NaturalGasPipelineLeaksandEmissions

AndyGreensponHEJC

March18,2015

ReferencesAlldataintheseslidescomesfromthefollowingsources: AmericaPaysforGasLeaks:NaturalGasPipelineLeaksCostConsumersBillions-AreportpreparedforSen.

EdwardJ.Markey. EnergyInformaOonAdministraOonwebsite.hQps://www.eia.gov/ CausesofClimateChange.hQp://www3.epa.gov/climatechange/science/causes.html NaturalgasproducOonchart:hQp://www.eia.gov/dnav/ng/hist/n9050us2a.htm NaturalgasproducOonanddistribuOonmap:hQp://www.eia.gov/pub/oil_gas/natural_gas/analysis_publicaOons/

ngpipeline/process.html hQp://www.iea.org/newsroomandevents/news/2015/march/global-energy-related-emissions-of-carbon-dioxide-

stalled-in-2014.html Nisbetetal.(2014),MethaneontheRiseAgain.Science,Vol343,31JANUARY2014. McKainetal.(2015),MethaneemissionsfromnaturalgasinfrastructureanduseintheurbanregionofBoston,

MassachuseQs.PNAS,vol.112,no.7,19411946,February17,2015. Brandtetal.(2014),MethaneLeaksfromNorthAmericanNaturalGasSystems.ScienceVol343,14February

2014. ConservaOonLawFoundaOon.hQp://clf.org/map/ GovernorPatrickSignsLegislaOontoAddressGasLeaks.Patch.hQp://patch.com/massachuseQs/salem/

governor-patrick-signs-legislaOon-to-address-gas-leaks MassachuseQsGovernorSignsGasLeaksBill;MoreWorkNeededtoAddressCost,ClimateImpact.CleanWater

Ac2on.hQp://www.cleanwateracOon.org/press/massachuseQs-governor-signs-gas-leaks-bill-more-work-needed-address-cost-climate-impact-0

BillH.2950,188th(2013-2014).AnActrelaOvetonaturalgasleaks.The189thGeneralCourtoftheCommonwealthofMassachuse?s.hQps://malegislature.gov/Bills/188/House/H2950

TennesseeGasPipelineCo.filesenvironmentalreportwithFederalEnergyRegulatoryCommission.MassLive.hQp://www.masslive.com/news/index.ssf/2015/03/tennessee_gas_pipeline_co_file.html

InfaceofopposiOon,companytoreroutegaspipeline.BostonGlobe.hQps://www.bostonglobe.com/business/2014/12/05/face-opposiOon-company-reroute-pipeline/wj0k4WbfYr5FFyyHtPmFGJ/story.html

NaturalGasvs.CO2 Naturalgas:primarilymethaneCH4. Globalaverageatmosphericlevels:

CO2~400ppm=400,000ppb. CH4~1,800ppb=0.45%ofCO2levels.

Lifespans: CO2:centuries CH4:decades(halflifeof7yearsinatmosphere)NaturalgasisamuchmorepotentgreenhousegasthanCO2

~20-25Omesmoreoverthelongterm(100years). ~72Omesmoreovera20yearhorizon.

NaturalgasproducOonandatmosphericmethanelevels.

MethanelevelshavesteadilyrisensincethestartoftheindustrialrevoluOonin1750.

Leveledoffintheearly2000s. Startedrisingagaininthelate2000s(duetothenaturalgasboom??)

Nisbetetal.(2014),MethaneontheRiseAgain.Science,Vol343,31JANUARY2014.

CausesofClimateChange.hQp://www3.epa.gov/climatechange/science/causes.html

U.S.naturalgasproducOonandpipelines

NaturalgasproducOonchart:hQp://www.eia.gov/dnav/ng/hist/n9050us2a.htm

Pipelinemap:hQp://www.nbcnews.com/id/39174246/ns/us_news/t/most-us-gas-lines-not-inspected-latest-technology/#.VQXfgxDF8T8

NaturalgasproducOonanddistribuOon

compressorstaOons

Pipelineinfrastructureisvastanddistributedcompaniescontrolmanydifferentgeographicregionsanddifferenttypesofpipes(material,age,size,milesofpiping).

NaturalgasproducOonanddistribuOonmap:hQp://www.eia.gov/pub/oil_gas/natural_gas/analysis_publicaOons/ngpipeline/process.html

Unaccountedfornaturalgas

GasdistribuOoncompaniesin2011reportedreleasing69billioncubicfeetofnaturalgastotheatmosphere. AlmostenoughtomeetthestateofMainesgasneedsforayear. Equivalentto~33.3millionmetrictonsofCO2.

EquivalenttoCO2emissionsof~6-7millionautomobiles. Reference:CO2emissionsin2014:32.3billionmetrictons

Naturalgasreleasedcontributesequivalentofonly~0.1%oftotalCO2emissions.

Naturalgasunaccountedforin2000-2011: U.S.:2.6trillioncubicfeet MassachuseQs:99-227billioncubicfeetofnaturalgas NaturalgasdistribuOonsystems(mainpipelinesandsmallerdistribuOon

networksandmains):19%oftotalCH4emissionsfromnaturalgassystems.

ConversionfactorassumpOons: 1billioncubicfeetCH4=19,300metrictonsCH4 1metrictonCH4=25metrictonsCO2equivalent 1billioncubicfeetCH4=482,500metrictonsCO2equivalent

AmericaPaysforGasLeaks:NaturalGasPipelineLeaksCostConsumersBillions-AreportpreparedforSen.EdwardJ.Markey.hQp://www.iea.org/newsroomandevents/news/2015/march/global-energy-related-emissions-of-carbon-dioxide-stalled-in-2014.html

Atmosphericnaturalgasmeasurementsvs.claimedemissionsfactors

www.sciencemag.org SCIENCE VOL 343 14 FEBRUARY 2014 733

POLICYFORUM

Natural gas (NG) is a potential bridge fuel during transition to a decarbon-ized energy system: It emits less car-

bon dioxide during combustion than other fos-sil fuels and can be used in many industries. However, because of the high global warming potential of methane (CH4, the major compo-nent of NG), climate benefi ts from NG use depend on system leakage rates. Some recent estimates of leakage have challenged the ben-efi ts of switching from coal to NG, a large near-term greenhouse gas (GHG) reduction opportunity ( 1 3). Also, global atmospheric CH4 concentrations are on the rise, with the causes still poorly understood ( 4).

To improve understanding of leakage rates for policy-makers, investors, and other decision-makers, we review 20 years of tech-nical literature on NG emissions in the United States and Canada [see supplementary mate-rials (SM) for details]. We fi nd (i) measure-ments at all scales show that offi cial inven-tories consistently underestimate actual CH4 emissions, with the NG and oil sectors as important contributors; (ii) many indepen-dent experiments suggest that a small number of superemitters could be responsible for a large fraction of leakage; (iii) recent regional atmospheric studies with very high emissions rates are unlikely to be representative of typi-cal NG system leakage rates; and (iv) assess-ments using 100-year impact indicators show system-wide leakage is unlikely to be large enough to negate climate benefi ts of coal-to-NG substitution.

UnderestimationDevice to ContinentThis study presents a fi rst effort to system-atically compare published CH4 emissions estimates at scales ranging from device-level (>103 g/year) to continental-scale atmospheric studies (>1013 g/year). Studies known to us that (i) report measurement-based emissions estimates and (ii) compare those estimates with inventories or estab-lished emission factors (EFs) are shown in the fi rst chart.

Studies that measure emissions directly from devices or facilities (bottom-up stud-ies) typically compare results to emissions factors (EFs; e.g., emissions per device). Large-scale inventories are created by multi-plying EFs by activity factors (e.g., number of devices).

Studies that estimate emissions after atmospheric mixing occurs (atmospheric studies) typically compare measurements to emissions inventories, such as the U.S. Envi-

Methane Leaks from North American Natural Gas Systems

ENERGY AND ENVIRONMENT

A. R. Brandt, 1 * G. A. Heath, 2 E. A. Kort, 3 F. OSullivan, 4 G. Ptron, 5, 6 S. M. Jordaan, 7 P. Tans, 5 J. Wilcox, 1 A. M. Gopstein, 8 D. Arent, 2, 9 S. Wofsy, 10 N. J. Brown, 11 R. Bradley, 12 G. D. Stucky, 13 D. Eardley, 13 R. Harriss 14

Methane emissions from U.S. and Canadian

natural gas systems appear larger than offi cial

estimates.

0.01 0.1 1.0 10 100 1000

Em

issi

ons

mag

nitu

de (

g C

H4/

year

)

Ratio: measured/inventory or measured/EF [unitless]

103

106

109

1012

1015

(26): Production + HF(18): Compressor(15): Gas plants(21): Gas plants, other(16): Distribution

(6): US(5): US + Can(8): US energy(13): US energy

(12): SoCAB I(1): UT(12): SoCAB II(7): all sources

(6): SC-US(9): SC-US(6): SC-US, NG + Petrol.

(2): SoCAB(2): SoCAB, NG + Pet.(10): LA county(3): DJ basin(25): DJ basin

(14): Gas plants

Scale of measurement

National or continentalMulti-stateRegional or air basinFacilityDevice or component

AttributionAttributed to oil and gas ormeasured at facillityAttributed to energy indust.or not attributed

1010

1011

1012

1013

1014Ratios as publishedRatios with common baseline (EPA GHGI)

0 1 2 3 4 5 6 7 8

Inventories and emissions factors consistently underestimate actual measured CH4 emissions across scales. Ratios >1 indicate measured emissions are larger than expected from EFs or inventory. Main graph compares results to the EF or inventory estimate chosen by each study author. Inset compares results to regionally scaled common denominator ( 17), scaled to region of study and (in some cases) the sector under examination. Multiple points for each study correspond to different device classes or different cases mea-sured in a single study. De nitions of error bar bounds vary between studies. (US, United States; Can, Canada; SC, South Central; Petrol. and Pet., petroleum; SoCAB, South Coast Air Basin; LA, Los Angeles; DJ, Denver-Julesberg; UT, Utah; HF, hydraulic fracturing). See SM for gure construction details.

1Stanford University, Stanford, CA. 2National Renewable Energy Laboratory, Golden, CO. 3University of Michigan, Ann Arbor, MI. 4Massachusetts Institute of Technology, Cam-bridge, MA. 5National Oceanic and Atmospheric Adminis-tration, Boulder, CO. 6University of Colorado, Boulder, CO. 7University of Calgary, Calgary, Canada. 8U.S. Department of State, Washington, DC. 9Joint Institute for Strategic Energy Analysis, Golden, CO. 10Harvard University, Cambridge, MA. 11Lawrence Berkeley National Laboratory, Berkeley, CA. 12Independent consultant, Gaithersburg, MD. 13University of California, Santa Barbara, Santa Barbara, CA. 14Environmen-tal Defense Fund, Boulder, CO.

*Full af liations for all authors are included in the supple-mentary materials. Views expressed in this article are those of the author and do not necessarily re ect those of the U.S. Department of State or the U.S. Government. Correspond-ing author. abrandt@stanford.edu

Published by AAAS

on

Mar

ch 1

5, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

5, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Mar

ch 1

5, 2

015

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

Emissionsfactor(EF):esOmatedemissionsperdevice.

Inventory:EFxnumberofdevices RaOo>1indicatesemissionsare

largerthanexpectedfromemissionsfactororinventoryclaimed.

Summaryofstudies: EmissionsareoverallunderesOmated. Stateandregionalstudiespredictlarger

underesOmaOonthannaOonalstudies. NaOonalstudies,whichaverageoutliers

beQer,suggest1.25-1.75OmestheemissionsthanexpectedfromthegreenhousegasinventoryoftheEPA.Challenge:aQribuOngCH4emissionstomulOple

potenOalsources(anthropogenicandnatural).

Brandtetal.(2014),MethaneLeaksfromNorthAmericanNaturalGasSystems.ScienceVol343,14February2014.

ReportedgasleaksinMassachuseQs

Grade1hazardousGrade2potenOallyhazardousGrade3non-hazardous

Companiesowenignoregrade3leaks.ButinaggregatetheycanmakesignificantcontribuOonstoCH4emissions.

ConservaOonLawFoundaOon.hQp://clf.org/map/

QuanOtaOvestudyofgasleaksintheBostonarea

Methane Concentrations in the Boston AtmosphereAtmospheric CH4 concentrations were measured continuouslyfrom September 2012 through August 2013 at two locations nearthe urban center [Boston University (BU) and Copley Square(COP)] and two locations outside of Boston [Harvard Forest(HF) and Nahant (NHT)] (Fig. 1 and SI Appendix, Table S2 andsection S1). Background concentrations in air flowing into thecity were estimated by randomly sampling from a range (5th to35th) of lower percentiles of CH4 measurements from two up-wind stations (HF or NHT, depending on the direction of sim-ulated air trajectories; SI Appendix, section S3.1), averaged overa 48-h moving window, to capture synoptic-scale variability andremove possible influences of small nearby sources (SI Appendix,section S3.3). Values of CH4 were calculated by subtractingbackground from urban concentrations. Hourly average CH4data were aggregated into daily afternoon (1116 h EST, 1621 hUTC) means to remove autocovariance and focus the analysis onperiods of well-mixed atmospheric conditions.

Methane concentrations in Boston were consistently elevatedover background (Fig. 2 and SI Appendix, Figs. S1 and S2) andfollowed a distinct daily pattern (Fig. 3 A and C, and SI Ap-pendix, Fig. S16), associated with growth and decay of theplanetary boundary layer. Concentrations fluctuated over shorttimescales (SI Appendix, Fig. S1) due to small-scale atmosphericcirculations and heterogeneous sources in the urban environ-ment. Methane concentrations were higher in winter than theother seasons at both sites, but CH4 varied less with season(Fig. 2). The average annual afternoon values of CH4 at BUand COP were 45.9 (37.3, 58.5) ppb and 30.5 (23.6, 39.3) ppb,respectively (Fig. 2), reflecting different sampling altitudes (30and 215 m, respectively; SI Appendix, Table S2). All errorsreported throughout the paper are 95% confidence intervals.Uncertainties in CH4 (Fig. 2) were calculated through a boot-strap analysis that included background concentrations and af-ternoon hourly, daily, and seasonally averaged CH4 measurements(SI Appendix, section S3.3).

Contribution of NG to Elevated CH4 ConcentrationsTo quantify the fraction of the observed CH4 that was due toNG emissions, we compared ratios of C2H6 and CH4 measuredin the atmosphere and NG pipelines serving the region. Ethaneis a significant component of NG, whereas microbial CH4 sources,such as landfills, sewage, and wetlands, produce little or no C2H6(15). Because Boston has no geologic CH4 seeps, no oil and gasproduction or refining, and low rates of biomass burning, thereare no known significant sources of C2H6 in the region otherthan NG.Ethane concentrations were measured with a laser spectrom-

eter (15) at BU for 3 mo in the fall and winter of 201213 and1 mo in the late spring of 2014 (SI Appendix, Fig. S6). Covariancesbetween atmospheric C2H6 and CH4 observations were de-termined from the daily slopes of a linear model that minimizes 2(16) of 5-min median afternoon data (Fig. 4 and SI Appendix,section S2.1). The median of the daily slopes of atmospheric C2H6versus CH4 was 2.6 (2.5, 2.8) % during the cool season and 1.6(1.4, 1.7) % during the warm season, obtained from days witha coefficient of determination (R2) > 0.75 (50% of the days).The average C2H6 and CH4 ratio in the NG flowing into the

region during the two atmospheric measurement periods was2.7 0.0% in the fall and winter of 20122013 and 2.4 0.1%in the spring of 2014, determined from hourly gas quality datafrom the three main pipelines that serve the region (17, 18) (SIAppendix, Figs. S7 and S8, and section S2.2). The quotient of theC2H6 and CH4 ratios in the atmosphere and pipeline demon-strates that NG contributed 98 (92, 105) and 67 (59, 72) % of theCH4 in Boston in the cool and warm seasons, respectively.This result is insensitive to assumptions about the relativecontribution of the three pipelines that supply the region and

72.5 71.5 70.5

41.5

42.0

42.5

43.0

43.5

Longitude

Latit

ude

0

10

100

500

4000

NHTHF

COPBU

0 50 km

Housing unitswith NG km! 2

Measurement sitesStudy BoundaryBU 50% influence contourCOP 50% influence contour

Fig. 1. Location of two city [Boston University (BU), 29-m height; CopleySquare (COP), 215-m height] and two peripheral [Harvard Forest (HF);Nahant (NHT)] measurement stations (black points) in Boston, and the sur-rounding area, overlaid on a map of the number of housing units with NGper square kilometer (14). The 90-km radius circle delineates the 18,000-km2 land area for which CH 4 emissions and the NG loss rate were calculated.The magenta and purple contours enclose 50% of the average footprint(sensitivity area) of the BU and COP afternoon measurements, respectively.The two city sites are difficult to distinguish at this scale because the hori-zontal distance between them is 2 km. The influence area is 80% largerfor COP than BU because the former station is higher. See SI Appendix, TableS2, for additional measurement site location information.

BU

CH

4 (p

pb)

1880

1920

1960

2000

SON DJF MAM JJA ANN

2040

6080

100

120

"C

H4

(ppb

)

COP

CH

4 (p

pb)

1880

1920

1960

2000

SON DJF MAM JJA ANN

2040

6080

100

120

"C

H4

(ppb

)

A B

Fig. 2. Average (95% confidence intervals) afternoon (11 16 h EST) CH4 (black; left y axis) and CH4 (blue; right y axis) by season and for the whole year at(A) BU and (B) COP.

1942 | www.pnas.org/cgi/doi/10.1073/pnas.1416261112 McKain et al.

AtmosphericCH4concentraOonsmeasuredconOnuouslyfromSep2012toAug2013atfourlocaOons:

Twourbancenters:BUandCopley TwolocaOonsoutsideBoston:Harvard

ForestandNahant- Randomsamplingover48hperiodsto

getbackgroundconcentraOons

ValuesofCH4calculatedbysubtracOngbackgroundfromurbanconcentraOons. HourlyaverageCH4dataaggregatedintodailyawernoonmeans(11-16hEST).

Methane Concentrations in the Boston AtmosphereAtmospheric CH4 concentrations were measured continuouslyfrom September 2012 through August 2013 at two locations nearthe urban center [Boston University (BU) and Copley Square(COP)] and two locations outside of Boston [Harvard Forest(HF) and Nahant (NHT)] (Fig. 1 and SI Appendix, Table S2 andsection S1). Background concentrations in air flowing into thecity were estimated by randomly sampling from a range (5th to35th) of lower percentiles of CH4 measurements from two up-wind stations (HF or NHT, depending on the direction of sim-ulated air trajectories; SI Appendix, section S3.1), averaged overa 48-h moving window, to capture synoptic-scale variability andremove possible influences of small nearby sources (SI Appendix,section S3.3). Values of ! CH4 were calculated by subtractingbackground from urban concentrations. Hourly average ! CH4data were aggregated into daily afternoon (1116 h EST, 1621 hUTC) means to remove autocovariance and focus the analysis onperiods of well-mixed atmospheric conditions.

Methane concentrations in Boston were consistently elevatedover background (Fig. 2 and SI Appendix, Figs. S1 and S2) andfollowed a distinct daily pattern (Fig. 3 A and C, and SI Ap-pendix, Fig. S16), associated with growth and decay of theplanetary boundary layer. Concentrations fluctuated over shorttimescales (SI Appendix, Fig. S1) due to small-scale atmosphericcirculations and heterogeneous sources in the urban environ-ment. Methane concentrations were higher in winter than theother seasons at both sites, but ! CH4 varied less with season(Fig. 2). The average annual afternoon values of ! CH4 at BUand COP were 45.9 (37.3, 58.5) ppb and 30.5 (23.6, 39.3) ppb,respectively (Fig. 2), reflecting different sampling altitudes (30and 215 m, respectively; SI Appendix, Table S2). All errorsreported throughout the paper are 95% confidence intervals.Uncertainties in ! CH4 (Fig. 2) were calculated through a boot-strap analysis that included background concentrations and af-ternoon hourly, daily, and seasonally averaged CH4 measurements(SI Appendix, section S3.3).

Contribution of NG to Elevated CH4 ConcentrationsTo quantify the fraction of the observed ! CH4 that was due toNG emissions, we compared ratios of C2H6 and CH4 measuredin the atmosphere and NG pipelines serving the region. Ethaneis a significant component of NG, whereas microbial CH4 sources,such as landfills, sewage, and wetlands, produce little or no C2H6(15). Because Boston has no geologic CH4 seeps, no oil and gasproduction or refining, and low rates of biomass burning, thereare no known significant sources of C2H6 in the region otherthan NG.Ethane concentrations were measured with a laser spectrom-

eter (15) at BU for 3 mo in the fall and winter of 201213 and1 mo in the late spring of 2014 (SI Appendix, Fig. S6). Covariancesbetween atmospheric C2H6 and CH4 observations were de-termined from the daily slopes of a linear model that minimizes " 2(16) of 5-min median afternoon data (Fig. 4 and SI Appendix,section S2.1). The median of the daily slopes of atmospheric C2H6versus CH4 was 2.6 (2.5, 2.8) % during the cool season and 1.6(1.4, 1.7) % during the warm season, obtained from days witha coefficient of determination (R2) > 0.75 (! 50% of the days).The average C2H6 and CH4 ratio in the NG flowing into the

region during the two atmospheric measurement periods was2.7 0.0% in the fall and winter of 20122013 and 2.4 0.1%in the spring of 2014, determined from hourly gas quality datafrom the three main pipelines that serve the region (17, 18) (SIAppendix, Figs. S7 and S8, and section S2.2). The quotient of theC2H6 and CH4 ratios in the atmosphere and pipeline demon-strates that NG contributed 98 (92, 105) and 67 (59, 72) % of the! CH4 in Boston in the cool and warm seasons, respectively.This result is insensitive to assumptions about the relativecontribution of the three pipelines that supply the region and

! 72.5 ! 71.5 ! 70.5

41.5

42.0

42.5

43.0

43.5

Longitude

Latit

ude

0

10

100

500

4000

NHTHF

COPBU

0 50 km

Housing unitswith NG km! 2

Measurement sitesStudy BoundaryBU 50% influence contourCOP 50% influence contour

Fig. 1. Location of two city [Boston University (BU), 29-m height; CopleySquare (COP), 215-m height] and two peripheral [Harvard Forest (HF);Nahant (NHT)] measurement stations (black points) in Boston, and the sur-rounding area, overlaid on a map of the number of housing units with NGper square kilometer (14). The 90-km radius circle delineates the ! 18,000-km2 land area for which CH 4 emissions and the NG loss rate were calculated.The magenta and purple contours enclose 50% of the average footprint(sensitivity area) of the BU and COP afternoon measurements, respectively.The two city sites are difficult to distinguish at this scale because the hori-zontal distance between them is ! 2 km. The influence area is ! 80% largerfor COP than BU because the former station is higher. See SI Appendix , TableS2, for additional measurement site location information.

BU

CH

4 (p

pb)

1880

1920

1960

2000

SON DJF MAM JJA ANN

2040

6080

100

120

"C

H4

(ppb

)

COPC

H4

(ppb

)

1880

1920

1960

2000

SON DJF MAM JJA ANN

2040

6080

100

120

"C

H4

(ppb

)

A B

Fig. 2. Average (95% confidence intervals) afternoon (11 16 h EST) CH4 (black; left y axis) and ! CH4 (blue; right y axis) by season and for the whole year at(A) BU and (B) COP.

1942 | www.pnas.org/cgi/doi/10.1073/pnas.1416261112 McKain et al.

McKainetal.(2015),MethaneemissionsfromnaturalgasinfrastructureanduseintheurbanregionofBoston,MassachuseQs.PNAS,vol.112,no.7,19411946,February17,2015.hQp://www.bostonglobe.com/metro/2015/01/22/natural-gas-leaks-boston-area-are-far-more-extensive-than-thought/5BykQrnaGRr2XLtxpHqLIM/story.html

QuanOtaOvestudyofgasleaksintheBostonarea

errors for the average emissions, atmospheric NG fraction, andNG consumption terms (SI Appendix, section S3.2.1).

The modest seasonality of the inferred NG loss rate (Fig. 5C)is driven by the small seasonal variability in total NG consumption(Fig. 5B). Our analysis makes no assumptions about the relativecontribution to emissions of specific NG-consuming sectors oremission processes (SI Appendix, section S3.2.1), which couldindividually have very different loss rates than the aggregate

estimate generated by this study. Our finding that the regionalaverage NG emission rate was seasonally invariant may indicatethat it does not strongly depend on the seasonally varying com-ponents of the NG system, or could result from multiplecompensating processes.

Comparison with Atmospheric Studies and InventoriesTwo recent studies in Los Angeles covering2 mo provide theonly previous atmosphere-based (top-down) estimates of emis-sions from NG in an urban area, 12% (0.73% when accountingfor the error ranges) of total NG consumed in the basin (10, 11).However, attribution of CH4 emissions to pipeline gas in LosAngeles is complicated by the presence of current and abandonedoil and gas wells, refinery operations, and natural CH4 seeps, inaddition to NG consumption. Other studies have estimated totalCH4 emission fluxes from a number of urban areas around theworld (SI Appendix, Table S1), using atmospheric data-modelframeworks of varying sophistication, but have not quantitativelyattributed fluxes to NG. Our value for total CH4 emissions inBoston is at the low end of the overall range of fluxes reported forother urban areas (SI Appendix, Table S1), suggesting that totalCH4 emission rates in Boston are not anomalous.

The US greenhouse gas (GHG) inventory (23) attributes 3,302Gg of CH4 emissions to NG transmission, storage, and distri-bution in 2012, equal to0.7% of the NG delivered to con-sumers (24). The key input data for NG distribution systems inthe national inventory are emissions factors developed from in-dustry measurements (25) and activity data on miles of pipelineby material and counts of metering and regulating stations,customer meters, and pipeline maintenance events and mishaps(23). Emissions of NG in our study area are equal to8% of USemissions attributed to distribution, transport, and storage, and23% of national emissions from distribution alone, a notablyhigher fraction than the3% of US residential and commercial

1800 2000 2200 2400 2600

05

1015

2025

CH4 (ppb)

C2H

6 (p

pb)

fallwinterspringatmospherepipeline

Fig. 4. Five-minute median atmospheric C 2H6 and CH4 measurement pointsat BU in fall and winter of 2012 2013 (black) and spring of 2014 (blue), 2

optimization lines fit to each day (light lines), average fit lines for bothseasons from all days with R2 > 0.75 (bold solid lines), and lines with slopes ofpipeline C 2H6/CH4 (dashed lines).

CH

4 E

mis

sion

s (g

m 2

yr

1 )

05

1015

2025

300

510

1520

2530

SON DJF MAM JJA ANN

TotalNat Gas

Nat

. Gas

Con

sum

ptio

n (g

CH

4 m2

yr

1 )

020

040

060

080

0

SON DJF MAM JJA ANN

Elec PowerRes & ComOther

Nat

. Gas

Los

s R

ate

(%)

01

23

4

SON DJF MAM JJA ANN

A B

C

Fig. 5. Seasonal and annual average (95% confidence intervals) ( A) optimized CH 4 emissions in total and from NG, ( B) NG consumption by sector, and ( C)NG loss rates, derived from CH 4 concentration observations from the BU and COP sites together. ( B) Consumption categories are electric power, residentialand commercial, and other, which is comprised of industrial, vehicle fuel, and pipeline and distribution use ( SI Appendix, section 3.2.1).

1944 | www.pnas.org/cgi/doi/10.1073/pnas.1416261112 McKain et al.

errors for the average emissions, atmospheric NG fraction, andNG consumption terms (SI Appendix, section S3.2.1).The modest seasonality of the inferred NG loss rate (Fig. 5C)

is driven by the small seasonal variability in total NG consumption(Fig. 5B). Our analysis makes no assumptions about the relativecontribution to emissions of specific NG-consuming sectors oremission processes (SI Appendix, section S3.2.1), which couldindividually have very different loss rates than the aggregate

estimate generated by this study. Our finding that the regionalaverage NG emission rate was seasonally invariant may indicatethat it does not strongly depend on the seasonally varying com-ponents of the NG system, or could result from multiplecompensating processes.

Comparison with Atmospheric Studies and InventoriesTwo recent studies in Los Angeles covering 2 mo provide theonly previous atmosphere-based (top-down) estimates of emis-sions from NG in an urban area, 12% (0.73% when accountingfor the error ranges) of total NG consumed in the basin (10, 11).However, attribution of CH4 emissions to pipeline gas in LosAngeles is complicated by the presence of current and abandonedoil and gas wells, refinery operations, and natural CH4 seeps, inaddition to NG consumption. Other studies have estimated totalCH4 emission fluxes from a number of urban areas around theworld (SI Appendix, Table S1), using atmospheric data-modelframeworks of varying sophistication, but have not quantitativelyattributed fluxes to NG. Our value for total CH4 emissions inBoston is at the low end of the overall range of fluxes reported forother urban areas (SI Appendix, Table S1), suggesting that totalCH4 emission rates in Boston are not anomalous.The US greenhouse gas (GHG) inventory (23) attributes 3,302

Gg of CH4 emissions to NG transmission, storage, and distri-bution in 2012, equal to 0.7% of the NG delivered to con-sumers (24). The key input data for NG distribution systems inthe national inventory are emissions factors developed from in-dustry measurements (25) and activity data on miles of pipelineby material and counts of metering and regulating stations,customer meters, and pipeline maintenance events and mishaps(23). Emissions of NG in our study area are equal to 8% of USemissions attributed to distribution, transport, and storage, and23% of national emissions from distribution alone, a notablyhigher fraction than the 3% of US residential and commercial

1800 2000 2200 2400 2600

05

1015

2025

CH4 (ppb)

C2H

6 (p

pb)

fallwinterspringatmospherepipeline

Fig. 4. Five-minute median atmospheric C2H6 and CH4 measurement pointsat BU in fall and winter of 20122013 (black) and spring of 2014 (blue), 2

optimization lines fit to each day (light lines), average fit lines for bothseasons from all days with R2 > 0.75 (bold solid lines), and lines with slopes ofpipeline C2H6/CH4 (dashed lines).

CH

4 E

mis

sion

s (g

m! 2

yr!

1 )

05

1015

2025

300

510

1520

2530

SON DJF MAM JJA ANN

TotalNat Gas

Nat

. Gas

Con

sum

ptio

n (g

CH

4 m

!2 y

r!1 )

020

040

060

080

0

SON DJF MAM JJA ANN

Elec PowerRes & ComOther

Nat

. Gas

Los

s R

ate

(%)

01

23

4

SON DJF MAM JJA ANN

A B

C

Fig. 5. Seasonal and annual average (95% confidence intervals) (A) optimized CH4 emissions in total and from NG, (B) NG consumption by sector, and (C)NG loss rates, derived from CH4 concentration observations from the BU and COP sites together. (B) Consumption categories are electric power, residentialand commercial, and other, which is comprised of industrial, vehicle fuel, and pipeline and distribution use (SI Appendix, section 3.2.1).

1944 | www.pnas.org/cgi/doi/10.1073/pnas.1416261112 McKain et al.

Leakratecorrespondsto~300,000metrictonsofnaturalgasleakedoverthe2012-2013yearstudiedabout2.7%ofallnaturalgasdeliveredtotheregionofstudy.

- 7.5millionmetrictonsCO2equivalentorCO2emissionsfrom~1.5millionpassengervehicles.

- Gasvaluedat$90millionandcouldheat200,000homesinayear. StateandfederalauthoriOespreviousesOmate:1.1%ofnaturalgaswasbeing

losttoleaksfromarangeofsourcesinthearea,includinghomes,businesses,andelectricitygeneraOonfaciliOes.

Ifcorrect,BostonareawouldbecontribuOng9%ofU.S.methanefromnaturalgasimpliesnaOonalesOmateisalsolow.

RaOoofC2H6toCH4determinesproporOonofnaturalgascontribuOontoCH4emissions

McKainetal.(2015),MethaneemissionsfromnaturalgasinfrastructureanduseintheurbanregionofBoston,MassachuseQs.PNAS,vol.112,no.7,19411946,February17,2015.hQp://www.bostonglobe.com/metro/2015/01/22/natural-gas-leaks-boston-area-are-far-more-extensive-than-thought/5BykQrnaGRr2XLtxpHqLIM/story.html

Primarycauseofnaturalgasleaksoldinfrastructure

Castironandbaresteel: Leak18OmesmoregasthanplasOcpipes. Leak57Omesmoregasthanprotectedsteel.

In2012,MassachuseQshad: 5,482milesofleak-pronemains. 194,326milesofleak-proneservicelines.

2013MassachuseAsStateRankforPipelineMaterial

Rank Item

2 Mostmilesofcastironservicelines

3 Mostmilesofcastironmains

4 Mostmilesofbaresteelservicelines

9 Mostmilesofbaresteelmains

6 Mostmilesofpipelinefromcastironandbaresteel

AmericaPaysforGasLeaks:NaturalGasPipelineLeaksCostConsumersBillions-AreportpreparedforSen.EdwardJ.Markey.

LackofincenOvestorepairminorleaks

Inmanystates,gascompaniespassonthecostoflostgastocustomers. MassachuseAscustomerslost$640millionto$1.5billionfrom2000-2011duetoleakedgas.

Replacingoldpipesrequiressignificantupfrontcapital. 33states,includingMassachuseQs,haveinfrastructurereplacementprograms. ButsOllliQleincenOvetoacceleratepipelinereplacementsolongascompaniescansOllpass

costsontocustomersforlostgas. Onlytwostates,PennsylvaniaandTexas,haveestablishedlimitsontheamount

companiescanchargecustomersforlostgas. Texas:2010to2012gascompaniesreducedtheirinventoryofleak-proneservicelinesby55

percent(101,790lines). InthissameOmeperiod,gascompaniesinMassachuseQsreducedtheirleak-proneservice

linesbyjust4percent(8,278lines). Asof2013,onlyfivestatesrequiredallnon-hazardousleakstoberepairedwithin

acertainOmeframe.

!!

Middleborough Gas & Electric Main Miles 5 -1% -90% 795% 0% 0% 0% -17% -7%

11

Middleborough Gas & Electric Service Lines 86 -7% -3% -9% -5% -3% -3% -3% -10% 149

New England Gas Main Milesb 17 -1% -1% -1% -2% -3% -3% 8% -4% 185

New England Gas Service Linesb -5,637 -2% -5% -2% -11% -13% -19% 448% -5% 8,813

Nstar Gas Main Miles 145 -3% -2% -2% -2% -3% -2% -3% -3% 716

Nstar Gas Service Lines 9,303 -2% -4% -2% -3% -4% -4% -4% -5% 26,514

Wakefield Municipal Gas & Light Main Miles 11 -1% -3% -4% -1% -1% -9% -4% -3% 37 Wakefield Municipal Gas & Light Service Lines 529 0% -4% -4% -1% -4% -7% -8% -2% 1,463 Westfield Gas & Electric Main Miles 15 -4% -2% -2% -2% -5% -4% -1% -10% 42 Westfield Gas & Electric Service Lines 493 14% -7% -6% -8% -6% -4% -4% -5%

1,543

!"##"$%((#)*)!"+,)!+-# ) 1,293 -3% -3% 0% -2% -2% -3% -3% -4% 5,571 !"##"$%((#)*)./0+$)1+,#) 28,419 -2% -4% -2% -3% -4% -3% 0% -4% 194,326 2"(+3,"- *)!"+,)!+-# ) 20,944 -5% -3% -3% -2% -4% 3% -4% -3% 93,705 2"(+3,"- *)./0+$# ) 2,036,032 -10% 4% -2% -2% -4% -35% -4% -12% 2,568,279

Source: SWDIIDQDO\VLVRI3+06$VCast and Wrought Iron and Bare Steel Pipeline Inventory, available at: http://opsweb.phmsa.dot.gov/pipeline_replacement/cast_iron_inventory.asp and http://opsweb.phmsa.dot.gov/pipeline_replacement/bare_steel_inventory.asp respectively. Notes: According to PHMSA officials, changes in replacement rates are generally due to three factors: (1) pipeline replacement, (2) acquisition of or selling off part of a distribution pipeline, or (3) changes in pipeline classification due to updated information or recordkeeping. a Owned by National Grid. Essex Gas was merged into Boston Gas in 2010. b 3DUWLFLSDWLQJLQ0DVVDFKXVHWWV7DUJHWHG,QIUDVWUXFWXUH5HSODFHPHQW3URJUDP)LWFKEXUJ*DV(OHFWULF applied in 2011.

cColumbia Gas is a subsidiary of NiSource.

17

Table 6: Replacement rates for leak-prone pipeline in Massachusetts, by company, 2004-2012

Replacement rate

Company Name

Leak-prone Pipeline Replaced Since 2004 2005 2006 2007 2008 2009 2010 2011 2012

Leak-prone Pipeline Remaining in 2012

Berkshire Gas - Main Miles 23 -2% -73% 251% -2% -2% -2% -3% -4% 115

Berkshire Gas - Service Lines 1,088 -3% -3% -3% -2% -2% -4% -2% -5% 3,864 Blackstone Gas - Main Miles 2 0% 0% 0% 0% 0% 0% 100% 0% 0 Blackstone Gas - Service Lines 0 0% 0% 0% 0% 0% 0% 0% 0% 0

Boston Gas - Main Milesa,b 496 -1% -2% -1% -1% -2% -3% -2% -3% 2,997

Boston Gas - Service Linesa,b 6,609 -2% 13% -1% -2% -3% -5% -3% -3% 90,523

Colonial Gas - Main Milesa,b 189 -1% -2% -2% -3% -3% -5% -19% -17% 253

Colonial Gas - Service Linesa,b 1,078 26% 1% -6% -8% -6% -3% -10% -10% 4,466

Columbia Gas - Main Milesb,c 344 -5% -5% -4% -4% -3% -2% -4% -4% 979

Columbia Gas - Service Linesb,c 13,907 -3% -4% -3% -3% -3% -3% -3% -4% 46,622

Essex Gas - Main Milesa,b 23 -2% -4% -1% -1% -3% -6% 1% -3% 111

Essex Gas - Service Linesa,b 533 -1% 4% -2% -2% -3% -2% -3% -3% 4,433 Fitchburg Gas & Electric - Main Milesb 21 -83% 433% -3% -3% -3% -3% -3% -4%

66

Fitchburg Gas & Electric - Service Linesb -490 -6% -14% -8% -9% -8% 119% -6% -8%

3,379

City of Holyoke Main Miles 6 0% -2% 0% -3% -3% 0% -2% 0% 58

City of Holyoke Service Lines 1,127 -2% -4% -2% -2% -5% -3% -7% -10% 2,557

16

AmericaPaysforGasLeaks:NaturalGasPipelineLeaksCostConsumersBillions-AreportpreparedforSen.EdwardJ.Markey.

Methodstodetectandreducepipelineleaks

Includeallemissionssourcesininventoryforpossibleleaks,including: downstreamofcustomermeters industrialfaciliOes residenOalandcommercialsezngs.

Improvesamplingprotocolsanddevelopmorecomprehensiveleaksurveys. negaOveunaccountedforgasvolumesbycompaniesindicatecalculaOngorreporOngerrors

infrequenthighemissioneventsareunder-sampled. smallleaksrequiremoresensiOveequipmenttodetect

Replaceoldmainsandservicelinessoonerratherthanlater.

McKainetal.(2015),MethaneemissionsfromnaturalgasinfrastructureanduseintheurbanregionofBoston,MassachuseQs.PNAS,vol.112,no.7,19411946,February17,2015.

NewMAlawtopromoterepairofpipelineinfrastructure

PassedinJuly2014:AnactrelaOvetonaturalgasleaks Grade1(hazardous)leaksmustberepairedunOlhazardiseliminated.

Grade2(potenOallyhazardous)leaksrequiredtoberepairedwithinayear.

Grade3(non-hazardous)leaksmustbereevaluated. Gascompaniesaccountableforplanstoremoveleak-proneinfrastructure.

WhatssOllmissing: RatepayerssOllpaythecostoflostgas. Grade3leaksdontactuallyhavetoberepairedonanyOmetable.

NorequirementtoacOvelyreplaceoldcastironandbaresteelpipeswithoutleaks.

BillH.2950,188th(2013-2014).AnActrelaOvetonaturalgasleaks.The189thGeneralCourtoftheCommonwealthofMassachuse?s.hQps://malegislature.gov/Bills/188/House/H2950TennesseeGasPipelineCo.filesenvironmentalreportwithFederalEnergyRegulatoryCommission.MassLive.hQp://www.masslive.com/news/index.ssf/2015/03/tennessee_gas_pipeline_co_file.htmlInfaceofopposiOon,companytoreroutegaspipeline.BostonGlobe.hQps://www.bostonglobe.com/business/2014/12/05/face-opposiOon-company-reroute-pipeline/wj0k4WbfYr5FFyyHtPmFGJ/story.html

OpOmismforthefuture?

BasedonEPAassumpOons,MassachuseQsresidentsstandtorealize$156millioninnetbenefitsover10yearsfromthecompaniesparOcipaOnginMAinfrastructurereplacementprogram.

StatelawrequiresMassachuseQstoreduceGHGemissionsto25percentbelow1990levelsby2020.

By2010,MassachuseQshadalreadysucceededinreducingmethaneemissionsfromthenaturalgasdistribuOonsystemby14percentbelow1990levels.

AmericaPaysforGasLeaks:NaturalGasPipelineLeaksCostConsumersBillions-AreportpreparedforSen.EdwardJ.Markey.

NewpipelineproposalthroughMA/NH2018?

Capacitytotransportupto2.2billioncubicfeetofnaturalgasperdayfromwellsinPennsylvaniatomarketsintheNortheast.

Co-locaOngwithexisOngrightofwayuOlitycorridors. 65and90%ofaffectedlandownersinMAandNHrespecOvelyhavenotgrantedpermission

toentertheirlandforsurveyingpurposes.- PossibleeminentdomainauthoritytopursueaccesstothosedeniedproperOesif

pipelinewinsacerOficatefromfederalregulators.TennesseeGasPipelineCo.filesenvironmentalreportwithFederalEnergyRegulatoryCommission.MassLive.hQp://www.masslive.com/news/index.ssf/2015/03/tennessee_gas_pipeline_co_file.htmlInfaceofopposiOon,companytoreroutegaspipeline.BostonGlobe.hQps://www.bostonglobe.com/business/2014/12/05/face-opposiOon-company-reroute-pipeline/wj0k4WbfYr5FFyyHtPmFGJ/story.html

QuesOons?