Emissions and Air Quality Implications of Upstream and Midstream Oil and Gas Operations in Mexico

: Mexico approved amendments to its constitution in December 2013 that initiated transfor-mational changes to its energy sector. This study developed a 2016 bottom-up emissions inventory for volatile organic compounds (VOCs), nitrogen oxides (NO x ), sulfur dioxide (SO 2 ), carbon monoxide (CO), and ﬁne particulate matter (PM 2.5 ) from upstream and midstream sector sources, including onshore and offshore well sites, gas ﬂaring, natural gas processing facilities, and natural gas compressor stations, throughout Mexican basins. Crude oil storage tanks at onshore oil well sites and venting and fugitive sources at offshore oil production sites were the primary sources of VOC emissions. Key contributions to NO x , CO, and PM 2.5 emissions were from internal combustion engines at offshore oil well sites and midstream operations. SO 2 emissions were associated with onshore and offshore gas ﬂaring and boilers and process heaters at natural gas processing facilities. Application of the inventory with the Comprehensive Air Quality Model with Extensions (CAMx) indicated that oil and gas production operations could contribute to ozone and PM 2.5 concentrations in Mexican and U.S. states under favorable transport patterns. This study provides a foundation for assessing the implications of Mexico’s future energy policies on emissions and domestic and cross-border air quality and public health.


Introduction
Mexico has been among the world's major exporters of crude oil, which has been crucial to its economy. Oil production peaked in 2004 with the supergiant Cantarell field in the southern Gulf of Mexico but has declined by 50% since then [1] due to constraints in investment resources and technical expertise required to fully exploit its hydrocarbon resources [2]. Energy reform was part of the Pacto por México that required amendments to the Mexican Constitution approved in December 2013 with secondary implementing legislation in 2014 [2][3][4]. A significant outcome was the allowance for private and foreign investment and participation under different contract modalities for oil and gas exploration and extraction, which had previously been restricted to the state-owned oil company Petróleos Mexicanos (Pemex), Mexico City, Mexico [2][3][4]. Between 2015 and 2018, Mexico awarded more than 100 contracts to companies within Mexico and 19 other countries for exploration and extraction of its onshore, shallow water, and deepwater hydrocarbon resources [5], which remain the property of the nation [4]. Following a transition in presidential administrations in December 2018, national priorities have emphasized energy sovereignty and increasing oil production and refining capacity with the prioritization and strengthening of Pemex [6,7]. Figure 1 shows the locations of 10,458 wells active during 2016 by basin [40]. Annual oil and gas production volumes were 788,738 thousand barrels (Mbbl) and 2,127,142 million cubic feet (MMcf) [40]. Offshore production in the shallow waters (<500 ft) of the Sureste Basin, which include the Ku-Maloob-Zaap (KMZ), Cantarell, Abkatún-Pol-Chuc, and Litoral de Tabasco fields, accounted for nearly 80% of total national oil production and 54% of natural gas production [40]. Onshore oil production is located in the Sureste and Tampico-Misantla basins. Most non-associated gas production occurs in the Burgos, Veracruz, and Sureste basins [41], and accounted for 22% of domestic production in 2016 [42].
Bottom-up emissions estimates were developed using activity metrics and per unit activity-based emission factors for the upstream and midstream sector sources addressed in this study. For onshore basins in Mexico, locations and activity metrics, including active well counts, oil and gas production volumes, and spud counts, were obtained from Mexico's National Hydrocarbons Commission (CNH) [40]. Activity metrics were mapped to U.S. source classification codes (SCCs) associated with onshore oil and gas operations (Table S1). An oil well was classified as having a gas-oil production ratio (GOR) <6000 cf/b and conversely for a natural gas well [43]. Because well site equipment configurations across Mexican basins were not readily available, activity metrics for Mexico were applied with emission factors developed for onshore basins in Texas that had similar operations and hydrocarbon resources. Texas-based emission control assumptions were removed as a conservative estimate. The Sabinas and Burgos basins share a common border with the Western Gulf Basin, and primarily include legacy vertical wells producing natural gas with no condensate production. Emission factors for these Mexican basins (Table S2) were based on emission estimates in the U.S. Environmental Protection Agency (EPA) National Emissions Inventory (2014NEIv2) [44] normalized by oil and gas production from the EPA Oil and Gas Emission Estimation Tool [45] for the Western Gulf Basin. Emission factors for the Tampico-Misantla, Veracruz, and Sureste basins (Table S3), which include legacy vertical wells producing both oil and natural gas, were based on the Palo Duro Basin using a similar approach. Offshore wells in Mexico produced oil and gas during 2016, with the exception of less than 1% in the Tampico-Misantla Basin, which produced only natural gas. Activity from offshore shallow water platforms was based on oil and/or gas production volumes from the CNH. Emission factors were developed using emissions [46] and production data for shallow water leases with gas or oil and gas production under the jurisdiction of the U.S. Bureau of Ocean Energy Management [47] in 2014 (Table S4). The analysis included only those leases with a GOR between 0 to 130 Mcf/bbl for consistency with the range reported by the CNH for offshore oil production wells in the Sureste Basin during 2016.
Shah et al. [48] identified flaring locations and flared gas volume from Visible Infrared Imaging Radiometer Suite (VIIRS) Nightfire (VNF) detections that were applied with emission factors from the EPA's AP-42 compilation [49] and Oil and Gas Emission Estimation Tool [45] [50] showed good spatial agreement with the locations of flaring activity in the Sureste Basin identified by Shah et al. [48]. Projection factors were developed from CNH onshore and offshore gas production volumes [40] to estimate emissions in 2016.

Midstream Emissions
Pemex [42] identified eleven natural gas processing facilities (Figure 1) in operation during 2016. Eight were included in the 2008 Mexico National Emissions Inventory [51], and emissions were projected based on the ratio of 2016 and 2008 natural gas intake volumes from Pemex [42]. Locations of the remaining three facilities were identified from the North American Cooperation on Energy Information (NACEI) [52]. Emissions were estimated using linear regressions between facility-wide emissions and 2016 petrochemical production from the eight INEM facilities reported by Pemex [42]. Figure 1 shows 22 central compressor stations in operation along SISTRANGAS pipelines [53]. Mexico's Ministry of Energy [54] and Eduardo [55] reported installed horsepower for 18 of the 22 compressor stations. The average installed horsepower across the 18 compressors was assumed for the four that lacked information. In the absence of station-specific engine type data, AP-42 [49] emission factors for uncontrolled 4-stroke rich burn engines were applied with installed horsepower to estimate emissions for each of the 22 compressor stations.

Air Quality Modeling Configuration
This study adapted a CAMx air quality modeling platform from the Texas Commission on Environmental Quality [56], which was based on the 2016v1 National Collaborative Emissions Modeling Platform [39]. CAMx is an open-source Eulerian photochemical grid modeling system for gas and particulate air pollution that has been applied across a range of spatial and temporal scales to support air quality research and regulatory assessments throughout the world [57]. The 36 km × 36 km horizontal domain included most of Canada, the continental United States, and almost all of Mexico ( Figure S1). Simulations were conducted for the 15 December 2015-1 January 2017 time period. The modeling configuration is described further by McDonald-Buller et al. [58]. Point source emissions for Mexico's upstream and midstream oil and gas sectors (NAICS categories 211110, 325110, 221210) and electricity sector (NAICS 221110) were replaced with our estimates; all other emissions remained identical. CAMx simulations examined the contributions of midstream sources and different geographic regions with upstream oil and gas operations to maximum daily 8 h average (MDA8) ozone concentrations and 24 h average PM 2.5 concentrations across Mexican states and U.S. border regions using an emissions zero-out approach.

Base Year Emissions Profiles
Annual emission estimates from onshore and offshore well sites, flaring, natural gas processing facilities, and natural gas compressor stations for 2016 are shown in Figure 2.
Total NO x , CO, VOC, SO 2 , and PM 2.5 emissions from these sources across Mexican basins were approximately 355,000, 427,000, 869,000, 141,000 and 5100 tons, respectively. Figure 3 shows the disaggregated contributions of emissions from offshore oil well and onshore oil and gas well sites and natural gas processing facilities by SCC-based categories in order to assess contributions to the annual totals shown in Figure 2. Emissions from flaring and natural gas compressor stations were each represented by a single SCC category, as described above. Offshore and onshore well site operations accounted for 69%-78% of the total annual NO x , CO, and PM 2.5 emissions, as shown in Figure 2, with the largest contributions from offshore oil production coming from the Sureste Basin ( Figure S2). Natural gas and diesel-fired internal combustion engines at offshore oil well production sites ( Figure 3a) contributed 55%-65% of total NO x , CO, and PM 2.5 emissions. Natural gas-fired compressor engines at onshore gas well sites (Figure 3b), primarily in the Burgos and Sureste basins, accounted for 11% of total NO x emissions. Gas-fired internal combustion engines at compressor stations contributed 15% of total NO x and 21% of total CO emissions. Approximately 22% of total PM 2.5 emissions were attributed to natural gas-fired boilers and turbines ( Figure 3c) at natural gas processing facilities.
Onshore and offshore well site operations in the Sureste and Tampico-Misantla basins ( Figure S2) were the primary sources of VOC emissions. Crude oil storage tanks from onshore oil well site operations ( Figure 3d) accounted for 63% of total VOC emissions. Collectively, well casing vents and fugitive sources from offshore oil production ( Figure 3a) in the Sureste Basin accounted for 28% of total VOC emissions.
Gas flaring and natural gas-fired boilers and process heaters at natural gas processing facilities contributed 52% and 47%, respectively, of total SO 2 emissions, as shown in Figure 2. Almost all SO 2 emissions from natural gas processing were concentrated in the onshore region of the Sureste Basin, which includes the Nuevo Pemex, Ciudad Pemex, and Cactus facilities in the states of Tabasco and Chiapas. SO 2 emissions from flaring occurred offshore (59%) and onshore (24%) in the Sureste Basin and onshore in the Tampico-Misantla Basin (15%), as shown in Figure S2.

Intercomparison of Emission Source Contributions
Key sources of criteria pollutant and precursor emissions identified by previous studies are similar to those found in our study, although these studies spanned time periods with different oil and gas production volumes and likely operational practices in Mexico. Most of these studies had a geographic focus on operations in the Bay of Campeche (Sureste Basin), in particular those in the northeastern region associated with the Cantarell and KMZ fields. Several bottom-up studies were conducted circa 1999-2006. Villaseñor et al. [34] attributed 63-79% of SO x , CO, and nonmethane hydrocarbons (NMHC) emissions to Cantarell platforms, with the offshore Cayo de Arcas terminal and Dos Bocas terminal in Tabasco identified as the largest sources of NO x . Schifter et al. [35] found flaring and combustion sources, including process heaters, boilers, and diesel engines, to be the primary sources of NO x , SO x , CO, HC, and PM emissions. Mendoza-Domínguez and Graniel-Peralta [36] identified seasonal differences in extraction and operational processes and meteorological conditions that contributed to low flare combustion efficiencies. Bottomup emissions estimates associated with the Nuevo Pemex, Cuidad Pemex, Cactus, and La Venta natural gas processing facilities were developed by Bauer et al. [37]. Among the dominant contributions to emissions were sulfur recovery units as sources of SO 2 , compression stations as sources of CO and NO x , boilers and power generation as sources of NO x , and oil batteries and compression stations as sources of total hydrocarbons.
Our examination indicated that the magnitude and spatial distribution of SO 2 and NO x emissions of Mexico's offshore oil and gas operations in the Sureste Basin were not represented in the more recent Emissions Database for Global Atmospheric Research (EDGAR) version 5.0 bottom-up inventory [59][60][61][62]. Missing emissions or large underrepresentation of these emissions is an important, persistent gap that has also been noted by Zhang et al. [63] in the EDGAR version 4.3.1 inventory.
We compared our bottom-up emission estimates with those in the 2016v1 National Collaborative Emissions Modeling Platform, which represented an interpolation of 2014 and 2018 projections of the 2008 INEM [39]. Upstream oil and gas well sites in the 2016v1 platform shown in Figure S3 were sparse relative to those identified in our study. Nationwide VOC, CO, and NO x emissions in the 2016v1 platform from upstream sources were lower than our estimates (Table S5). In contrast, emissions of SO 2 in the 2016v1 inventory were greater, with 96% of the nationwide upstream sector total attributed to a single offshore location in the Sureste Basin possibly intended to represent the Cantarell and/or KMZ complexes. Compressor stations along SISTRANGAS pipelines, as well as three natural gas processing facilities, La Congrejera, Parajitos, and Morelo, were not included in the 2016v1 platform.
A recent bottom-up study by ICF [64] identified vented emissions (e.g., offshore venting, stranded gas venting, venting from oil tanks and condensate tanks), fugitives, and flaring as sources of methane emissions from Mexico's oil and natural gas industry and significant opportunities for abatement. Top-down approaches have applied airborne-or satellite-based measurements to examine emission trends in the Sureste Basin. Fioletev et al. [65] attributed changes in an offshore SO 2 hotspot between 2005-2007 and 2008-2010 to rising production of heavier crude oil in the KMZ fields as Cantarell declined. A secondary SO 2 onshore hotspot was attributed to the Nuevo Pemex gas processing center. Zhang et al. [63] found annual SO 2 and NO 2 emission rates between 2005-2017 over the KMZ and Cantarell offshore production cluster peaked in 2008 and have declined since due to expanded capacity for associated gas utilization, which reflects policy interventions to reduce flaring. Zavala-Ariaza et al. [66] suggested that associated gas from offshore production is being transported and flared at onshore midstream facilities. Inaccurate assumptions regarding flaring efficiencies in the Mexican national greenhouse gas inventory contributed to overestimation of methane emissions offshore but underestimation of emissions from Nuevo Pemex.

Improving Contemporary Inventories
Current country-specific data are an ongoing need for refining bottom-up emissions inventories for Mexico. Activity metrics and locations of oil and gas sector emission sources were available through the mining of data from Mexico's federal agencies and cooperative international initiatives, but emission factors in this and prior studies have been drawn from U.S. resources, including the EPA and BOEM. In assessments of earlier bottom-up inventories, Muriel-García et al. [38] noted the needs for emission measurements and the homogenization of emission factors. Zavala-Ariaza et al. [66] suggested the main driver for inaccuracies in the Mexican greenhouse gas inventory to be the use of emission factors that are not specific to Mexico.
Uncertainties identified in this study included oil and gas well site process and equipment configurations, temporal variations in emissions, and emission control technologies and strategies across Mexican basins.
Mexico has been among the world's top countries for gas flaring [67,68]. Flare combustion efficiencies and smoke formation are influenced by factors such as the heating values and chemical compositions of flared gases, as well as operating practices (e.g., air or steam assisting) [69][70][71][72][73], and these should be more fully characterized in Mexico. CNH issued guidelines for the avoidance or reduction of natural gas flaring and venting in 2008 that led to investments in gas treatment and handling and reinjection capacity by Pemex [68]. Guidelines issued in 2016 focused on implementing methods for measuring associated gas flaring and maximizing gas utilization and conservation by operators [67,68]. Federal guidelines targeting methane emissions in 2018 included specifications for the efficiency of destruction equipment [10].
Although the U.S. BOEM inventory [47] included non-platform emissions (e.g., mobile vessels, helicopters, pipelaying operations), we did not extrapolate for Mexico. Nonplatform emissions are dependent on spatial domain and travel patterns, which were not expected to be consistent between the U.S. and Mexico. All offshore activity in Mexico during 2016 occurred in shallow waters. It was not feasible to isolate non-platform emissions in the BOEM inventory associated with shallow water platforms. Information collection using an approach similar to the Gulfwide Offshore Activities Data System (GOADS) by BOEM would facilitate development or refinement of these emissions in Mexico.

Air Quality Impacts and Implications of Future Development
CAMx predictions provided a perspective on the contributions of upstream and midstream sector emission sources to air quality across the Mexican states and U.S. border regions shown in Figure 4. Percentile differences in MDA8 ozone and 24 h average PM 2.5 concentrations by region from zeroing all upstream and midstream emissions relative to the 2016 base case are shown in Figure 5. Spatial patterns in the differences in MDA8 ozone and 24 h PM 2.5 concentrations were found to be similar.
Emissions from offshore well sites in the Sureste Basin were the primary influences on MDA8 ozone and PM 2.5 concentrations among the source types and regions considered in this study. Figure 6 and Figure S4 indicate that these emissions potentially contribute to air quality throughout other areas of Mexico, as well as in U.S. states such as Texas under favorable transport patterns. Impacts on average occur in states along or near Mexico's eastern coastline, including Tabasco, Oaxaca, Chiapas, Veracruz, and Tamaulipas.
The HYSPLIT forward trajectories initiated offshore within the Sureste Basin, as shown in Figure 7, illustrate how seasonal differences in transport patterns contributed to the spatial footprint of MDA8 ozone and PM 2.5 impacts in downwind areas during 2016 as predominantly southeasterly wind flow patterns in the spring shifted to northeasterly by the fall.  Emissions from offshore well sites in the Sureste Basin were the primary influences on MDA8 ozone and PM2.5 concentrations among the source types and regions considered in this study. Figure 6 and Figure S4 indicate that these emissions potentially contribute to air quality throughout other areas of Mexico, as well as in U.S. states such as Texas under favorable transport patterns. Impacts on average occur in states along or near Mexico's eastern coastline, including Tabasco, Oaxaca, Chiapas, Veracruz, and Tamaulipas. The HYSPLIT forward trajectories initiated offshore within the Sureste Basin, as shown in Figure 7, illustrate how seasonal differences in transport patterns contributed to the spatial footprint of MDA8 ozone and PM2.5 impacts in downwind areas during 2016 as predominantly southeasterly wind flow patterns in the spring shifted to northeasterly by the fall.
(a) Emissions from onshore oil and gas well sites in the Tampico-Misantla, Veracruz, and Burgos and Sabinas basins, as shown in Figure S5, had more modest impacts on MDA8 ozone concentrations. Activity in the Burgos/Sabinas Basin could influence air quality in Texas during southeasterly transport patterns. Onshore upstream sources in the Sureste Basin and midstream sources, as shown in Figure S5, typically had localized ozone impacts in Tabasco and southern Veracruz, but could also contribute to MDA8 ozone concentrations in states along Mexico's eastern coastline.
Atmosphere 2021, 12, x FOR PEER REVIEW 11 of 18 Emissions from offshore well sites in the Sureste Basin were the primary influences on MDA8 ozone and PM2.5 concentrations among the source types and regions considered in this study. Figure 6 and Figure S4 indicate that these emissions potentially contribute to air quality throughout other areas of Mexico, as well as in U.S. states such as Texas under favorable transport patterns. Impacts on average occur in states along or near Mexico's eastern coastline, including Tabasco, Oaxaca, Chiapas, Veracruz, and Tamaulipas. The HYSPLIT forward trajectories initiated offshore within the Sureste Basin, as shown in Figure 7, illustrate how seasonal differences in transport patterns contributed to the spatial footprint of MDA8 ozone and PM2.5 impacts in downwind areas during 2016 as predominantly southeasterly wind flow patterns in the spring shifted to northeasterly by the fall.
(a) Emissions from onshore oil and gas well sites in the Tampico-Misantla, Veracruz, and Burgos and Sabinas basins, as shown in Figure S5, had more modest impacts on MDA8 ozone concentrations. Activity in the Burgos/Sabinas Basin could influence air quality in Texas during southeasterly transport patterns. Onshore upstream sources in the Sureste Basin and midstream sources, as shown in Figure S5, typically had localized ozone Mexico conducted nine bidding cycles between 2015-2018 for the onshore, shallow water and deepwater blocks shown in Figure 8, which attracted domestic and international private sector investment [5]. Following a transition in presidential administrations, a moratorium was placed on future rounds, accompanied by a renewed focus on Pemex. Projects awarded under previous rounds have continued. The CNH reported production from 33 contracts during September 2021 (137.8 Mbpd of oil; 213.3 MMcfpd of natural gas) [5].
Although uncertainty for the future direction of private sector participation has increased, Figure 8 provides an indication of key geographic locations where future oil and gas production could expand depending on investment and technical resources. For example, the Zama oil field discovery (Round 1.1, Block 7) announced in July 2017 represented one of the largest shallow water discoveries in the world over the last 20 years (approximately 670-1010 MMboe [74]). Mexico has so far not pursued substantial development of its unconventional resources, although it has become increasingly reliant on U.S. pipeline imports of natural gas for its electricity sector. The Burgos Basin has promising technically recoverable shale gas resources [4] that could be poised for development, similar to the Eagle Ford Shale in Texas, if current policies change.

Conclusions
This study developed a 2016 bottom-up emissions inventory for VOCs, NO x , SO 2 , CO, and PM 2.5 from onshore and offshore well sites, gas flaring, natural gas processing facilities, and natural gas compressor stations throughout Mexican basins. Crude oil storage tanks at onshore oil well sites (63%) and venting and fugitive sources at offshore oil production sites (28%) were the primary sources of VOC emissions. Natural gas and diesel-fired internal combustion engines at offshore oil well sites accounted for 55%-65% of NO x , CO, and PM 2.5 emissions. Midstream operations represented 15%-21% of NO x and CO emissions. Onshore and offshore gas flaring (52%) and natural gas-fired boilers and process heaters at natural gas processing facilities (47%) accounted for almost all SO 2 emissions. Identification of process and equipment configurations, temporal variations in emissions, super-emitting sources, and emission control implementation and effectiveness are ongoing needs for Mexico.
CAMx simulations identified geographic areas within Mexico and U.S. border states where emissions from oil and gas operations could contribute to MDA8 ozone and 24 h average PM2.5 concentrations. Among the source types and regions considered in this study, offshore oil well site operations in the Sureste Basin were the primary influence on air quality along the eastern coastline and other areas of Mexico and in Texas under favorable transport patterns. Exploration and development of Mexico's hydrocarbon resources could lead to changes in emissions profiles and air quality in the coming years.
Mexico has recently expressed its commitment to reducing greenhouse gas emissions but also to increasing oil production and refining capacity. Development of a photochemical modeling platform with high spatial granularity coupled with intensive surface and airborne measurements across Mexican basins would facilitate an improved understanding of the impacts of Mexico's future energy sector transitions. Top-down approaches that routinely use satellite retrievals can be used to track emissions and air quality trends in Mexico and U.S. border regions. Continued refinement of bottom-up emission inventories and coordinated atmospheric modeling can support the design of optimum emission control strategies for existing and future operations.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/atmos12121696/s1, Figure S1: CAMx 36 km × 36 km horizontal modeling domain., Figure S2: Contributions to NO x , CO, SO 2 , VOC, and PM2.5 emissions from onshore and offshore well sites, natural gas processing facilities, and flaring by basin, Figure S3: Locations of active 2016 oil and/or gas wells by basin and flaring in this study and oil and gas sector emission sources identified in the 2016v1 National Collaborative Emissions Modeling Platform, Figure S4: Predicted annual average and maximum differences in 24 h PM 2.5 concentrations by grid cell when emissions from offshore well sites in the Sureste Basin were zeroed relative to the 2016 base case, Figure S5: Predicted annual average differences and maximum differences in MDA8 ozone concentrations by grid cell when emissions from onshore well site operations in the Sureste, Tampico-Misantla, Veracruz, and Burgos and Sabinas basins and from midstream sources were zeroed relative to the 2016 base case. Table S1: Oil and gas activity metrics mapped to EPA emission source classification codes; Table S2: Emission factors for onshore oil and gas production well sites in the Sabinas and Burgos basins., Table S3: Emission factors for onshore oil and gas production well sites in the Sureste, Tampico-Misantla, and Veracruz basins, Table S4: Emissions per unit of production (lb/Mbbl/yr) from offshore oil and gas well sites, Table S5: Annual emissions of CO, NO x , PM 2.5 , SO 2 , and VOC from this study and the 2016v1 inventory for oil and gas exploration and extraction, natural gas processing facilities, and natural gas compressor stations.