Estimating Total Methane Emissions from the Denver-Julesburg Basin Using Bottom-Up Approaches

Abstract

Methane is a powerful greenhouse gas with a 25 times higher 100-year warming potential than carbon dioxide and is a target for mitigation to achieve climate goals. To control and curb methane emissions, estimates are required from the sources and sectors which are typically generated using bottom-up methods. However, recent studies have shown that national and international bottom-up approaches can significantly underestimate emissions. In this study, we present three bottom-up approaches used to estimate methane emissions from all emission sectors in the Denver-Julesburg basin, CO, USA. Our data show emissions generated from all three methods are lower than historic measurements. A Tier 1/2 approach using IPCC emission factors estimated 2022 methane emissions of 358 Gg (0.8% of produced methane lost by the energy sector), while a Tier 3 EPA-based approach estimated emissions of 269 Gg (0.2%). Using emission factors informed by contemporary and region-specific measurement studies, emissions of 212 Gg (0.2%) were calculated. The largest difference in emissions estimates were a result of using the Mechanistic Air Emissions Simulator (MAES) for the production and transport of oil and gas in the DJ basin. The MAES accounts for changes to regulatory practice in the DJ basin, which include comprehensive requirements for compressors, pneumatics, equipment leaks, and fugitive emissions, which were implemented to reduce emissions starting in 2014. The measurement revealed that normalized gas loss is predicted to have been reduced by a factor of 20 when compared to 10-year-old normalization loss measurements and a factor of 10 less than a nearby oil and production area (Delaware basin, TX); however, we suggest that more measurements should be made to ensure that the long-tail emission distribution has been captured by the modeling. This study suggests that regulations implemented by the Colorado Department of Public Health and Environment could have reduced emissions by a factor of 20, but contemporary regional measurements should be made to ensure these bottom-up calculations are realistic.

1. Introduction

Methane (CH₄) is a powerful greenhouse gas (25 times higher 100-year warming potential than carbon dioxide); emissions have been linked to a changing global climate, and it is considered to be the most important gas for short term climate benefits [1,2,3]. In recent years, the importance of accurately quantifying methane emissions from the surface of the Earth has increased, as these provide data on the largest emission sectors that can then be targeted for mitigation. Nearly all global greenhouse gas emissions are calculated using a bottom-up approach, where emissions are generated from the activity (number of emission sources per sector) multiplied by an emission factor [4,5,6,7]. However, many recent studies have reported that these bottom-up derived estimates may be significantly underestimated [8,9,10,11].

The main shortcomings of bottom-up emission approaches are that the inventory of emission sources may be incomplete, i.e., the activity data are wrong, or the emission factors may be incorrect. Taking the energy sector as an example, the current US EPA estimates do not include methane emissions from the gathering systems or pipelines from production sites to processing facilities; further, emission factors representing one million production sites across the US [12] were based on measurements at only 334 facilities [13,14]. The relatively small sample size is unlikely to capture the long-tail emission distribution typical of oil and gas production facilities, where the average emission is dominated by a small number of very large emitting sites [15,16,17,18]. Both omission and underestimation result in a likely systematic underestimation of the total regional emission.

Currently, the amount that bottom-up derived methods underestimate emissions is typically indicated when directly compared to emissions calculated by top-down methods. Top-down methods include drive-by surveys with a methane-specific analyzer mounted in a car [16,19], but they are more commonly conducted using aircraft surveys [17,20,21,22,23,24,25,26,27] or satellite observations [10,28,29]. Airborne surveys are typically mass balance experiments or use a downward looking camera that can detect methane to collect methane plume images; the measured methane mass is then coupled with meteorological data and the emission rate of the source is calculated [20]. Assuming a long-tail distribution and all plumes above the detection limit of the instrument are observed, the emission estimate of a region can be calculated. Next-generation aircraft-based platforms have a detection limit approaching 1 kg CH₄ h⁻¹ [20], while satellite-based detection limits are around 100 times more [24,28]. One key shortcoming of top-down approaches is that they survey the emissions for a finite period of time—minutes for aircrafts and seconds for satellites. This means that any longer-duration (days, weeks, years) emission quantified using a top-down observation must either assume it is seeing a constant emission from the source or use some statistical approach to fill in the gaps between observations. Another shortcoming of top-down observations is that they typically take place during the day and are potentially conducted at the same time as peak human activity at the site, i.e., maintenance, cleaning, or bulldozing, which is likely to result in observing higher than normal emissions. Regardless of the approach, it is likely that top-down methods that use observations during the day to calculate emissions overestimate total emissions.

Despite shortcomings of both approaches, reconciling regional bottom-up and top-down emission estimates remains the key challenge for mitigation to meet climate goals [30].

The majority of countries globally are required to report anthropogenic greenhouse gas emissions to the Intergovernmental Panel on Climate Change (IPCC) as part of the Paris Agreement [31]. Emissions can be calculated using a tier method by complexity, where Tier 1 methods are simple methods that use default values, Tier 2 methods use single country-specific emission factors, and Tier 3 methods are complex approaches that often use modeling to calculate emissions [2,6,32]. In the US, the Environmental Protection Agency (EPA) uses Tier 3 approaches to estimate that 29% of methane emissions come from natural gas and petroleum systems, 25% from enteric fermentation, 15% from landfills, 8% from manure management, 6% from coal mining, 6% from flooded land, 3% from land use and management, and 9% from others [33].

Even though the EPA uses the most complex methods, the IPCC stipulates that for reporting, it is still unclear how regionally representative of the actual emissions these methods are. A recent study [11] reported that emissions from oil and gas operations in the Delaware basin in Texas and New Mexico calculated using EPA Tier 3 emission factors were potentially an underestimate by a factor of five when compared to regional measurement-informed emission factors. The normalized losses were estimated at 0.57% and 2.8% for the Tier 3 and measurement-informed approaches, respectively. The main sources of uncertainty were identified as emissions from flaring-associated gas, the gathering pipeline system, maintenance events, and large fugitive events.

Methane emissions in the Denver-Julesburg (DJ) basin have been quantified by several top-down studies with normalized production loss estimates ranging from 2.1 to 4.1%. The emissions from the DJ basin are more complicated than the Delaware basin, as there are also emissions from the waste, agriculture, and natural sectors (Figure 1). In 2008, tower measurements suggested natural gas emissions were underestimated by a factor of 2, with the normalized loss estimated at 4% (range 2.3% to 7.7%) compared to a bottom-up inventory estimated at 1.6% [25]. In 2012, an aircraft-based mass–balance flux approach estimated total methane emissions in the DJ basin at 26,000 kg CH₄ h⁻¹ with agriculture, natural, and waste sources contributing 26%, and a gas normalized loss of 4.1 ± 1.5% [26]. In 2015, another aircraft-based mass balance experiment estimated methane emission from the DJ Basin at 24,000 kg CH₄ h⁻¹, ethane at 7000 kg h⁻¹, and normalized production loss at 2.1% of the 1.18 × 10⁹ m³ natural gas produced per month (equivalent to ~1400 MMscf NG day⁻¹) [27]. In 2020, a driving survey estimated total methane emissions from oil and gas operations in the DJ basin at 56 Gg, corresponding to a normalized gas loss of 0.26% [19].

Since the top-down measurements in 2012, environmental regulations for oil and gas production activities in the DJ basin have changed a great deal. The Colorado Department of Public Health and Environment (CDPHE) has been relatively rigorous in regulation implementation, which includes the following: in 2013, the CDPHE identified that regular leak detection could reduce emission rates by 31% and adopted leak detection and repair (LDAR) requirements in 2014; in 2014, the CDPHE established the first national regulatory requirements for oil and gas methane; in 2017–2023, the CDPHE performed on-going revisions to the Air Quality Control Commission Rulemaking, and Regulation 7 introduced requirements for compressors, pneumatics, equipment leaks, and fugitive emissions; in 2018, the CDPHE implemented regulations for flowline testing and guidelines to reduce venting from storage tanks; in 2019, a senate bill granted the CDPHE authority to regulate greenhouse gases, collect inventory data, and establish reduction targets; in 2022, the CDPHE approved funds to conduct aerial and ground-based methane monitoring. With the introduction of these regulations to oil and gas operations in the DJ basin, it is assumed that emissions have reduced, but it is currently unclear by how much.

In this study, we investigate the representativeness of IPCC and EPA emission factors for use in the DJ basin. For the purposes of this study, we set the boundaries of the study area between latitudes of 39.9 and 40.7, and longitudes between −105.3 and −104.2. Specifically, we aim to compare emission factors generated by three bottom-up inventories using 1. IPCC default Tier 1/2 emission factors; 2. Colorado-specific EPA Tier 3 emission factors; and 3. regional measurement-informed emission factors. We will compare the normalized emissions from oil and gas production to both the Delaware basin studies [9,10,11] and those generated by top-down approaches [25,26,27]. The overall aim of this study is to investigate the representativeness of regional oil and gas emissions, the sources of the largest emissions (production, midstream, etc.), and how regulation can affect emissions.

2. IPCC Emission Estimate

2.1. Methods

2.1.1. Energy Sector

For the production of oil and gas and midstream processes, methane emissions are calculated using the rate of gas and oil production rates. The Tier 1 methane emission estimates from the energy sector in the DJ basin use the 2022 production values from 14,047 oil and gas wells, which produce 253 million bbls of oil and 1.92 billion Mscf of natural gas [34], and IPCC emission factors for “Developed Countries” [35]. Emission factors and uncertainties are categorized as fugitive, vented, or flared emissions from gas production, oil production, servicing, gas transport, condensate transport, oil transport, and gas processing [1,2,6]. Following the methods of Riddick et al. (2024) [11], the Tier 1 methane emission estimate from the DJ basin is our ‘best-guess’ and the average of each emission factors’ maximum and minimum values. The methane content of the DJ basin’s natural gas was assumed to be 77.5% (range 70–85%) [19], but does not account for gas composition changes along the processing chain.

2.1.2. Agriculture Sector

Enteric fermentation

Location-specific, IPCC Tier 2 methane emissions from enteric fermentation in the DJ basin were calculated by multiplying the activity data (number of animals) and a regional emission factor. The livestock population data within the DJ basin were taken as the 2021 values for cattle, heifers, horses, sheep, swine, and poultry from the Colorado Department of Public Health and Environment (CDPHE) website (

Table 1. Tier2 IPCC emission factors for the animal types in the DJ basin.

Animal Type	Population	Emission Factor (kg CH4 head−1 Year−1)
Cattle	426,636	64
Dairy Heifer	285,403	138
Sheep	32,500	9
Poultry	65,430	0

) [36]. The CDPHE livestock population dataset was chosen as it is the only dataset that provides high resolution counts on the number of beasts in specific facilities across the Colorado county. This gives us the ability to generate animal populations for all facilities within the study area and not just county-aggregated values. The emission factors (kg CH₄ head⁻¹ year⁻¹) for enteric fermentation for livestock other than cattle were based on the US national average. The emission factor for cattle was taken as the Colorado state average presented in the Intergovernmental Panel on Climate Change (IPCC) 2006 and IPCC 2019 reports on greenhouse gas (GHG) emissions [35].

Manure management

For each species of animal, Tier 2 methane emissions from manure management in the US are generated for two productivity systems: high and low productivity (p). High and low productivity systems are defined for dairy cattle, other cattle, and the rest of the livestock separately (

Table 2. Volatile solids (VS, g VS kg animal mass⁻¹ day⁻¹), percentage use (%), and emission factors (EF, g CH₄ g VS⁻¹) of high-productivity management strategies used to calculate emissions by animal type; UAL—uncovered anaerobic lagoon; LS—liquid/slurry, >1 month; SS—solid storage; DL—dry lot; DS—daily spread; P—pasture/range/paddock; PM—poultry manure with litter.

Animal	VS	High-Productivity Manure Management Strategy
		UAL		LS		SS		DL		DS		P		PM
		%	EF	%	EF	%	EF	%	EF	%	EF	%	EF	%	EF
Cattle	7.6	-	-	1	31	43	42	14	1.2	-	-	42	0.6	-	-
Heifer	9.2	26	108	24	42	24	15	-	-	11	0.2	15	0.6	-	-
Swine	3.3	28	202	65	78	4	6	3	3	-	-	-	-	-	-
Sheep	8.2	-	-	-	-	54	46	-	-	-	-	46	0.6	-	-
Horse	5.6	-	-	-	-	50	50	-	-	-	-	50	0.6	-	-
Chicken	14.5	1	175	29	68	70	5	-	-	-	-	-	-	-	-
Broiler	16.8	-	-	-	-	-	-	-	-	-	-	-	-	100	0.6
Pullets	5.9	-	-	-	-	-	-	-	-	-	-	-	-	100	0.6

). For dairy cattle, the high-productivity systems consist of cows that are concentrated in confinement production systems or grazing on high quality pastures with supplements. The low productivity cattle system consists of low-yielding dairy cows, grazing on non-improved pastures, with locally produced roughage and agro-industrial by-products. For other cattle, high-productivity systems are based on animal feeding systems based on forage and concentrates in confinement producing systems or grazing with supplements or on improved pastures, producing high rates of daily weight gain, and vice-versa for low-productivity systems amounting to low weight gain. For other livestock species, high-productivity systems are 100% market-oriented with a high level of capital input requirements and high level of overall herd performance. The low-productivity system is meant for local markets or self-consumption with low capital input requirements and a low level of herd performance. For each productivity system, there are species-specific manure management systems which have an associated emission factor (EF, g CH₄ g VS⁻¹). The manure management systems used here were calculated for high-productivity systems and are recommended by IPCC as default for high income countries, especially the North American nations and the US. For each species-specific manure management system within each productivity system, the methane emission is calculated by multiplying the animal population (P), the annual average volatile solids per species head (VS, g Volatile Solid kg animal mass⁻¹ day⁻¹), and the emission factor.

2.1.3. Waste Sector

Municipal landfill

The IPCC Tier 2 approach for calculating methane emissions from landfill uses US-specific default values, as presented in the IPCC guidelines for National Greenhouse Gas Inventories [35], in a first-order decay equation (Equation (1)). Methane emissions (Q, g CH₄ y⁻¹) are calculated for each year (x) that the landfill has been open (t, years) using the first-order decay factor (A, Equation (2)), the default methane generation rate constant (k = 0.05 yr⁻¹), the total municipal solid waste generated in year x (M_T(x), Gg/yr), the fraction of total municipal solid waste disposed at the landfill in year x (M_F(x)), and the methane generation potential (L₀, g CH₄ g waste⁻¹). For the DJ basin, it was assumed that each landfill had been open for 30 years [37].

$$Q={\sum }_{x=0}^{t}A\times k\times M_T\left(x\right)\times M_F\left(x\right)\times L_0\times e^{-k\left(t-x\right)}$$

(1)

$$A=(1-e^{-k})/\mathrm{k}$$

(2)

The M_T(x) was calculated by multiplying the population in the DJ basin and the US IPCC per capita waste generation potential (1.14 tons waste capita⁻¹ year⁻¹) [35,37]. The M_F(x) was taken as the default value of 0.55. The L₀ is calculated by multiplying the degradable organic carbon in year (0.1583 g C Gg waste⁻¹), the fraction of degradable organic carbon dissimilated (0.55 for “developed countries”), the fraction by volume of methane in landfill gas (0.5 default), and a conversion from carbon to methane (1.33).

Wastewater treatment

The methane estimates presented here were made using a Tier 2 approach. The total methane emission (Q, g CH₄ yr⁻¹) from wastewater treatment facilities in the DJ basin study area was calculated from the annual total organic matter in wastewater (O, kg BOD yr⁻¹), the organic matter removed from wastewater in the inventory year (S, kg BOD yr⁻¹), the emission factor for the treatment/discharge pathway or system (E, kg CH4 kg BOD⁻¹), and the amount of methane recovered or flared (R, kg CH₄ yr⁻¹) (Equation (3)). The BOD refers to the biological oxygen demand. To generate the methane emission, the biological oxygen demand was assumed to be 5–25 mg L⁻¹; it was also assumed that no organic matter was removed (IPCC default), that there was an emission factor of 0.068 kg CH₄ kg BOD⁻¹, and that there was a removal value of zero (IPCC default) [35].

$$Q=\left[\left(O-S\right)\times E\right]-R$$

(3)

2.2. Results

2.2.1. Energy Sector

Using Tier 1 emission factors and associated uncertainties [35], the total methane loss calculated by the production-based bottom-up approach was estimated to be 244 Gg CH₄ y⁻¹ with an upper and lower range of 23 and 798 CH₄ y⁻¹, respectively (

Table 3. Methane emission estimates from the energy sector using IPCC emission factors. Unit * denotes kg CH₄ per 10⁶ m³ gas produced; ** denotes kg CH₄ per 10³ m³ oil production. Source—Fu denotes fugitive, Fl denotes flaring, and V denotes venting.

Category	Best Guess EF	Source	Best (Gg CH4 y−1)	Min (Gg CH4 y−1)	Max (Gg CH4 y−1)
Production
Gas	1340 *	Fu	72.9	0.00	250
Gas	0.8 *	Fl	0.04	0.03	0.05
Oil	1800 **	Fu	72.9	0.00	291
Oil	720 **	V	29.1	14.6	43.7
Oil	25 **	Fl	1.0	0.51	1.5
Well servicing	110 **	Fl & V	4.5	2.4	6.7
Transmission
Gas	123 *	Fu	6.7	0.0	19.6
Gas	182 *	V	9.9	4.4	30.5
Liquids	110 **	All	4.5	0.0	8.9
Oil	5.4 **	All	0.22	0.0	0.44
Processing
Gas	590 *	Fu	32.1	0.0	112
Gas	2.0 *	Fl	0.11	0.08	0.14
Total (Gg CH4 y−1)			234	22	765
Normalized loss (%)			0.84	0.08	2.76

). This corresponds to a loss of 0.84% (range: 0.08% to 2.76%) of production. The Tier 1 emission factors for production originate from reports and studies made around 20 years ago [13,14,38,39,40]; there can be orders of magnitude between the maximum and minimum emission factors.

2.2.2. Agriculture

Using the IPCC Tier 1/2 approaches, emissions from agriculture primarily come from livestock management, i.e., from enteric fermentation and manure management. Enteric fermentation contributed 67 Gg CH₄ year⁻¹, while manure management emitted 25 Gg CH₄ year⁻¹. More than 95% of the methane emissions from agricultural emissions come from cattle (28 Gg CH₄ y⁻¹) and dairy (64 Gg CH₄ y⁻¹), while horses, sheep, and poultry contribute a small amount (<1 Gg CH₄ y⁻¹).

2.2.3. Waste Sector

Using the IPCC methods, landfills contribute 31 Gg CH₄ yr⁻¹, equivalent to 84% of the methane emissions from the waste sector. There are two types of landfills: municipal and industrial. The calculations and estimates are made for the municipal landfills and a factor was used to account for the industrial emissions based on the municipal to industrial waste emissions from the Colorado average for the DJ basin [40]. Methane, which is produced in anaerobic environments, is emitted during wastewater collection and treatment. Methane from wastewater treatment is significant. It contributes around <2% of the total methane emissions. The estimated emissions from the wastewater sector are 0.6 Gg CH₄ y⁻¹ based on the IPCC method.

3. EPA Emission Estimates

3.1. Methods

3.1.1. Energy Sector

The CH₄ estimate for the oil and gas system was calculated using the EPA greenhouse gas inventory that is publicly available. The EPA inventory consists of both basin-specific and non-basin-specific data. The inventory estimates emissions separately, i.e., petroleum (oil) systems are estimated separately from the natural gas (NG) systems [41]. The Colorado Department of Public Health and Environment (CDHPE) 2022 well head production data were filtered to our DJ basin study-area boundary and classified each well as either a petroleum (oil) or gas well based on the gas-to-oil ratio (GOR; scf bbl⁻¹); emissions are then calculated using the appropriate emission factors [41]. There is uncertainty over which GOR should be used, as it varies between states; for the purpose of this study, we will use the US Energy Information Administration (EIA) value of 6000 scf bbl⁻¹ [42]. The barrel oil equivalent (BOE) for each well was calculated (using 1 BOE is equal to 1 bbl of oil and 6 Mcf NG) to identify marginal wells (production less than 16 BOE); these were classified as non-producing. In total, there were 14,047 well heads: 9243 gas well heads, 4127 oil well heads, and 678 non-producing well heads.

To estimate emissions from the non-basin-specific AFs in the EPA NG and petroleum systems inventories, the ratio of gas and oil wells to the total wells in the US (410,246 gas wells and 529,419 oil wells) was used to estimate DJ AFs. The EPA petroleum inventory has basin-specific information only for tanks, pneumatic devices, well heads, separators, heaters, and headers. For the NG systems inventory, the EPA provides basin-specific information for well pad equipment (heaters, separators, dehydrators, meters/piping, and compressors), pneumatic device vents, chemical injection pumps, condensate tank vents, and well clean-ups. For basin-specific data, data were directly extracted for the Denver region, basin ID 540, i.e., sheet 3.5–16 for the petroleum system and 3.6–20 for the NG system. These sheets provided fractions either per well or production, which were converted to DJ using the number of wells and oil/gas production where necessary. Emission factors for the non-basin-specific data remained as they were, while for basin-specific data, EFs were specific to the Denver region. Emissions were calculated for the whole oil and gas sector, i.e., classified as production, gas processing plants, transmission, and storage.

3.1.2. Agriculture Sector

The activity data for animal populations (

Table 4. Emission factors used to generate the EPA-based and measurement-based emission estimates for manure management and enteric fermentation. Emissions from poultry were not observed during the measurement-informed campaign and have been omitted from this estimate.

		EPA Tier 2				Measurement-Informed
Animal Type	Population (Head)	Enteric Fermentation EF (kg head−1 y−1)	Manure Management EF (kg head−1 y−1)	Emission Enteric Fermentation (Gg y−1)	Emission Manure Management (Gg y−1)	EF (kg head−1 y−1)	Emission (Gg y−1)
Cattle	426,636	100.5	2	42.9	0.850	46.4	19.8
Heifers	285,401	151.5	107.9	43.2	30.8	271.6	77.5
Poultry	65,430	0.0	0.7	0	0.046	NA	NA
Sheep	32,500	8.0	0.2	0.26	0.0065	7.9	0.26

) were extracted from 2023 CDHPE data. The data were filtered to the DJ boundary. The activity data for CAFOs in the DJ are 426,636 cattle, 285,403 dairy cows, 65,430 poultry, and 32,500 sheep (Table 4). The EFs for CAFOs were extracted from the EPA’s 2021 State Inventory Tool (SIT), Colorado. The 2021 SIT provides EFs from enteric fermentation and manure management, which is more inclusive compared to previous downwind measurements studies which could not distinguish the source of CAFOs emissions. Enteric fermentation EFs are 100.5, 151.5, 0, and 8 kg CH₄ head⁻¹ yr⁻¹ for cattle, dairy, poultry, and sheep, respectively. Manure management EFs were 2, 107.9, 0.7, and 0.2 kg CH₄ head⁻¹ yr⁻¹ for cattle, dairy, poultry, and sheep, respectively.

3.1.3. Waste Sector

To calculate methane emissions from landfill and wastewater treatment facilities, the EPA and CDPHE follow the IPCC Tier 2 approach, as described in Section 2.1.3 above [37,43].

3.2. Results

The total CH₄ emission from the oil and gas sector in the DJ basin was 120 Gg CH₄ yr⁻¹; 83% was from the natural gas system and 17% was from the petroleum system. The analysis showed that most CH₄ emissions came from the production sector (69%) (

Table 5. Methane emission estimates from the energy sector using regional EPA emission factors.

Sector	Petroleum Emission (Gg CH4 y−1)	NG Emission (Gg CH4 y−1)
Production	20.8	60.3
Gathering and boosting	0.07	35.9
Processing		2.78
Total	20.9	99.1
Normalized loss (%)	0.08	0.36

). Emissions from pneumatic devices contributed 43% of the total emissions from the energy sector, while emissions from gas engines and compressors contributed 25% and tanks contributed 8%. Summing the methane losses from the oil and gas wells results in a normalized contribution of 0.44%.

Total methane emissions from agricultural sources in 2022 were estimated at 118 Gg, which comprises 44 Gg CH₄ from CAFOs, 74 Gg CH₄ from dairy farms, and less than 300 Mg CH₄ from both poultry and sheep. Enteric fermentation is responsible for 86 Gg CH₄, while manure management emits 32 Gg CH₄. As with the IPCC method above, landfills contribute 31 Gg CH₄ yr⁻¹ and emissions from the wastewater sector are estimated at 0.6 Gg CH₄ y⁻¹.

4. Measurement-Informed Inventory

4.1. Methods

4.1.1. Energy Sector Emissions

Many studies have reported the problems that bottom-up models have in simulating methane emissions from oil and gas production, transport, and processing [3,9,10,11,44,45,46]. The main shortcomings of bottom-up models are that they fail to include all sources and that a single-value emission factor cannot reproduce the complex emission distributions resulting from maintenance, management decisions, state-specific regulation, and the stochastic nature of upset conditions that result in large emissions. Many of these problems have been addressed in the recently created MAES (Mechanistic Air Emissions Simulator), which uses a stochastic model to quantify temporally resolved probabilistic emissions from oil and gas production and midstream sites [47].

Oil and gas production

The MAES uses a Monte Carlo approach to generate spatiotemporal variability into emissions estimates with a 1 s interval using both traditional methods and mechanistic models. While traditional methods multiply emission factors and activity factors to estimate emissions, the mechanistic sub-models simulate emissions using equipment-level chemical and physical processes. The understanding of fluid flow through the equipment and the equipment’s operating state is essential for understanding emissions and estimating emissions under different operational conditions. For instance, flares may function in operating, malfunctioning, or unlit states, each with a different destruction efficiency. This efficiency is considered when multiplied by the equipment throughput to estimate emissions resulting from the incomplete combustion of natural gas. This approach, combined with the Monte Carlo iterations, provides a range of realistic emission estimates specific to the equipment, which is substantially more comprehensive than utilizing an average emission factor for that equipment. As such, the MAES incorporates equipment-specific data, which are essential to predicting realistic multi-species emissions, which are in turn essential for understanding source apportionment.

Unlike traditional bottom-up models, the MAES considers the well head gas composition, which varies within and across basins, to estimate emissions. It also incorporates emissions from upset conditions. Long simulations can be conducted to capture low-probability events such as operational failures, thereby representing emissions from the typical long-tail distribution. Consequently, emissions from the MAES tend to be slightly higher than reported emissions submitted by operators, as they may likely not include emissions from upset conditions. Santos et al. found that in the DJ basin, emissions from such events can be 17% higher than reported figures [48]. However, when emissions from upset conditions are excluded from the simulations, the simulated emissions from the MAES showed an error within 10% compared to the reported emissions from operators in the DJ basin in Colorado.

The area designated as the ‘DJ basin’ for the MAES simulation is constrained within the following coordinates: latitude 39.9 to 40.7, and longitude −105.3 to −104.2.

Table 6. Number of oil and gas facilities in the DJ basin per sector and their operating status (inventory year 2021).

Sector	Operating Status	Count
1. Production	1.1 Operating	2775
	1.2 Partially Operating	136
	1.3 Shut-in	1539
	1.4 Abandoned	248
	1.5 Other	10
2. Pre-production		49
3. Midstream	3.1 Operating	94
	3.2 Partially Operating	2
	3.3 Shut-in	12
	3.4 Other	1
Total		4866

below shows the number of facilities within this area by sector and their operating status. To simulate emissions for the DJ basin using the MAES, this study utilized 2021 inventory data submitted by operators in the basin to CDPHE. These data are published every year through CDPHE’s Oil and Natural Gas Annual Emission Inventory Reporting (ONGAEIR) program [49]. Additionally, aerial measurements made by Carbon Mapper were integrated into the simulation to account for abnormal emissions, which are often missed in annual inventory reports.

As the MAES uses mechanistic models, it is critical to inform the simulator how the equipment on each site is interconnected. Therefore, simulations utilized prototypical sites, which refer to how cohorts of facilities with similar configurations are classified. Each prototypical site has a pre-defined facility diagram establishing fluid flows among equipment. Partner operators voluntarily provided prototypical site matches for their sites in the basin. Mollel et al. (in review) describes the development and application of prototypical sites in the DJ basin [50], while Santos et al. (in review) describe inputs, outputs, and how MAES simulations are performed in detail [48]. For the production sector, we only simulated sites that were either operating or partially operating. Out of these, 605 sites did not report any hydrocarbon liquid or water production. Consequently, we excluded these sites from the simulation, resulting in a total of 2306 simulated active sites with production.

Maintenance

A measurement-based study in the DJ basin [51] estimated emissions from maintenance operations in 82% of the basin during working hours, 0.800 to 1600 MST, at 7.47 Mg CH₄ h⁻¹. Annual emissions were calculated by assuming that operations during six of the eight working hours per day result in emissions (one hour to set up and one hour to clear away); a working year of 5 days per week over 50 weeks was also assumed (one week for Thanksgiving and one week for Christmas).

Gathering systems

One significant uncertainty in both the IPCC and the EPA energy sector emission estimate methodologies are emissions from the gathering system [11,17]. The emissions from the ~4600 miles of gathering pipeline in the DJ basin are estimated at 6.9 Gg CH₄ y⁻¹ (1.5 Mg CH₄ mile⁻¹ y⁻¹) and 2.3 Gg CH₄ y⁻¹ (0.5 Mg CH₄ mile⁻¹ y⁻¹) for the IPCC and EPA methods, respectively. A recent aircraft study measured emissions from a gathering line system in the Permian basin at 10.4 Mg CH₄ mile⁻¹ y⁻¹ [17], suggesting both the IPCC and EPA could underestimate the gathering emissions by at least a factor of 10. A recent study in the DJ basin [11] identified total emissions from plumes from known pipeline leaks at 0.6 Mg CH₄ h⁻¹. For the purposes of this study, we will adopt these emissions as the pipeline emission factor, i.e., 1.1 Mg CH₄ mile⁻¹ y⁻¹ from 4600 miles.

Post-production

A source that is not currently included in either the IPCC or EPA emission inventory is the emission from abandoned oil and gas wells. Abandoned wells are defined as wells that have not produced oil or gas within the last 12 months. A recent study conducted in Colorado estimated average methane emissions from abandoned wells at 586 g CH₄ well⁻¹ h⁻¹ [52]. ECMC identifies that there were 2294 wells declared either temporarily abandoned (TA) or shut-in (SI) wells in our study area before 2022 (

Table 7. Number of temporarily abandoned (TA) or shut-in (SI) wells and decade declared either TA or SI on the ECMC database.

Year	Number TA or SI Wells
1920	2
1970	3
1980	2
1990	2
2000	12
2010	742
2020 and 2021	1531
Total	2294

) [53].

4.1.2. Agriculture Emissions

Agriculture emissions are based on a driving survey conducted on 2260 km of roads in the DJ basin in 2022 [19]. When emissions were normalized by animal population, this study identified average emissions of 5.3, 31, and 0.9 kg CH₄ head⁻¹ h⁻¹ from forty-two CAFOs, twenty dairy farms, and two sheep farms. As data were not collected on management practices, these average per animal emissions are the combined methane measurements from enteric fermentation and manure management. Animal population data were taken from the 2023 CDHPE data (Table 4).

4.1.3. Waste Sector Emissions

Landfill

Average emissions from landfills were measured at 1.02 g CH₄ m⁻² h⁻¹ from a landfill in Nebraska [54]. We assume that as NE and CO are relatively close, these measured emissions and waste contents are relatively similar. Using a GIS, the uncapped emission area of the eight landfills within the DJ basin study region was estimated at 1100 acres (average 140 acres). Here, we assume that emissions measured by Xu et al. (2010) were from the uncapped emission area only and that no methane is emitted from capped areas of the landfill. We also assume that the area in the satellite image is representative of the uncapped emission area at any point in time.

Wastewater treatment sites

Wastewater passing through the 55 wastewater treatment facilities in the study area was taken from 2022 CDPHE data and is estimated at 245 million m³ per year with an average 4.5 million m³ of wastewater flowing through each facility per year. From satellite images, the majority of the 55 wastewater treatment facilities in the study area have on-site anerobic digestors. A recent study reported methane emissions at 12.5 g CH₄ m⁻³ of wastewater flow from facilities with anaerobic digestors [22].

4.1.4. Emissions from Natural Sources

The driving survey described in Section 4.1.2 also measured average emissions of 15 mg CH₄ m⁻² h⁻¹ from six freshwater bodies (five reservoirs and one lake) with a total emission area of 16.00 acres. From satellite images, it is estimated that there are 26,329 acres of water in the DJ study region. The largest water body is the Riverside Reservoir (40.337° N, 104.257° W), which has a surface area of ~4250 acres, and the average surface area of the 806 water bodies is 32 acres (median 5.2 acres). Many of the lakes (~30) dry up in the summer.

4.2. Results

4.2.1. Energy Sector

Operating and partially operating production sites in the DJ basin were estimated by the MAES to have a total annual methane emission of 14.2 Gg. The operating and partially operating sites that did not report any hydrocarbon liquid production sum up to a total emission of 89 Mg of methane per year, with an average emission per site of 0.15 Mg y⁻¹. The sites classified as shut-in, abandoned, and other, sum up an annual emission of 80.37 Mg CH₄ y⁻¹, with an average emission per site of 0.04 Mg CH₄ y⁻¹. Therefore, in summary, combining the MAES-simulated emissions and reported emissions for the sites that were not included in the simulation, the production sector accounted for 14.4 Gg CH₄ y⁻¹.

In the midstream sector, we focused our simulation on the 73 gas plants and compressor stations that were either in operating or partially operating states. The emissions from these sites were estimated by the MAES to be 13.0 Gg CH₄ y⁻¹. The remaining midstream sites not included in the simulation reported a combined annual emission of 45.3 Mg CH₄ y⁻¹, with an average of 1.5 Mg CH₄ y⁻¹ site⁻¹. In total, the midstream sector is estimated to have a total annual emission of 13 Gg CH₄ y⁻¹. For maintenance activities, the 4600 miles of gathering pipelines, and 6795 abandoned wells, we estimate emissions of 13.3, 5.3, and 11.8 Gg CH₄ y⁻¹, respectively.

4.2.2. Agriculture Sector

Using emission factors measured from operations in the DJ basin, the methane emissions from agriculture in 2022 are estimated at 98 Gg. A total of 20 Gg of methane are emitted from CAFOS, 78 Gg are emitted from dairy farms, and a negligible amount is emitted from sheep farms (Table 4).

4.2.3. Waste Sector

We estimate the total emissions from the 1100 acres of uncapped landfills at 40.3 Gg CH₄ y⁻¹. Emissions from the largest and smallest emitting landfills are estimated at 13 and 0.4 Gg CH₄ y⁻¹, respectively, with an average emission of 5 Gg CH₄ y⁻¹. Methane emissions from the 55 wastewater facilities in the study area are estimated at 3.1 Gg CH₄ y⁻¹, with an average emission from each site of 56 Mg CH₄ y⁻¹. The maximum and minimum emissions are estimated at 430 and 0.2 Mg CH₄ y⁻¹ from the City of Boulder 75th Street WWTP and Town of Mead Lake Thomas subdivision WWTF, respectively.

4.2.4. Natural Sector

Using the freshwater area data, we estimate the methane emission from freshwater sources in our study area at 14.0 Gg CH₄ y⁻¹. The average emission is estimated at 17 Mg CH₄ y⁻¹, while the maximum is 2.26 GG CH₄ y⁻¹ from the Riverside Reservoir.

5. Discussion

This study presents the results of three bottom-up approaches which were used to generate methane emission inventories for the Denver-Julesburg basin in Colorado, US. The first method, a Tier 1/2 approach, used “developed world” IPCC emission factors and US-specific emission factors to generate a 2022 basin-wide emission estimate of 358 Gg CH₄ (

Table 8. Total methane emissions (Gg CH₄ yr⁻¹) from energy, agriculture, waste, and natural emission sources within the DJ basin as calculated using the Tier 1/2 IPCC emission factors (EFs), the EPA Colorado-specific emission factors, and measurement-informed emission factors. The source apportionment is calculated for the emissions using the measurement-informed emission factors.

Sector	Sub-Sector	Tier 1/2 IPCC EFs (Gg CH4 yr−1)	EPA CO-Specific EFs (Gg CH4 yr−1)	Measurement-Informed EFs (Gg CH4 yr−1)	Source Apportionment (%) ***
Energy	Production	180	81	14	27
	Midstream	54	39	13
	Maintenance			13
	Gathering lines			5
	Abandoned			12
Agriculture	Enteric F.	67	86	98 *	46
	Manure M.	25	32
Waste	Landfill	31	31	40	20
	Wastewater	0.6	0.6	3
Natural	Water	**	**	14	7
Total		358	269	212
O&G normalized loss (%)		0.84	0.44	0.22

* Combined emissions from enteric fermentation and manure management. ** Not calculated in these inventories. *** Source apportionment based on the “measurement-informed” calculations.

), with the largest emissions coming from the energy sector (234 Gg CH₄; 65%), followed by agriculture (92 Gg CH₄; 26%) and waste (32 Gg CH₄; 9%). The second approach used Colorado-specific emission factors generated by the US Environmental Protection Agency (EPA), which estimated total emissions of 269 Gg CH₄, but emissions were more evenly split between the energy (120 Gg CH₄; 45%) and agriculture sectors (118 Gg CH₄; 45%). The third approach used measurement-informed emission factors taken from peer-reviewed papers that reported emissions from the DJ basin. Total emissions, which included estimates from other sources (abandoned wells and freshwater sources), were similar to those generated using the EPA’s Colorado-specific emission factors (212 Gg CH₄), but emissions from the agriculture sector (98 Gg CH₄; 45%) were higher than the energy sector (57 Gg CH₄; 29%).

One major shortcoming of using a single emission factor to generate an emission estimate is that the sample size used to generate the number may be too small and not adequately account for any long-tail emission distribution. For generating the measurement-informed emission inventory, the agricultural emission factors used were based on measurements of forty-two CAFOs, twenty dairy farms, and two sheep farms (as specified in L407). For the waste sector, the landfill emission factor was taken from only one site, but the wastewater estimate was taken from a synthesis of 101 site measurements. The emission factor for freshwater was taken as the average of six individual measurements. As the drivers of these sources are mainly biogeochemical-normalized, emissions are likely to be more normally distributed than the long-tail distribution commonly seen for emissions from oil and gas infrastructure. For the largest emission sources (CAFOS, dairy farms), the sample size is a large fraction of the number of facilities in the study area (42 out of 64 CAFOs and 20 out of 59 dairy farms). Of the other large emission sources (landfill and freshwater), more measurements would improve the accuracy of the emission factor used; however, at this time, measurements do not exist, and we suggest that further research focus should be put on making measurements of these emission sources. Another observation from our analysis is that the methane generation potentials and decay rates published by the IPCC can produce reasonably precise emission estimates. Again, we highlight that data measuring emission rates from landfills are relatively rare and more focus should be put onto investigating regional variation in emissions resulting from waste content and environmental variability.

Recent site modifications and regulation enforcement have contributed to a significant decrease in emissions in the DJ basin. The CDPHE defined standards for new well production facilities in 2019, which must comply with storage tank emission control requirements, gas pneumatic emissions, and redirect flashed gas from separators to the sales line; or, emissions must be controlled with a piece of equipment that has an average efficiency of 95% [55]. Additionally, operators in the basin are transitioning from reciprocating and centrifugal compressors to screw drive compressors, which are known for their lower emissions. They are also replacing natural gas-fired compressor drivers with electric engines, a shift that has contributed to a reduction in emissions. The regulations imposed on oil and gas operations by the CDPHE has introduced requirements for compressors, pneumatics, equipment leaks, and fugitive emissions, flowline testing, and guidelines to reduce venting from storage tanks. If this was simulated in the emission estimate generated using CO-specific emission factors, methane emissions would be reduced by 46 Gg CH₄ y⁻¹ by converting all pneumatic devices to air pneumatics, 8 Gg CH₄ y⁻¹ by addressing rod packing emissions from compressors, and 9 Gg CH₄ y⁻¹ by controlling emissions from tanks. Overall, accounting for these mitigation strategies in the bottom-up inventory would reduce the 81 Gg CH₄ y⁻¹ from production (Table 8) to 18 Gg CH₄ y⁻¹.

In all cases, the normalized gas loss from oil and gas operations, which is calculated by dividing methane emissions by the methane produced, is much lower than the historic normalized gas loss values. Top-down approaches estimated DJ basin normalized gas loss at 4.1% in 2012 [26] and 2% in 2015 [27], while bottom-up approaches generated estimates between 0.84 and 0.22% with the measurement-informed emission factors generating the lower value (Table 8; Figure 2). Of particular interest is the decrease in normalized gas loss while the production rates increase, resulting in a decrease in absolute methane emissions (absolute emissions: 228 Gg CH₄ in 2012 to 57 Gg CH₄ in 2022) despite production increasing by over a factor of seven (methane production: 3.6 Tg CH₄ in 2012 to 27.5 Tg CH₄ in 2022).

The findings presented here are in contrast to a similar study generated for the Delaware basin in Texas, US [11], where the normalized loss using the IPCC (1.1%) and EPA Texas-specific (0.57%) emission factors were both lower than the measurement-informed emission factor-based estimate (2.8%). This suggests that the regulatory changes since 2012, which include comprehensive requirements for compressors, pneumatics, equipment leaks, and fugitive emissions, are likely to have had a significant effect on methane emissions.

Even though modeling informed by measurement suggests comprehensive emissions reductions within the DJ basin, it is currently unclear if this is the case. Modeling using the MAES could underestimate emissions if the long-tail distribution is not appropriately represented in the model, and this can only be accounted for using direct measurements. Currently, the frequency, duration, and size of emissions events have been generated using measurements with relatively high detection limits, and we suggest that the MAES emission estimates could be improved by observing more emission events. This is also true of the emission estimates from the gathering lines and abandoned wells, where data are taken from a relatively small sample size.

6. Conclusions

In this study, we present methane emission estimates derived using three bottom-up approaches from all sources within the Denver-Julesburg basin, CO, USA in 2022. The Tier 1/2 approach that used emission factors published by the IPCC estimated methane emissions of 358 Gg. The Tier 3 approach that used EPA-published emission factors for Colorado estimated total emissions of 269 Gg. Using regionally specific measurement-informed emission factors, the basin-wide methane emissions were calculated at 212 Gg.

The largest difference in these calculated emissions resulted from using the Mechanistic Air Emissions Simulator (MAES) to simulate emissions from the production and transport of oil and gas. The MAES accounts for changes to regulatory practice in the DJ basin, which include requirements for compressors, pneumatics, equipment leaks, and fugitive emissions, which were implemented to reduce emissions starting in 2014. This study suggests that regulations implemented by the Colorado Department of Public Health and Environment could have significantly reduced production-related emissions, but contemporary regional measurements should be made to ensure these bottom-up calculations are realistic.