Regional source profiles of sources of SO x at the Grand Canyon during Project MOHAVE. J Air Waste Manage Assoc 1997;47:101᎐118. Eatough DJ, Green M, Moran W, Farber R. A case study of particulate impacts on the Grand Canyon from northern Mexico. Proceedings, 92nd Annual A& WMA Meeting Paper No 99-497, 1999. Eatough DJ, Farber RJ, Watson JG. Second generation chemical mass balance source apportionment of sulfur oxides and sulfate at the Grand Canyon during the Project summer intensive. J Air Waste Manage Assoc 2000a; 50:759᎐774. Eatough DJ, Cui W, Hull J. Fine particulate chemical composition and light extinction at Meadview, Arizona Žin preparation. 2000b. Eldred RA, Cahill TA, Wilkinson LK, Feeney PJ, Chow JC, Malm WC. Measurement of fine particles and their chemical components in the IMPROVErNPS networks. In: Mathai CV, editor. Transactions: visibility and fine particles. Pittsburgh, PA: Air & Waste Management Association, 1990:187᎐196.
Gebhart KA, Malm WC, Flores M. A preliminary look at source᎐receptor relationships in the Texas᎐Mexico border area. J Air Waste Manage Assoc 2000a;50:759᎐774. Gebhart KA, Kriedenweis SM, Malm WC. Visibility at Big Bend National Park: a summary of source attribution analyses of the historical database and the 1996 scoping study. Sci Total Environ 2000b:276. Green, MC, Pai P, Ashbaugh L, Farber RJ. Evaluation of wind fields used in the Grand Canyon Visibility Transport Commission analyses. J Air Waste Manage Assoc 2000;50:809᎐817. Heffter JL. Air Resources Laboratories Atmospheric Transport and Dispersion Model ŽARL-ATAD., Technical Memorandum ERL ARL-81, NOAA, Rockville, 1980. Pitchford M, Green M, Kuhns H, Farber R. Characterization of regional transport dispersion using Project MOHAVE tracer data. Visual air quality: aerosols and global radiation balance. Air Waste Manage Assoc VIP 1997;76Ž1.:181᎐200. Pitchford M, et al. Ž1999.. Project MOHAVE Final Report. Internet version Žwww.epa.govrregion09rairrmohaverreport.html.. Pitchford M, Green M, Kuhns H, Farber R. Characterization of regional transport and dispersion using Project MOHAVE tracer data. J Air Waste Manage Assoc 2000;50:733᎐745. Watson JG, Green M, Hoffer TE, Lawson DR, Eatough DJ, Farber RJ, Malm WC, McDade CE, Pitchford M. Project MOHAVE data analysis plan, Proceedings, 86th Annual A& WMA Meeting 1993, Paper No. 93-MP-4.10, 1993. Watson JG, Chow JC. Source characterization of major emission sources in the Imperial and Mexicali Valleys along the U.S.rMexico border. Sci Total Environ 2001;276:33᎐47.