Volatile organic compounds in the downtown area of Mexicali, México during the spring of 2005: analysis of ambient data and source–receptor modeling

 

A. MENDOZA
Departamento de Ingeniería Química, Tecnológico de Monterrey Ave. Eugenio Garza Sada 2501, Monterrey, N. L., México 64849.
Corresponding author; e–mail: mendoza.alberto@itesm.mx

A. A. GUTIÉRREZ, E. I. PARDO
Centro de Calidad Ambiental, Tecnológico de Monterrey Ave. Eugenio Garza Sada 2501, Monterrey, N. L., México 64849.

 

Received May 2, 2008; Accepted December 10, 2008

 

RESUMEN

La calidad del aire de la ciudad fronteriza de Mexicali no es satisfactoria, principalmente debido a los altos niveles de partículas suspendidas (PS) finas y ozono (O₃) que se registran ahí. Esto también influye en la calidad del aire de las ciudades del Valle Imperial en California. En abril de 2005 se llevó a cabo una campaña de monitoreo para obtener concentraciones de compuestos orgánicos volátiles (COV) en aire ambiente y así obtener información de los niveles y posibles fuentes de estos precursores de O₃ y PS finas. Contenedores de seis litros (canisters) se emplearon para tal fin en un sitio localizado en la región centro de la ciudad, para posteriormente analizar las muestras y obtener concentraciones de 54 especies preseleccionadas. Las muestras se tomaron diariamente durante tres períodos de tiempo: 6–9, 13–16 y 20–23 h tiempo local. La concentración promedio de compuestos orgánicos no metánicos totales (CONMT) fue de 560 ppbC, siendo las muestras de la mañana las que reportaron los mayores valores promedio (795) y las muestras del mediodía los menores valores (257 ppbC). Los compuestos aromáticos contribuyeron con alrededor de 24% del valor de CONMT reportado. Correlaciones entre especies indican una importante influencia de fuentes antrópicas locales en las mediciones (p. ej., una proporción de 2.1 entre xilenos y benceno). Adicionalmente, una alta correlación entre productos típicamente emitidos por motores de combustión interna (p. ej., acetileno–etileno R² = 0.90, etileno–propileno R² = 0.94, benceno–etileno R² = 0.75) y una contribución consistente de más del 15% de estas especies (etileno, acetileno, propileno y benceno) indican un importante aporte de fuentes móviles. Esto se comprueba a través de la aplicación de un modelo receptor, el Chemical Mass Balance, a los datos observados. Las estimaciones indican que las fuentes móviles a gasolina contribuyen con un 56% del nivel de COV observado, mientras que las emisiones de gas LP contribuyen con un 18, las emisiones de escape de vehículos diesel con un 6 y productos generales de consumo con un 5.

 

ABSTRACT

The border city of Mexicali has some of the worst air pollution in México, mainly due to the high levels of fine particulate matter (PM) and O₃ registered. This also impacts the air quality of communities across the border in California's Imperial Valley. A field campaign was conducted in April of 2005 to obtain ambient concentrations of volatile organic compounds (VOC), and thus obtain insight on the levels and sources of these O₃ and fine PM precursors. Six–liter stainless steel canisters were deployed at a site located in the downtown area of the city to obtain ambient air samples that were then analyzed for 54 selected target species. Samples were collected on a daily basis at three different sampling times: 6–9, 13–16, and 20–23 h LT. Average total non–methane organic compounds (TNMOC) concentration was 560 ppbC, with the morning samples having the highest reported average concentration (795) and the afternoon the lowest (257 ppbC). Aromatics contributed around 24% to the TNMOC concentration. Correlation between different species indicates strong influence of local anthropogenic sources (e.g., xylenes–benzene ratio of 2.1). In addition, high correlations between typical products from internal combustion engines (e.g., acetylene–ethene R² = 0.90, ethene–propylene R² = 0.94, benzene–ethene R² = 0.75) across all samples and a consistent contribution of more than 15% of these species (ethene, acetylene, propylene and benzene) indicate major contribution from mobile sources. This is supported by receptor modeling results obtained through the application of the Chemical Mass Balance model to the ambient data. Source apportionment estimates indicate a 56% contribution of gasoline–related mobile source emissions to the VOC that were measured, 18 contribution from LPG emissions, 6 from diesel exhaust, and 5 from consumer products.

Keywords: Air pollution, emission sources, ambient air monitoring, Chemical Mass Balance, US–México border region.

 

1. Introduction

Volatile organic compounds (VOC) are typically abundant in urban atmospheres. The main anthropogenic sources of these chemical species are emissions from mobile sources, fugitive gasoline emissions, chemical and petrochemical industry, gaseous fuels leaks, and in general operations where solvents are used (e.g., architectonic coatings, industrial and commercial painting processes, printing processes, dry cleaning, degreasing processes, graphic arts, etc.). VOC can have a variety of harmful health effects. At high exposure levels, several VOC can cause central nervous system depression (e.g., drowsiness, stupor) (Maroni et al., 1995), and can be irritating upon contact with the skin (Hodgson et al., 1991) or to the mucus membranes if inhaled (Mølhave, 1991), while other are proven carcinogenic substances (Jones, 1999). In addition to the health risks, VOC contribute to ozone and fine particulate matter formation in the troposphere (Russell et al., 1995; Robinson et al., 2007). In the oxidation process of VOC that leads to tropospheric ozone formation, semi–volatile organic compounds are formed which then can condensate onto preexisting aerosols. This process is relevant since the hygroscopic properties of such aerosols change as the ratios of inorganic, organic and inert species present in the individual aerosols change, with corresponding impacts on the optical and radiative–forcing properties of the polluted air (Kanakidou et al., 2005).

The city of Mexicali, Baja California, is located in the western United States–México border (Fig. 1). Mexicali is a very dynamic city, with intense industrial and commercial activities. From a more regional perspective, Mexicali is located in the valley of Mexicali, adjacent to California's Imperial Valley, where the cities of Calexico, El Centro and Brawley are located. This region as a whole (Imperial Valley and the Valley of Mexicali) is also characterized by intense agricultural activity. Its location, coupled with an ongoing increase in population, and in industrial and commercial facilities, makes of the Mexicali–Imperial Valley border area a key region in the national as well as the international economic arenas. Currently, this border region is home to more than 1 million people, most of them living in the Mexican side of the border (Quintero Núñez et al., 2006). However, this growth has derived in important environmental problems for Mexicali. The city has had some of the worst air pollution in the border region of México and this also impacts the air quality of communities across the border in California's Imperial Valley (INE, 1999; Quintero Núñez et al., 2006). In a recent report from México's Instituto Nacional de Ecología, Mexicali has been placed as the most polluted city in the country with respect to PM10 and CO, and the sixth with respect to O3 (Zuk et al., 2007). In addition, Mendoza et al. (2006) found that PM2.5 levels in Mexicali can be also quite high, with a strong organic carbon component.

In the city of Mexicali the prevailing weather, as well as for most of the valley, is very arid. Typical conditions include clear skies, low humidity, extremely hot summers, mild winters, and low precipitation usually occurring during the winter and summer months (GEBC, 1999). These conditions are determined to a great extent by the presence of a semi–permanent subtropical cyclonic system in the Pacific Ocean near the California coast and the topography of the valley. Near–sea level altitude and a mountain range to the west of the valley acts as a shadow to the humid air flow from the ocean. During the spring months, the area is influenced by high pressure cells that foster stagnant wind conditions, which provoke the accumulation of air pollutants in the urban area (Vanoye García, 2007). Ground–level wind–flow patterns are dictated by the topographic orientation of the valley, which is northwest to southeast. About two–thirds of the time winds are from the northwest and one–third from the southeast; rarely winds are from other directions (GEBC, 1999). A past study also found a 2 m/s wind–speed threshold below which wind direction tended to be erratic (with the corresponding accumulation of pollutants), and above which pollutants dispersion was favored (Chow et al., 2000).

Past air quality studies in Mexicali includes a VOC monitoring campaign during the summer of 1997 as part of the Southern California Ozone Study (SCOS97; Zielinska et al., 1999). The main goal of that study was to establish boundary conditions for the regional air quality modeling domain set for SCOS97, and provide some information on VOC levels in the main cities of the California–México border region. A key finding of the ambient data analysis includes a large contribution from mobile sources to the total VOC observed. A second monitoring campaign was conducted during the summer of 1999, and the data collected was used for more detailed source apportionment calculations (Fujita et al., 2004). Results of this second study indicate a contribution of nearly 40% of gasoline vehicles to the total ambient non–methane hydrocarbons, followed by natural gas fugitive emissions, propane combustion and diesel exhaust. Solvent emissions from surface coatings and biogenic emissions were generally insignificant. The present study provides updated information on the levels of VOC in the ambient air of downtown Mexicali during a spring episode, and an analysis on the main sources that contribute to the observed concentrations through receptor modeling. The information is relevant to address the impacts of the growth of the city on air pollution, asses eventually the effectiveness of the air quality improvement programs set forth for the region by the state of Baja California (GEBC, 1999) and the Border 2012 Program (Quintero–Núñez and Sweedler, 2004), and help decision–makers orient further emission control policies for the region.

 

2. Experimental and modeling methods

2.1 Sampling campaign

Ambient air samples were collected at the Universidad Autónoma de Baja California (UABC) site from the routine air quality monitoring network of the city of Mexicali (Fig. 2). The UABC site is located within a University campus. The samples were obtained at the same location where the rest of the instruments of the site grab their samples: on top of a two–story building, mainly surrounded by open spaces, a few buildings and a small parking lot. The area where the UABC is located is predominantly residential, although several commercial facilities are located in the district.

A field campaign was conducted in April of 2005 to obtain ambient concentrations of VOC. Sampling activities started April 13 and ended April 22. Samples were collected on a daily basis at three different sampling times: 6–9, 13–16 and 20–23 h LT. The monitoring periods were selected to capture early morning, fresh VOC emissions, afternoon aged air masses, and early night emissions. The samples were collected in six–liter stainless steel SUMMA®–polished canisters. The sampling train included the following components, all constructed of stainless steel: a sampling inlet, a sintered metal particle filter, a critical orifice, a mass–flow controlling device calibrated for 3–hour sampling periods, a vacuum gauge and a canister. In addition, the sampling inlet was connected to a 1.5 m teflon sampling line to elevate the sampling point to a distance above the building top similar to the sampling points of the rest of the station monitoring devices. Prior to sampling, a leak test was conducted to insure proper connection between the components of the sampling train. A total of 33 valid samples were collected during the monitoring campaign.

The chemical analysis of the samples was based on methods contained in the United States Environmental Protection Agency's (US EPA) technical guidance documents (US EPA, 1998) using GC/dual capillary columns and dual FID, and was performed by Air Toxics LTD (Folsom, California). The method involves cryogenic cooling of the ambient air samples to obtain up to 50 mL of a condensate. After concentrating the sample, the concentrated aliquot was flash vaporized and swept trough a water management system to remove water vapor. Following dehumidification, the sample was passed directly in to the GC system for analysis. Fifty–four target species were selected for analysis (Table I). The list resembles the Photochemical Assessment Monitoring Stations (PAMS) program target list of the US EPA. Each analysis was reported in ppbC using propane as the calibrate. Total non–methane organic compounds (TNMOC) were determined by summing the area count of all target and non–target peaks in the C2 through C12 range and applying the response factor of propane. Quality assurance/quality control specifications indicated by the analytical laboratory included 100% of laboratory blanks and 10% sample duplicates. Reporting limit was between 10 and 20 ppbC.

 

2.2 Source–receptor modeling

Receptor models are mathematical procedures for identifying and quantifying the contribution of sources of air pollutants to a given receptor location. Receptor models do not use pollutant emissions, meteorological data or chemical transformation mechanisms to estimate the contribution of sources to receptor concentrations as chemical–transport models do (Russell and Dennis, 2000). Instead, receptor models use chemical and physical characteristics of gases and particles measured at the source and receptor to both identify the presence of and to quantify source contributions to receptor concentrations (Watson, 1984).

In this study, the Chemical Mass Balance (CMB) receptor model version 8 (US EPA, 2004), was employed to obtain source contribution estimates to the VOC ambient air concentrations observed at the UABC monitoring site. CMB has been described in the literature (e.g., Cooper et al., 1984; Gordon, 1988) and has been used extensively in source apportionment studies that involve VOC species (Watson et al., 2001). Briefly, CMB solves a set of linear equations that express the chemical concentration at each receptor site as a linear combination of the product of source profile abundances and source contributions. Thus, known concentrations of component i (e.g., a given VOC species), C_i, will be:

where F_ij are the known source composition factors (source profiles) of species i in source j, and S_j the unknown contribution of source j. The number of chemical species (I) must be greater than or equal to the number of sources (J) for a unique solution to these equations. CMB uses the effective variance weighted solution, which can be superior to the classic weighted least–squares, and has been described by Cooper et al. (1984) and Henry (1992). This technique provides realistic estimates of the uncertainties of the source contributions (owing to its incorporation of both source profile and receptor data uncertainties) and gives greater influence to chemical species with smaller values of uncertainty in both the source and receptor measurements than to species with higher values of uncertainty.

Source profiles employed in this study were the ones derived and reported by Fujita et al. (2004) for a VOC source apportionment study conducted for the same region in the '90s. Source profiles database was enhanced by profiles derived by others for California (Censullo, 1996; Fujita et al., 2002) and México City (Mugica et al., 1998) CMB applications. One–sigma uncertainties of the original data were derived from variations among multiple measurements for a particular source type or a nominal analytical uncertainty of 15%. The assigned uncertainties were the larger of the two values based on recommendations by Fujita et al. (2004). Table II lists the source profiles that were used for source apportionment in this study with a brief description and assigned mnemonics. Categories of source profiles include gasoline and diesel exhaust, liquid and vapor gasoline, commercial natural gas and liquefied petroleum gas, surface coatings, consumer products, and isoprene as a biogenic marker. The actual source composition profiles used are presented in Table III.

 

3. Results

3.1 TNMOC Results

Results of the campaign show an average TNMOC concentration of 560 ppbC (standard deviation = 489 ppbC). Table IV presents the basic statistics for the TNMOC concentrations measured during the three monitoring periods: morning, afternoon and night. Putting into context the level of VOC present in Mexicali's ambient air, morning average TNMOC concentration obtained in Mexicali during the campaign was approximately half of the average reported morning (also 6–9 h samples) total non–methane hydrocarbons concentration reported for México City during a spring of 2003 campaign (Molina et al., 2007).

Results indicate that, in general, TNMOC concentrations were higher during the morning hours, followed by the nighttime and lower during the afternoon. The variation of individual chemical species tended to follow the same pattern. From a temporal perspective, Figure 3 presents the variation of the morning, afternoon and nighttime TNMOC concentrations and wind conditions (speed and direction) during the whole monitoring campaign. A valley on the TNMOC time series plot between the night of April 16 and the morning of April 20 (Saturday through Wednesday) indicates an episode of prevailing relatively low concentrations. Of note, the sample taken during the nighttime period of April 19 was invalidated due to sampling problems. Meteorological conditions during this period included winds that were on average higher than those of the previous and latter days. This condition could have favored lower concentrations by diluting the pollutants at a rate higher than the one at which they were being emitted. Figure 4 presents a scatter plot of TNMOC versus wind speed data. It can be observed that a threshold exists between wind speeds of 1 and 2 m/s above which concentrations fall below 500 ppbC; the great majority of these events occurred precisely during the April 16–20 period (Fig. 4). These findings are in line with the ~2 m/s wind speed threshold reported by Chow et al. (2000) and described earlier. Wind direction during the first part of the referred period was dominated by sustained southeasterly winds, while the latter part was dominated by sustained northwesterly winds. Less stable wind direction behavior was observed during the other days where winds were generally below 1 m/s. This is also in line with the conditions mentioned earlier that favor pollutants accumulation.

Given that the low–TNMOC period occurred during a weekend–weekday transition, it is possible that traffic changes could have influenced the observed concentrations. A traffic study conducted in Mexicali during June 2004 (Tejeda–Honstein et al., 2005) reported that, considering the 22 most important roadways in Mexicali, any given day between Monday and Friday represents 14.6% ± 0.2% of the total weekly traffic. Traffic on Saturdays and Sundays averaged 14.5 and 12.4% of the total weekly traffic, respectively. Thus, it is not expected that weekend–weekday traffic differences could be a major factor influencing the observed drastic changes in VOC concentrations. In addition, Tejeda–Honstein et al. (2005) also report a rather constant hourly traffic density profile from ~8 to ~20 h, with highest traffic occurring between 13 and 19 h. Thus daily variation in the results appears to be, again, mainly due to meteorological factors rather than possible changes in traffic.

 

3.2 Individual species results

From a species–wise perspective, of the 54 species analyzed only 27 were detected (including TNMOC) in the ambient air samples, and in particular during the afternoon periods only four species were detected: propane, isopentane, 3–methyl hexane and styrene. Figure 5 shows the concentration distribution of individual organic compounds observed during the morning and night sampling periods; the afternoon period is not presented due to the small number of species detected. The same trend, observed for the concentration levels of the TNMOC data, was observed for the individual species and is not presented to be brief: high levels detected for the morning period, while the afternoon concentrations were the lowest detected. The species consistently with the highest concentration (on a ppbC basis) were propane, isopentane, toluene, and m,p–xylene. In addition, the aromatic species, mainly benzene, toluene, ethylbenzene (BTEX), xylenes and 1,2,4–trimethylbenzene, contributed approximately 24% of the total average of TNMOC concentration.

Vehicles fueled by liquid petroleum gas (LPG) are known to represent an important source of propane emissions (Duffy and Nelson, 1996; Na et al., 2004). However LPG leaks in transportation lines, refueling and containers, as well as commercial and domestic consumption, have proven to be also a considerable source of propane in Mexican cities (Blake and Rowland, 1995; Gamas et al., 2000; Shifter et al., 2000). These last sources can be thought to be major contributors to the high propane concentrations observed, given that LPG is not used extensively in Mexicali's vehicular fleet. Regarding isopentane, it is a major component of gasoline and even though it is present in exhaust emissions, the main source of isopentane in TNMOC ambient concentrations can be related to gasoline vapors (McLaren et al., 1996).

 

3.3 Correlation between species

A normal correlation analysis between individual species was conducted with the purpose of identifying, in a preliminary way, the sources of the pollutants identified. For that matter, species were classified in two groups: aromatic and aliphatic. For each case, the correlations between TNMOC and individual species were also considered. Correlation coefficients (R²) were evaluated into three correlation ranges: good (0.80 < R² < 1.0), moderate (0.50 < R² < 0.80), and weak (R² < 0.50) (Devore, 2000). The strongest correlation between aromatic species was obtained for the xylenes (Table V). Moderate correlations were obtained for toluene and the xylenes, and benzene and 1,2,4–trimethylbenzene. The rest of the correlations for this chemical family were weak. The correlations between TNMOC and the aromatics indicate moderate correlations for toluene, the xylenes and ethylbenzene, while for 1,2,4–trimethylbenzene and benzene the correlation was weak.

Correlation coefficients between aliphatic species are shown in Table VI. The species that present the stronger correlations are those associated typically with emissions from internal combustion engines. For example, the correlations between acetylene and ethene (R² = 0.90) and ethene and propylene (R² = 0.94) indicate a high probability that the origin of these species could be the same and could be attributed to exhaust of gasoline vehicles (Barletta et al., 2002; Chan et al., 2002; Na et al., 2004). Other correlations that support this hypothesis are the ones between aromatic and aliphatic species, e.g. benzene–ethene (R² = 0.75) and acetylene–benzene (R² = 0.84) (Zielinska et al., 1999). Even more, a consistent contribution of more than 15% of these mobile emissions tracers (i.e., ethene, acetylene, propylene and benzene) to the TNMOC concentration indicates major contribution from mobile sources.

 

3.4 Source–receptor modeling

CMB was applied to valid ambient VOC samples that were obtained at the UABC monitoring site in Mexicali. VOC concentrations obtained in the field campaign were reported by the analytical laboratory in ppbC, and were converted to µg/m³ at local temperature and pressure conditions during the sampling period to be consistent with the source profiles used.

When using receptor models, the relative proportions of chemical species must change little between the source and the receptor, otherwise, this fundamental assumption in the formulation of the receptor modeling would be violated. VOC are typically oxidized in the lowest two kilometers of the troposphere and can have lifetimes ranging from minutes to several months. Thus, in CMB calculations only species with lifetimes in the atmosphere greater than that of toluene (~9 hours during the summer) are recommended as fitting species in samples from urban source areas (Fujita et al., 2004). An exception to this is isoprene. This species is included as a fitting species despite its high reactivity because it serves as a marker for biogenic emissions. Selection of fitting species in our application was based on recommendations by Fujita et al. (2004). For the particular case of the afternoon samples, no source apportionment was attempted since not enough data were available (i.e., not enough species were identified in the samples to apply the CMB model).

Acceptable fits for the source profiles used were achieved for only 13 samples: seven morning samples (April 13–16, 18, 21 and 22) and six nighttime samples (April 13–15, 17, 21 and 22). The other 20 samples did not have enough species reported to complete the requirements of the receptor model and calculate the source contributions. Figures 6 and 7 show the source contributions to TNMOC estimated in this study for the morning and night periods, respectively, in percents of the measured concentration. The most significant source for the morning period was gasoline mobile sources (exhaust emissions plus fugitive gasoline emissions); it represents, on average, 42% of the contributions for the morning period, except for April 18 (a Monday) where the most important source were consumer products. For the night period, the most significant source was again gasoline mobile sources (~70% average contribution). However, it is interesting to note that for the morning samples CMB got a better fit using Mexicali–derived profiles (exhaust, fugitive and composite emission profiles), while for the night samples CMB got a better fit using US–derived profiles. Only in one day (April 13), CMB got a better fit in the night sample using the profile obtained from Mexican vehicle exhausts. The "Mexican gasoline vehicle exhaust" profile (MEXI_GEX) was derived from '90s fleet. The result might be an indication that the current fleet has modernized and has become similar to the US fleet from the '90s. Another aspect to note is that on April 17, a Sunday, CMB estimates that the main source affecting the site (~50% contribution) was LPG fugitive emissions. This can be an indication of the strong commercial activity from restaurants and other food establishments during that day, as well as a decrease in traffic. In addition, recall that during the April 17–20 period rather low VOC concentrations were obtained; some of the day–to–day variation observed during this period could be a result of such condition.

 

4. Discussion

4.1 VOC observations

Plausible origins of the VOC detected in monitoring site can be further explored analyzing the relationship between VOC and CO concentrations, since CO can typically be associated with mobile emissions (exhaust emissions; Barletta et al., 2002). A moderate correlation (R² = 0.66) was found between average CO and TNMOC concentrations (Fig. 8). CO concentrations used were those registered also at the UABC site, and the average periods were the same as the VOC sampling periods used. Higher correlations were obtained for the nighttime data (R² = 0.93) and the afternoon data (R² = 0.75). Correlation for the morning data resulted in R2 = 0.61. This is an indication that pollutants sampled at the UABC site have a high probability of having their source in combustion processes, and particularly, mobile sources, across all times.

Vehicle exhausts typically contain aromatic compounds, in particular benzene, toluene and xylenes. However, toluene and xylenes are also found in solvent emissions while benzene is not (Na et al., 2004, 2005). This explains the moderate correlations observed among benzene and the rest of the BTEX species, i.e. toluene and xylene emissions cannot be completely attributed to mobile sources. Nonetheless, the average ratio between benzene (B) and toluene (T) concentrations can give further insight into the characteristics of the main source of pollutants at the site. B:T ratios of 0.41 and 0.42 for the morning and night periods, respectively, were obtained. With this, it can be inferred again that vehicle emissions, and in particular those from non–catalytic vehicles, were the major sources of TNMOC in Mexicali (Barletta et al., 2002; Chan et al., 2002). The ethene–acetylene ratio is also a well–known indicator to distinguish between emissions from catalytic and non–catalytic converter vehicles. Emissions from non–catalytic converter units usually have an ethene–acetylene ratio close to unity, while catalyst–equipped vehicles have a ratio of 3 or greater (Duffy et al., 1996; Sagebiel et al., 1996; Barletta et al., 2002). For the ambient air samples obtained in this study, ethene to acetylene ratios of 0.82 and 0.79 were obtained for the morning and night periods, respectively, confirming that non–catalytic converter vehicles constitute the main TNMOC sources in Mexicali. This is in line with the fact that Mexicali's vehicular fleet is quite old and that there is evidence that a majority of the vehicles have tampered pollution control devices. For example, in the study of GEBC (1999) it was detected that 95% of the vehicles were from 1990 or older.

Of particular interest is also to compare the results obtained here with those reported in past studies were VOC concentrations and possible sources were also analyzed at the same monitoring location. Zielinska et al. (1999) analyzed VOC obtained from ambient air samples in Mexicali during a field campaign in the summer of 1997, and also concluded that motor vehicle emissions were the main source of these pollutants. Fujita et al. (2004) conducted a field campaign in the summer of 1997, used source apportionment techniques to analyze the data, and concluded that gasoline vehicles were consistently the largest contributor to ambient total non–methane hydrocarbons (TNMHC). Table VII shows a comparison of pollutant concentrations in Mexicali between the three studies. It can be noted that the ethene–acetylene ratio has increased from 0.62 in 1997 to 0.82 in 2005. This can be an indication of a slight decline in the relative contribution of non–catalytic converter vehicles to the total emissions. Also, consistently the three studies report a xylenes–benzene ratio close or greater than 2, which indicates the influence of fresh emissions at the site (Roberts et al., 1997). Thus, results suggest that the main source of VOC in downtown Mexicali continues to be gasoline–powered vehicles from nearby sources.

Finally, a correlation analysis between early morning TNMOC measurements and O₃ concentrations can give some insight on the influence of the VOC emissions in the generation of this secondary species. In general, relatively weak to weak correlations between early morning TNMOC measurements at the UABC site and O₃ at the same site and at a monitoring site in Calexico, CA were obtained. The highest correlations were for average O3 at UABC (R2 = 0.49) and peak O₃ at Calexico (R² = 0.37) (Fig. 9). TNMOC versus peak ozone at UABC yielded an R² of 0.21, and for average O₃ at UABC during the afternoon monitoring period yielded an R² of 0.25. TNMOC at UABC versus average afternoon O₃ at Calexico resulted in an R² of 0.30. These results imply that TNMOC emitted in downtown Mexicali contribute to the chronic pollution of O₃ in Mexicali, and are one of the possible causes of ozone peaks in Calexico which is far enough for the pollutants to react. A detailed source–receptor analysis is needed to determine more accurately the impact of the transport of these emissions.

 

4.2 Source–receptor results comparisons

The CMB receptor model was used to determine the contributions of sources in Mexicali to ambient VOC at the UABC site, for the period April 13 to 22, 2005. In a past source apportionment study by Fujita et al. (2004) that used data obtained at the same site but for different period (September 22 to 27, 1999), specific source profiles explained 75% of the mass on average while the rest was attributed to unidentified sources. In our study, on average, 86% of the mass was explained by the source profiles used. In addition, Fujita et al. (2004) estimated that gasoline vehicles were consistently the largest contributor, accounting for about 40% of ambient TNMOC during the morning periods, followed by natural gas of LPG sources with 18% contribution, and negligible contributions from other sources. In our study, gasoline mobile sources contribution was estimated at 56%, followed by fugitive LPG emissions (18%), diesel vehicles (6), and consumer products (5). In the Fujita et al. (2004) study, the "background source" did not showed up as a contributor in the morning periods, and for the afternoon periods it was assigned a small contribution (1.5 to 4.4%). In our study the background source was not an important source, probably indicating significant changes in the source profiles of surroundings sources in the last six years.

The previous figures can be compared to the relative contribution of mobile sources to the total hydrocarbons or VOC emissions reported in emissions inventories for the city or municipality of Mexicali. For example, the 1996 official inventory for the city (GEBC, 1999) indicated that 60% of the total emitted hydrocarbons came from mobile sources (70% if emissions from fuel transportation and storage activities are included), while 15% came from LPG sources; 35% of the emissions came directly from light–duty vehicles. In more recent inventory efforts, data from base years 1999 (ERG et al., 2004) and 2004 (Tejeda–Honstein et al., 2005) estimated ~95% contribution of gasoline–powered vehicles to total mobile VOC emissions, the rest being diesel sources. According to the CMB results obtained here, ~90% of the mobile VOC emissions had their source in gasoline sources and the rest in diesel sources. Even though the inventories are from different years and different methodologies were used to derive them, the contribution of mobile sources and LPG sources appear to be in the same range proving consistency in the trends observed by the different studies.

Comparing the results obtained in our CMB application with others, it can be seen that the source contributions for VOC species in downtown Mexicali resembles that of other major cities. For example, Brown et al. (2007) report that in Los Angeles, CA mobile–related emissions accounted for 70–80% of the fitted contribution, while solvent use accounted for 17% or less of the fitted contribution. In a study conducted in Beijing, China in August of 2005 (Song et al., 2007), source contributions to ambient VOC observations were as follow: 52% gasoline–related emissions, 20 petrochemicals, 11% LPG, 5% natural gas, 5 painting, 3 diesel vehicles, and 2 biogenic emissions. Jorquera and Rappenglueck (2004) estimated that near 70% of ambient VOC concentrations observed in Santiago, Chile were due to mobile sources, ~11% were due to LPG losses, and ~6% to diesel exhaust. Finally, CMB application to data obtained in México City (Vega et al., 2000; Mugica et al., 2002) resulted in contributions in the range of 50 to 64% from motor vehicle exhaust, 15–25% from LPG sources, and 9–12% from asphalting operations. In all previous cases, the receptor sites were in a great majority located in residential or commercial areas with important traffic influence, as was the case for the UABC site. In a review paper by Watson et al. (2001), it was reported that, considering the results of 103 references, gasoline–related mobile emissions (exhaust plus evaporative and fugitive emissions) contributed in general to up to 50% or more of the ambient VOC measured in most of those studies.

 

5. Conclusions

A monitoring campaign was conducted in April of 2005 to quantify the levels of VOC at a site located near the downtown district of the border city of Mexicali. TNMOC concentrations obtained were higher in the early morning samples, followed by the nighttime samples and lower for the afternoon samples. This same pattern was followed by the individual species detected. The species registered with the highest concentrations (on a ppbC basis) were propane, isopentane, toluene and the xylenes. The presence of propane and isopentane can indicate an important contribution of LPG–related source as well as gasoline evaporation. However, the main source of the organic species sampled appears to be vehicle exhaust emissions, and in particular vehicles with no catalytic converter. This is supported by the ethene–acetylene ratios obtained (~0.8), and the high correlation between ethene, propylene and benzene. This hypothesis is also supported by a benzene–toluene ratio of ~0.4, and a strong correlation between TNMOC and CO measurements obtained at the same site. A xylenes–benzene ratio close to 2 indicates influence of nearby sources to the observed ambient VOC concentrations, most probably vehicular traffic and commercial activities given the location of the monitoring site. In addition, source–receptor modeling results also indicate that mobile sources (56%) and fugitive LPG sources (18) can be pointed out as main contributors to the observed VOC at the UABC site. In general, main sources of VOC in downtown Mexicali appear to have not changed in the past ten years given the results of two other studies conducted in 1997 and 1999, though the increase in the ethene to acetylene ratio might indicate slight decline in the relative contribution of non–catalytic converter vehicles to the total emissions. The results obtained in this study can help further efforts on assessing the effectiveness of air quality improvement programs of the region, and help orient future policies. Also, results can serve to more detailed top–down emissions inventory validation studies, such that those inventories can be used with more confidence in more detailed air quality modeling efforts.

 

Aknowledgements

This work was supported by the US EPA under Federal Grant/Contract No. XA–83239201–0 through subcontract No. 06–602 to Tecnológico de Monterrey by Arizona State University, as part of the Southwest Consortium for Environmental Research and Policy programs. Although the research described has been funded by the US EPA, it has not been subject to the Agency's peer and policy review and, therefore, does not necessarily reflect the views of the Agency, and no official endorsement should be inferred. The authors greatly appreciate the support of the Municipality of Mexicali, Mexicali's air quality monitoring network and the Universidad Autónoma de Baja California during the execution of the monitoring campaigns. This work was also partially supported by Tecnológico de Monterrey through grant number CAT–052.

 

References

Barletta B., S. Meinardi, I. J. Simpson, H. A. Khwaja, D. R. Blake and F. S. Rowland, 2002. Mixing ratios of volatile organic compounds (VOC's) in the atmosphere of Karachi, Pakistan. Atmos. Environ. 36, 3429–3443.        [ Links ]

Blake D. R. and F. S. Rowland, 1995. Urban leakage of liquefied petroleum gas and its impact on México–City air–quality. Science 269, 953–956.        [ Links ]

Brown S. G., A. Frankel and H. R. Hafner, 2007. Source apportionment of VOCs in the Los Angeles area using positive matrix factorization. Atmos. Environ. 41, 227–237.        [ Links ]

Censullo A. C., D. R. Jones and M. T. Wills, 1996. Improvement of speciation profiles for architectural and industrial maintenance coating operations. Final report prepared by California Polytechnic State University for the California Air Resources Board, San Luis Obispo, CA. 156 pp.        [ Links ]

Chan C. Y., L. Y. Chan, X. M. Wang, Y. M. Liu, S. C. Lee, S. C. Zou, G. Y. Sheng and J. M. Fu, 2002. Volatile organic compounds in roadside microenvironments of metropolitan Hong Kong. Atmos. Environ. 36, 2039–2047.        [ Links ]

Chow J. C., J. G. Watson, M. C. Green, D. H. Lowenthal, B. Bates, W. Oslund and G. Torres, 2000. Cross–border transport and spatial variability of suspended particles in Mexicali and California's Imperial Valley. Atmos. Environ. 34, 1833–1843.        [ Links ]

Cooper J. A., J. G. Watson and J. J. Huntzicker, 1984. The effective variance weighting for least squares calculations applied to the mass balance receptor model. Atmos. Environ. 18, 1347–1355.        [ Links ]

Devore J. L., 2000. Probabilidad y estadística para ingeniería y ciencias. 7th. Ed. Thomson Learning, San Luis Obispo, CA, 768 pp.        [ Links ]

Duffy B. L. and P. F. Nelson, 1996. Non–methane exhaust composition in the Sydney Harbour Tunnel: A focus on benzene and 1,3–butadiene. Atmos. Environ. 30, 2759–2768.        [ Links ]

ERG, Acosta y Asociados and TransEngineering. 2004. México National Emissions Inventory, 1999: Six Northern States. Final Report prepared for Secretariat of the Environment and Natural Resources and the National Institute of Ecology of México, United States Environmental Protection Agency, Western Governors Association, and North American Commission for Environmental Cooperation. 3393–00–011–002, Sacramento CA, 132 pp.        [ Links ]

Fujita E. M., D. E. Campbell, W. Stockwell, R. E. Keislar, B. Zielinska, J. C. Sagebiel, W. Goliff, M. Keith and J. L. Bowen, 2002. Weekend/weekday ozone observations in the south coast air basin. Final Report prepared for the National Renewable Energy Laboratory by the Desert Research Institute, Reno, NV, 16 pp.        [ Links ]

Fujita E. M., D. E. Campbell, R. Keislar and B. Zielinska, 2004. Analysis of ambient air quality data collected in and near northern Baja California, México. Task 4: VOC Sampling and Source Apportionment. Final Report prepared for the California Air Resources Board by the Desert Research Institute. Reno, NV, 83 pp.        [ Links ]

Gamas E. D., M. Magdalena, L. Díaz, I. Schifter, L. Ontiveros and G. Álvarez–Cansino, 2000. Contribution of liquefied petroleum gas to air pollution in the Metropolitan Area of México City. J. Air Waste Manage. Assoc. 50, 188–198.        [ Links ]

GEBC, 1999. Programa para mejorar la calidad del aire de Mexicali 2000–2005. Gobierno del Estado de Baja California, México, 245 pp.        [ Links ]

Gordon G. E., 1988. Receptor models. Environ. Sci. Technol. 22, 1132–1142.        [ Links ]

Henry R. C., 1992. Dealing with near collinearity in chemical mass balance receptor models. Atmos. Environ. 26A, 993–938.        [ Links ]

Hodgson M. J., J. Frohlinger, E. Permar, C. Tidwell, N. D. Traven, S. A. Olenchock and M. Karpf, 1991. Symptoms and micro–environmental measures in non–problem buildings. J. Occupational Medicine 33, 527–533.        [ Links ]

INE, 1999. Tercer informe sobre la calidad del aire en ciudades mexicanas 1998. Dirección de Gestión e Información Ambiental, Instituto Nacional de Ecología, México, D. F., 135 pp.        [ Links ]

Jones A. P., 1999. Indoor air quality and health. Atmos. Environ. 33, 4535–4564.        [ Links ]

Jorquera H. and B. Rappenglueck, 2004. Receptor modeling of ambient VOC at Santiago, Chile. Atmos. Environ. 38, 4243–4263.        [ Links ]

Kanakidou M., J. H. Seinfeld, S. N. Pandis, I. Barnes, F. J. Dentener, M. C. Facchini, R. Van Dingenen, B. Ervens, A. Nenes, C. J. Nielsen, E. Swietlicki, J. P. Putaud, Y. Balkanski, S. Fuzzi, J. Horth, G. K. Moortgat, R. Winterhalter, C. E. L. Myhre, K. Tsigaridis, E. Vignati, E. G. Stephanou and J. Wilson, 2005. Organic aerosol and global climate modelling: a review. Atmos. Chem. Phys. 5, 1053–1123.        [ Links ]

Maroni M., B. Seifert and T. Lindvall (Eds.), 1995. Indoor Air Quality – a Comprehensive Reference Book. Elsevier, Amsterdam, 1049 pp.        [ Links ]

McLaren R., D. L. Singleton, J. Y. K. Lai, B. Khouw, E. Singer, Z. Wu and H. Niki, 1996. Analysis of motor vehicle sources and their contribution to ambient hydrocarbon distributions at urban sites in Toronto during the Southern Ontario oxidants study. Atmos. Environ. 30, 2219–2232.        [ Links ]

Mendoza A., E. I. Pardo, A. A. Gutiérrez, P. Caballero and J. Martínez, 2006. Characterization of PM_2.5 composition and concentration levels in the Mexicali/Imperial Valley Border Area during Fall of 2004 and Spring of 2005. 99th Annual Conference of the Air and Waste Management Association, New Orleans, LA, Compack disk.        [ Links ]

Molina L. T., C. E. Kolb, B. de Foy, B. K. Lamb, W. H. Brune, J. L. Jiménez, R. Ramos–Villegas, J. Sarmiento, V. H. Páramo–Figueroa, B. Cárdenas, V. Gutiérrez–Avedoy and M. J. Molina, 2007. Air quality in North America's most populous city — overview of the MCMA–2003 campaign. Atmos. Chem. Phys. 7, 2447–2473.        [ Links ]

Mølhave L., 1991. Indoor climate, air pollution, and human comfort. J. Expo. Anal. Env. Epid. 1, 63–81.        [ Links ]

Mugica V., E. Vega, J. L. Arriaga and M. E. Ruiz, 1998. Determination of motor vehicle profiles for nonmethane organic compounds in the Mexico City metropolitan area. J. Air Waste Manage. Assoc. 48, 1060–1068.        [ Links ]

Mugica V., J. Watson, E. Vega, E. Reyes, M. Ruiz and J. Chow, 2002. Source apportionment of NMHC in México City. The Scientific World Journal 2, 844–860.        [ Links ]

Na K., Y. P. Kim, I. Moon and K. Moon, 2004. Chemical composition of major VOC emission sources in the Seoul atmosphere. Chemosphere 55, 585–594.        [ Links ]

Na K., K. Moon and Y. P. Kim, 2005. Source contribution to aromatic VOC concentration and ozone formation potential in the atmosphere of Seoul. Atmos. Environ. 39, 5517–5524.        [ Links ]

Quintero–Núñez A. and A. Sweedler, 2004. Air quality evaluation in the Mexicali and Imperial Valleys as an element for an Outreach Program. In: Imperial–Mexicali Valleys: Development and environmental of the U. S. Mexican Border Region. (K. Collins, P. Ganster, C. Mason, E. Sánchez López, M. Quintero–Núñez, Eds.). San Diego State University Press, San Diego, 263–279.        [ Links ]

Quintero Núñez M., M. A. Reyna, K. Collins, S. Guzmán, B. Powers and A. Mendoza, 2006. Issues related to air quality and health in the California–Baja California Border Region. In: The U.S.–Mexican Border Environment: Binational Air Quality Management. (R. Pumfrey, Ed.) SCERP Monograph Series, No. 14, San Diego State University Press, San Diego, 1–46.        [ Links ]

Roberts P. T., C. P. MacDonald, H. H. Main, T. S. Dye, D. L. Coe and T. L. Haste, 1997. Analysis of meteorological and air quality data for the 1996 Paso del Norte ozone study, STI–997330–1754–FR. Sonoma Technology Inc., Santa Rosa, CA, 281 pp.        [ Links ]

Robinson A. L., N. M. Donahue, M. K. Shrivastava, E. A. Weitkamp, A. M. Sage, A. P. Grieshop, T. E. Lane, J. R. Pierce and S. N. Pandis, 2007. Rethinking organic aerosols: Semivolatile emissions and photochemical aging. Science 315, 1259–1262.        [ Links ]

Russell A. G., J. B. Milford, M. S. Bergin, S. McBride, L. McNair, Y. Yang, W. Stockwell and B. E. Croes, 1995. Urban ozone control and atmospheric reactivity of organic gases. Science 269, 491–495.        [ Links ]

Russell A. and R. Dennis, 2000. NARSTO critical review of photochemical models and modeling. Atmos. Environ. 34, 2283–2324.        [ Links ]

Sagebiel J. C., B. Zielinska, W. R. Pierson and A. W. Gertler, 1996. Real–world emissions and calculated reactivities of organic species from motor vehicles. Atmos. Environ. 30, 2287–2296.        [ Links ]

Shifter I., M. Magdalena, L. Díaz, L. A. Melgarejo, A. Barrera, B. Krüger, J. L. Arriaga and E. López–Salinas, 2000. Contribution of LPG–derived emissions to air pollution in the Metropolitan Area of Guadalajara City. J. Air Waste Manage. Assoc. 50, 1857–1864.        [ Links ]

Song Y., M. Shao, Y. Liu, S. Lu, W. Kuster, P. Goldan and X. Shaodong, 2007. Source apportionment of ambient volatile organic compounds in Beijing. Environ. Sci. Technol. 41, 4348–4353.        [ Links ]

Tejeda–Honstein D. D., P. C. Montúfar–Ortiz, J. A. Aguilar–Gómez and I. Gonzáles–Merino, 2005. Inventario de emisiones a la atmósfera de fuentes móviles en la Zona Metropolitana de Mexicali. Reporte final preparado por el Tecnológico de Monterrey, Campus Estado de México, Mexicali, México, 51 pp.        [ Links ]

US EPA, 1998. Technical assistance document for sampling and analysis of ozone precursors, EPA/600–R–98/161. United States Environmental Protection Agency. National Exposure Research Laboratory, Research Triangle Park, NC, 493 pp.        [ Links ]

US EPA, 2004. EPA–CMB8.2 Users Manual, EPA–452/R–04–011. United States Environmental Protection Agency. Air Quality Model Group, Emissions Monitoring and Analysis Division, Office of Air Quality Planning and Standards, Research Triangle Park, NC, 123 pp.        [ Links ]

Vanoye García A. Y., 2007. Determinación de patrones meteorológicos conducentes a episodios de contaminación atmosférica en Monterrey y Mexicali empleando un modelo meteorológico de mesoescala. Master Thesis, Tecnológico de Monterrey. Monterrey, México, 425 pp.        [ Links ]

Vega E., V. Mugica, R. Carmona and E. Valencia, 2000. Hydrocarbon source apportionment in México City using the chemical mass balance receptor model. Atmos. Environ. 34, 4121–4129.        [ Links ]

Watson J. G., 1984. Overview of receptor model principles. J. Phys. Chem. A 34, 619–623.        [ Links ]

Watson J. G., J. C. Chow and E. M. Fujita, 2001. Review of volatile organic compound source apportionment by chemical mass balance. Atmos. Environ. 35, 1567–1584.        [ Links ]

Zielinska B., J. Sagebiel, G. Harshfield and R. Pasek, 1999. Volatile organic compound measurements in the California/Mexico border region during SCOS97. Sci. Tot. Environ. 276, 19–31.        [ Links ]

Zuk M., L. Rojas Bracho and M. G. Tzintzun Cervantes, 2007. Tercer almanaque de datos y tendencias de la calidad del aire en nueve ciudades mexicanas. Instituto Nacional de Ecología, México, D. F., 116 pp.        [ Links ]