Long-term decline of global atmospheric ethane concentrations and implications for methane

The longest continuous record of global atmospheric ethane levels is presented, showing that global ethane emission rates decreased by 21 per cent from 1984 to 2010, probably owing to decreased venting and flaring of natural gas in oil fields; decreased venting and flaring also account for at least 30 to 70 per cent of the decrease in methane emissions over the same period. Ethane is the most abundant non-methane hydrocarbon in the remote atmosphere and is a precursor to tropospheric ozone. This paper presents the longest continuous record of global atmospheric ethane levels assembled so far and finds that global ethane-emission rates decreased by 21% between 1984 and 2010. This can probably be attributed to a decrease in fugitive emissions, such as the venting and flaring of natural gas from oil fields, rather than a decline in its other major sources, biofuel use and biomass burning. Because methane shares ethane's main sources of emissions, this new long-term ethane record can be used to investigate changes in global methane levels. This leads the authors to suggest that reduced fugitive fossil-fuel emissions also account for 30–70% of the decrease in global methane emissions. After methane, ethane is the most abundant hydrocarbon in the remote atmosphere. It is a precursor to tropospheric ozone and it influences the atmosphere’s oxidative capacity through its reaction with the hydroxyl radical, ethane’s primary atmospheric sink1,2,3. Here we present the longest continuous record of global atmospheric ethane levels. We show that global ethane emission rates decreased from 14.3 to 11.3 teragrams per year, or by 21 per cent, from 1984 to 2010. We attribute this to decreasing fugitive emissions from ethane’s fossil fuel source—most probably decreased venting and flaring of natural gas in oil fields—rather than a decline in its other major sources, biofuel use and biomass burning. Ethane’s major emission sources are shared with methane, and recent studies have disagreed on whether reduced fossil fuel or microbial emissions have caused methane’s atmospheric growth rate to slow4,5. Our findings suggest that reduced fugitive fossil fuel emissions account for at least 10–21 teragrams per year (30–70 per cent) of the decrease in methane’s global emissions, significantly contributing to methane’s slowing atmospheric growth rate since the mid-1980s.

After methane, ethane is the most abundant hydrocarbon in the remote atmosphere. It is a precursor to tropospheric ozone and it influences the atmosphere's oxidative capacity through its reaction with the hydroxyl radical, ethane's primary atmospheric sink [1][2][3] .
Here we present the longest continuous record of global atmospheric ethane levels. We show that global ethane emission rates decreased from 14.3 to 11.3 teragrams per year, or by 21 per cent, from 1984 to 2010. We attribute this to decreasing fugitive emissions from ethane's fossil fuel source-most probably decreased venting and flaring of natural gas in oil fields-rather than a decline in its other major sources, biofuel use and biomass burning. Ethane's major emission sources are shared with methane, and recent studies have disagreed on whether reduced fossil fuel or microbial emissions have caused methane's atmospheric growth rate to slow 4,5 . Our findings suggest that reduced fugitive fossil fuel emissions account for at least 10-21 teragrams per year (30-70 per cent) of the decrease in methane's global emissions, significantly contributing to methane's slowing atmospheric growth rate since the mid-1980s.
The estimated emissions budget for ethane (C 2 H 6 ) is approximately 13 teragrams (1 Tg 510 12 g) per year. Its primary sources are fossil fuels (mainly evaporative emissions from their production, transmission and processing; 8.0-9.2 Tg yr 21 ), biomass burning (2.4-2.8 Tg yr 21 ) and biofuel use (2.6 Tg yr 21 ), with minor oceanic and biogenic sources and a possible geological source 3,6-8 . Ethane has a strong interhemispheric gradient because its major sources are primarily located in the Northern Hemisphere and its atmospheric lifetime (about two months) is shorter than the interhemispheric mixing time (about a year) (refs 1, 2). Ethane emissions in the Northern Hemisphere have been estimated to be about 12 Tg yr 21 , of which 1.7-2.0 Tg yr 21 are transported to the Southern Hemisphere via interhemispheric transport 1,7 . An additional 1.0 Tg yr 21 is released within the Southern Hemisphere, primarily from biomass burning 7 .
Whereas early measurements of atmospheric ethane in Switzerland  recorded an average increase of 0.8 6 0.3% yr 21 , subsequent field studies  have reported either steady or declining ethane levels in both hemispheres (Supplementary Table 1). The present work covers latitudes from 71u N (Barrow, Alaska) to 47u S (Slope Point, New Zealand) and represents the world's longest-running global atmospheric ethane monitoring program. The measurements began in 1984 as part of our established network to monitor global concentrations of trace gases including methane, chlorofluorocarbons and other light hydrocarbons 2,9,10 . Each season between 60-80 air samples were collected at 40-45 remote surface locations in the Pacific Basin (Fig. 1). These samples were analysed at our UCI laboratory using gas chromatography and were used to calculate a globally averaged ethane mixing ratio (see Methods Summary). Based on an intercomparison with ethane data from 2006-2010 from the NOAA/INSTAAR global non-methane hydrocarbon monitoring programme, which is not limited to the Pacific basin, we estimate that our data realistically approximate the average global tropospheric ethane concentration to within 15-20%, which is within the combined uncertainty margin of both records ( Supplementary Figs 1-2).
Consistent with its source distribution, ethane mixing ratios are greatest at latitudes north of 30u N, then drop precipitously from 30u N to the Intertropical Convergence Zone (Fig. 2a). Ethane is fairly well-mixed south of the Intertropical Convergence Zone, with typical variability of less than 30% among the latitudinal bands shown in  In addition to pronounced latitudinal variations, atmospheric ethane concentrations vary both seasonally and interannually. Ethane shows a late winter maximum and summer minimum in each hemisphere, with a crossover point near the Equator (Fig. 2a). Ethane's fossil fuel source is not believed to have a strong seasonal cycle 3 , so the clear, anticorrelated seasonal signals in each hemisphere are principally attributed to enhanced summertime photochemical sinks. From 1984 to 2010 the average (61s) amplitude of ethane's seasonal signal was much larger in the high Northern Hemisphere (8006 59 p.p.t.v.) than in the mid-Southern Hemisphere (35 6 11 p.p.t.v.), which is again consistent with ethane's source distribution. The northern seasonal signal also shows significant interannual variability, with several large positive short-term anomalies of up to 30% that typically occur every 3-5 years (Fig. 2a) and have been linked to fluctuations in biomass burning emissions 10 .

0°-
Atmospheric ethane mixing ratios have declined significantly since 1984, with the strongest decrease at high northern latitudes (Fig. 2b). From a linear fit to running annual averages from 1984 to 2010, we conclude that ethane declined by an average of -12.4 6 1.  21 . The good agreement between our top-down emission rates and the recent bottom-up estimates suggests that the ethane budget is well balanced. Therefore, major additional sources are unlikely, making it improbable that ethane has a large geological source. However, confirmation of this would require detailed analysis of the ethane budget, which is beyond the scope of this work.
To investigate the dominant processes that control the abundance of ethane in each hemisphere, simulations were run using the TM5 atmospheric tracer transport model (see Methods Summary) together with ethane data collected from 2000 to 2010 at selected sites, namely Barrow (71u N), Majuro (7u N) and Norfolk Island (29u S). The model reproduced the observational data well at Barrow and Norfolk Island (Fig. 3). Poorer agreement was found in Majuro, probably because it lies near the transitional zone between the two hemispheres ( Supplementary Fig. 3). Anthropogenic sources (that is, fossil fuel sources) were the main driver of the ethane signal in Barrow, with a smaller biomass burning component that peaks during summer (Fig. 3a). The observations and model both showed maximum concentrations in air arriving at Norfolk Island during September and October (Fig. 3b), consistent with a predominant biomass burning signal in the Southern Hemisphere 11 .
The model simulations discussed above inform our investigation of the causes of ethane's decline over the past 25 years. It is unlikely that changes in global levels of the hydroxyl radical (OH) were responsible for ethane's long-term global decline, because global OH levels have shown little interannual variability since the late 1990s and are generally well buffered against changes on interannual timescales 12 . Instead, because anthropogenic emissions dominate the ethane budget in the high Northern Hemisphere, where the strongest ethane decline has occurred, we attribute ethane's decline primarily to reduced emissions from the fossil fuel sector, specifically fugitive emissions of

LETTER RESEARCH
natural gas, which is ethane's main fossil fuel source. Fugitive emissions include venting and flaring, evaporative losses, and equipment leaks and failures, but exclude the combustion of fuels. Natural gas associated with oil fields, also known as associated gas, is commonly vented or flared when it cannot be used directly or transported away 13,14 . Although no reliable global data are available, most natural gas venting occurs at upstream production facilities 14 . Flaring is estimated to have peaked in the 1970s, after which high energy prices encouraged the use of more natural gas as a fuel 15 , and global gas flaring has remained fairly constant since the mid-1990s (ref. 13).
Although increased use of catalytic converters on internal combustion engines has effectively reduced the emissions of many hydrocarbons, it does not appear to have similarly affected ethane levels (see Supplementary Information). Quantifying and understanding ethane's long-term global decline provides valuable constraints on how we understand the long-term slowdown in the global growth rate of atmospheric methane 10,16 , an important greenhouse gas and tropospheric ozone (O 3 ) precursor. Following strong growth of 1.0-1.2% yr 21 in the early and mid-1980s, the methane growth rate slowed to 0.40 6 0.31% yr 21 in the 1990s and to 0.11 6 0.15% yr 21 in the 2000s, with strong interannual variability 10 . Because methane has a relatively long atmospheric lifetime (about nine years), its atmospheric concentrations respond more slowly to changes in emissions than is the case for ethane. Consequently, methane's atmospheric growth rate is a more sensitive indicator of fluctuations in methane's emissions. Although a change in methane emission rate is strongly connected to a change in growth rate, we note that the two are not equivalent.
A graph of atmospheric ethane mixing ratio versus methane growth rate reveals a remarkably strong correlation between the two gases over the past 25 years (Fig. 4) and indicates that there has been a long-term change in a source common to both compounds. The above discussion suggests that this common source is fugitive emissions from the oil and natural gas industries. Because methane and ethane are emitted from fossil fuel sources with characteristic emission ratios 7,17,18 , our longterm ethane record can be used to quantitatively investigate methane's slowing growth rate. We note that this analysis does not exclude the possibility that changes in other sources have also contributed to methane's long-term growth rate decline.
Using  Even though the TM5 model is capable of much higher spatial resolution (that is, 1u 3 1u), for this work the model used nested grids at a lower resolution (6u 3 4u) that is more appropriate to our observational network, which represents well-mixed background air. Ethane simulations (summed sources) are shown as a black line, ethane observations as red circles, fossil fuel simulations as a blue line, biomass burning simulations as a red line, biofuel simulations as a green line, and background simulations as a pink line.

RESEARCH LETTER
Our global data suggests that this was not the case. Although the decline in ethane emissions has been greatest at high northern latitudes (1. were based on emission ratios measured near oil storage and processing facilities, and they lie at the lower end of MER values representing the fossil fuel industry as a whole 17,18,25 . For example, the MER of natural gas varies considerably depending on the type of deposit, with mass-based values of the order of 3 for gas associated with oil fields, 6 for gas condensate (or 'wet' gas), and 10 or more for 'dry' gas fields, although there is substantial overlap among the categories 17,18 . Further, the MER for coalbed gas can exceed 5,000 in certain basins 25 . Because the MER of associated gas (that is, approximately 3) is on the lower end of the mass-based MER scale, our above estimate of 10-21 Tg yr 21 should be a fairly accurate assessment of the component of methane's decline associated with fossil fuel use. However, if more methaneenriched fossil fuel sources also contributed to the decline in methane emissions, then 10-21 Tg yr 21 may be an underestimate.
Our estimates of global ethane concentrations and emissions indicate a significant decrease in fugitive fossil fuel methane emissions of at least 10-21 Tg yr 21 , or more than 30%-70% of the total decline in global methane emissions from 1984-2010 (approximately 30 Tg yr 21 ; ref. 5). That is, our results are incompatible with a roughly constant fossil fuel source of methane since the 1980s, as put forward in a recent scenario 20 . Further, we have demonstrated the utility of long-term co-measurement of ethane and methane, and we encourage close scrutiny of ethane levels and corresponding methane growth rates in the future. For example, signals such as a significant upturn in methane growth, without a corresponding increase in global ethane levels, may indicate releases of methane from methane-rich sources such as wetlands or melting permafrost.

METHODS SUMMARY
Air sampling and analysis. Ethane measurements were made using our established technique of whole-air sampling followed by analysis using gas chromatography with flame ionization detection 2,9,10 . The samples were analysed at our UCI laboratory within one month of collection. Rigorous tests have shown that light alkanes are stable in the canisters over this period. The measurements for each trace gas use an internally consistent, internationally recognized calibration scale and detailed quality control procedures. The precision of the ethane measurements is 1%, the accuracy 5%, and the detection limit 3 p.p.t.v. Ethane mixing ratios that appear contaminated by local sources are removed from the data set (typically 2-6 per season), and the remaining data are used to construct a seasonal global ethane average using an equal surface area weighting method (see Supplementary Information). Running annual global averages are in turn calculated from these seasonal means. The measurements are archived at http://cdiac. ornl.gov/tracegases.html. Atmospheric modelling. Ethane simulations were run using the TM5 model of atmospheric tracer transport 26 . Assimilated meteorology data from the European Centre for Medium Range Weather Forecasting were used to calculate the atmospheric transport. A repeating seasonal cycle of OH (ref. 27) and the Jet Propulsion Laboratory kinetics data evaluation for the photochemical destruction of ethane (ref. 28) were used to calculate the reaction rate between ethane and OH. Anthropogenic ethane sources were estimated using the Emission Database for Global Atmospheric Research data set for methane, release version 4.0 (http:// edgar.jrc.ec.europa.eu, 2009) together with the appropriate methane-to-ethane emission ratios 7 . Ethane emissions from biomass burning were estimated from the Global Fire Emissions Database version 2 and published emission ratios 29 . Biofuel emissions were retrieved from ref. 30. The relatively small ethane emissions from biogenic sources and the ocean were not included because of concern about their uncertainty.