A gridded inventory of anthropogenic methane emissions from Mexico based on Mexico's national inventory of greenhouse gases and compounds

Table 1 gives the anthropogenic methane emissions from Mexico's national inventory for 2015 as reported to the UNFCCC [14]. Here and elsewhere, we refer to sectors as the primary entries in table 1 (e.g. Fuel combustion activities) and subsectors as the secondary entries (e.g. Road transportation). Figure 1 displays the relative contributions of major sectors and subsectors. Total national emission is 5.1 Tg a⁻¹ with dominant sectoral contributions from livestock (47%), fugitive emissions from fuels (20%), solid waste management and disposal (15%), and wastewater treatment and discharge (14%).

Zoom In Zoom Out Reset image size

Table 2 provides additional breakdown for oil/gas emissions by source type based on information embedded in the national inventory [14]. This additional breakdown is necessary for the implementation of mitigation measures and to track the effectiveness of existing oil/gas regulations in Mexico.

The national inventory has no intra-annual variability and we do not introduce any in our gridded inventory. Seasonal variability is expected to be small except for manure emissions, which can be distributed over the year on the basis of local temperature [21]. Emissions from rice cultivation are seasonal [27] but are very small in Mexico.

We spatially allocate the national inventory emissions to a 0.1° × 0.1° grid using a number of datasets with national coverage, most of which can be accessed via the federal government's online data portal (https://datos.gob.mx, last access: January 2020). We rely heavily on Mexico's Directorio Estadístico Nacional de Unidades Económicas (DENUE) which includes the point locations of major industrial and commercial facilities [28], and for oil and gas we primarily use the Sistema de Información de Hidrocarburos (SIH) which includes geospatial information for a majority of Mexico's oil and gas infrastructure [29]. Additional details on the spatial datasets used are included in the relevant sections below.

For methane point sources we distribute emissions using either the number or the activity of individual facilities. For non-point sources we distribute emissions using the count of sources per grid cell (e.g. number of cows) or categorical information (e.g. source present or not in the grid cell). For some subsectors we first estimate state or municipal emissions and then allocate to finer scale if possible. Mexico has 32 states subdivided into 2465 municipalities.

2.2.1. Fuel combustion activities and chemical industry.

2.2.2. Fugitive emissions from fuels.

2.2.3. Livestock and aggregate sources.

2.2.4. Solid waste disposal.

2.2.5. Biological treatment of solid waste and incineration/open burning of waste.

2.2.6. Wastewater treatment and discharge.

Figure 2 shows the spatially allocated total emissions from Mexico's national inventory for 2015, and the sectoral contributions from livestock, fuels, solid waste disposal, and wastewater, accounting for over 97% of total emissions. Total emissions are highest in central Mexico and along the east coast, reflecting contributions from all four sectors.

Zoom In Zoom Out Reset image size

Livestock emissions total 2.4 Tg a⁻¹ with cattle enteric fermentation and manure management contributing 88% and pig manure management contributing 7%. Emissions are highest in central Mexico and along the east coast, reflecting dense livestock populations which we capture by restricting municipal livestock populations to the pasture and agricultural lands in these regions. Emissions are especially concentrated in the states of Jalisco, Durango, and Veracruz due to their large cattle populations with Jalisco and Veracruz having the greatest number of milk and beef cows in 2015, respectively.

Oil/gas emissions total 0.7 Tg a⁻¹ with flaring and venting during oil production as the two largest sources accounting for 32% and 24% of oil/gas emissions, respectively. Emissions are concentrated along the east coast, and offshore activities account for 51% of oil/gas emissions. Emissions are highest in regions of intense oil production, reflecting our use of oil and gas production per well to allocate emissions. Coal emissions total 0.3 Tg a⁻¹ and are concentrated in the state of Coahuila in northern Mexico.

Solid waste emissions total 0.8 Tg a⁻¹. Managed waste disposal sites (landfills) contribute 78% while unmanaged sites and open dumpsites each contribute 11%. The spatial distribution of solid waste emissions reflects the facility locations we use to map emissions. These largely follow population density although landfill sites tend to be located on the peripheral of cities rather than in city centers. Emissions are highest in central Mexico where there are a number of large facilities. Hotspot facilities are discussed in section 3.2.

Wastewater treatment and discharge emissions total 0.7 Tg a⁻¹ of which 28% are domestic and 76% are industrial. The spatial distribution of domestic wastewater emissions follows population density, reflecting both the locations of treatment plants and the discharge of untreated wastewater. Industrial wastewater emissions are spread out spatially based on the locations of industries associated with significant water usage and organic waste generation, in particular the food and beverage industries. Regions of high emissions include central Mexico, the Yucatán Peninsula, and the Baja California Peninsula. Emissions from domestic and industrial wastewater treatment plants are more concentrated than untreated wastewater discharges because we allocate emissions to specific facilities.

Figure 3 shows the hotspots in our inventory exceeding 20 Gg a⁻¹ (2.3 tons h⁻¹) per 0.1° × 0.1° grid cell. Table 3 gives the locations and magnitudes of these hotspots. There are 16 such hotspots, accounting together for 13% of national emissions. The largest hotspot is the Mina VII coal mine (1; 119 Gg a⁻¹) of the MICARE Unit in Nava, Coahuila. Coal production in Nava accounts for 42% of Mexico's coal production, while the nearby municipality of Múzquiz (including hotspots 4, 8) accounts for 39% [32]. The second largest hotspot is the Neza III landfill just outside of Mexico City (2; 116 Gg a⁻¹) which is surrounded by four additional landfill hotspots (11, 12, 14, 15). The third largest hotspot is from offshore oil/gas production in the Ku/Maloob/Zaap oil field (3; 81 Gg a⁻¹) and is part of four clustered offshore emission hotspots (3, 6, 9, 10) that together account for 37% of national oil production and 29% of associated gas production in 2015 [34]. We also find large emissions from the nearby onshore Cactus gas processing complex (7; 28 Gg a⁻¹). Other emission hotspots include landfills in Monterrey, Nuevo León (5; 34 Gg a⁻¹) and Tijuana, Baja California (16; 20 Gg a⁻¹), and the SuKarne cattle processing facility in Vista Hermosa, Michoacán (13; 21 Gg a⁻¹).

Zoom In Zoom Out Reset image size

Uncertainty estimates for national emissions by subsector are reported by INECC as ±2σ relative error standard deviations (table 1) [14]. For the oil/gas subsectors some uncertainty estimates are asymmetrical so they are reported as (−2σ, +2σ) relative error standard deviations (table 2). Uncertainties are relatively large (38%–66%) for the fuel subsectors but very small (<10%) for the other major emission subsectors including cattle enteric fermentation, cattle/pig manure management, managed solid waste disposal (landfills), and industrial wastewater, resulting in a very low uncertainty of 4.9% on the total national emission. The latter uncertainty is likely too low given that other North American countries estimate much higher subsector uncertainties. For example, the average uncertainty estimate (±2σ relative error standard deviations) for cattle enteric fermentation is 15%–20% for the US and Canada, while the manure management uncertainty estimate range is 19%–50% [33, 49]. Hristov et al [50] independently estimated an average uncertainty of 16% for US enteric fermentation emissions and 64% for manure management. The landfill and industrial wastewater uncertainties are also much higher for Canada and the US (∼40%–50%).

Our gridded inventory has additional uncertainty related to the spatial allocation of emissions. Maasakkers et al [21] evaluated their 0.1° × 0.1° gridded national inventory for the US with a more detailed 4 × 4 km² inventory constructed by Lyon et al [22] for northeastern Texas including local information on sources and direct methane emission measurements. They found uncertainties (±1σ) of 50%–100% for all subsectors on the 0.1° × 0.1° grid. Uncertainties on some sources decreased with spatial averaging, in particular for livestock, but remained above 30% for all subsectors even at 0.5° × 0.5° spatial resolution. In absence of empirical estimates in Mexico similar to those of Lyon et al for evaluation of our inventory's spatial uncertainty we recommend using the Maasakkers et al relative uncertainty estimates for our gridded inventory.

The locations of our hotspots are well established, but there is uncertainty in the allocation of emissions between large facilities, especially when sources have skewed emissions distributions [51]. We are not able to capture hotspot emissions from conditions that may be related to abnormal operating conditions or anomalous activity [51, 52].

Figure 4 compares Mexico's national inventory of anthropogenic methane emissions to EDGAR v5 [20] for 2015. EDGAR estimates a total emission of 6.7 Tg a⁻¹, which is 25% higher than the national inventory. Livestock (2.8 Tg a⁻¹) and wastewater (0.7 Tg a⁻¹) emissions in EDGAR are similar to the national inventory. Solid waste disposal emissions (1.6 Tg a⁻¹) are a factor of two greater, coal emissions (0.05 Tg a⁻¹) are 81% lower, and oil/gas emissions (1.3 Tg a⁻¹) are 78% higher. EDGAR uses global datasets of country activity data (e.g. the number of cattle) to calculate emissions, so these differences in sector emissions reflect the use of country-specific activity data and emission factors in Mexico's national inventory.

Zoom In Zoom Out Reset image size

Also shown in figure 4 is a comparison to a 2016 emission inventory for the Metropolitan Area of the Valley Mexico, Inventario de Emisiones de la Ciudad de México (IE CDMX) [53]. This includes both Mexico City proper and surrounding municipalities. Total anthropogenic emission in that inventory is 319 Gg a⁻¹, close to our inventory (344 Gg a⁻¹) and much higher than EDGAR (152 Gg a⁻¹). The IE CDMX inventory has 70% of its emission from solid waste, consistent with our inventory (74%), while EDGAR has 74% of its emission from wastewater.

Figure 5 compares the spatial distribution of emissions in our work and EDGAR. Figure S1 (available online at stacks.iop.org/ERL/15/105015/mmedia) shows the comparison by sector. EDGAR livestock emissions are more diffuse which is likely due to our use of municipal livestock statistics. EDGAR wastewater emissions follow population whereas many of our high emission regions do not follow population because they are associated with industrial wastewater. The discharge of industrial wastewater to water bodies has been recognized as both a climate and water pollution concern in Mexico [54]. EDGAR's solid waste emissions are concentrated in fewer grid cells than our inventory likely because we include unmanaged sites and dumpsites. Offshore production accounts for only 16% of total oil/gas emissions in EDGAR but 51% in our inventory. Gas transmission emissions are concentrated at compressor stations in our work but are uniform along pipelines in EDGAR.

Zoom In Zoom Out Reset image size

EDGAR is also very different in the distribution and magnitude of emission hotspots. The two largest hotspots in EDGAR are landfills with grid cell emissions of 840 Gg a⁻¹ and 420 Gg a⁻¹, higher than any of our hotspots and not coincident spatially. The oil/gas hotspots in EDGAR have higher emissions (maximum emission of 120 Gg a⁻¹) and different locations compared to our inventory's oil/gas hotspots. There are very few coal emission hotspots in EDGAR (maximum emission of 20 Gg a⁻¹) which are located in the state of Coahuila but again do not coincide with our inventory's coal hotspots. The differences in landfill and coal hotspots are likely due to missing facility locations in EDGAR while the differences in oil/gas hotspots are likely due to EDGAR's use of different spatial datasets, like satellite observed flaring, to allocate oil/gas emissions.

Figure 5 also compares our inventory to the 2012 oil/gas emission inventory for Canada and Mexico from Sheng et al [55] which relied on an older version of the national inventory. Their oil/gas emissions are 1.2 Tg a⁻¹ with 75% from oil and 25% from gas, as compared to 0.8 Tg a⁻¹ in the national inventory with 58% from oil and 42% from gas. Their gas transmission emissions are more uniformly distributed along pipelines, similar to EDGAR, and their gas production emissions are limited to regions of non-associated gas production whereas gas production in our inventory is highest offshore because of associated gas production. Offshore activities account for 56% of total oil/gas emissions in their inventory, as compared to 51% in our work. There is a checkerboard pattern in the comparison map (figure 5) because we allocate emissions by well production whereas they do this allocation by well count.

Finally, figure 5 compares our inventory to the 2013 Wolf et al global livestock emission inventory [56]. Wolf et al have total livestock emissions in Mexico of 2.7 Tg a⁻¹, including 2.3 Tg a⁻¹ from enteric fermentation (21% higher than Mexico's national inventory) and 0.4 Tg a⁻¹ from manure management (7% lower). Their emissions are more diffuse than our inventory because they spread livestock emissions over shrublands and grasslands as defined by MODIS satellite-based land cover data [57] whereas we use more specific land cover data for pasture and agricultural lands. We are also better able to resolve areas of high emission than Wolf et al because we use municipal data to spatially allocate emissions whereas they only use state-level data.