An intensive coordinated airborne and ground-based measurement study was conducted in the Fayetteville Shale in northwestern Arkansas during September and October 2015 to compare and explain potential discrepancies between top-down and bottom-up estimates of regional natural gas (NG) methane (CH4) emissions. In situ mobile downwind measurements are used to document the ethane to methane enhancement ratios (ERs) in emission plumes from NG operations in the region. Enhancement ratios are low (<2% for 87% of NG sources sampled) in this dry gas-producing region and normally distributed around 1.3% in the western half of the study area. A few sampled landfills emitted CH4 but no ethane (C2H6). Sampling drives around large chicken farms, prevalent in the region, did not detect significant downwind CH4 enhancements. In situ airborne measurements of C2H6 and CH4 from area-scale surveys over and downwind of the region documented the resulting ERs from a mix of CH4 sources. Based on these measurements, we show that on average during the measurement windows 85–95% of total CH4 emissions in the western half of the Fayetteville Shale originated from NG sources, which agrees well with bottom-up estimates from the same field study. Lower mixing ratios measured over the eastern half of the region did not support the ER analysis due to the low signal-to-noise on C2H6 measurements.

## 1. Introduction

Methane (CH4) is the primary component of natural gas (NG). Despite CH4 being a potent greenhouse gas, when used as an alternative for power generation, NG can have a smaller climate impact than coal and oil, which each emit about twice as much carbon dioxide as NG per unit of energy delivered (EIA, 2016). However, losses to the atmosphere, which are estimated to be 2.3% of gross U.S. natural gas production (Alvarez et al., 2018) along the NG supply chain, can reduce its advantages. Natural gas systems were estimated to account for 25% of total U.S. anthropogenic CH4 emissions in 2015 (EPA, 2017), making them the second largest national anthropogenic CH4 source, behind enteric fermentation and above landfills (EPA, 2017). Within the NG sector, the largest fraction of CH4 emissions, 66%, originates during the production phase (EPA, 2017).

Studies comparing top-down (atmospheric measurement based) and bottom-up (inventory/activity based) estimates of regional or national CH4 emissions from oil and natural gas (O&NG) operations in the U.S. have pointed to systematic discrepancies between the two methods (Brandt et al., 2014; Zavala-Araiza et al., 2015). Aircraft campaigns in several U.S. O&NG basins have derived larger emissions of CH4 and ethane (C2H6), another major component of NG, than estimated using available inventory data (Karion et al., 2013; Karion et al., 2015; Kort et al., 2016; Pétron et al., 2012; Pétron et al., 2014).

A major challenge for comparing regional top-down studies with O&NG emission inventories has been the quantitative attribution of top-down regional emission estimates between different sources of CH4. The mass balance method used with aircraft data provides total CH4 emissions for an area, which then must be apportioned between O&NG and other sources. To do so, previous top-down regional and national studies have often relied completely or partly on bottom-up information for CH4 emissions apportionment or non-O&NG emissions estimation (Caulton et al., 2014; Karion et al., 2013; Pétron et al., 2012), even in cases where C2H6 was used to distinguish between sources (Peischl et al., 2015, 2016).

Atmospheric measurements of CH4 alone, unless directly downwind of a point source (Conley et al., 2017; Frankenberg et al., 2016), are usually not enough to distinguish and separate contributions from different CH4 sources. A few apportionment studies in O&NG fields have used alkane ratios or CH4 isotopic signatures to distinguish between emissions from various CH4 sources (Hopkins et al., 2016; Smith et al., 2015, 2017; Townsend-Small et al., 2015). However, results from these studies can be limited by a relatively small number of discrete samples or days of measurements. Additionally, mass balance and attribution methods using alkane or isotope enhancement ratios require subtracting local background values, which have not always been sufficiently documented (Schwietzke et al., 2017).

The work presented here is part of a larger coordinated intensive study conducted in the Fayetteville Shale dry gas play in north-central Arkansas in September and October 2015 to compare state of the science bottom-up and top-down regional NG CH4 emission estimates. Emissions quantification at the facility or equipment level is presented by Bell et al. (2017), Conley et al. (2017), Robertson et al. (2017), Vaughn et al. (2017), Yacovitch et al. (2017), and Zimmerle et al. (2017). An aircraft mass balance estimate of spatially resolved CH4 emissions from the study area on two consecutive days is described in Schwietzke et al. (2017), and a reconciliation between this top-down estimate and a bottom-up CH4 emissions estimates is provided by Vaughn et al. (2018). The bottom-up model presented by Vaughn et al. revealed a strong diurnal variability in total NG CH4 emissions, with the peak occurring during the mid-afternoon due to manual liquid unloading events at 107 wells over the two days modeled (Vaughn et al., 2018). Modeled bottom-up emissions estimates captured the same spatial variability in CH4 emissions throughout the study area with a reasonable agreement in emissions estimates between the two approaches for the total study area, eastern half, and western half. NG CH4 emissions normalized by natural gas production were found to be two times higher in the western half than in the eastern half (Schwietzke et al., 2017).

Presented here are atmospheric in situ CH4 and C2H6 measurements from the ground and aircraft to apportion CH4 emissions between NG operations and other sources in the Fayetteville dry shale gas field. C2H6 to CH4 enhancement ratios in observed emission plumes (ERs, the molar ratio of C2H6 and CH4 relative to their respective local background values) are compiled downwind of NG facilities in the Fayetteville Shale to characterize the spatial variability of NG ERs throughout the study area. These source ERs are then used together with total emission ERs from aircraft measurements to attribute CH4 emissions between NG and other sources.

High frequency in situ measurements permitted the determination of enhancement ratios at more locations than would have been possible with discrete flask samples. The ability to quickly make measurements at a given location eliminated the need to correct for the species background, which can vary over time and space. The use of data collected on multiple days allows an assessment of the variability in ERs at different NG facilities and variability among ERs on flights across different days. This method is atmospheric measurement-based with incorporation of publicly available data and does not require inventory or operator-provided data. The unique combination of ground and aircraft data incorporates information about both source signatures and the mix of sources in the area into the attribution. Our work is an effort to further refine top-down measurement and attribution methods to help improve independent constraints on regional long-lived greenhouse gas emission attribution.

## 2. Methods

### 2.1. Overview

The study area (Figure 1) comprised the core of the Fayetteville Shale production basin in Arkansas, USA, and spanned Cleburne, Conway, Faulkner, Independence, Jackson, Van Buren, and White counties to form an area ~65 km (north to south) by ~150 km (west to east). This area is a dry NG production region (Arkansas Geological Survey, 2015) with no oil production or processing of NG (Vaughn et al., 2018), and accounted for ~3.1% of national dry NG production in September 2015 (EIA, 2017). NG facilities in the area in 2015 totaled ~5500 active NG wells, 125 gathering stations, and six transmission stations. Variability in the C2H6 to CH4 emission ratio for NG sources is expected to be due to variability in raw gas composition, as there is no gas processing at production facilities or at centralized locations. In addition to NG operations, potential CH4 sources included agriculture (cattle and poultry farms), wetlands, geologic seeps, stationary combustion, and one small landfill in Faulkner County (Schwietzke et al., 2017). In the absence of biomass burning, only NG related sources in the study area are expected to co-emit C2H6 with CH4. A fire ban was in effect in much of the study area during the study, so biomass burning is not expected to have occurred in the majority of the study area.

Figure 1

Study area map. Map of the study area marked with routes of NOAA ML drives with continuous measurements from September 23, 2015 through October 8, 2015. County outlines are shown in green. DOI: https://doi.org/10.1525/elementa.351.f1

Figure 1

Study area map. Map of the study area marked with routes of NOAA ML drives with continuous measurements from September 23, 2015 through October 8, 2015. County outlines are shown in green. DOI: https://doi.org/10.1525/elementa.351.f1

Fast response in situ CH4 and C2H6 measurements were obtained using both ground-based and airborne platforms. An instrumented van (NOAA Global Monitoring Division (GMD) mobile laboratory) measured CH4 and C2H6 in CH4 emission plumes from public roads downwind of NG facilities in order to gather data needed to calculate representative ERs. C2H6 and CH4 data were also collected by two mobile laboratories operated by Aerodyne Research, Inc. performing dual tracer flux ratio measurements, a downwind method used to determine facility-level CH4 emissions (Roscioli et al., 2015; Yacovitch et al., 2017). A single-engine instrumented aircraft operated by Scientific Aviation was used to obtain the C2H6 and CH4 data (Schwietzke et al., 2017) for regional-scale ERs that reflected the mix of CH4 emissions in the study area. Each platform’s instrumentation and sampling strategy are described below.

### 2.2. Mobile laboratory and aircraft instrumentation and sampling strategy

#### NOAA mobile laboratory

The NOAA mobile laboratory (ML), a 2005 Chevrolet 15 passenger van, was equipped with a four species cavity ring down spectrometer (CRDS) (Picarro model #G2401-m) that provided sequential measurements of CH4, CO2, and CO mole fractions, and measurements of water vapor every three seconds. A tunable infrared laser direct absorption spectroscopy (TILDAS) analyzer (Aerodyne Research, Inc.) with a 2990 cm–1 laser measured C2H6, CH4, and water vapor at 1 Hz. Stationary 1-sigma noise over 30 seconds under the deployment conditions was approximately 0.30 ppb on CH4 for the CRDS, 0.05 ppb on C2H6, and 0.40 ppb on CH4 for the TILDAS. Prior to and following ML drives each day, three gas standards prepared by NOAA GMD in Boulder, Colorado were measured by the CRDS and TILDAS instruments to identify potential instrument drift (see Supplemental Material). The ML was also equipped with a global positioning system (GPS) and NOAA GMD flask air sampling system. Further details of the ML set-up can be found in the Supplemental Material.

Discrete air samples, used to compare with in situ measurements, were collected using a programmable compressor package (PCP) and programmable flask packages (PFPs). The PCP contains the pump, batteries, and a microprocessor to control sampling, and the PFP consists of a manifold and twelve cylindrical, 0.7 L borosilicate glass flasks that are filled one by one (Aircraft Program Methods, 2018). These discrete air samples were analyzed by NOAA GMD using the same systems used for measurements in the Global Greenhouse Gas Reference Network (ESRL, 2017). CH4 was analyzed by gas chromatography with flame ionization detection (Dlugokencky et al., 1994), and C2H6, by gas chromatography mass spectrometry (see Supplemental Material). C2H6 and CH4 dry air mole fractions have 1-sigma uncertainties of 11 ppt (NOAA 2016 scale) and 1.1 ppb (NOAA X2004A scale), respectively.

#### Aerodyne mobile laboratory

Aerodyne Research, Inc. operated two mobile laboratories during the time when the NOAA ML was also deployed. Each housed TILDAS analyzers used for measuring CH4, C2H6 and the trace gases used in their tracer release method for facility-level quantification of CH4 emissions. Details of the mobile laboratory set-ups can be found in the supplemental material of Yacovitch et al. (2017).

#### Scientific aviation aircraft

Instrumentation onboard the Mooney Ovation aircraft included a NOAA GMD flask air sampling system, a cavity ring down spectrometer (Picarro model # 2301-f) with 0.5 Hz water-corrected measurements of CH4 (<2 ppb uncertainty (Karion et al., 2013) CO2, and water vapor, and a TILDAS C2H6 mini trace gas monitor (Aerodyne Research, Inc.). This instrument, with a 2997 cm–1 laser, (~200 ppt noise in the configuration used in this study) was used to obtain 1 Hz measurements of C2H6. Though C2H6 values were not water-corrected (not removing water during calculations ignores dilution and results in a lower mole fraction than the dry air mole fraction), the highest water content observed on any flight (<2%) would necessitate a data correction that is small relative to the C2H6 signal observed on flights. Additional instrumentation included a Vaisala relative humidity and temperature probe and GPS (Hemisphere VS101). Details of the aircraft set-up can be found in Conley et al. (2017).

### 2.3. Methane emission source attribution model

A top-down attribution of CH4 emissions between NG and other CH4 sources was accomplished using the approach described below. Atmospheric ER measurements are assumed to be equivalent to emission ratios from CH4 sources since the average global atmospheric lifetimes of C2H6 and CH4 (approximately 2 months and 9 years, respectively (Rudolph and Enhalt, 1981; Nisbet et al., 2016)) make chemical losses negligible when examining plumes and air masses downwind of a facility or source region over timescales on the order of minutes to a few hours. It is assumed that the ER from a given source in the study area does not change from hour-to-hour or day-to-day due to the fact that the produced gas is dry and there are no gas treatment facilities or liquid condensate collection activities in this basin. Results presented in a later section support this assumption.

For an area of interest, the total CH4 and C2H6 emissions (E) can be separated into the contributions from NG and other sources (denoted “other”).

$E(CH4)total=E(CH4)NG+E(CH4)other$
$E{\left( {C{H_4}} \right)_{total}} = E{\left( {C{H_4}} \right)_{NG}} + E{\left( {C{H_4}} \right)_{other}}$
Eq. 1
$E(C2H6)total=E(C2H6)NG+E(C2H6)other$
$E{\left( {{C_2}{H_6}} \right)_{total}} = E{\left( {{C_2}{H_6}} \right)_{NG}} + E{\left( {{C_2}{H_6}} \right)_{other}}$
Eq. 2

C2H6 emissions from other sources, such as biomass and biofuel burning, are assumed to be negligible within the study area. There was a fire ban in effect for a large part of the study area at the time of the study. Therefore,

$E(C2H6)other≈0$
$E{\left( {{C_2}{H_6}} \right)_{other}} \approx 0$
Eq. 3
$E(C2H6)total=E(C2H6)NG$
$E{\left( {{C_2}{H_6}} \right)_{total}} = E{\left( {{C_2}{H_6}} \right)_{NG}}$
Eq. 4

Equations 1 and 2 are rewritten below to use area mean quantities constrained with field measurements. The NG facility plume ERs determined from ML data are used to constrain an area mean ER[C2H6/CH4]NG in Eq. 6. The area-scale plume ERs from aircraft data, which represent the mix of CH4 sources in the study area, are used to constrain an area mean ER[C2H6/CH4]area in Eq. 5.

$E(CH4)total=E(C2H6)total*1ER[C2H6CH4]area=E(C2H6)NG*1ER[C2H6CH4]area$
$\begin{array}{*{20}{c}}{E{{({{\rm{CH}}_4})}_{total}}} \ { = E{{({C_2}{H_6})}_{total}}*\frac{1}{{ER{{\left[ {\frac{{{C_2}{H_6}}}{{{CH_4}}}} \right]}_{area}}}}}\\\qquad\qquad\quad{} \ { = E{{({C_2}{H_6})}_{NG}}*\frac{1}{{ER{{\left[ {\frac{{{C_2}{H_6}}}{{{CH_4}}}} \right]}_{area}}}}}\end{array}$
Eq. 5
$E{\left( {\rm{CH}_4} \right)_{NG}} = E{\left( {{C_2}{H_6}} \right)_{NG}}*\frac{1}{{ER{{\left[ {\frac{{{C_2}{H_6}}}{CH_4}} \right]}_{NG}}}}$
Eq. 6

The relative contribution of CH4 from NG related sources in the study area can thus be calculated using Eq. 7, the ratio of Eq. 6 to Eq. 5.

$E(CH4)NGE(CH4)total=ER[C2H6CH4]areaER[C2H6CH4]NG$
$\frac{{E{{(C{H_4})}_{NG}}}}{{E{{(C{H_4})}_{total}}}} = \frac{{ER{{\left[ {\frac{{{C_2}{H_6}}}{{C{H_4}}}} \right]}_{area}}}}{{ER{{\left[ {\frac{{{C_2}{H_6}}}{{C{H_4}}}} \right]}_{NG}}}}$
Eq. 7

The result is a mean estimate of the fraction of the total CH4 emissions that originate from NG sources in the area of interest during the time of the campaign.

This model is only applicable in cases in which there are only two distinct CH4 source categories with a small range of ERs, as another CH4 source with its own characteristic ER would introduce another mode in the ER distribution and would require another independent trace gas marker and more equations to constrain the attribution. As previously stated, all C2H6 enhancements are assumed to originate from NG related sources, and the ER distribution for NG point source emissions and regional area source plumes is assumed constant over the time period of the study.

### 2.4. Mobile laboratories and aircraft sampling strategies

#### NOAA mobile laboratory

Twelve daytime drives were conducted with the NOAA ML, ranging from six to ten hours each (typically from about 10:00–18:00 local time), to compile ERs in CH4 plumes throughout the study area. Drives were made on public roads throughout extensive portions of the study area, as shown in Figure 1. Distances from the ML on the road to a facility were typically 0.05 to 0.30 km.

When a CH4 emission plume (at least 50 ppb above local background) was observed downwind of a potential emission source, the ML was intentionally positioned to acquire 30 seconds to a few minutes of in situ data to measure multiple CH4 plumes to calculate the ER, check for temporal stability in the ER at the source, and in some instances, collect flask air samples in the local background air and in the facility emission plume. When collecting in situ data, the ML was driven slowly out of the CH4 plume to allow CH4 to return to its background level. In this way, a range of CH4 and C2H6 mole fractions could be measured and a slope calculated to determine an ER. Section 3.1 provides an assessment of the temporal consistency of ERs when measurements at the same facility were repeated on multiple days.

To collect flask air samples, the ML engine was first turned off. Then, the PFP sampling manifold and flask were first flushed (at ~10 and ~17 SLPM, respectively) for 45 seconds and, when complete, air sample collection was manually triggered using the real-time display of in situ measurements to target different mole fractions. Fill times were ~10 seconds. The core GMD chemical analysis instrumentation focuses on tracking background air composition, and to avoid contamination of the GC-FID columns used to measure samples in the Global Greenhouse Gas Reference Network, we intentionally only collected PFP air samples with CH4 mole fractions below 5 ppm.

#### Aerodyne mobile laboratory

Included in our analysis are ERs from Aerodyne Research, Inc. for which: (i) the CH4 emissions at the NG facility could be determined using the dual tracer release method, (ii) there was a coefficient of determination of 0.50 or greater, and (iii) manual inspection of the CH4 and C2H6 plumes did not indicate any problems with the Aerodyne dual tracer release at the target site, as described in Section 6 of the Supplemental Material in Yacovitch et al. (2017). Here we use only ERs for which the reported 95% confidence interval is less than or equal to 30% of the mean ER, leaving 46 of the original 49 facilities at which ERs were obtained by Yacovitch et al. (2017).

#### Scientific aviation aircraft

Fifteen flights were conducted between September 21 and October 14, 2015. Data from October 1 and October 2 were used by Schwietzke et al. (2017) to calculate a total CH4 emission estimate for the study area using a mass balance approach. Flights with box, raster, or spiral sampling patterns were conducted over the entirety and subsections of the study area to gather data for CH4 emission quantification, CH4 source attribution, and individual facility CH4 emission quantification, respectively (see Supplemental Material). Aircraft measurements from five raster flights, the paths of which are shown in Figure 2, were used to investigate mean ERs for the western and eastern halves of the study area.

Figure 2

Map of raster flights used for ER calculations. Five raster flights, conducted on September 22, October 2, October 5, October 6, and October 7, 2015, were used to obtain area-scale ERs. DOI: https://doi.org/10.1525/elementa.351.f2

Figure 2

Map of raster flights used for ER calculations. Five raster flights, conducted on September 22, October 2, October 5, October 6, and October 7, 2015, were used to obtain area-scale ERs. DOI: https://doi.org/10.1525/elementa.351.f2

### 2.5. Analysis of ground measurements in facility-level plumes

Mobile laboratory data were used to determine representative ERs for NG sources in the study area. CH4 plumes measured by the ML were identified in CH4 data from the CRDS using an instantaneous 50 ppb CH4 enhancement threshold above the local CH4 background. This threshold was chosen to distinguish enhancements caused by local emissions from instrument noise and background variability without biasing plume selection toward relatively large CH4 plumes. The start time of each plume was determined by the time immediately before CH4 increased by 50 ppb and the end time was determined by the time CH4 returned to its start time CH4 value, with low variability in CH4 prior to and following the plume. Plumes were identified by manually screening the time series multiple times to prevent omissions and errors. If data were acquired when the ML was moving, only plumes less than one minute in duration were included in the analysis to avoid including multiple facility plumes and potentially varying background levels. Using code described by Yacovitch et al. (2014), a C2H6 vs. CH4 scatterplot was created for each plume with data from the TILDAS, and the ER for each plume was determined using the slope calculated by orthogonal distance regression. Because the C2H6 and CH4 data were acquired from the same absorption spectra on the same instrument, no sampling time offset optimization was required. Additionally, no assignment of C2H6 and CH4 background was needed for each plume because in the method described above, the background is accounted for in the intercept of the scatterplot and the slope determined by enhancements alone, reflecting the C2H6 to CH4 ratio in isolated plumes.

An ER was assigned to a source only if the coefficient of determination, R2, was at least 0.65. For CH4 sources where more than one ER was determined, a time-weighted average was calculated to assign a single mean ER to the source. Only results for which a facility location could be readily identified as the emission plume source are presented. CH4 plumes that could not be attributed to a specific source are not included in the analysis. Plumes from 71 NG facilities measured by the NOAA ML passed the filtering criteria and though the minimum R2 required was 0.65, 85% of the NG plumes had an R2 associated with the ER that was 0.95 or greater.

### 2.6. Analysis of aircraft measurements for area-scale plumes

Aircraft measurements in the boundary layer over the study area were used to determine area-scale plume ERs (see Eq. 5). The portion of the data for which CH4 was enhanced above background was identified on downwind transects (for box pattern flights) and legs (for raster pattern flights) and an ER was calculated using orthogonal distance regression weighted by the airborne instrumentation noise of 1.3 ppb for CH4 (Karion et al., 2015) and 200 ppt for C2H6. We divide the study area into western and eastern halves using –92.1° longitude according to observed spatial differences in aircraft-based CH4 emission estimates normalized by NG production (Schwietzke et al., 2017). In the event that multiple transects or legs from a single flight were downwind of the study area, the ERs for all transects or legs were averaged. Typical ranges observed while flying transects were approximately 1940 to 2300 ppb for CH4 and 1.6 to 4.0 ppb for C2H6. Mole fractions observed in the aircraft data are lower than those observed in the ML, where CH4 and C2H6 enhancements very close to sources could be up to several thousand ppb CH4 and several hundred ppb C2H6. As the aircraft measured downwind of different regions of the study area, background CH4 levels sometimes changed. To distinguish background variability from emission sources, ERs were determined only for those subsections of a leg or transect that included CH4 enhancements of at least 50 ppb.

Criteria for acceptance of ERs were that measurements were made in the planetary boundary layer, a CH4 enhancement in area plumes of at least 50 ppb above background and a C2H6 enhancement of at least 1 ppb to ensure a high enough signal to noise ratio, and a coefficient of determination of at least 0.65. After filtering using the above criteria, flight pattern, and wind direction, five flights could be used in the determination of an area ER for the western half of the study area. Several transects in these flight and others had low signal to noise and low enhancements, disqualifying them for use in ER calculations. Low signal to noise occurred due to low CH4 and C2H6 enhancements in the planetary boundary layer and was observed more often over the eastern half of the study area, where total CH4 emissions were found to be lower than in the western half (Schwietzke et al., 2017).

## 3. Results and Discussion

### 3.1. Mobile laboratory drives

#### Facility plume C2H6 to CH4 enhancement ratios on different days

Seven of eight facilities at which plume measurements were repeated over multiple days show consistent (difference <20%) ERs across days, as shown in Table 1. This excludes 106 facilities with measurements on only one day (see next subsection). ERs shown in Table 1 are for each plume measured at the facility rather than the time-weighted average for the facility. All ERs have an R2 of 0.65 or greater, and 85% of the NG plumes had an R2 associated with the ER that was 0.95 or greater. The repeatability of ERs on different days at these facilities suggests that an ER measured at a facility by the ML on any given day can be used as a representative value for that facility even without repeated measurements on different days and that emission ratios at NG facilities were likely constant over the course of the study. The spread in ERs observed at Production Pad 7 on September 24 suggests another emission source in addition to vented gas. CH4 plumes at this pad that are enhanced in C2H6 also show a strong correlation with the combustion product, CO2 (see Supplemental Material), though no correlation between CO and CH4 was observed. The most likely sources of combustion at production pads in this dry gas basin are on-site NG-powered compressor engines, used at some production pads to increase flow from a well. It is possible that the natural gas-fueled compressor was running at the time of the ML measurements, though the activity data needed for confirmation are not available. Using ERs only from plumes for which the CH4 enhancement was at least 50 ppb, the time-averaged ER for production pad 7 is 1.7%.

Table 1

NOAA ML repeat ER measurements. All in situ C2H6 to CH4 ERs measured by the ML at the same facilities over multiple days. DOI: https://doi.org/10.1525/elementa.351.t1

Facility typeDateIn situ C2H6 to CH4 ER

10/3/2015 1.3%
10/2/2015 0.9%
9/30/2015 1.6%
10/5/2015 1.9%
10/2/2015 1.3%
10/5/2015 1.1%, 1.1%
10/5/2015 1.5%
Production pad 7 9/24/2015 0.8%, 0.9%, 0.9%, 0.9%, 1.0%, 5.3%, 5.9%, 11.6%
9/28/2015 0.9%, 0.9%, 0.9%, 0.9%, 0.9%, 1.0%
10/2/2015 0.9%
Gathering station 9/29/2015 1.4%, 1.4%
10/3/2015 1.4%, 1.4%, 1.4%
Facility typeDateIn situ C2H6 to CH4 ER

10/3/2015 1.3%
10/2/2015 0.9%
9/30/2015 1.6%
10/5/2015 1.9%
10/2/2015 1.3%
10/5/2015 1.1%, 1.1%
10/5/2015 1.5%
Production pad 7 9/24/2015 0.8%, 0.9%, 0.9%, 0.9%, 1.0%, 5.3%, 5.9%, 11.6%
9/28/2015 0.9%, 0.9%, 0.9%, 0.9%, 0.9%, 1.0%
10/2/2015 0.9%
Gathering station 9/29/2015 1.4%, 1.4%
10/3/2015 1.4%, 1.4%, 1.4%

Low temporal variability between ERs at a NG facility is expected because the ER from mostly fugitive (in this region) emissions is dependent upon the composition of the raw natural gas being extracted, which remains constant over the timescale at which this study was conducted. However, variability in natural gas composition due to geologic differences throughout a region can lead to spatial variability in ERs.

#### Range and spatial distribution of C2H6 to CH4 ERs observed at different NG facility types

In this section, we present ER results from the 114 NG facilities measured by the NOAA ML and the Aerodyne mobile laboratories (AML) (Yacovitch et al., 2017). For the three facilities that were measured by both NOAA and the AML, the ERs at both gathering stations are identical (1.0% and 1.4% for the two gathering stations) and within 10% for the production pad (1.1% measured by NOAA and 1.0% measured by the AML). The dataset presented from this point forward consists of results from both mobile laboratories.

ERs in NG facility plumes, including production pads, gathering stations, transmission stations, pipelines and dehydrator stations, display some spatial variability throughout the study area, especially on the eastern side, as shown in Figure 3. Table 2 shows the range of ERs observed at each type of NG facility measured on the ground in the western and eastern halves of the study area. The sample size of three of the NG source types (pipeline junctions, dehydrator stations, and transmission stations) is small, making it difficult to draw conclusions about the observed ER ranges at these types of facilities. More measurements were made throughout the study area at production pads and gathering stations, which represent the majority (~98%) of the active facilities and about 97% of estimated CH4 emissions (Vaughn et al., 2018) in the study area. Production pads and gathering stations display similar ER ranges of 0.8 to 7.8% and 1.0 to 7.9%, respectively. The similar ER ranges between these two types of facilities are to be expected given the lack of gas processing in this dry gas field, which has very low water and non-methane hydrocarbon content and no co-produced natural gas condensate or oil.

Figure 3

Spatial variability of ERs in the study area. Mean facility-level NG ERs determined from ML measurements. DOI: https://doi.org/10.1525/elementa.351.f3

Figure 3

Spatial variability of ERs in the study area. Mean facility-level NG ERs determined from ML measurements. DOI: https://doi.org/10.1525/elementa.351.f3

Table 2

ER ranges by facility type. Range of ERs in plumes sampled downwind of NG facilities in the western and eastern halves of the study area by the NOAA and Aerodyne MLs. DOI: https://doi.org/10.1525/elementa.351.t2

Source typeWest ERSample sizeEast ERSample size

Production pads 0.8–2.1% 52 0.9–7.8% 20
Gathering stations 1.0–1.7% 26 1.1–7.9% 10
Pipeline junction N/A 3.9%
Dehydrator station N/A 4.0%
Transmission station 2.0% 1.6–5.7%
Source typeWest ERSample sizeEast ERSample size

Production pads 0.8–2.1% 52 0.9–7.8% 20
Gathering stations 1.0–1.7% 26 1.1–7.9% 10
Pipeline junction N/A 3.9%
Dehydrator station N/A 4.0%
Transmission station 2.0% 1.6–5.7%

The ranges of ERs observed at production pads are similar to gathering stations within the western half and within the eastern half of the study area. In the western half, production pad ERs range from 0.8 to 2.1% and gathering stations from 1.0 to 1.7%. In the eastern half, production pad ERs range from 0.9 to 7.8% and gathering stations from 1.1 to 7.9%. Facilities in the eastern half have a greater range of ERs that tend toward higher values. Only one of 13 facilities with an ER of 2.1% or greater is located in the western half. Of the remaining facilities with relatively high ERs, all of which are located in the eastern half, eight are located in White county (2.1–5.7%), and four are located in Independence county (6.2–7.9%). Though the range of ERs observed at NG facilities in the eastern half is larger, facilities with relatively high ERs tend to be in the same geographic subregion of the study area (see Figure 3).

At the gathering station at which a 7.9% ER was observed, combustion was evident, as CH4 plumes correlated with both CO2 and CO. At gathering stations, which have on-site compressors, combustion processes are expected but CH4 correlation with combustion tracers was observed at only one of the eight gathering stations surveyed by the NOAA ML. It is possible that at the other stations, hot combustion plumes rose too high and that the ML was not far enough downwind to be able to measure the plume where it reached the surface. However, even though combustion occurs at compressor stations, higher ERs at these facilities are not necessarily expected because combustion slip (unburned fuel in the compressor engine exhaust) has been shown to be the main source of CH4 emissions at gathering stations in the study area (Vaughn et al., 2017).

Figure 4 shows the distribution of ERs measured at NG facilities by the MLs. The majority (64%) of NG ERs observed in the western half range from 1.2 to 1.5%, with a mode of 1.3%. The eastern half has a similar range of frequent ERs of 1.2 to 1.5% with a peak in the distribution at 1.3% also. However, there is significant variability in ER in the eastern half, where 34% of measured facilities have ERs of 2% or greater. Higher ERs in the eastern half are confirmed by comparison with C2H6 to CH4 ratios from gas composition data provided by operators for a subset of production pads and gathering stations (Figure 5). In cases where results from multiple gas composition samples were provided for a facility, the mean C2H6 to CH4 ratio was used. Gas composition data exhibit modes for both the western and eastern halves of 1.3 to 1.5% respectively, consistent with ML plume data. Composition data also reveal a portion of facilities in the eastern half with C2H6 to CH4 ratios greater than 2%. This portion is 28% compared with the 34% determined by ML measurements. Like the ML results, all facilities in the eastern half with C2H6 to CH4 ratios of at least 2% are located in White and Independence counties. Higher ERs in the eastern half are likely a result of the thermal maturity of the geologic formation being drilled. Thermal maturity refers to the extent to which reactions involving heat that convert organic matter to oil and natural gas have occurred. Generally, natural gas is expected to be wetter (higher C2H6 to CH4 ratio) as the thermal maturity of its source rock decreases (Visschedijk et al., 2018).

Figure 4

ER distributions from ML measurements. (a) Distribution of the 114 NG ERs compiled by the NOAA and Aerodyne MLs. (b) Same as (a) but for NG ERs ≥2%. DOI: https://doi.org/10.1525/elementa.351.f4

Figure 4

ER distributions from ML measurements. (a) Distribution of the 114 NG ERs compiled by the NOAA and Aerodyne MLs. (b) Same as (a) but for NG ERs ≥2%. DOI: https://doi.org/10.1525/elementa.351.f4

Figure 5

Ratio distributions from operator gas composition data. (a) Distribution of gas composition C2H6 to CH4 ratios at production pads and gathering stations using 2015 gas composition data provided by local NG facility operators. (b) same as (a) but for ratios ≥2%. DOI: https://doi.org/10.1525/elementa.351.f5

Figure 5

Ratio distributions from operator gas composition data. (a) Distribution of gas composition C2H6 to CH4 ratios at production pads and gathering stations using 2015 gas composition data provided by local NG facility operators. (b) same as (a) but for ratios ≥2%. DOI: https://doi.org/10.1525/elementa.351.f5

In the Barnett Shale play in Texas, where natural gas is produced throughout the entirety of the study area, the distribution of 172 NG ERs measured by Yacovitch et al. (2015) showed a peak at approximately 1.5%. The highest ERs were observed in the northwest, where oil is co-produced.

The aircraft performed spiral flight patterns around several potential CH4 sources, including non-NG sources, and ERs were calculated. Although results are not used in the attribution, they can be found in the Supplemental Material.

#### Measurements at poultry farms

Arkansas is among the top states for the production of poultry in the United States, ranking third and fourth for broiler production in terms of head of chickens and pounds produced, respectively, in 2015 (USDA, 2015). While chickens are not known to be a CH4 source as a result of enteric fermentation like cattle, CH4 could be emitted from their manure (Monteny et al., 2001). Chicken litter, a combination of manure and the bedding in the chicken house, is often temporarily stored on-site until it can be removed and applied as a fertilizer for crops. However, best management practices recommend covered (dry) storage, which serves to minimize the potential for CH4 and ammonia production (Ogejo et al., 2009).

Commercial poultry farms in the study area were long enclosed buildings with fans for ventilation and no visible liquid manure storage ponds. Measurements were made downwind of nine poultry farms (see Supplemental Material for locations), identified by the strong downwind localized smell of ammonia. At the time of the study, 292 poultry farms were known to be in the study area. Using the same 50 ppb threshold above the local background as was used to identify NG CH4 plumes, poultry farms exhibited no substantial CH4 enhancements. For a poultry farm measurement shown in Figure 6, CH4 enhancements were approximately 10 ppb during each ML transect. For comparison, background variability for surface CH4 on ML drives had a range of approximately 300 ppb. Based on these data, it is not expected that poultry contributed significantly to total CH4 emissions in the study area. The bottom-up CH4 emissions estimate for poultry farms provided by Schwietzke et al. (2017) of <200 kg CH4/hr for the entire study area also suggests that poultry farms are a small source of CH4 emissions when compared to the total (30.6 tons CH4/hr).

Figure 6

Poultry farm ML downwind measurements. Time series of measurements collected downwind of a poultry farm at which the ML performed three successive downwind transects. CH4 (red, blue and turquoise) is from the CRDS and C2H6 (green) from the TILDAS. Peak CH4 enhancements are on the order of 10 ppb, not a significant enhancement compared to the 50 ppb threshold used for this study and the levels detected downwind of NG facility plumes. DOI: https://doi.org/10.1525/elementa.351.f6

Figure 6

Poultry farm ML downwind measurements. Time series of measurements collected downwind of a poultry farm at which the ML performed three successive downwind transects. CH4 (red, blue and turquoise) is from the CRDS and C2H6 (green) from the TILDAS. Peak CH4 enhancements are on the order of 10 ppb, not a significant enhancement compared to the 50 ppb threshold used for this study and the levels detected downwind of NG facility plumes. DOI: https://doi.org/10.1525/elementa.351.f6

#### Determination of mean NG C2H6 to CH4 ERs

To derive mean ERs for the full study area as well as the western and eastern halves, we calculate spatially distinct means of the ML NG facility plume ERs weighted by NG production volumes using a Delaunay Triangulation (see Supplemental Material). This method accounts for spatial variability in the ER across each half and for sections of the study area with less measurement coverage than others. The Delaunay Triangulation algorithm in MATLAB connects the ER measurement locations into triangles (using their GPS coordinates) so that no other measurement location lies in a circumcircle of a given triangle and the value of the minimum angle of each triangle is maximized (See Figure S5). The varying sizes of the resulting triangles account for disparity in measurement density in the study area and ensure that there are no triangles without associated ER measurements. Each vertex of a triangle represents a measured ER. For each triangle, the three ERs are averaged and the mean value is assigned as the mean ER for all NG emissions in that triangle. Using September 2015 NG production data from the Arkansas Oil and Gas Commission (AOGC) (AOGC, 2015), the NG production of all production pads contained within each triangle was summed. With this production data and the mean ER for each triangle, a production weighted mean NG ER was calculated for the study area, western half, and eastern half.

A simulation was performed in which a simple random sample of 80% of the NOAA and AML facility ER data points was selected without replacement and the Delaunay Triangulation method was applied. Performing this simulation 30 times generated a mean ER and standard deviation, which was used to assign an uncertainty to the mean NG ERs generated by Delaunay Triangulation. The resulting mean NG ERs are 1.68 ± 0.10% for the full study area, 1.34 ± 0.03% for the western half, and 2.10 ± 0.22% for the eastern half.

### 3.2. Determination of area-scale C2H6 to CH4 ERs from aircraft data

While ML data were used to determine ERs characteristic of NG operations in the study area, aircraft data were used to calculate area-scale ERs that include CH4 emissions from all sources in the study area. CH4 plumes measured by the aircraft are from a combination of multiple point sources after emissions have been transported from their sources and mixing has occurred downwind. NG source ERs and area-scale ERs were then combined (Eq. 7) to estimate the portion of the total CH4 emissions in the study area originating from NG operations.

Area-scale ERs were calculated by fitting orthogonal distance regression slopes through the C2H6 vs. CH4 in situ aircraft measurements for multiple flight legs in the boundary layer downwind of or over the study area and calculating a mean for each flight. The resulting aircraft ERs are shown in Table 3. ERs for each transect or leg of each flight used for attribution can be found in the Supplemental Material.

Table 3

Area-scale ERs determined from aircraft measurements. Area-scale ERs for the western half determined using aircraft data for each flight for which analysis was possible. DOI: https://doi.org/10.1525/elementa.351.t3

DateTime window of measurements (local time)Altitude range (meters above ground level)CH4 range (ppb)C2H6 range (ppb)R2 range and number of legsMean C2H6 to CH4 ER (in situ)

9/22/2015 3:56 PM–4:12 PM 403–735 1910–2050 1.3–3.7 R2 = 0.69 for 1 leg 1.2 ± 0.03%
10/2/2015 12:47 PM–1:13 PM 244–460 1940–2080 1.2–2.9 R2 = 0.64 for 1 leg 1.1 ± 0.04%
10/5/2015 12:09 PM–2:05 PM 157–444 1999–2337 2.3–8.4 R2 from 0.67 to 0.91 for 7 legs 1.2 ± 0.03%
10/6/2015 11:04 AM–12:09 PM 198–466 1990–2236 2.3–5.7 R2 from 0.65 to 0.91 for 4 legs 1.3 ± 0.05%
10/7/2015 12:41 PM–3:19 PM 138–560 1939–3500 1.2–11.4 R2 from 0.67 to 0.90 for 10 legs 1.3 ± 0.04%
DateTime window of measurements (local time)Altitude range (meters above ground level)CH4 range (ppb)C2H6 range (ppb)R2 range and number of legsMean C2H6 to CH4 ER (in situ)

9/22/2015 3:56 PM–4:12 PM 403–735 1910–2050 1.3–3.7 R2 = 0.69 for 1 leg 1.2 ± 0.03%
10/2/2015 12:47 PM–1:13 PM 244–460 1940–2080 1.2–2.9 R2 = 0.64 for 1 leg 1.1 ± 0.04%
10/5/2015 12:09 PM–2:05 PM 157–444 1999–2337 2.3–8.4 R2 from 0.67 to 0.91 for 7 legs 1.2 ± 0.03%
10/6/2015 11:04 AM–12:09 PM 198–466 1990–2236 2.3–5.7 R2 from 0.65 to 0.91 for 4 legs 1.3 ± 0.05%
10/7/2015 12:41 PM–3:19 PM 138–560 1939–3500 1.2–11.4 R2 from 0.67 to 0.90 for 10 legs 1.3 ± 0.04%

Figure 7A shows an example raster flight track for the determination of an ER for one leg (leg 4). For this flight on October 5, 2015 with wind from the North, CH4 enhancements were observed over the western half of the study area, beginning with leg 4 (starting upwind in the northwestern corner and moving further southeast). The portion of leg 4 for which CH4 is enhanced is shown with a box in Figure 7B and the corresponding C2H6 to CH4 scatterplot in Figure 7C. The slope is 1.25 ± 0.04% (R2 = 0.88).

Figure 7

Example of ER determination process for an aircraft leg. (a) October 5, 2015 flight track color-coded by CH4. Leg 4 is labeled. (b) The C2H6 and CH4 time series for leg 4. The box indicates the region where CH4 is enhanced over background. (c) The C2H6 vs. CH4 scatterplot of the data from the CH4 enhanced portion of leg 4 fit with an orthogonal distance regression line. DOI: https://doi.org/10.1525/elementa.351.f7

Figure 7

Example of ER determination process for an aircraft leg. (a) October 5, 2015 flight track color-coded by CH4. Leg 4 is labeled. (b) The C2H6 and CH4 time series for leg 4. The box indicates the region where CH4 is enhanced over background. (c) The C2H6 vs. CH4 scatterplot of the data from the CH4 enhanced portion of leg 4 fit with an orthogonal distance regression line. DOI: https://doi.org/10.1525/elementa.351.f7

Representative area-scale ERs were calculated for the western half using five raster flights conducted on September 22, October 2, October 5, October 6 and October 7, 2015 (Table 3), and a mean C2H6 to CH4 slope for the enhancements observed over and/or downwind of the western half of 1.21 ± 0.02% was calculated.

While this result is higher than the 0.6% determined by Peischl et al. (2015) from two flights over the Fayetteville Shale, the ER reported in this study is only for the western half of the study area, whereas the 0.6% reported by Peischl et al. is for the entire region. The mean area-scale ER of 1.21 ± 0.02% from this study falls within the range of ~1–3% from dry gas sources with a mode of 1.8% determined from flights in the Barnett Shale (Smith et al., 2015).

Larger enhancements over the western half permitted calculation of area-scale ERs whereas in the eastern half, C2H6 signal with the same magnitude as the measurement noise made calculating ERs impossible. CH4 emissions from NG operations normalized by NG production were found to be about one half in the eastern half relative to the western half on two days (Schwietzke et al., 2017). Since C2H6 is co-emitted with CH4, it is likely that C2H6 emissions are also lower in the eastern half, and a spatially-resolved bottom-up CH4 emissions inventory of the total study area showed no substantial differences in non-NG related CH4 emissions between the western and eastern half. Assuming similar atmospheric mixing and dispersion conditions in the eastern half as in the western half, lower emissions would produce smaller enhancements, which matches observations.

Though C2H6 mixing ratios over the eastern half of the study area were generally lower than those observed over the western half, a raster flight on October 6 revealed C2H6 enhancements in the northeastern section of the study area (Figure 8). ERs on the four legs of this flight with C2H6 mixing ratios above 3.5 ppb (shown in light blue on Figure 8) were 1.5 ± 0.1%, 2.0 ± 0.1%, 2.5 ± 0.2%, and 7.1 ± 0.2%. The plume with an ER of 7.1% (shown in red on Figure 8) represents a localized enhancement in both CH4 and C2H6.

Figure 8

CH4 and C2H6 mole fractions on flights used for CH4 attribution. (a) Flight tracks from five raster flights on September 22 and October 2, 5, 6, and 7, 2015 color-coded by CH4 mole fraction. (b) Flight tracks color-coded by C2H6 mole fraction. DOI: https://doi.org/10.1525/elementa.351.f8

Figure 8

CH4 and C2H6 mole fractions on flights used for CH4 attribution. (a) Flight tracks from five raster flights on September 22 and October 2, 5, 6, and 7, 2015 color-coded by CH4 mole fraction. (b) Flight tracks color-coded by C2H6 mole fraction. DOI: https://doi.org/10.1525/elementa.351.f8

The mean fraction of CH4 emissions attributable to NG sources in the western half of the study area was calculated using Eq. 7. With [C2H6/CH4]area = 1.21 ± 0.02% and [C2H6/CH4]NG = 1.34 ± 0.03%, the mean estimate of the portion of the total CH4 emissions originating from NG related sources in the western half of the study area was 90 ± 5% (1-sigma), and the portion of the total CH4 emissions originating from other sources was estimated to be 10 ± 5%.

Our estimate for the portion of total CH4 emissions originating from NG sources compares well with bottom-up estimates made for the western half of the study area by Vaughn et al. (2018). Bottom-up emission estimates were made for the two afternoons during which the mass balance flights used by Schwietzke et al. (2017) to calculate total CH4 emissions occurred. Vaughn et al. (2018) estimated 89.9% and 87.8% of total CH4 emissions in the western half to have originated from NG sources on October 1 and October 2, 2015, respectively, yielding a mean of 88.8%. The highest estimated CH4 emissions from non-NG sources were from beef cattle and geologic seeps (Schwietzke et al., 2017).

Several factors prevented attribution of CH4 emissions in the eastern half. First, CH4 and C2H6 enhancements measured by the aircraft in the planetary boundary layer over the eastern half were low compared to those in the western half (as shown in Figure 8 and by Schwietzke et al. (2017). In particular, the signal to noise ratio for the C2H6 enhancements onboard the aircraft was too low over the eastern half to allow estimation of ERs for aerial plumes mapped over the subregion. The C2H6 enhancements were also too low along the downwind transects of the two mass balance flights analyzed by Schwietzke et al. (2017) to determine a mass balance estimation of the C2H6 emissions for the entire study region, with C2H6 ranges on the downwind transects on October 1st and 2nd of 0.8 to 2.4 ppb and 1.0 to 2.0 ppb, respectively.

## 4. Conclusions

Ground-based, atmospheric measurements were used to determine a representative range of C2H6 to CH4 facility-level plume ERs, confirmed by produced gas composition data. Using airborne and mobile surface atmospheric measurements of C2H6 and CH4 and a simple two source mixing model, we show that 85–95% of total CH4 emissions in the western half of the study area originate from NG operations. The western half represents 76% of total study area CH4 emissions based on aircraft measurements (Schwietzke et al., 2017). Our result agrees with the NG CH4 emissions estimate from a bottom-up model combining field measurements and spatiotemporal NG operations data by Vaughn et al. (2018).

Attribution of CH4 emissions in the eastern half of the study area was not possible due to low boundary layer enhancements in CH4 and C2H6. Low signal to noise for C2H6 also prohibited the use of the aircraft mass balance model for obtaining a C2H6 flux in both halves of the study area. Additionally, the assumption of a single mode NG ER distribution is not valid for the eastern half due to the more complex NG ER distribution.

The spatial variability in ER across the study area, as demonstrated by the difference in ER distributions in the western half compared with the eastern half, is important to account for when developing aggregated source emission signatures in a NG producing region. Although the majority of NG ERs observed in both the western and eastern halves fall in the same range of 1.2 to 1.5%, the eastern half has a second mode in its distribution, in which 34% of measured facilities have ERs of 2% or greater.

The Fayetteville Shale was chosen for the field campaign partly because CH4 emissions from sources other than NG systems were expected to be low, making CH4 source attribution less complicated than in areas with many potential CH4 sources. Results from this study confirm that NG operations are the primary source of CH4 in the western portion of the study area. Poultry farms were not found to be a significant source of CH4.

Use of a ML facilitated the collection of an extensive amount of data on specific types of NG facilities and their C2H6 to CH4 source signatures. For facility emission sampling, the ML was positioned directly adjacent to and downwind of the target facility such that ERs could be assigned to specific facilities. Fast response C2H6 and CH4 in situ measurements made it possible to determine ERs in multiple plumes downwind of individual emission sources, yielding robust facility-level statistics compared to those from a simple linear regression using a small number of discrete flask samples collected in sources plumes or background conditions at different locations and times. Since ML routes were restricted only by public road access, it was possible to perform repeat measurements downwind of several facilities and examine day-to-day ER variability. With the exception of one production pad at which some combustion was likely occurring, ERs at all seven facilities with repeated measurements over several days were consistent over time (<10% variation), suggesting that measurements of C2H6 and CH4 by ML surveys represent daily ERs at a given facility during the campaign.

Aircraft measurements in the atmospheric boundary layer were critical for obtaining information about area-integrated emission signals. By flying over and downwind of the entire western half of the study area, where CH4 and C2H6 enhancements were greatest, the aircraft sampled area plumes in the western half and provided a representative ER for the mix of emission sources.

The attribution model used in this study assumes that all C2H6 emissions originate from NG sources. Additionally, mean ERs for NG and total emissions are assumed to remain constant from day to day. The two-source model can only be applied in a simple case with two source categories with a single mode each, and would not be appropriate for use in an area with more complicated mix of sources and source signatures, including wetter gas and oil production, which could introduce other modes in the ER distribution. A complex mix of co-located sources with several different source signatures can present a challenge in that multiple tracers or constraints may be required for attribution. In such a case, a representative sample population of facility plumes with sufficient signal to noise may be used as a proxy for the mix of emissions as in Smith et al. (2015), preferably for multiple days and different flight paths. To further advance the value of top-down methods, it would be beneficial to test how stable the emission estimates and attribution solution are for different flight patterns, wind configurations, and dates/seasons, and to have access to composition data for key emission sources in a study region.

## Data Accessibility Statement

Enhancement ratios obtained by the mobile laboratories are uploaded in Microsoft Excel .xlsx format. NOAA mobile laboratory C2H6 and CH4 enhancements relative to the local background are provided for ten different facilities (Sites 1–10) as text files. Files for the remaining 104 facilities can be provided upon request. Due to a governance agreement with project sponsors, locations of high frequency in situ ground measurements cannot be published. Aircraft data for the flights used for attribution and all discrete flask data are provided. Data are available at the following address: ftp://aftp.cmdl.noaa.gov/data/campaign/mls/RPSEA2015/.

## Acknowledgments

We thank members of our Technical Review Committee, F. George, G. Jersey, P. Lacey, T. Rufael, and D. Sundararajan, for their feedback and guidance during the data analysis and writing process. We also thank Southwestern Energy for providing a parking site with power for the mobile laboratory during the field campaign.

## Funding information

Funding for this work was provided by the Research Partnership to Secure Energy for America/National Energy Technology Laboratory contract no. 12122-95/DE-AC26-07NT42677 to the Colorado School of Mines, with additional equal contributions from the Colorado Energy Research Collaboratory, the National Oceanic and Atmospheric Administration, Southwestern Energy, XTO Energy (a subsidiary of ExxonMobil), Chevron, Statoil, and the American Gas Association.

## Competing interests

The authors have no competing interests to declare.

## Author contributions

• Substantial contributions to conception and design: IMM, GP, DZ, SS, SC, TV, CB, RS

• Acquisition of data: GP, ET, JK, PH, SCH, SC, SS, BM, BH, ED, PL, SW, EM, MC, AC, MR, DK

• Analysis and interpretation of data: GP, SS, TIY, DZ

• Drafting the article or revising it critically for important intellectual content: GP, SS, TIY, BM

• Final approval of the version to be published: All authors

1
Aircraft program methods
.
2018
. Available at: https://www.esrl.noaa.gov/gmd/ccgg/aircraft/sampling.html. Accessed 2018 August 6.
2
Alvarez
,
RA
,
Zavala-Araiza
,
D
,
Lyon
,
DR
,
Allen
,
DT
,
Barkley
,
ZR
,
Brandt
,
AR
,
Davis
,
KJ
,
Herndon
,
SC
,
Jacob
,
DJ
,
Karion
,
A
,
Kort
,
EA
,
Lamb
,
BK
,
Lauvaux
,
T
,
Maasakkers
,
JD
,
Marchese
,
AJ
,
Omara
,
M
,
Pacala
,
SW
,
Peischl
,
J
,
Robinson
,
AL
,
Shepson
,
PB
,
Sweeney
,
C
,
Townsend-Small
,
A
,
Wofsy
,
SC
and
Hamburg
,
SP
.
2018
.
Assessment of methane emissions from the U.S. oil and gas supply chain
.
Science
361
(
1693
):
1
8
. DOI:
3
Arkansas Geological Survey
.
2015
.
Natural gas
. Available at: https://www.geology.arkansas.gov/energy/natural-gas-in-arkansas.html. Accessed 2018 August 6.
4
Arkansas Oil and Gas Commission (AOGC)
.
2015
.
Fayetteville shale gas sales
. Available at: http://www.aogc.state.ar.us/sales/shale.aspx. Accessed 2018 August 6.
5
Bell
,
CS
,
Vaughn
,
TL
,
Zimmerle
,
DJ
,
Herndon
,
SC
,
Yacovitch
,
TI
,
Heath
,
GA
,
Pétron
,
G
,
Edie
,
R
,
Field
,
RA
,
Murphy
,
SM
,
Robertson
,
AM
and
Soltis
,
J
.
2017
.
Comparison of methane emission estimates from multiple measurement techniques at natural gas production pads
.
Elem Sci Anth
5
:
79
. DOI:
6
Brandt
,
AR
,
Heath
,
GA
,
Kort
,
EA
,
O’Sullivan
,
F
,
Pétron
,
G
,
Jordaan
,
SM
,
Tans
,
P
,
Wilcox
,
J
,
Gopstein
,
AM
,
Arent
,
D
,
Wofsy
,
S
,
Brown
,
NJ
,
,
R
,
Stucky
,
GD
,
Eardley
,
D
and
Harriss
,
R
.
2014
.
Methane leaks from North American natural gas systems
.
Science
343
(
6172
):
733
735
. DOI:
7
Caulton
,
DR
,
Shepson
,
PB
,
Santoro
,
RL
,
Sparks
,
JP
,
Howarth
,
RW
,
Ingraffea
,
AR
,
Cambaliza
,
MOL
,
Sweeney
,
C
,
Karion
,
A
,
Davis
,
KJ
,
Stirm
,
BH
,
Montzka
,
SA
and
Miller
,
BR
.
2014
.
Toward a better understanding and quantification of methane emissions from shale gas development
.
111
(
17
):
6237
6242
. DOI:
8
Conley
,
S
,
Faloona
,
I
,
Mehrotra
,
S
,
Suard
,
M
,
Lenschow
,
DH
,
Sweeney
,
C
,
Herndon
,
SC
,
Schwietzke
,
S
,
Pétron
,
G
,
Pifer
,
J
,
Kort
,
EA
and
Schnell
,
R
.
2017
.
Application of Gauss’s Theorem to quantify localized surface emissions from airborne measurements of wind and trace gases
.
Atmos Meas Tech
10
:
3345
3358
. DOI:
9
Dlugokencky
,
EJ
,
Steele
,
LP
,
Lang
,
PM
and
Masarie
,
KA
.
1994
.
The growth rate and distribution of atmospheric methane
.
J Geophys Res Atmos
99
:
17021
17043
. DOI:
10
Earth System Research Laboratory (ESRL)
.
2017
.
Global greenhouse gas reference network
. Available at: http://www.esrl.noaa.gov/gmd/ccgg. Accessed 2018 August 6.
11
EIA
.
2016
.
Carbon dioxide emissions coefficients
. Available at: https://www.eia.gov/environment/emissions/co2_vol_mass.php. Accessed 2018 August 6.
12
EIA
.
2017
.
U.S. dry natural gas production
. Available at: https://www.eia.gov/dnav/ng/hist/n9070us2A.htm. Accessed 2018 August 6.
13
EPA
.
2017
.
Inventory of U.S. greenhouse gas emissions and sinks: 1990–2015
. Available at: https://www.epa.gov/sites/production/files/2017-02/documents/2017_complete_report.pdf. Accessed 2018 August 6.
14
Frankenberg
,
C
,
Thorpe
,
AK
,
Thompson
,
DR
,
Hulley
,
G
,
Kort
,
EA
,
Vance
,
N
,
Borchardt
,
J
,
Krings
,
T
,
Gerilowski
,
K
,
Sweeney
,
C
,
Conley
,
S
,
Bue
,
BD
,
Aubrey
,
,
Hook
,
S
and
Green
,
RO
.
2016
.
Airborne methane remote measurements reveal heavy-tail flux distribution in Four Corners region
.
35
(
35
):
9734
9739
. DOI:
15
Hopkins
,
FM
,
Kort
,
EA
,
Bush
,
SE
,
Ehleringer
,
JR
,
Lai
,
C-T
,
Blake
,
DR
and
Randerson
,
JT
.
2016
.
Spatial patterns and source attribution of urban methane in the Los Angeles Basin
.
J Geophys Res-Atmos
121
(
5
):
2490
2507
. DOI:
16
Karion
,
A
,
Sweeney
,
C
,
Kort
,
EA
,
Shepson
,
PB
,
Brewer
,
A
,
Cambaliza
,
M
,
Conley
,
S
,
Davis
,
K
,
Deng
,
A
Hardesty
,
M
,
Herndon
,
SC
,
Lauvaux
,
T
,
Lavoie
,
T
,
Lyon
,
D
,
Newberger
,
T
,
Pétron
,
G
,
Rella
,
C
,
Smith
,
M
,
Wolter
,
S
,
Yacovitch
,
TI
and
Tans
,
P
.
2015
.
Aircraft-based estimate of total methane emissions from the Barnett Shale region
.
Environ Sci Technol
49
(
13
):
8124
8131
. DOI:
17
Karion
,
A
,
Sweeney
,
C
,
Pétron
,
G
,
Frost
,
G
,
Hardesty
,
RM
,
Kofler
,
J
,
Miller
,
BR
,
Newberger
,
T
,
Wolter
,
S
,
Banta
,
R
,
Brewer
,
A
,
Dlugokencky
,
E
,
Lang
,
PM
,
Montzka
,
SA
,
Schnell
,
R
,
Tans
,
P
,
Trainer
,
M
,
Zamora
,
R
and
Conley
,
S
.
2013
.
Methane emissions estimate from airborne measurements over a western United States natural gas field
.
Geophys Res Lett
40
(
16
):
4393
4397
. DOI:
18
Kort
,
EA
,
Smith
,
ML
,
Murray
,
LT
,
Gvakharia
,
A
,
Brandt
,
AR
,
Peischl
,
J
,
Ryerson
,
TB
,
Sweeney
,
C
and
Travis
,
K
.
2016
.
Fugitive emissions from the Bakken shale illustrate role of shale production in global ethane shift
.
Geophys Res Lett
43
(
9
):
4617
4623
. DOI:
19
Monteny
,
GJ
,
Groenestein
,
CM
and
Hilhorst
,
MA
.
2001
.
Interactions and coupling between emissions of methane and nitrous oxide from animal husbandry
.
Nutr Cycl Agroecosys
60
:
123
132
. DOI:
20
Nisbet
,
EG
,
Dlugokencky
,
EJ
,
Manning
,
MR
,
Lowry
,
D
,
Fisher
,
RE
,
France
,
JL
,
Michel
,
SE
,
Miller
,
JB
,
White
,
JWC
,
Vaughn
,
B
,
Bousquet
,
P
,
Pyle
,
JA
,
Warwick
,
NJ
,
Cain
,
M
,
Brownlow
,
R
,
Zazzeri
,
G
,
Lanoisellé
,
M
,
Manning
,
AC
,
Gloor
,
E
,
Worthy
,
DEJ
,
Brunke
,
E-G
,
Labuschagne
,
C
,
Wolff
,
EW
and
Ganesan
,
AL
.
2016
.
Rising atmospheric methane: 2007–2014 growth and isotopic shift
.
Global Biogeochem Cy
30
(
9
):
1356
1370
. DOI:
21
Ogejo
,
JA
and
Collins
,
ER
, Jr.
2009
. Storing and handling poultry litter.
Virginia Polytechnic Institute and State Unversity, Virginia Cooperative Extension
.
422
054
. Available at: https://pubs.ext.vt.edu/content/dam/pubs_ext_vt_edu/442/442-054/442-054_pdf.pdf. Accessed 2018 December 21.
22
Peischl
,
J
,
Karion
,
A
,
Sweeney
,
C
,
Kort
,
EA
,
Smith
,
ML
,
Brandt
,
AR
,
Yeskoo
,
T
,
Aikin
,
KC
,
Conley
,
S
,
Gvakharia
,
A
,
Trainer
,
M
,
Wolter
,
S
and
Ryerson
,
TB
.
2016
.
Quantifying atmospheric methane emissions from oil and natural gas production in the Bakken shale region of North Dakota
.
J Geophys Res Atmos
121
(
10
):
6101
6111
. DOI:
23
Peischl
,
J
,
Ryerson
,
TB
,
Aikin
,
KC
, de, Gouw, JA,
Gilman
,
JB
,
Holloway
,
JS
,
Lerner
,
BM
,
,
R
,
Neuman
,
JA
,
Nowak
,
JB
,
Trainer
,
M
,
Warneke
,
C
and
Parrish
,
DD
.
2015
.
Quantifying atmospheric methane emissions from the Haynesville, Fayetteville, and northeastern Marcellus shale gas production regions
.
J Geophys Res Atmos
120
(
5
):
2119
2139
. DOI:
24
Pétron
,
G
,
Frost
,
G
,
Miller
,
BR
,
Hirsch
,
AI
,
Montzka
,
SA
,
Karion
,
A
,
Trainer
,
M
,
Sweeney
,
C
,
Andrews
,
AE
,
Miller
,
L
,
Kofler
,
J
, Bar-
Ilan
,
A
,
Dlugokencky
,
E
,
Patrick
,
L
,
Moore
,
CT
, Jr.
,
Ryerson
,
TB
,
Siso
,
C
,
Kolodzey
,
W
,
Lang
,
PM
,
Conway
,
T
,
Novelli
,
P
,
Masarie
,
K
,
Hall
,
B
,
Guenther
,
D
,
Kitzis
,
D
,
Miller
,
J
,
Welsh
,
D
,
Wolf
,
D
,
Neff
,
W
and
Tans
,
P
.
2012
.
Hydrocarbon emissions characterization in the Colorado Front Range: A pilot study
.
J Geophys Res
117
(
D4
):
D404304
. DOI:
25
Pétron
,
G
,
Karion
,
A
,
Sweeney
,
C
,
Miller
,
BR
,
Montzka
,
SA
,
Frost
,
G
,
Trainer
,
M
,
Tans
,
P
,
Andrews
,
AE
,
Kofler
,
J
,
Helmig
,
D
,
Guenther
,
D
,
Dlugokencky
,
E
,
Lang
,
PM
,
Newberger
,
T
,
Wolter
,
S
,
Hall
,
B
,
Novelli
,
P
,
Brewer
,
A
,
Conley
,
S
,
Hardesty
,
M
,
Banta
,
R
,
White
,
A
,
Noone
,
D
,
Wolfe
,
D
and
Schnell
,
R
.
2014
.
A new look at methane and nonmethane hydrocarbon emissions from oil and natural gas operations in the Colorado Denver-Julesburg Basin
.
J Geophys Res Atmos
119
(
11
):
6836
6852
. DOI:
26
Robertson
,
AM
,
Edie
,
R
,
Snare
,
D
,
Soltis
,
J
,
Field
,
RA
,
Burkhart
,
MD
,
Bell
,
CS
,
Zimmerle
,
DJ
and
Murphy
,
SM
.
2017
.
Variation in methane emission rates from well pads in four oil and gas basins with contrasting production volumes and compositions
.
Environ Sci Technol
51
(
15
):
8832
8840
. DOI:
27
Roscioli
,
JR
,
Yacovitch
,
TI
,
Floerchinger
,
C
,
Mitchell
,
AL
,
Tkacik
,
DS
,
Subramanian
,
R
,
Martinez
,
DM
,
Vaughn
,
TL
,
Williams
,
L
,
Zimmerle
,
DJ
,
Robinson
,
AL
,
Herndon
,
SC
and
Marchese
,
AJ
.
2015
.
Measurements of methane emissions from natural gas gathering facilities and processing plants: Measurement methods
.
Atmos Meas Tech
8
(
5
):
2017
2035
. DOI:
28
Rudolph
,
J
and
Enhalt
,
DH
.
1981
.
Measurements of C2-C5 hydrocarbons over the North Atlantic
.
J Geophys Res
86
(
C12
):
11959
11964
. DOI:
29
Schwietzke
,
S
,
Pétron
,
G
,
Conley
,
S
,
Pickering
,
C
,
,
I
,
Dlugokencky
,
E
,
Tans
,
P
,
Vaughn
,
TL
,
Bell
,
CS
,
Zimmerle
,
DJ
,
Wolter
,
S
,
King
,
CW
,
White
,
AB
,
Coleman
,
T
,
Biano
,
L
and
Schnell
,
R
.
2017
.
Improved mechanistic understanding of natural gas methane emissions from spatially resolved aircraft measurements
.
Environ Sci Technol
51
(
12
):
7286
7294
. DOI:
30
Smith
,
ML
,
Gvakharia
,
A
,
Kort
,
EA
,
Sweeney
,
C
,
Conley
,
S
,
Faloona
,
I
,
Newberger
,
T
,
Schnell
,
R
,
Schwietzke
,
S
and
Wolter
,
S
.
2017
.
Airborne quantification of methane emissions over the Four Corners region
.
Environ Sci Technol
51
(
10
):
5832
5837
. DOI:
31
Smith
,
ML
,
Kort
,
EA
,
Karion
,
A
,
Sweeney
,
C
,
Herndon
,
SC
and
Yacovitch
,
TI
.
2015
.
Airborne ethane observations in the Barnett Shale: Quantification of ethane flux and attribution of methane emissions
.
Environ Sci Technol
49
(
13
):
8158
8166
. DOI:
32
Townsend-Small
,
A
,
Marrero
,
JE
,
Lyon
,
DR
,
Simpson
,
IJ
,
Meinardi
,
S
and
Blake
,
DR
.
2015
.
Integrating source apportionment tracers into a bottom-up inventory of methane emissions in the Barnett Shale hydraulic fracturing region
.
Environ Sci Technol
49
(
13
):
8175
8182
. DOI:
33
USDA
.
2015
.
State-level poultry production data
. Available at: https://quickstats.nass.usda.gov/results/B2269209-0E29-3655-B48E-F8BCDDB62A65. Accessed 2018 August 6.
34
Vaughn
,
TL
,
Bell
,
CS
,
Pickering
,
CK
,
Schwietzke
,
S
,
Heath
,
G
, Pé
tron
,
G
,
Zimmerle
,
DJ
,
Schnell
,
RC
and
Nummedal
,
D
.
2018
.
Temporal variability largely explains top-down/bottom-up difference in methane emissions estimates from a natural gas production region
.
115
(
46
):
11712
11717
. DOI:
35
Vaughn
,
TL
,
Bell
,
CS
,
Yacovitch
,
TI
,
Roscioli
,
JR
,
Herndon
,
SC
,
Conley
,
S
,
Schwietzke
,
S
,
Health
,
GA
,
Pétron
,
G
and
Zimmerle
,
DJ
.
2017
.
Comparing facility-level methane emission rate estimates at natural gas gathering and boosting stations
.
Elem Sci Anth
,
5
:
71
. DOI:
36
Visschedijk
,
AJH
,
Denier van der Gon
,
HAC
,
Doornenbal
,
HC
and
Cremonese
,
L
.
2018
.
Methane and ethane emission scenarios for potential shale gas production in Europe
.
45
:
125
131
. DOI:
37
Yacovitch
,
TI
,
Daube
,
C
,
Vaughn
,
TL
,
Bell
,
CS
,
Roscioli
,
JR
,
Knighton
,
WB
,
Nelson
,
DD
,
Zimmerle
,
DJ
,
Pétron
,
G
and
Herndon
,
SC
.
2017
.
Natural gas facility methane emissions: Measurements by tracer flux ratio in two US natural gas producing basins
.
Elem Sci Anth
5
:
69
. DOI:
38
Yacovitch
,
TI
,
Herndon
,
SC
,
Pétron
,
G
,
Kofler
,
J
,
Lyon
,
D
,
Zahniser
,
MS
and
Kolb
,
CE
.
2015
.
Mobile laboratory observations of methane emissions in the Barnett Shale region
.
Environ Sci Technol
49
(
13
):
7889
7895
. DOI:
39
Yacovitch
,
TI
,
Herndon
,
SC
,
Roscioli
,
JR
,
Floerchinger
,
C
,
McGovern
,
RM
,
Agnese
,
M
,
Pétron
,
G
,
Kofler
,
J
,
Sweeney
,
C
,
Karion
,
A
,
Conley
,
S
,
Kort
,
EA
,
Nähle
,
L
,
Fischer
,
M
,
Hildebrandt
,
L
,
Koeth
,
J
,
McManus
,
JB
,
Nelson
,
DD
,
Zahniser
,
MS
and
Kolb
,
CE
.
2014
.
Demonstration of an ethane spectrometer for methane source identification
.
Environ Sci Technol
48
(
14
):
8028
8034
. DOI:
40
Zavala-Araiza
,
D
,
Lyon
,
DR
,
Alvarez
,
RA
,
Davis
,
KJ
,
Harriss
,
R
,
Herndon
,
SC
,
Karion
,
A
,
Kort
,
EA
,
Lamb
,
BK
,
Lan
,
X
,
Marchese
,
AJ
,
Pacala
,
SW
,
Robinson
,
AL
,
Shepson
,
PB
,
Sweeney
,
C
,
Talbot
,
R
,
Townsend-Small
,
A
,
Yacovitch
,
TI
,
Zimmerle
,
DJ
and
Hamburg
,
SP
.
2015
.
Reconciling divergent estimates of oil and gas methane emissions
.
112
(
51
):
15597
15602
. DOI:
41
Zimmerle
,
DJ
,
Pickering
,
CK
,
Bell
,
CS
,
Heath
,
GA
,
Nummedal
,
D
,
Pétron
,
P
and
Vaughn
,
TL
.
2017
.
Gathering pipeline methane emissions in Fayetteville shale pipelines and scoping guidelines for future pipeline measurement campaigns
.
Elem Sci Anth
5
:
70
. DOI:
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC-BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. See http://creativecommons.org/licenses/by/4.0/.