Current Inventories of Methane Emissions
Development of greenhouse gas (GHG) emission inventories is necessary for estimating the relative significance of emissions from various sources and evaluating the effects of mitigation efforts. Inventories are developed for specific purposes and can provide emission estimates at many scales, from facility level, to urban, regional, national, and global scales.
INVENTORY OF U.S. GREENHOUSE GAS EMISSIONS AND SINKS
The United States, like all developed countries, submits a national, annual inventory of anthropogenic GHG emissions and sinks to the United Nations (UN) as part of its treaty obligations under the UN Framework Convention on Climate Change (UNFCCC).1 The first Greenhouse Gas Inventory (GHGI) was submitted by the United States in 1994, covering the years 1990-1993 (EPA, 1994), and an inventory has been submitted every year since, with the latest submitted in 2017 and providing estimates for the years 1990-2015 (EPA, 2017b).
In developing the annual GHGI, the United States follows the 2006 Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories (IPCC, 2006).2 Key principles for development of the GHGI are inclusion of only anthropogenic GHG emissions, reporting of actual as opposed to potential emissions, ensuring full territorial coverage, and following a tiered methodological approach (Box 2.1) as contained in IPCC (2006) (see Appendix D for details). The GHGI of all developed countries is subject to an annual review by an international expert review team to ensure that each country follows the requirements in the relevant UNFCCC decisions and IPCC (2006). A final report of the review, which is coordinated by the UNFCCC secretariat, is made available online annually.3
1 See http://unfccc.int/files/essential_background/background_publications_htmlpdf/application/pdf/conveng.pdf, Article 4, paragraph 1(a).
2 The IPCC is in the process of developing a report to refine the 2006 IPCC Guidelines. This report is expected to be released in 2019. See https://www.ipcc-nggip.iges.or.jp/home/2019refinement.html.
3 The latest review report of the U.S. GHG inventory can be found at http://unfccc.int/national_reports/annex_i_ghg_inventories/inventory_review_reports/items/9916.php.
A key feature of the GHGI reporting to the UNFCCC is providing estimates from 1990,4 and accordingly, most countries have a base year of 1990 for reporting of methane emissions. Generally, recalculations of previously submitted GHGI estimates are made to reflect updated sources of information. These recalculations may be related to the use of a new emission calculation methodology, updated activity data, emission factors, or other parameters used for one or more years of the time series.
The U.S. Environmental Protection Agency (EPA) serves as the lead agency compiling and coordinating the inventory development, with input from other federal agencies, state agencies, research and academic institutions, industry trade associations, individual companies, and other experts (Appendix D). The EPA actively engages with the research and stakeholder communities to improve the GHGI. This has included multiple stakeholder engagements to improve the GHGI in recent years. The EPA also produces “memos”5 to disseminate methodological or emission factor and activity data updates in the GHGI and has solicited stakeholder feedback on the same.
4 See UNFCCC, Article 4, paragraph 2(b). http://unfccc.int/files/essential_background/background_publications_htmlpdf/application/pdf/conveng.pdf.
5 See “Archives: Previously Posted Memos and Other Information on Stakeholder Engagement” at https://www.epa.gov/ghgemissions/natural-gas-and-petroleum-systems.
GREENHOUSE GAS REPORTING PROGRAM
Unlike the GHGI, which strives to develop a total, national-level estimate of GHG emissions and sinks by GHG and by sector, the GHG Reporting Program6 (GHGRP) is a facility-level reporting program for larger GHG emitters from 41 industries across the United States and its territories. In general, facilities emitting ≥25,000 metric tons (or 0.025 Tg) of CO2 equivalents7 of GHGs (carbon dioxide, methane, nitrous oxide, and fluorinated gases) per year are subject to regulation under the GHGRP, thereby excluding reporting from smaller sources. As such, the total estimated GHG emissions reported within the GHGRP are lower than the GHGI. Facilities from several industries covered in this report are subject to reporting under the GHGRP, in particular, landfills, petroleum and natural gas operations, and underground coal mines. All of these operations are required to report their fugitive and combustion-related methane emissions, along with other GHGs. Methane emissions from enteric fermentation and manure management systems are not subject to reporting under the GHGRP. The program was designed to allow stakeholders to better track emissions and identify opportunities to reduce emissions, waste, and costs,8 and although it may improve the estimation of national GHG emissions for specific categories, it was not designed to produce an inventory of national emissions. The GHGRP reports are generated by the industrial owner or operator following specific methods outlined in the regulations.
Overall, the GHGRP has been a valuable tool to improve methane estimates in the GHGI and to the scientific community in various methane measurement studies. Although coverage under the GHGRP is not complete, the methane data reported under the program can provide additional insight beyond what can be obtained from the national GHGI, because emissions are estimated using a higher-tiered method and provide specific plant- and component-level emission data for sources across the United States. The EPA uses the GHGRP data to update activity data and emission factors in the GHGI and refine national estimates and trends. After facilities submit information to the GHGRP, the EPA conducts several checks.9 If potential errors are identified, the EPA will notify the reporter, who can resolve the issue either by providing an acceptable response describing why the flagged issue is not an error or by correcting the flagged issue and resubmitting their annual GHG report. Despite this process, data
6 The EPA finalized the GHGRP on October 30, 2009, and codified the regulations under 40 CFR Part 98. See https://www.epa.gov/ghgreporting.
7 GHGRP thresholds are based on the IPCC (2007) Fourth Assessment Report (AR4), 100-year global warming potential.
8 See https://www.epa.gov/ghgreporting/learn-about-greenhouse-gas-reporting-program-ghgrp.
9 See https://www.epa.gov/sites/production/files/2015-07/documents/ghgrp_verification_factsheet.pdf.
discrepancies have been noted in certain sectors (see sector-specific text in later section of this chapter for additional details).
STATE-LEVEL METHANE EMISSION INVENTORIES
Currently, methane inventories are generated by some states while others have methane reporting requirements only from certain industrial categories, similar to the GHGRP. The scope and coverage of categories varies from state to state, as does the frequency with which the state inventory is published. There are also many states that currently do not develop emission inventories.
The EPA has developed a “State Inventory Tool” (SIT)10 that can be used by individual states to estimate methane emissions for key categories. The SIT provides a common framework developed in a manner consistent with the GHGI methodologies, but allowing for the use of state-specific activity data and emission factors where available. However, only a few states report methane emissions either through their state regulatory reporting program or employing the SIT to estimate their methane emissions.
Some states, such as California, have developed their methane inventories by using state regulatory reporting standards.11 By maintaining consistency with IPCC methods, state-level inventories can generate state-level activity data and emission factors and provide useful data within their boundaries for continuous improvement of the national GHGI. These state-level inventories can also play an important role in the evaluation of state-level policies. For example, methane emission data for petroleum and natural gas systems are generally not collected at the state level, and there are significant regional differences in emissions (Allen et al., 2013; Lamb et al., 2015) due to differences in production practices (wet-gas versus dry-gas production), geology, infrastructure, and methane mitigation strategies. A state-level inventory can provide valuable information on state-specific activity data to reduce regional uncertainties.
States not currently developing a statewide methane inventory could use the EPA’s State Inventory Tool (SIT) and complementary methodologies to provide valuable information, such as activity data and emission factors, for continuous improvement of the U.S. Greenhouse Gas Inventory.
10 See https://www.epa.gov/statelocalclimate/download-state-inventory-and-projection-tool.
11 See https://www.arb.ca.gov/cc/inventory/data/data.htm; https://www.arb.ca.gov/cc/reporting/ghgrep/ghg-rep.htm.
GRIDDED METHANE INVENTORIES AND STUDIES
As noted previously, activity data in the GHGI are generally extrapolated to large regional scales (e.g., state or National Energy Modeling System regions from the Energy Information Administration) and emissions therefore are presented in similar spatial scales. Thus it is another challenge to verify12 the GHGI because field measurement campaigns are mostly conducted at a subregional or site-specific scale. However, spatially resolved inventories are verifiable because spatial scales can be established at 0.1° × 0.1° or finer resolution.13 The GHGI is also strongly limited in the range of science questions and policy issues to which it can be applied because of its current large scale. Spatially and temporally resolved, verifiable inventories could inform the development of emission mitigation policies, and they also have the potential to direct appropriate mitigation programs at certain regional hotspots. Success in this regard requires “completeness” in the broadest sense, so that all major sources are included in the inventory. These goals require the ability to test the emission rates prescribed in the inventory and to accurately attribute emissions to particular source types using direct measurements. Verifiability is the bedrock upon which inventories must be built if they are to be widely applicable to societal needs. It is also important to ensure that the datasets and computer codes associated with development of the inventories are made publicly available to allow verification of the results or further analysis.
The GHGI reports estimated annual emissions for the United States, thus achieving adequate resolution for UNFCCC reporting on long-term national trends. However, the objectives for measuring and monitoring of methane in the United States are broader than the UNFCCC objectives for which the GHGI was originally developed. The EPA and other federal agencies, the states, and the U.S. scientific community are strongly motivated to better understand the geographic and temporal distribution of methane emissions.
Moreover, for the consideration of different spatial and temporal scales, alternatives to the GHGI may be more appropriate. For instance, field studies that quantify local- to regional-scale emissions may report daily or hourly averages. Discrepancies between these subnational values and those reported by the GHGI may be difficult to explain given the differences in estimation and measurement approaches, as well as the complex spatial and temporal dynamics of methane emissions from various source categories.
12 In this report, “verify” means to test against top-down measurements.
13 0.1° × 0.1° is roughly 100 km2.
To address this challenge, several studies and entities have developed methane emission inventories at a finer spatial and temporal resolution than the GHGI. For example, the European Commission global dataset Emissions Database for Global Atmospheric Research (EDGAR)14 estimates global emissions on a per country basis as well as a 0.1° × 0.1° grid and a temporal resolution of 1 month. EDGAR is frequently used by the scientific community for comparisons with top-down study results. The Greenhouse gas and Air pollutant Interactions and Synergies (GAINS)15 model is used to evaluate cost-effective controls on greenhouse gases and other air pollutants to inform response strategies. This model is run at a global scale with a 0.5° × 0.5° spatial resolution and can produce annual and monthly estimates. GAINS distinguishes between 165 regions around the globe, including many countries. Maasakkers et al. (2016) published a spatially and temporally resolved gridded inventory of the 2012 GHGI (Figure 2.1). The motivation for developing this gridded inventory was to provide (1) a better a priori estimate (i.e., a first estimate of emissions to be improved by the application of inverse models) and (2) a better interpretation of the top-down inversion results relative to EDGAR. Furthermore, most of the dozens of published top-down studies from the last 5 years construct their own “downscaled” inventory to compare bottom-up inventory results to their atmospheric observations. A consistently generated gridded inventory will provide a standard baseline against which to compare top-down atmospheric results. Maasakkers et al. (2016) designed a national gridded inventory to address these issues. They disaggregated the GHGI data into a spatial resolution of 0.1° × 0.1° with a monthly temporal resolution of the emissions using data from various sources. Importantly, the inventory also provided an error characterization that was unavailable in the commonly used EDGAR.
The Maasakkers gridded inventory could be updated to later years as activity data become available. Comparisons with other gridded inventories suggest that revisions would be beneficial. When compared with EDGAR, the Maasakkers et al. (2016) gridded inventory results for the United States are about 2.6 Tg yr–1 higher and the distributions are different. Large spatial differences in emissions from petroleum and natural gas systems and manure management were also observed between the two inventories, suggesting that evaluating emissions using different spatial and temporal scales can influence estimates.
Hristov et al. (2017) produced a gridded 0.1° × 0.1° inventory of livestock methane emissions for the contiguous United States. Their analysis yielded total average livestock methane emissions of 8.916 Gg yr–1 (or ~0.01 Tg yr–1) for 2012 that were compa-
14 See http://edgar.jrc.ec.europa.eu/.
15 See http://www.iiasa.ac.at/web/home/research/researchPrograms/air/GAINS.html.
rable to 2012 estimates from both the GHGI and EDGAR of 9.044 Gg yr–1 (~0.01 Tg yr–1) and 8.657 Gg yr–1 (~0.01 Tg yr–1), respectively. However, the spatial distribution of emissions differed significantly from that of EDGAR.
The differences in enteric emission estimates between the Hristov et al. (2017) and Maasakkers et al. (2016) studies are mainly due to different emission factors for the various subcategories of cattle (Figure 2.2). These comparisons (although only for methane emissions from livestock) demonstrate that discrepancies in source attribution still exist in current bottom-up gridded inventories. As an example, methane emissions from livestock in Texas and California (highest contributors to the national total) in the Hristov et al. (2017) study were 36 percent lower and 100 percent greater, respectively, than estimated by EDGAR. Compared with the Maasakkers et al. (2016) analysis, total livestock emissions in the Hristov et al. (2017) study for these two states were 12 percent lower and 4 percent higher, respectively. Differences in spatial distribution will likely have a strong impact on posterior (i.e., following inverse modeling adjustment) emissions estimated by top-down approaches, within the contiguous United States, even if the overall magnitudes of these estimates are consistent.
Another example of a gridded inventory is the World Resources Institute’s (WRI) Climate Analysis Indicators Tool (CAIT),16 which is a state-level gridded allocation that uses some of the same principles as Maasakkers et al. (2016). CAIT is an online analytical tool that provides GHG emissions, including methane, and projected emission data across different source categories across the world. CAIT U.S. State GHG Emissions (CAIT-US) rely on EPA data, including the SIT tool.17 However, it does not include methane emissions from key petroleum and natural gas segments and from industrial wastewater. This tool helps users understand which states have the highest methane emissions and identify potentially relevant policies for the methane sources within these states (Figure 2.3).
The development of a gridded version of the national GHGI can aid policymaking by providing greater clarity on where emissions are occurring. The national GHGI and the gridded version meet different user needs. In general, policymakers, media, and indus-
16 CAIT Climate Data Explorer available at http://cait.wri.org.
17 See http://cait.wri.org/docs/CAIT2.0_US_Documentation.pdf. CAIT uses data from the GHGI published in 2015 for calendar years 1990-2013.
try have tended to focus historically on the central (average) emission estimates provided in the national GHGI, while researchers also focus on the associated uncertainties and the finer resolution offered by the gridded inventory. While meeting different user needs, the various methane inventories can be complementary and, together, increase overall knowledge.
Finer-scale, gridded inventories of national methane emissions provide significant value to the scientific community to better characterize and compare inventories and test against top-down emission estimates. Further, gridded inventories have the potential to inform mitigation action at spatial scales relevant to policymakers and industry. Improvements in the GHGI and state-level inventories will also support improvements in finer-scale gridded inventories.
OVERVIEW OF METHANE EMISSIONS IN THE UNITED STATES FROM KEY SOURCE CATEGORIES
According to the latest GHGI (EPA, 2017b), total methane emissions in the United States in 2015 were estimated to be 26.23 Tg excluding land use, land use change, and forestry (LULUCF), and 26.68 Tg including LULUCF. Overall, between 1990 and 2015, total anthropogenic methane emissions reported by the EPA in the GHGI decreased by 16.0 percent (excluding emissions from LULUCF). However, trends differed among source categories. The values below are according to EPA (2017b) using current methods unless otherwise noted. Uncertainties related to emission estimates from these sources are discussed in detail in Chapter 4.
- Enteric fermentation was the largest source of methane emissions in 2015, an increase of 1 percent from 1990 despite declining cattle population.
- Manure management methane emissions have increased by 78 percent since 1990, resulting partially from a shift from smaller dairies and swine production facilities to larger facilities and increasing use of liquid manure systems for both swine and dairy cows, which have higher methane emissions than solid storage or application to pasture, range, and paddock.
- Natural gas system emissions reflect the vented, fugitive, and flared emissions from hundreds of thousands of individual sources. Overall, methane emissions from this category have decreased by 16 percent since 1990, with the increase in emissions from field production offset by decreases in emissions from natural gas processing, transmission and storage, and distribution driven largely by reductions in emissions from compressors in the transmission and storage segment and replacement of cast iron and steel pipelines with plastic pipeline in the distribution segment.
- Petroleum system methane emissions stem from numerous small and large sources. Between 1990 and 2015, methane emissions decreased by 28 percent due in large part to a decrease in the venting and flaring of associated gas from petroleum wells.
- Landfills are the third largest source of methane emissions, exhibiting a decline of more than 35 percent since 1990 as a result of increased use of landfill biogas collection systems (879 landfill sites in 2015 according to the June 2017 GHGRP). Moreover, commercial biogas utilization has been installed at over 600 sites, primarily for onsite electrical generation.18 In addition, diversion of garden waste from landfills (required in 25 states) and, increasingly, food waste (in more than 180 communities) has resulted in greater than 17.63 Tg
18 See Landfill Methane Outreach Program (LMOP), https://www.epa.gov/lmop (June 2017).
annually diverted to more than 4,900 composting operations (Goldstein et al., 2014). These decreases were partially offset by increasing waste generation from an increased U.S. population.
- Coal-mining methane emissions result predominantly from degasification and ventilation systems at underground coal mines (66 percent of the total) with a lower contribution from surface mines (13 percent) and post-mining activities at underground and surface mines (11 percent). Abandoned mines were responsible for 9 percent of emissions from this category. Overall, methane emissions from coal mining have declined by 35 percent since 1990 due in part to the decline in the number of active mines that, becoming abandoned and sealed, emit significantly less methane.
Methane Emissions from the Agriculture Sector
The main sources of methane emissions from U.S. agriculture are generated by enteric fermentation in livestock, manure management including composting, rice cultivation, and field burning of agricultural residues. In 2015, it is estimated that these activities collectively contributed 9.77 Tg of methane emissions, or about 37 percent of anthropogenic methane emissions in the United States, excluding LULUCF (EPA, 2017b). This section discusses the two largest sources within the agriculture sector: enteric fermentation and manure management. Rice cultivation is discussed briefly in Appendix C.
Beef cattle are the largest contributor of enteric methane emissions (they accounted for 71 percent of 2015 emissions), followed by dairy cattle (26 percent), and the remaining emissions are from horses, sheep, swine, goats, American bison, mules, and asses (EPA, 2017b). Enteric methane emissions represent about 2-11 percent of dietary gross energy consumed by livestock (Moraes et al., 2014), and methane is a natural by-product of microbial fermentation of carbohydrates and amino acids in the rumen and the hindgut of farm animals (Hristov et al., 2013).
The amount of enteric methane emissions largely depends on the type of digestive system (ruminants versus nonruminants, with ruminants producing much more methane), feed intake (greater feed intake leads to more methane production), and feed composition, particularly the amount of fiber and lipid in the diet (Appuhamy et al., 2016). Feed intake is positively correlated with methane emissions and productivity (e.g., milk or average daily gain). Attempts to quantify enteric methane emissions show
that feed intake alone can account for most of the variability in enteric methane emissions (e.g., Hristov et al., 2013; Kebreab et al., 2008; Moraes et al., 2014).
For GHGI reporting, national estimates have relied on the IPCC (2006) methodologies to calculate enteric methane emissions based on type of livestock. The majority of the enteric methane emissions are from cattle, and so a more detailed approach (i.e., IPCC Tier 2) is applied for cattle, while a less detailed approach (i.e., IPCC Tier 1) is used for other livestock. The IPCC Tier 2 approach (IPCC, 2006) requires detailed information on livestock subcategories (e.g., calves, stockers, and cows) and feed intake estimates in each subcategory. In the Tier 2 methodology, animal performance and diet data are used to estimate feed intake as the daily amount of energy (gross energy intake [GEI], MJ day–1) that an animal needs for maintenance and for activities such as growth, lactation, and pregnancy (IPCC, 2006) considering feed digestibility. Annual cattle population data are obtained from the U.S. Department of Agriculture’s National Agricultural Statistics Service QuickStats database (USDA-NASS, 2016). Diet characteristics are estimated by region for dairy, beef cattle on pasture (i.e., cow-calf operations), and feedlot beef cattle. These diet characteristics are used to calculate digestible energy (DE) values. Enteric methane emissions are then calculated as a percentage of GEI.
The default emission factors (Ym) for cattle in the United States are 3 ± 1 percent for feedlot cattle (fed high-grain diets) and 6.5 ± 1 percent for dairy cows (cattle that are primarily fed forages, concentrate feeds, low-quality crop residues, and by-products) and grazing cattle (IPCC, 2006). However, because of the availability of detailed diet information for different regions and animal types in the United States, DE and Ym values that are unique to the specific regions have been developed. Diet characterizations and estimation of DE and Ym values are based on information from state agricultural extension specialists, a review of published forage quality studies and scientific literature, expert opinion, and modeling of animal physiology. A simulation using a process-based ruminant digestion model was used to estimate Ym for each diet evaluated from the literature, and a function was developed to adjust regional values over time based on the national trend (Kebreab et al., 2008). For feedlot animals, the DE and Ym values used have been continuously updated from the literature for 1990-2007 (EPA, 2014). For grazing beef cattle, Ym values are based on specific diet components, and the values have been continuously updated with weight and weight gains for cattle taken from the literature and expert opinion.
Most of the equations for calculating GEI in the IPCC (2006) guidelines were developed as an average for world conditions (not only the United States) using data that are
now over 30 years old.19 Recent research suggests that estimates could be improved by updating methodologies to reflect the current state of science. For example, the net energy requirement for maintenance during lactation used in the IPCC (2006) is greater than estimates from Moraes et al. (2015) commonly used for Holstein cows in the United States. The equation to calculate net energy requirement for lactation does not include protein, which can be variable in milk. Moreover, the Ym value for dairy cattle in the United States has been reported to be 12 percent lower than IPCC (2006) recommendations (e.g., Appuhamy et al., 2016). In addition to updating the methodologies, research is needed to define the probabilities of updated emission factors, given that, from a statistical perspective, emissions factors should follow certain probability distributions.
In an effort to evaluate IPCC Tier 2 recommendations and improve methodologies for estimating methane emissions from enteric fermentation, an international team of animal scientists (Global Network20) compiled a large database of available information on this topic. They reported that models that used dry matter intake (DMI) or DMI and fiber concentration had accuracy in predicting enteric methane emissions from lactating dairy cows similar to models that included a greater number of animal and feed-related variables. These findings suggest that enteric methane emissions can be predicted with a single or a few variables, which can be relatively easily generated for various animal categories and production systems. Further, it suggests that the existing, rather complicated Tier 2 methodology could be simplified without increasing uncertainty of estimations.
DMI of cattle can be estimated using extant models. Using such estimates in models predicted enteric methane emissions well in dairy cattle (Appuhamy et al., 2016). However, beef cattle are the main contributors to enteric methane production, and predicting enteric methane emissions from beef cattle on range and pasture is highly uncertain. This is mainly due to difficulty in measuring actual intake while cattle are on rangeland or pasture as well as changes in digestibility of forages in different seasons of the year. The Global Network is working to develop models to predict enteric emissions from beef cattle and, similar to dairy, DMI and fiber content of diet are the primary determinants of emissions. Therefore, better estimation of intake and forage quality is required to improve predictions. The analysis from the Global Network also indicated that the average Ym value for nonfeedlot dairy was around 7.1 percent of GEI, which is an increase of about 9 percent compared to IPCC (2006) recommendations.
19 The IPCC is currently revising methodologies for both the enteric and manure management sections.
20 The Global Network is a multinational collaborative effort funded by national governments. For more information, see http://animalscience.psu.edu/fnn/current-research/global-network-for-enteric-methanemitigation.
Updating and simplifying existing Tier 2 equations, including emission factors, based on synthesis of recent scientific studies would improve the methodology for estimating methane emissions from enteric fermentation. Additional research focused on predicting dry matter intake estimates for cattle on rangeland and pasture based on animal and feed characteristics could significantly improve emission estimates for this livestock category.
Estimated 2015 manure management methane emissions reported by the GHGI totaled 2.65 Tg, with the dairy (52 percent) and swine (37 percent) being the two largest contributors (Figure 2.4). Manure from livestock is managed in various systems, using practices that contribute to methane emissions. These include storing manure in pits under housing areas, manure treatment (e.g., solids separation, anaerobic digestion, and composting), and manure transport and storage in tanks, lagoons, static piles, etc. (Key et al., 2011; Leytem et al., 2013; MacDonald et al., 2007). National GHGI reporting for manure management is based on the IPCC (2006) methodologies with some modifications to account for variable methane conversion factors (MCFs) on a
state basis. The overall estimation is derived from the amount of volatile solids (VS) that is stored for each livestock category (beef cattle, dairy cattle, swine, etc.) in each manure management system (anaerobic lagoon, static pile, etc.) along with a maximum methane generation potential (Bo) for each manure type (swine, dairy, etc.) and an MCF. The sum of all livestock categories over each manure management system provides the final national GHGI estimate for manure emissions.
The VS excreted for each livestock category are estimated utilizing feed intake and digestibility, while Bo and MCF (which are based on regional average temperatures) values were derived from the literature, when available, or the expert opinion of the IPCC (2006) authors and are largely unverified using on-farm data. Therefore, there is great uncertainty as to whether estimated emissions are representative of actual on-farm emission rates. For example the Bo values for anaerobic lagoons used in the IPCC methodologies were derived from research on the biological activity of methane digesters (Bryant et al., 1977; Hashimoto, 1983; Hashimoto et al., 1981; Morris, 1976), which may not be representative of a typical anaerobic lagoon. Research suggests the broader microbial community, longer residence times, and lower loading rates of uncovered anaerobic lagoons may lead to higher VS degradation rates than those found in anaerobic digesters (Lory et al., 2010). In addition, methane emissions from liquid systems are greatly influenced by manure temperature. Although the MCFs attempt to account for changes in emissions due to regional average temperatures, they were based on very limited data that may not accurately capture these relationships.
Comparisons of on-farm emission data with EPA inventory estimates have indicated that in some instances, the EPA methodology underestimated emission from liquid manure systems in North America by approximately 50 percent (Baldé et al., 2016; Leytem et al., 2017). In addition, activity data are difficult to obtain, as there is little information available that accurately portrays the amount of manure VSs that are handled and stored in each manure management system.
Presently, manure VS distribution within manure management systems is assigned based on farm size, which has not been validated. Therefore, improvements in both activity data and methodologies are needed to improve manure emission estimates.
More accurate methane emission estimates from manure management could be achieved by collecting more accurate activity data related to the distribution of manure in different manure management systems, updating the maximum methane potential for the various manure groups and methane conversion factors (via taking into account the effect of temperature and storage time on emissions), as well as evaluating the accuracy of assigning manure distribution based on farm size.
Methane Emissions from Petroleum and Natural Gas Systems
Emissions of methane from the petroleum and natural gas systems are estimated in the 2015 EPA GHGI (EPA, 2017b) to total 8.10 Tg. This includes petroleum and natural gas supply chain sources from production to distribution (Figure 2.5), but does not include emissions associated with fuel use (e.g., unburned methane from electricity power generation) or end-use emissions (i.e., residential, commercial, industrial, or transport). Petroleum and natural gas emission estimates are generally based on multiplying estimated or actual counts of specific devices (e.g., wells, compressors, and pneumatic controllers) and operations (e.g., compressor blowdowns and well completions) by an average emission rate (or emission factor) for the equipment type or operation.
The major emission source categories for which methane measurements are estimated in the petroleum and natural gas category include (Figure 2.6)
- Natural gas well sites: Sources of emissions at well sites are inventoried at the equipment level for the national inventory. Sources include but are not limited to pneumatic devices, liquid unloadings, and tanks. Pneumatic devices, primarily pneumatic controllers, use gas pressure to open and close control valves, typically on sites that do not have electrical power. Liquid unloadings remove
- Natural gas gathering and boosting stations: These stations collect gas from multiple wells and perform some combination of compression, dehydration, and treatment; national emissions are inventoried at the facility level.
- Gas processing plants: These facilities receive gas from gathering and boosting stations and create pipeline-quality natural gas, removing contaminants and separating natural gas plant liquids (NGPLs, primarily ethane, propane, and butane) into separate NGPL sales streams; national inventories are performed at the facility level.
- Transmission facilities: Reciprocating and centrifugal compressors of various sizes, used at multiple points along the natural gas supply chain, are the primary sources of emissions during transmission. National inventories are performed at both the facility and component levels.
- Petroleum production well sites: Equipment and operations at petroleum well sites that are the largest sources of inventoried emissions include pneumatic devices, flares, and tanks. Flares are used to combust gases that might otherwise be vented, such as gases generated during well completions or
accumulated liquids from well bores; some liquid unloadings result in venting to the atmosphere. Tanks are used in storage of petroleum, condensate, and produced water and can vent methane.
gas produced in association with petroleum production (associated gas) that cannot be delivered to a pipeline system. Offshore platforms are inventoried separately from onshore well sites.
Studies have demonstrated that a significant fraction of the emissions from the petroleum and natural gas industry originate from a relatively small fraction of outlying sources with higher-than-expected emission factors, that some specific sources responsible for the emissions may change over time (including daily), and that the relative frequency of these high-emitting sources varies between petroleum and natural gas production regions. Additional discussions about specific methods, uncertainties, and unaccounted-for emission sources related to petroleum and gas inventories are discussed in Chapters 3 and 4.
The six highest-emitting sources within the natural gas systems collectively contribute over 68 percent of the total emissions from natural gas (Table 2.1). The emission factors for pneumatic controllers, normal fugitive emissions, gas engines, exhaust vents in production, and transmission segments are based on a comprehensive study by the Gas Research Institute and the EPA (Harrison et al., 1996). Emission estimates from these top six categories can be improved by conducting additional field studies at petroleum and gas production sites, natural gas gathering and boosting (G&B) stations, and natural gas transmission compressor stations. Recent studies by Marchese et al. (2015) and Zimmerle et al. (2015) have contributed to improvements in emission estimates from the G&B subsegment and also to updates to the transmission station emissions. The emission factors for other key sources such as pneumatic controllers, normal fugitives in the production sector, and gas engine exhausts can be improved by conducting additional studies and assessing incorporation of recent measurement studies by Allen (2014), Allen et al. (2015a,b), Prasino Group (2013), and Thoma et al. (2017). Fugitive emission factors might be improved by reviewing relevant data collected by leak detection and repair programs under state (e.g. Colorado), federal (New Source Performance Standard OOOOa), and voluntary programs (EPA Methane Challenge). Employing activity data from the 2017 GHGRP would improve the emission estimates for the largest emission source, G&B stations, as well as other key sources.
Recent measurement campaigns in the Barnett Shale in Texas (Zavala-Araiza et al., 2015) and Fayetteville Shale petroleum and gas production regions in Arkansas (Zimmerle et al., 2016) as well as measurements in the San Francisco Bay Area (Fischer et al., 2017) have demonstrated the importance of carefully designed, tiered-measurement campaigns incorporating and complementing the strengths of bottom-up and top-down approaches in improving the characterization of methane emissions from petroleum and natural gas supply chains. The emission factors for key emission
TABLE 2.1 Total Emissions and Basis of Emission Estimation for the Top Six Emission Sources Within the Natural Gas Systems
|GHGI Emission Source||Sector||2015 Emissions, Tg||Basis for Emission Estimation|
|Gathering & boosting (G&B) stations||Production||1.97||G&B emission factors are based on Marchese et al. (2015) (53,066 standard cubic feet day–1 (scfd) methane per G&B station, or 1,503 cubic meters day–1 (scmd) which relies on emission measurements by Mitchell et al. (2015) at 114 G&B stations).|
|Pneumatic controllersa||Production||1.02||Pneumatic controller emission factors are based on Gas Research Institute (GRI)/EPA data (EPA, 1996b), but adjusted for methane content and operating hours from GHGRP Subpart Wb data reported in 2014.|
|Pneumatic device vents, intermittent bleed (IB)||0.90||The IB emission factor is based on 323 scfd (9 scmd) whole gas. This factor is taken from GRI/EPA. As indicated above, the EPA adjusts the GHGI factor based on the average methane content and operating hours reported for the GHGRP Subpart W in 2014.|
|Pneumatic device vents, high bleed (HB)||0.09||The HB emission factor is based on 896 scfd (25 scmd) whole gas. This factor was derived from GRI/EPA study data split between HB and LB based on 6 scfh (0.2 scmh). The derivation is provided in Table B-14 of the American Petroleum Institute Compendium (API, 2009). The EPA adjusts the GHGI factor based on the average methane content and operating hours reported for the GHGRP Subpart W requirements in 2014.|
|Pneumatic device vents, low bleed (LB)||0.03||The current LB emission factor is based on 33.4 scfd (0.9 scmd) whole gas. This factor was derived from GRI/EPA study (EPA, 1996b) data split between HB and LB based on 6 scfh (0.2 scmh). The derivation is provided in Table B-14 of the American Petroleum Institute Compendium (API, 2009). The EPA adjusts the GHGI factor based on the average methane content and operating hours reported for the GHGRP Subpart W requirements in 2014.|
|GHGI Emission Source||Sector||2015 Emissions, Tg||Basis for Emission Estimation|
|Transmission station total emissions||Transmission and storage||0.57||The emission factors are based on the Zimmerle et al. (2015) analysis of the Subramanian et al. (2015) measurements at 37 transmission stations in 16 states.|
|Normal fugitives||Production||0.34||Normal fugitives consist of fugitive emissions from gas wells and well-pad equipment. Emission factors are based on the EPA/GRI, methane emissions from the natural gas industry (EPA, 1996a).|
|Gas engines (compressor exhaust vent)||Production||0.23||The compressor exhaust emission factor is based on the EPA/GRI methane emissions from the natural gas industry (EPA, 1996a).|
|Engines (compressor exhaust in transmission)||Transmission and storage||0.25||The compressor emission factor is based on EPA/GRI methane emissions from the natural gas industry (EPA, 1996a).|
a Devices used in petroleum and natural gas systems to regulate liquid levels, valves, and gas pressure. Controllers powered by natural gas pressure when open, release methane (EPA, 2016b).
bSubpart W refers to the section in the Code of Federal Regulations that outlines the regulations related to the calculation, monitoring, and reporting of GHG emissions from petroleum and natural gas facilities under EPA’s Greenhouse Gas Reporting Program. The Code of Federal Regulations is divided into 50 titles; Subpart W is contained in Title 40, Chapter 1, Subchapter C, Part 98.
sources such as pneumatic controllers and engine exhaust emissions still rely on factors that were generated over two decades ago. In many cases, these emission factors do not represent the population of emission sources or the current facility designs or work practices. The Gas Research Institute (GRI)/EPA 1996 factors were largely developed at the component level within a facility (e.g., number of high-bleed, low-bleed, and intermittent-bleed pneumatic vents).
In the petroleum and natural gas industry, there are thousands and in some cases millions of individual emission components. Obtaining accurate activity data at such a granular component level is a significant challenge. This is a leading cause of uncertainties with the petroleum and natural gas industry methane estimates (Chapter 4),
and updated counts have caused revisions to prior estimates and methods as part of the annual recalculation process within the GHGI. It is expensive to conduct national-scale, component-level bottom-up measurement campaigns that have enough samples to represent an industry that is complex and spatially and temporally varying and that comprises potentially hundreds of thousands of discrete components nationwide (Chapter 3). Although the EPA publishes the activity data associated with the GHGI, some of these data conflict with other agencies’ data. For example, in the 2015 GHGI, the EPA lists the number of active gas wells in the United States as 421,893, while the Energy Information Administration (EIA) lists the same as 574,530. Similarly, the EPA identifies 668 natural gas processing plants in the 2014 GHGI, while the EIA identifies 551 such plants.21 Hence, a publicly available, consistent, and centralized repository, similar to the National Oil and Gas Gateway,22 of key facility-level and component-level activity data at an appropriate spatial scale could improve national methane emission estimates while supporting improved gridded inventories.
The EPA has in recent years employed activity data reported through the GHGRP to improve the emission estimates for some of the emission sources in the GHGI. For example, component-level emission factors and activity data have been improved in recent years due to the GHGRP reports by individual facilities by essentially scaling up the reported activity data for certain emission sources to a national total for use in the GHGI (EPA, 2017a,c). As noted earlier, the GHGRP reports emissions from a subset of the national total emission sources, generally considered as larger facilities or facilities with higher throughput. For example, Zimmerle et al. (2015) estimated that the 2012 GHGRP reported emissions from 25 percent of transmission and storage facilities, accounting for about 15 percent of the modeled emissions from the study. The 2015 total GHGRP methane emissions reported is only about 35 percent of the 2015 GHGI estimates for the petroleum and natural gas industry. However, the activity data are generated independently by each reporting facility and therefore provide a finer-resolution estimate than was possible prior to the availability of GHGRP data, which strengthens the value of this information. Although this new information is an improvement, uncertainties related to data errors have been noted.23 For example, the 2015 total volume of natural gas produced at wells for all facilities reporting to the GHGRP exceeds volumes reported by the EIA.24 Other data errors include data incon-
21 Natural Gas Annual Respondent Query System (EIA-757 Data through 2014).
22 See http://www.noggateway.org/.
23 The issues identified here are not based on a comprehensive review of all of the GHGRP data available through Envirofacts, but rather are an illustration of issues identified in conducting various data analysis activities. As a result, there may be additional issues that are not included here.
24 EIA reports 32,894,727 million cubic feet (931,466 million cubic meters) in 2015.
sistencies where reported hours of operations exceeded the total operating hours in a year. Further, since the GHGRP facilities are a subset of the total national population, the use of GHGRP activity data or developing national-level emission factors from the GHGRP for the entire population of emission sources needs to be done with caution.
Component-level estimation methods support granular-level component or subcomponent-level (e.g., different types of storage tanks or pneumatic controllers or different types of meters/regulators) emission inventory development. Therefore, year-to-year change in emissions profiles can be better explained by component-level estimates. Some recent methane studies, namely Marchese et al. (2015) and Zimmerle et al. (2015), report methane emissions from various segments of the petroleum and natural gas industry, presenting facility and subfacility levels instead of discrete component-level estimates. The results have now been incorporated into the GHGI calculations, reflecting EPA efforts to improve methodologies based on new information, when information is presented in a format that allows for incorporation. In addition to providing facility-level estimates, these methods could also potentially be used to provide a mechanistic understanding of emissions (i.e., using operational data and parameters along with observed emissions from the facility to characterize the temporal and spatial profile of the resulting emissions). Along with site access and data, a mechanistic understanding allows scientists to characterize and explain the emissions associated with the specific time duration of the study and the top-down and bottom-up emission estimation differences, thus negating the need to draw statistical assumptions related to frequency of emission events from limited observations (e.g., Bell et al., 2017; Schwietzke et al., 2017; Zimmerle et al., 2016, 2017).
Top-down studies also have crucial value in evaluating whether improvements to component-level emission estimates actually capture the bulk of total facility emissions. Recent studies have indicated that there may be unaccounted-for categories in current, component-level inventories (e.g., the GHGI), implying that component-level inventories may be incomplete (see further discussion on unaccounted-for emissions later in this chapter). These unaccounted-for emissions may be due to inaccurate activity data or high-emitting sources that are not accounted for in current emission factors; they may also be due to missing emission source categories. Prior unaccounted-for emissions from gathering and boosting stations now form the single largest emission source from natural gas systems, accounting for about 30 percent of the total natural gas emissions.
Littlefield et al. (2017) employed the term “unassigned” emissions to define the difference between observed or measured emissions at the facility level and the sum of component-level measurements (i.e., observed but unassigned to specific emission
components) at natural gas production sites. Using data from one petroleum and gas production area (Barnett) in the United States, Littlefield et al. estimated these unassigned methane emissions to be about 19 percent of the entire natural gas supply chain emissions. Similarly, Zavala-Araiza et al. (2017) concluded that there is a 52 percent difference between site-based estimates from the Barnett and component-based methane estimates of the petroleum and gas production sector due to existence of high-emitting sources (also known as “super-emitters”; see Box 2.2 and Chapter 3 for more details) due to abnormal conditions. Zimmerle et al. (2015) found that while the study model estimate (SME) of 1,237 Gg yr–1 [range = 950 to 1,680] (or 1.24 Tg yr–1) was lower and not statically different from the 2012 GHGI estimates from transmission and storage facilities (1,805 Gg yr–1 [range = 1,460 to 2,350], or 1.81 Tg yr–1), the author
estimated that about 23 percent of the SME attributable to such high-emitting sources resulted from abnormal events that are not accounted for in the GHGI.
More recent studies (Bell et al., 2017; Schwietzke et al., 2017; Zimmerle et al., 2016) provide a mechanistic understanding of the emission differences and indicate that finer temporal and spatial contemporaneous measurements, along with site access and activity data, can explain discrepancies between observed “peak emissions” (existence of a diurnal cycle with maximum emission intensity during mid-day) and annual average emissions, the latter being what forms the basis of the GHGI.
The need for component-level inventory estimates is justified mainly to support methane mitigation actions. Component-level methods provide better characterization of changes in emissions and therefore support appropriate operational and policy methane mitigation actions. The GHGRP’s Subpart W is designed to provide facility-scale estimates by aggregating the component-level estimates at a facility that relies on multiple methods ranging from engineering estimates using default activity or emission factors to estimates from direct measurements. However, facility-level methods have the advantage of providing “completeness” of inventories by reducing uncertainties associated with activity data counts in the petroleum and gas industry, and are able to employ multiple measurements (satellites, aircrafts, towers, mobile vans, tracers) to verify and characterize facility-level emission factors.
The extensive network and complex composition of petroleum and natural gas infrastructure in the United States warrants careful consideration of how emission factors and activity data are developed and aggregated for use in inventories. Priority emission subcategories for improvement are those whose emissions estimation primarily relies on activity data and/or emission factors that originated from the 1996 EPA/Gas Research Institute study, including pneumatic controllers, fugitive emissions, and engine exhaust emissions through comprehensive data measurement and analysis.
Employing additional research to investigate the benefits and accuracy resulting from the use of facility- versus component-level emission data would further the development of national-scale emission inventories for petroleum and natural gas sources.
Methane Emissions from Landfills
Estimated 2015 waste sector methane emissions reported by the GHGI totaled 5.30 Tg. The dominant source was landfill methane (87 percent) with small estimated contribu-
tions from wastewater and the composting of garden and food waste. Current methodologies for the GHGI and GHGRP for all three sources have high uncertainties (IPCC, 2006). This section focuses on landfilling as the major source of methane from the waste sector (see Appendix C for a brief discussion of wastewater emissions). As discussed below, current methodologies (i.e., IPCC, 2006) assume a primary dependency for landfill emissions on the mass of landfilled waste that is not supported by current literature.
Biogenic methane in landfills is the end product of anaerobic microbial decomposition of paper products, garden, and food waste via methanogenic pathways. Most small, marginally engineered landfills closed in the mid-1980s prior to the implementation of EPA Subtitle D (Resource Conservation and Recovery Act) engineering and operational standards. Thus, at present, U.S. landfills are highly engineered and highly regulated facilities requiring monitoring and control of liquids and biogas. In addition, quarterly gridded sitewide surveys are required to detect and remediate elevated surface concentrations of methane. Engineered biogas recovery is routinely mandated at larger sites under Clean Air Act regulations to mitigate emissions of nonmethane organic compounds, which are trace components of landfill biogas. The recovered gas is either flared or, depending on site-specific economics, utilized for local energy needs. The commercial use of landfill biogas in the United States began in 1975 (Palos Verdes, California landfill). Encouraged by renewable energy tax credits, other incentives, and the EPA Landfill Methane Outreach Program (LMOP), utilization now occurs at more than 600 U.S. sites, primarily for onsite electrical generation and sale to the local grid.25 Larger landfills must also report to the GHGRP, which as of 2015, included 879 sites with engineered biogas recovery systems to capture and combust methane and nonmethane organic compounds. Importantly, emissions of methane are also reduced via oxidation of methane to carbon dioxide + water vapor in landfill cover soils by indigenous methanotrophic microorganisms. Oxidation and net emission rates can vary greatly depending on site-specific cover soils and seasonal climate (Chapter 3).
The IPCC (1996, 2006) methodologies have historically been used for both GHGI and GHGRP estimates for landfill methane emissions. At the time of IPCC (1996), there were few comprehensive field measurements for landfill methane emissions (see discussion in Spokas et al., 2011, 2015). The IPCC (2006) revisions added calculation tools and defaults but did not fundamentally alter the underlying first-order decay (FOD) model, a first-order kinetic equation for estimation of methane generation from landfilled waste as the starting point for estimating emissions. In general, because of the necessity to consider complex soil gas transport and microbial methane oxidation
processes, simple “emission factor” multiplied by “activity data” (i.e., landfilled waste) calculations are inappropriate for landfill methane emissions. Moreover, based largely on 1990s science prior to most field measurement campaigns, the current IPCC (2006) methodology includes several assumptions of questionable validity, as discussed below.
Modeling of Methane Generation Using First-Order Kinetic Equation
The IPCC (2006) methodology relies on a single first-order kinetic equation (termed the FOD model) to estimate methane generation at all engineered landfills worldwide. In this model, the kinetic constant (k, 1/t) is assumed to be related to climate (i.e., highest value for warm humid sites). Also, biogas generation from a given mass of buried degradable carbon is assumed to peak in the year of disposal and decline exponentially thereafter. This approach is based on the assumption that a landfill functions like an anaerobic digester and was specifically based on an empirical model originally applied to biogas generation at a single California landfill (e.g., Scholl Canyon Model, EMCON Associates, 1980). Validation for the original IPCC (1996) methodology was limited to a comparison of modeled biogas generation to measured biogas recovery (not emissions) at nine Dutch landfills (Oonk and Boom, 1995; Van Zanten and Scheepers, 1995). A simple mass balance is then applied to derive the emitted methane; for engineered landfills, this is assumed to be equal to 90 percent of the modeled methane generation minus the measured or assumed methane recovery. The 90 percent allows for 10 percent annual oxidation in cover soils and has often been coupled in the United States (e.g., California GHG reporting) with an assumption of 75 percent biogas recovery “efficiency.” Concerns with this approach include the reliance on limited data, the 10 percent annual oxidation (discussed below), global application of the FOD model (also discussed below), and the lack of field validation. Historically, the United States has also not compiled site-specific data on landfilled waste mass; thus, national, regional, and state data have typically been used as the basis for GHGI reporting to the UNFCCC.
Use of Historic U.S. Landfilled Waste Estimates for GHGI Reporting
There has been a long-recognized discrepancy between two disparate estimates for annual U.S. waste generation that are the current basis for estimating landfill emissions. These are the EPA material flow model approach developed in the 1970s by Franklin and Associates and the much larger (greater than 50 percent) annual values reported in periodic Biocycle Magazine/Columbia University “State of Garbage” reports
compiled from state-level reporting.26 Recently, 2012 landfilled waste (summed from specific sites for the U.S. GHGRP) was documented to be more than double the number reported by the EPA (Powell et al., 2016). In general, the differences have to do with inclusion of landfilled construction and demolition debris (which generally does not lead to significant methane emissions) as well as other waste streams not captured by the current EPA estimates. For the future, harmonization of U.S. waste generation and management data, including annual statistical summaries for various subsources (i.e., household, construction, business/commercial, industrial, and forestry) could greatly improve tracking of national and regional trends for waste generation, recycling, and disposal. Moreover, achieving consistency and coordination with Eurostat’s waste data methodology27 could also be a desirable strategy to eventually achieve waste data harmonization among the United States, Europe, and other highly developed countries. At a minimum, more rigorous accounting of diverse waste generation and management activities could enable better local tracking of biodegradable waste diversion from landfills (typically to composting and anaerobic digestion; e.g., Brown, 2016); this is hampered by the high uncertainties associated with current national methods.
IPCC (2006) Assumption of 10 Percent Soil Oxidation
Many factors influence temporal soil oxidation rates; thus, relying on a fixed value is not appropriate. The 10 percent annualized value in IPCC (2006) was derived from the first study in the literature (Czepiel et al., 1996a,b) and was appropriate only for the specific New Hampshire study site. Realistically, methanotrophic oxidation rates in landfill cover soils are related to soil temperature and moisture variations over an annual cycle in each site-specific cover soil (e.g., Spokas et al., 2015, Fig. S9; see also Bogner et al., 2011; Scheutz et al., 2009; Spokas and Bogner, 2011; Spokas et al., 2011). Thus, soil oxidation is a climate-related feedback on landfill emissions. Overall, cover-specific oxidation and “net” emission rates on a unit area basis (e.g., grams of methane per meter per day) can vary by five to six orders of magnitude. Soil oxidation rates can range from negligible (too hot or cold, too wet or dry) to greater than 100 percent of the methane transport rate from the anaerobic zone (e.g., net uptake of atmospheric methane by cover soils with high oxidation capacities). Optimal soil temperature for oxidation is about 35°C with optimal soil moisture potential near the water-holding capacity (about 10 kPa) (Spokas and Bogner, 2011).
26 See https://www.biocycle.net/2010/10/26/the-state-of-garbage-in-america-4.
27 For example, europa.eu/eurostat/statistics-explained/index.php/Waste_statistics.
Assumed Dependency of Methane Emissions on Mass of Landfilled Waste
The direct application of IPCC (2006) at the facility level can result in a robust linear correlation between waste in place (WIP) and estimated emissions. To give one example, for 2011 California emissions (372 sites), plotting 2011 estimated emissions from the California Air Resources Board versus WIP (using 75 percent recovery efficiency and 10 percent soil oxidation), this regression indicates 191 Mg CH4 emitted per Tg WIP (r2 = 0.88) (Spokas et al, 2015; see Chapter 4, Figure 4.2, for highest-emitting sites using IPCC methodology). This means that large sites cannot reduce their reported emissions below a certain threshold as defined by this relationship, regardless of additional mitigation measures such as increased density of biogas recovery wells, installation of horizontal collectors concurrent with filling, or construction of thicker cover soils. This is not realistic and tends to stifle site-specific mitigation strategies because no measurable emission reductions can be derived using the IPCC (2006) method.
Conversely, a simple linear dependency for methane generation and recovery from a given mass of landfilled waste can now be readily demonstrated using 2010 data from 129 California landfill sites inclusive of small/large, new/old, open/closed sites from all climatic regions of California. Measured site-specific methane recovery was simply and linearly related to the mass of WIP at the rate of approximately 125 Nm3 hr–1 methane recovered per Tg WIP (r2 = 0.85) (Spokas et al., 2015).28 This indicates demonstrable steady-state methane generation based on a large California dataset spanning more than half a century of controlled landfilling practices, in contrast to the FOD assumptions of IPCC (2006) where methane generation and emissions peak shortly after waste deposition. This is a surprising finding, given the diversity of California landfill sites, and requires further investigation. However, a similar relationship can be derived from the 2015 GHGRP database containing 879 sites with biogas recovery (96 Nm3 hr–1 methane recovered per Tg WIP; r2 = 0.59). In general, this suggests that landfill processes are longer term and more akin to (controlled) geological burial of organic carbon than the “anaerobic digester” analog assumed by the current FOD model.
Increased Complexity of Recent GHGRP Calculations
In recent years, the GHGRP has also applied IPCC (2006) for site-specific methane emission estimates for reporting sites. However, there are four different sets of re-
28 “N” is for normal conditions of 0°C temperature and 101 kPa pressure used globally.
ported emissions (equations HH-5 through HH-8 in 40 CFR Part 98). These calculations depend on (1) whether biogas collection is considered and (2) what assumptions are made for biogas “collection efficiency,” as well as (3) variable assumptions for percent oxidation from negligible to 35 percent (generally representative of average oxidation in the literature, e.g., Chanton et al., 2011). These more complex estimation methods have never been field validated for site-specific application and, in general, tend to magnify rather than reduce the shortcomings of the FOD method for the GHGI, as discussed above.
Need for a Field-Validated Process-Based Emission Model
The shortcomings of the current methodology indicate that a bottom-up process-based model is needed that can specifically address cover-specific methane emissions inclusive of site-specific climate. One such model has been developed and field validated, both for California and internationally, during the last decade. California Landfill Methane Inventory Model (CALMIM) 5.429 requires minimal site-specific data for inventory purposes and incorporates both the known dependencies for landfill methane emissions (soils, operational factors) with 0.5° × 0.5° U.S. Department of Agriculture climate models. It is also compatible with IPCC Tier 3 guidelines, has already been applied to a revised 2010 inventory for California (Spokas et al., 2015), and is being further evaluated in a current California research project by California Polytechnic State University and the University of California Irvine, supported by the California Air Resources Board and Department of Resources Recovery and Recycling (Calrecycles). See additional discussion in Chapters 3 and 4 as well as further information available in Bogner et al. (2011, 2014), Spokas and Bogner (2011), and Spokas et al. (2011, 2015). As discussed in Chapter 3, the GHGI and GHGRP might investigate an improved methodology focusing on the processes now known from literature to control emissions, for example, site-specific climate and operational factors (cover soil area, thickness, and composition; extent of engineered biogas recovery). A primary focus on site operational data as well as site-specific measurements for critical variables (e.g., WIP, average annual biogas recovery rate, and average annual methane concentration) has the potential to greatly improve and harmonize inventory calculations with the recent scientific literature. For example, de la Cruz et al. (2016) recently determined that IPCC (2006) can result in methane emission estimates ranging from several times to as much as 30-fold too high.
National harmonization and statistical analysis of waste generation, diversion,
29 See ars.usda.gov.
processing, and disposal data are needed to better quantify national trends for consistency with international norms (i.e., Eurostat).
The current landfill methane U.S. Greenhouse Gas Inventory methodology based on IPCC (2006) was never field validated for emissions, relies on the mass of waste in place as the major driver for emissions, assumes that methane generation rates peak in the year following disposal, and does not directly consider major mitigation strategies such as increased density of biogas recovery wells or increased seasonal oxidation in thicker cover materials. Utilizing a field-validated, process-based model for inventory development has the potential to inform and incentivize improved landfill design and operational strategies to yield quantifiable reductions of landfill methane emissions.
Methane Emissions from Coal Mines
In 2015, total U.S. methane emissions related to coal mining (including abandoned mines) were estimated to be 2.69 Tg (EPA, 2017b). The United States was the second largest coal producer in the world in 2015, producing 812 Tg, following only China (EIA, 2016). In 2015, there were 834 coal mines in the United States: 529 surface and 305 underground (EIA, 2016; see Figure 2.7). Methane is emitted to the atmosphere from three types of coal-mining–related activities: underground mining (active and abandoned), surface mining, and post-mining activities (e.g., coal handling). Active underground mines are the largest source of coal methane emissions, responsible for approximately 66 percent of total emissions from this category in 2015 (Figure 2.8). Surface mines account for a large fraction of coal production in the United States (66 percent in 2015) but, due to the typically lower gas content of shallow surface-mined coals compared to deeper coals mined underground, were responsible for just 13 percent of emissions. Post-mining activities (surface plus underground) and abandoned underground mines make smaller contributions to emissions (11 and 9 percent, respectively, in 2015) (EPA, 2017b; see Figure 2.8). The EPA estimates of methane emissions from coal mines discussed below follow the 2006 IPCC Guidelines.
Methane emissions from underground mines come from two sources: (1) ventilation systems and (2) degasification systems. Emissions from active underground mines are estimated based on data collected in individual mines. If methane is recovered and utilized (which occurs at some gassy mines), then recovered volumes are subtracted
from the total, resulting in the estimate of net emissions to the atmosphere. Ventilation systems are responsible for the majority of coal mine emissions (Figure 2.8), even though the methane is not highly concentrated from this source. In contrast, degasification systems release highly concentrated gas, and often in large volumes, but because they are used only in selected gassy mines, their total national emissions are much lower than those from the ventilation systems.
Ventilation-system methane emissions in the GHGI are estimated using information from the GHGRP, the U.S. Mine Safety and Health Administration (MSHA), and occasionally from site-specific sources (e.g., state gas production databases; EPA, 2017b). Since 2011, the gassiest mines (that emit more than about 36,500,000 cubic feet [or 1,033,565 cubic meters] of methane per year) report their emissions to the GHGRP and have the option of employing their own measurements or using measurements collected by the MSHA as part of required quarterly inspections. Currently, the GHGI estimates from this source are based on GHGRP data obtained from underground mines that report the emissions and quarterly measurement data obtained from MSHA for the remaining mines. For ventilation systems, methane emission calculations rely on the measurements of the flow rate and the methane content in the ventilation
air. Both of these measurements come from underground observations in entries connected to the ventilation shafts. Because the emission assessment relies on individual measurements in the entries rather than the total output on the surface, accurate reporting can be obtained by visiting and measuring all entries. Usually, there are no measurements conducted at the surface to confirm the values obtained underground, yet such measurements are needed to verify the current techniques.
Degasification systems are composed of wells drilled from the surface to the mine or boreholes drilled inside the mine that remove mine gases before, during, or after mining. Gassy underground mines report weekly measurements on the emissions from degasification systems to the GHGRP, and these data are summed to obtain yearly emission estimates (EPA, 2017b). For the 16 mines that have methane recovery and use projects, the amounts of methane recovered or destroyed (by flaring or thermal oxidation) is subtracted from the total volume of degasification system emissions.
For underground mines, measurements conducted at the surface are needed to verify the values obtained underground.
Surface Mining and Post-Mining Activities
Methane emissions in the GHGI from surface mining and post-mining activities30 are estimated by multiplying basin-specific surface mine coal production by basin-specific gas content of the surface-mined coal (occurring up to about 76 meters deep) and by one emission factor for the United States. Basin-specific in situ gas content data used come from relatively outdated compilations by Rightmire et al. (1984) and Diamond et al. (1986). Including additional gas content data collected from various coal basins over the last 30 years is needed to decrease uncertainties. The emission factor currently used in the United States is based on 150 percent of the in situ gas content of the coal (EPA, 2017b) because mine-specific data are not available. This approach is limited in that it cannot account for variations in methane content of coal across different coal basins and different mines.
Post-mining activities utilize basin-specific coal production multiplied by basin-specific gas content and an emission factor of 32.5 percent of the in situ gas content of the coal (EPA, 2017b) to account for emissions during coal transportation and storage, based on a study of British coals (Creedy, 1993). For comparison, a 20 percent factor is used by Australia, after data by Williams et al. (1993). Similar to surface mines, because of variations in gas content between coals within the same basin (e.g., Strąpoć et al.,
30 The GHGRP does not include reporting from operations at surface mines or post-mining activities.
2008), both basin-specific gas content and an emission factor of 32.5 percent can be sources of significant uncertainties in methane estimates from post-mining activities.
Updated and expanded information on coal gas content at mined depth for individual mines would allow for the development of more precise and representative methodologies for estimating surface coal mine methane emissions.
Abandoned Underground Coal Mines
Methane emissions from abandoned underground coal mines vary depending upon many factors including gas content of the coal, mine flooding, presence of conduits, the quality of mine seals, and the time since mine closure. The EPA produced a comprehensive inventory of abandoned underground coal mines in the United States (EPA, 2004) and developed a methodology to evaluate their methane emissions, which was subsequently incorporated into the 2006 IPCC Guidelines. The EPA classifies abandoned underground mines into three categories: venting mines, flooding mines, and sealed mines, with the estimation methods modified for each category to reflect different emission levels. Emission estimates from abandoned underground mines take into account the emissions during the active phase of the mine and use emission rate decline curves that have been developed for abandoned coal mines. These decline curves are used to evaluate gas emission potential and duration (Karacan et al., 2011). For the mines where venting is reduced because of mine flooding, decline curves are adjusted based on field measurements, and the modified equations are used to calculate gas flow rates (EPA, 2004). For mines with very good seals, the emission predictions are similar to the vented mines, but using lower initial rates dependent on the degree of sealing (high, middle, or low values; EPA, 2004). The main challenge with the estimation of methane emissions from the abandoned underground mines is generation of an accurate decline curve. Methane adsorption isotherms, coal permeability, and pressure at mine abandonment are needed to establish a reliable decline curve, and typically, there is uncertainty in all three parameters.
Unaccounted-for Methane Emission Sources in the GHGI
Although several methane emission sources are already accounted for in inventories, recent studies indicate that for certain end-use sectors, certain discrete emission sources may be missed or underestimated in the current GHGI (Fischer et al., 2017;
Lamb et al., 2016; Lavoie et al., 2017; McKain et al., 2015; Mehrotra et al., 2017).31 Broadly, these unaccounted-for sources could be divided into two groups.
The first group is known sources that may not be fully accounted for in inventories. Examples include high-emitting sources (Box 2.2) and sources in certain natural gas end-use sectors such as residential and commercial operations, power plants,32 refineries,33 and transportation.34 Recent research indicates that some source categories may not fully account for emissions from all sources within these facilities (Clark et al., 2017; Fischer et al., 2017; Lamb et al., 2016; Lavoie et al., 2017; McKain et al., 2015; Mehrotra et al., 2017). Lavoie et al. (2017) presented the first facility-scale methane measurements from three natural gas power plants (NGPPs) and three petroleum refineries. Their estimates were larger than the GHGRP submittals for these facilities by factors of 21 to 120 (NGPPs) and 11 to 90 (refineries), pointing to potential methane sources unaccounted for in the GHGRP and the GHGI. The authors scaled up these emissions to annual methane emissions of NGPPs and refineries and estimated that these sources contribute about 0.61 ± 0.18 Tg methane yr–1 to the United States. In another study, Fischer et al. (2017) estimated methane emissions for three refineries in California and suggested that the GHGRP estimates were underreported by a factor of 14. In the transportation sector, Clark et al. (2017) estimated “pump-to-wheels” (PTW) methane emissions from the heavy-duty transportation sector fueled by natural gas. Although the study does not compare results to the GHGI or GHGRP, it does indicate that engine crankcase and tailpipe emissions contribute to about 70 percent of the total PTW emissions.
Another example of another known and likely not fully accounted-for source is the dispersed, diffuse emissions occurring in urban areas. In addition to the more familiar “street leaks” from pipelines in the distribution system (Lamb et al., 2016; McKain et al., 2015; von Fischer et al., 2017), these emissions may also arise “after the meter,” that is from leaks in homes and commercial buildings due to unburned gas from appliances or venting of pressure regulators, or due to equipment such as ovens and hot water heaters (Fischer et al., 2017; Mehrotra et al., 2017).
31 The GHGI does not account for abandoned petroleum and natural gas wells within the petroleum and natural gas system inventories. EPA (2017b) acknowledged this limitation and has proposed revisions in June 2017 for incorporation in the 2018 GHGI to account for abandoned petroleum and natural gas wells.
32 The methane emissions from fossil fuel combustion at power plants in the 2015 GHGI were estimated to be about 0.02 Tg (EPA, 2017b).
33 The 2015 methane emissions from petroleum refining were estimated to be approximately 0.03 Tg in the GHGI or 0.3 percent of the total estimated emissions from petroleum and natural gas systems (EPA, 2017b).
34 The 2015 total methane emissions from the transportation sector were estimated to be about 0.08 Tg, according to the GHGI (EPA, 2017b).
Urban settings are particularly challenging areas in which to assess methane emissions because of the presence of a multitude of commercial and residential sources. Many of these are of small scale, and their pattern of emissions is difficult to predict. In such settings, a multifaceted approach is especially desired to identify the sources and find a linkage between sources and emissions. McKain et al. (2015) analyzed data from a long-term measurement network consisting of four measurement tower sites in eastern Massachusetts and estimated an annual average methane emission rate from the study area to be two to three times higher than the Massachusetts state inventory. The authors attributed the difference between the study estimates and other inventories to unaccounted-for methane emission sources “downstream of customer meters” at industrial, residential, and commercial end-use facilities; however, they did not provide an estimate of such emissions. In another example, Fischer et al. (2017) measured methane emissions at single-family homes in the San Francisco Bay Area, California. They found methane leakage at nearly every home, even when the natural gas appliances were not used. Extrapolating these results statewide, the authors estimated unaccounted-for methane emissions from the residential sector to be about 0.2 percent of residential natural gas consumption. These emissions would add an additional 10 percent to the petroleum and gas methane emissions currently reported in the California state inventory.
The second group of unaccounted-for sources is previously unknown sources that have been unrecognized because of their scale, complexity of attribution, or other factors and are not currently incorporated in the GHGI. An example is previously unrecognized microbial coal-bed methane from active mining identified as part of a study in the Four Corners region (e.g., Arata et al., 2016).
All of these methane sources contribute to underestimating of bottom-up emissions, and their presence creates emission uncertainties. Though these studies of unaccounted-for emissions are limited in sample size to develop national emission factors or extrapolate to develop a national estimate, these studies suggest that further research is necessary to better characterize emissions from these sources. Increased, comprehensive research and methane measurement and monitoring could provide important insights and evaluation of the magnitude of these emissions for potential inclusion of these emission sources in future methane inventories, in particular, the GHGI (Chapter 4).
Additional research to better characterize emissions from unaccounted-for emission sources would likely further the incorporation of these sources into the U.S. Greenhouse Gas Inventory and other methane inventories. Given that observations of unaccounted-for methane emissions are limited, any extrapolation to national totals needs to be done with caution.
RECALCULATIONS IN THE GHGI
As indicated earlier in this chapter, the United States and other countries must use 1990 as a base year when estimating methane emissions for submission to the UNFCCC. Emission estimates from previous years are often recalculated in the GHGI based on new information so as to ensure that observed trends in emissions are real and not an artifact of the use of different data sources and methods. New information to better estimate emissions is becoming available due to the GHGRP and several methane measurements and activity data studies conducted recently which supports recalculations of methane emissions for certain sources. In general, recent revisions to the GHGI are more accurate than previous estimates and are therefore encouraged. The requirement to recalculate emissions back to 1990 using consistent data source and methods, however, poses distinct challenges for the United States (and other countries) to accurately estimate methane emissions. New research and estimation techniques may provide a more robust estimate of methane emissions. However, application of current emissions and activity data to develop a consistent time series back to 1990 can be difficult if the updated activity data and emission factors are not available for the entire time series or if new emission factors are not applicable to earlier years due to changes in technology and/or practices in the industry.
In addition to the technical challenge of developing a consistent time series of data, there is a challenge in communicating the results of such recalculations. Use of updated data can result in significant revisions in estimates for any given year when recalculated using new information. Figure 2.9 illustrates this, showing, for instance, that estimated emissions for 1990 were reported to be higher when calculated using information available in 2017 when compared to information available in 1998. Drawing inferences related to changes in emissions from key categories from prior years should be done with caution. Improvements in emission factors and activity data are encouraged and revisions reflect the use of updated technical data.
Although the 1990 baseline has remained for the UNFCCC and therefore GHGI development, recent global and U.S. policy commitments have used either 200535 or 201236 as a base year for emission reduction strategies.
The challenge of recalculating emissions back to 1990 should not be a barrier for utilizing the most up-to-date methodologies and information in developing
35 U.S. Nationally Determined Contribution under the Paris Agreement. http://www4.unfccc.int/ndcregistry/PublishedDocuments/United%20States%20of%20America%20First/U.S.A.%20First%20NDC%20Submission.pdf.
36 The Obama administration adopted a 40-45 percent reduction from 2012 levels by 2025. This commitment is also shared by Canada and Mexico.
the U.S. Greenhouse Gas Inventory. The United States and the broader international community could consider adoption of the use of an alternative base year or period for reporting of national GHG inventories consistent with more recent national and international policies and commitments.
FUTURE U.S. METHANE EMISSIONS
Projecting future methane emissions is extremely challenging. Accurate projections of anthropogenic methane emissions are a key foundation for planning national policies or goals, but these projections are dependent on many factors that are difficult to
predict, including future energy and agricultural policies, methane mitigation policies, development of natural resources, population migration, etc. Unlike projections of carbon dioxide, which are strongly correlated with fuel combustion, methane emissions stem from several sources across various segments of the economy and are influenced by many distinct factors. As previously described in this report, many challenges and knowledge gaps remain in quantifying current emissions, adding complexity to future projections. Finally, climate change during the next few decades will also directly affect future methane emissions from several sources.
The United States reports sectoral projections through its National Communications37 every 4 years, and every 2 years through its Biennial Reports issued by the U.S. Department of State to the UNFCCC. The reporting guidelines specify that countries should report projections (1) by gas and by sector, (2) without mitigation measures, (3) with measures (encompasses currently implemented and adopted policies and measures), and (4) with additional measures (also includes planned measures). However, the guidelines do not specify a method for projecting these emissions. U.S. projections are based on the methods outlined in Methodologies for U.S. Greenhouse Gas Emissions Projections: Non-CO2 and Non-Energy CO2 Sources (EPA, 2013). Emission estimates are calculated by scaling the most recent GHGI with macroeconomic data (e.g., population and gross domestic product), projections of activity data (e.g., natural gas production from the Energy Information Administration), and current mitigation policies and programs (e.g., Landfill Methane Outreach Program, Natural Gas STAR, Clean Air Act, and Clean Fuels Programs, as well as state-level programs). The most recent national projections are presented in the 2016 Biennial Report (DOS, 2016), which includes projections of total U.S. methane in 2020 (26.80 Tg methane), 2025 (26.96 Tg methane), and 2030 (27.28 Tg methane), as well as emissions by major source category. The 2025 and 2030 values are about 1 to 2 percent lower than 2015 emission values.
The quality of these national projections can only be as good as the quality of the most recent inventory estimates combined with the emission factors and activity data applied for future estimates. Moreover, activity data represented in the EPA 2015 GHGI (for calendar year 2013) projections which were employed in the last biennial report may not be sufficiently robust. As noted in earlier sections, methane inventories have been recalculated in recent years to account for improved activity data, emission fac-
37 See http://unfccc.int/national_reports/items/1408.php. The last U.S. biennial report was submitted in December 2015 and employed the GHGI published in April 2015 covering emissions from 1990 to 2013 (EPA, 2016a).
tors, and methods. This illustrates the added uncertainty of using current inventories to develop future methane projections to 2030 and beyond.
The current methods established for projecting methane emissions from key categories may not be the best predictor of future emissions. For example, while the trends in emissions from livestock generally follow changes in cattle populations (e.g., Alemu et al., 2011), cattle population trends alone cannot be a predictor for enteric methane emissions. Animal live weight and feed dry matter intake have been steadily increasing. Both of these factors result in greater enteric emissions. Animal productivity (milk or daily gain) has also been increasing, which results in decreased emission intensity (i.e., emission per unit of product). Similarly, petroleum and natural gas emissions have significant regional variability with regard to technological improvements and voluntary actions, resulting in variable emission projections and activity data correlations. For the waste sector, improvements in the collection and statistical analysis of relevant waste data are needed (including waste generation from households and businesses, internal diversions/recycling/reuse within sectors, and the mass of biodegradable waste disposed to landfill), as are improvements to current methodologies discussed in this chapter.
For several sources where methane is generated, transported, and/or oxidized in soils and sediments, future climate is also an important determinant for future emissions at specific global locations. Sources include methane emissions due to subsurface leakages from petroleum/gas piping, land application of manures, some coal-bed leakages, and seasonally variable emissions from landfill cover soils. Subsurface methane transport to the atmosphere can be largely diffusive (from higher to lower methane concentrations) with both methane transport and soil oxidation rates dependent on transient soil temperature and moisture. Future climate could be factored into future emission estimates using a combination of climate projections and field-validated process-based modeling for specific sources.
Activity data and emission factors that serve as proxies for projecting future methane emissions, as well as current inventory methodologies, need to be robustly investigated regarding correlations to measured emissions from key categories. Such investigations could include updating the EPA report Methodologies for U.S. Greenhouse Gas Emissions Projections: Non-CO2 and Non-Energy CO2 Sources in collaboration with experts from federal and state agencies, municipalities, industry, and the research community. Future climate projections also need to be factored into future emission estimates for sources where methane is generated, transported, and oxidized in soils and sediments.