Transcript
Anthropogenic and natural methane emissions from a shale gasexploration area of Quebec, Canada Daniele L. Pinti a, ⁎ , Yves Gelinas b , Anja M. Moritz b , Marie Larocque a , Yuji Sano c,d a GEOTOP and Département des sciences de la Terre et de l'atmosphère, Université du Québec à Montréal, CP 8888, Succ. Centre-Ville, Montréal, QC H3C 1P8, Canada b GEOTOP and Department of Chemistry and Biochemistry, Concordia University, 7141 Sherbrooke St. West, Montreal, QC H4B 1R6, Canada c Atmosphere and Ocean Research Institute, University of Tokyo, Kashiwa, Chiba 277-8564, Japan d Department of Geosciences, National Taiwan University, Roosevelt Road, Taipei 106, Taiwan H I G H L I G H T S • CH 4 baselines in groundwater used forre fi ning emissions to the atmosphere • Methane fl uxes from Quebec Uticashale prospection area evaluated • CH 4 emissions from human abstractionare 300% those of all United Kingdomaquifers • Groundwater discharge is the main CH 4 emission natural source • CH 4 emitted from groundwater could behigher than that from fracking, yearlyG R A P H I C A L A B S T R A C T a b s t r a c ta r t i c l e i n f o Article history: Received 1 April 2016Received in revised form 16 May 2016Accepted 26 May 2016Available online 3 June 2016Editor: D. Barcelo The increasing number of studies on the determination of natural methane in groundwater of shale gasprospection areas offers a unique opportunity for re fi ning the quanti fi cation of natural methane emissions.Here methane emissions, computed from four potential sources, are reported for an area of ca. 16,500 km 2 of the St. Lawrence Lowlands, Quebec (Canada), where Utica shales are targeted by the petroleum industry.Methane emissions can be caused by 1) groundwater degassing as a result of groundwater abstraction fordomestic and municipal uses; 2) groundwater discharge along rivers; 3) migration to the surface by (macro-and micro-) diffuse seepage; 4) degassing of hydraulic fracturing fl uids during fi rst phases of drilling. Methaneemissions relatedto groundwaterdischarge to rivers(2.47 × 10 − 4 to 9.35 × 10 − 3 Tgyr − 1 ) surpass those of dif-fuse seepage (4.13 × 10 − 6 to 7.14 × 10 − 5 Tg yr − 1 ) and groundwater abstraction (6.35 × 10 − 6 to2.49 × 10 − 4 Tg yr − 1 ). The methane emission from the degassing of fl owback waters during drilling of theUtica shaleovera 10- to20-yearhorizonisestimatedfrom2.55×10 − 3 to1.62×10 − 2 Tgyr − 1 .Theseemissionsare from one third to sixty-six times the methane emissions from groundwater discharge to rivers. This studyshowsthatdifferentmethaneemissionsourcesneedtobeconsideredinenvironmentalassessmentsofmethaneexploitation projects to better understand their impacts.© 2016 Elsevier B.V. All rights reserved. Keywords: Methane (CH 4 )Shale gasGroundwaterEmissionsQuebec (Canada)Science of the Total Environment 566 – 567 (2016) 1329 – 1338 ⁎ Corresponding author. E-mail address:
[email protected] (D.L. Pinti).http://dx.doi.org/10.1016/j.scitotenv.2016.05.1930048-9697/© 2016 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv 1. Introduction Methane is the third most abundant greenhouse gas (GHG) in theEarth's atmosphere (ca. 5000 Tg; 1 Tg = 10 12 g) after H 2 O and CO 2 ,and the second radiative forcing anthropogenic gas after CO 2 . In 2011the concentration of methane in the atmosphere was 1.29 mg m − 3 (or 1803 ppb) and exceeded the pre-industrial levels by about 150%(IPCC, 2013), surpassing the highest levels ever recorded in ice coresin the last 800 kyrs (Sapart et al., 2012).Fossil fuel emissions contribute largely to atmospheric methane.Measurementsofatmospheric 14 CH 4 byAcceleratorMassSpectrometry(AMS) have shown that 30 ± 2% of atmospheric methane is derivedfrom fossil carbon sources, either natural from terrestrial and marinegasseeps(Etiopeetal.,2008)oranthropogenicfromoilandcoalmining(Lasseyetal.,2007).Methaneemissionstotheatmospherearedestinedto rise owingtotheincrease in theexploitation of shale gasworldwide.Several studies have focused on the contribution of shale gas methaneemissions to GHG, but their conclusions vary. Howarth et al. (2011)and Caulton et al. (2014) suggested that over a 20-year period, shalegas emissions could contribute from 40 to 60 times more to the green-house footprint than those from coal. Hultman et al. (2011) comparedthe life-cycle emissions of conventional gas, shale gas and coal, andfound that the impact of shale gas on GHG emissions is 11% higherthan conventional gas but 44% lower than coal. Burnham et al. (2011)showed that shale gas life-cycle GHG emissions are 6% lower than con-ventionalnaturalgas,23%lowerthangasoline,and33%lowerthancoal.Shale gas exploitation by hydraulic fracturing is also a major sourceofpublicconcernduetopotentialwatercontaminationbymethaneandfracturing fl uids, through leaky well casings, spillage, or other contami-nation pathways (e.g., Allen et al., 2013). These human threats con-vinced several governmental agencies to fund studies aiming to obtainreliable data on source, concentration and migration mechanismsof methane in groundwater, prior to any shale gas exploitation(Hamilton, 2011; Boltom and Pham, 2013; Drage and Kennedy, 2014;Moritz et al., 2015; McIntosh et al., 2014; Molofsky et al., 2013;McPhillips et al., 2014; Pinti et al., 2013; Humez et al., 2016). Thesesurveys are of great importance as little data is available on naturalmethane in groundwater systems (Dyck, 1980; Barker and Fritz, 1981;Aravena and Wassenaar, 1993; Darling and Gooddy, 2006;Dochartaigh et al., 2011).In the Province of Quebec (Canada), 130 wells tapping shallowgroundwater of aquifers of the St. Lawrence Lowlands, a relatively fl atarea located between Montreal and Quebec City, were analyzed forthe methane content and its isotopic composition (Moritz et al.,2015). In this region, Ordovician methane-prone Utica shale has beenthe focus of intense exploration for its unconventional gas productionpotential until 2010, when a moratorium halted hydraulic fracturingat its very beginning. The focus of the study by Moritz et al. (2015)was to document baseline concentrations and sources of dissolvedmethane in order to evaluate potential environmental risks in case of Utica shale gas exploitation.Thegoalofthisstudyistoshowhowmethanedatafromgroundwa-ter can be used to estimate anthropogenic and natural emissions of methane in shale gas prospection areas (Fig. 1). Emissions of methaneare here assumed to be: 1) groundwater abstraction for domestic andmunicipal uses; 2) methane degassed during groundwater dischargein streams and rivers; 3) macroseepage and microseepage over naturalgas source rocks; 4) potential methane degassing during well comple-tion fl owback. This last methane source is here evaluated to comparemethaneemissions prior to shale gasexploitation, with potential emis-sions related to the drilling and hydraulic fracturing or “ fracking ” oper-ations. Potential fugitive methane emissions produced during shale gasexploitation (source 5 in Fig. 1), which could dramatically increasemethane emissions to the atmosphere (Schneising et al., 2014), shouldbe added to this fi gure. The evaluation of fugitive emissions is not pos-sible without measured data on well casing quality (the major sourcefor theseemissions;e.g.,Darrahetal.,2014),andthusisnotconsideredfor this area where exploitation is not in progress. 2. Geology and hydrogeology of the study area Thestudyareacovers16,537km 2 in theSt.Lawrence Lowlands.Itisdelimited by the Appalachian Mountains to the southeast and by thenorth shore of the St. Lawrence River to the northwest (Fig. 2). This re-gion was the target of detailed hydrological, geochemical and isotopicstudies centered on the Becancour River watershed (Meyzonnat et al.,2016; Vautour et al., 2015; Méjean et al., 2016), the Nicolet and lowerSaint-François River watersheds (Larocque et al., 2015; Saby et al.,2016) and the eastern portion of the Monteregie region (Carrier et al.,2013) (Fig. 2). These studies targeted the understanding of aquifer Fig. 1. Sources of anthropogenic and natural methane emissions in a watershed where shale gas geological formations are present or are exploited. See text for details. Numerals inhexagons represent anthropogenic emission sources, numerals in circles represent natural emission sources.1330 D.L. Pinti et al. / Science of the Total Environment 566 – 567 (2016) 1329 – 1338 geometry, watercirculationand residence timein order to evaluate thewaterresourcesavailable.Theworkwascarriedoutundertheguidanceof the Quebec Ministry of Environment in the framework of the 2009 – 2015 PACES aquifer characterization program ( Programme d'acquisitiondes connaissances surleseauxsouterraines ).Thethreezonesofstudyaredesignated as BEC (Becancour), NSF (Nicolet St. François) and MONT(Monteregie Est) in the discussion below.Thestudyareacoverstwogeologicalprovinces:theCambrian-Ordo-vicianSt.LawrencePlatformandtheOrdovician-DevonianAppalachianMountains, the two provinces are delimited by the regional low-anglefault named the Logan Line (Fig. 2). The St. Lawrence Platform corre-sponds to a Cambrian-Lower Ordovician siliciclastic and carbonateplatformhavingamaximumthicknessofca.1200m,overlainbyamin-imum of ca. 1800 m of Middle-Late Ordovician foreland carbonate-clastic deposits (Lavoie et al., 2013), the latter constituting the regionalfractured rock aquifer targeted by this groundwater methane survey.The Middle Ordovician Utica Shale (or its facies-equivalent StonyPoint), which is the main target for the oil and gas industry, is a limymudstone with total organic carbon (TOC) concentrations between 1and 1.5% (Lavoie et al., 2013). Another secondary target is the Sainte-RosalieGroupandtheLorraineGroup(TOC=0.5 – 1%)thataresiltstone,mudstoneandsiltyshales,whichrepresentsthemostoutcroppinggeo-logical formation in the St. Lawrence Lowlands (Globensky, 1987). Theterrains outcropping on the Appalachian Mountains correspond toimbricated thrust sheets composed of Cambrian-Ordovicianmetasediments (Globensky, 1993). These Cambrian-Ordovician unitsare partially covered by discontinuous Quaternary fl uvio-glacial sand-stones that constitute the shallower granular aquifer. This semi- Fig. 2. Interpolated spatial distribution of the methane concentrations in groundwater for the study area. Boundaries of the three areas where measurements were available (BecancourRiver/BEC,NicoletandlowerSaintFrançoisrivers/NSFandeasternportionoftheMonteregieregion/MONT)arereportedindottedlines.Thefrequencyhistogramsrepresentthemethaneconcentration distribution in each of the above-mentioned watersheds. Modi fi ed from Moritz et al. (2015). The white diamonds indicate the location of the 130 sampled wells.1331 D.L. Pinti et al. / Science of the Total Environment 566 – 567 (2016) 1329 – 1338 con fi ned aquifer is isolated at its base by impermeable tills units(Lamothe, 1989).Groundwaterchemistryshowstheoccurrenceoflow-salinitywater,dominantlyof Ca-Mg-HCO 3 type,close to themainrechargearea of theAppalachian Mountains. This water evolves by ion exchange into a Na-HCO 3 type along the fl ow line, with electrical conductivity rangingfrom 88 to 4466 μ S cm − 1 in the study area. Saline groundwater (con-ductivity from 717 to 31,500 μ S cm − 1 ) is found in a 10 km wide zonebordering the St. Lawrence River (Meyzonnat et al., 2016). Groundwa-ter is clearly brackish in a 2200 km 2 area to the north of the MONTarea, where salinity reaches 7 g L − 1 (Beaudry, 2013). The source of sa-linityderivesfromexchangesoffreshwaterwithporeseawatertrappedinto the thick Champlain Sea clays that con fi ne partially or totally thefractured bedrock aquifer in a narrow band along the St. LawrenceRiver (Beaudry, 2013).Water is homogeneously abstracted in the region because watersupply is dominated byprivate wells to supply single-family dwellings.In a few areas, small- to medium-size municipalities (from a few hun-dreds to few thousand people) are supplied by municipal wells(Carrier et al., 2013; Larocque et al., 2013, 2015; Lavoie et al., 2013).For example, in the administrative area of Centre-du-Quebec, whichcorresponds to the BEC and NSF regions(Fig. 2), there are 71municipalwater distribution systems against a total number of single wells rang-ing between 9000 and 13,000 (MDDELCC, 2015).Thewaterusage(bothsurfaceandgroundwater)isquitevariableinthe three studied regions. In the BEC region, the percentages of wateruseforagriculture,industryandresidentialare87.7%,2.9%and9.4%,re-spectively (Larocque et al., 2013). These percentages are 22%, 38% and40%, for the NSF region (Larocque et al., 2015) and 3.5%, 77.4% and19.1%, for the MONT region (Carrier et al., 2013).Aquifersinthestudyareaarehistoricallyknownforcontaininglargeamounts of methane with more than 600 methane traces detected ingroundwater wells (e.g., Séjourné et al., 2013). Macroseeps of methaneare also known in the region and the best example is the “ Fontaine dudiable ” (The Devil's Fountain), an eternal fl ame source located nearthe town of Trois-Rivieres (Fig. 2). Here methane from microbial srcinis produced locally in the Quaternary sandy deposits (St-Antoine andHéroux,1993).Theworkof Moritzetal.(2015)wasthe fi rsttopreciselymeasure CH 4 concentrations and identify its sources using δ 13 C andwetness index C 1 /C 2 + C 3 in the region. Methane concentrations weremeasured in 130 groundwater wells at depths between 6 and 120 min the southern shore of the St. Lawrence River, together with ethane(C 2 H 6 ), propane (C 3 H 8 ) (Moritz et al., 2015), radiogenic helium ( 4 He)(Pinti et al., 2013) and 222 Rn (Pinti et al., 2014). Details of the samplingand analyses techniques for methane are reported in Moritz et al.(2015). 3. Methane in groundwater of the studied area Moritzetal.(2015)publishedadetailedreportonmethaneconcen-trations and sources in the groundwater of the St. Lawrence Lowlands.Here, the results needed to understand the presence of methane ingroundwater are summarized.A total of 110 of the 123 sampled wells in the BEC, NSF and MONTregions contained measurable concentrations of methane. A few morewells (n = 7) sampled in the north shore of the St. Lawrence Riverwillbenotconsideredinthefollowingdiscussion,becausetheirnumberis not statistically signi fi cantcomparedto thearea investigated (Fig. 2).Atotalof118methanemeasurements(including8duplicates)wereobtainedfromwellsofBEC,NSFandMONTregion.Methaneconcentra-tionrangesfrom0.7to45.9×10 3 μ gL − 1 .Themethanemedianconcen-tration for the St. Lawrence Lowlands is 111 μ g L − 1 while the averageconcentrationis4417 μ gL − 1 (Table 1).Themethaneconcentrationfre-quency histograms for the three studied regions show clearly a lognor-mal distribution with a positive skew (Fig. 2). In order to take intoaccount these distributions, both the median and the average CH 4 con-centrations are used in the following emission estimates.There is a dichotomy in the methane distribution in the area, withhigher concentrations (median value of 510 μ g L − 1 ) measured in theSt. Lawrence Platform (the region N-NW of the Logan Line; Fig. 2)where organic-rich Ordovician shales outcrops (Utica, Lorraine and St.Rosalie Groups), compared to the Appalachian Mountains (medianvalueof24 μ gL − 1 )whereCambrio-Devonianagemetasedimentsdom-inate (Fig. 2). The higher methane concentrations were found in wellstapping groundwater from the Lorraine shales (Moritz et al., 2015),the most widespread geological formation outcropping in the St. Law-rence Lowlands (Lavoie et al., 2013) and a secondary potential targetfor the gas exploration companies.Acomparisonofthemedianandaverageconcentrationsofmethanein the St. Lawrence Lowlands with those from other shale gasprospection areas (Table 1; Fig. 3) reveals that the Ordovician regional aquifer of the St. Lawrence Lowlands is the most methanogenicamong those studied until now for determining methane baselinelevels. When average concentrations are considered, only the SouthernOntarioaquifers,whicharesimilarinlithologyandshaleformationsen-countered,havemethaneconcentrationscomparable to those of the St.Lawrence Lowlands. A methane survey by Humez et al. (2016) in 186observation wells in the Alberta Basin (Canada) shows a higher meth-ane baseline than in the St. Lawrence Lowlands, with values closer tothe ones observed in groundwater wells located in exploited areas of the Marcellus shale (Pennsylvania and New York states) and Barnettshale (Texas state) (Darrah et al., 2014). The area of Alberta surveyedbyHumezetal.(2016)cannotreallybeconsideredpristinesinceitcor-responds to the oil and gas productive zone of the Petroleum WesternCanadaBasin.ThedatafromtheAlbertaBasin,aswellasfortheMarcel-lus and Barnett areas (Fig. 3) does not re fl ect natural baseline methaneconcentrations in groundwater but possibly a mix of natural and an-thropogenic methane sources.The δ 13 C signatures of methane in the St. Lawrence Lowlands weremeasured for 73 wells that had a suf fi ciently high methane concentra-tiontoallowprecisemeasurementofitsstablecarbonisotopesignature.The values ranged between − 105.1 and − 24.8 ‰ , indicating the pres-enceoftwosources,the fi rstbeingmoreenrichedin 13 Cofthermogenicsrcin(i.e.,fromthematurationofkerogen;Schoell,1980),andthesec-ond being more depleted in 13 C and microbial in srcin, i.e., from theproduction of methane by methanogens (Whiticar, 1999). The isotopicsignature of methane in groundwater from the Lorraine shales rangedfrom − 103 to − 55 ‰ with a median value of − 63 ‰ for samples withCH 4 higher than 5 μ g L − 1 , indicating a predominantly microbial source(Moritz et al., 2015).The occurrence of widespread biogenic sources of methane is inagreement with the redox conditions of the fractured bedrock and thenatureoftheaquifers.Microbialmethaneismoreabundantincon fi nedaquifers with strongly reducing conditions (Aravena and Wassenaar,1993; Darling and Gooddy, 2006), which are favorable for acetate fer-mentation and/or CO 2 reduction (Whiticar, 1999). Moritz et al. (2015) have shown statistical evidence that the methane concentration ingroundwateroftheSt.LawrenceLowlandsincreaseswiththeevolutionofgroundwaterchemistry,withhighervaluesmeasuredinevolvedNa-HCO 3 and NaCl waters con fi ned in the plain compared to Ca-Mg-HCO 3 type waters that represent the recharge. Méjean et al. (2016) havemeasured very low dissolved uranium concentrations in groundwaterand highly fractionated 234 U/ 238 U signatures from the NSF region,con fi rming that the Ordovician regional aquifer is under strongreducing conditions. 4. Methodology The equations used to calculate potential methane emissions in thestudy area for groundwater abstraction (source 1; Fig. 1), groundwaterdischarge (source2; Fig. 1)and fromwellcompletion fl owback(source 1332 D.L. Pinti et al. / Science of the Total Environment 566 – 567 (2016) 1329 – 1338 4;Fig.1)areherepresented.Theequationsneededtocalculatetheoret-ically methane seepage emissions are detailed in the discussion.Methane emissions from groundwater abstraction for domestic andmunicipal uses are calculated using the following equation:Q abs ¼ X ji C CH4 T C ð Þ i … j ð 1 Þ where Q abs is the quantity of CH 4 emitted to the atmosphere annually(Tg yr − 1 ), C CH4 is the concentration of methane in groundwater( μ gL − 1 )foreachregionoraquiferconsideredandT C isthetotalvolumeof water pumped yearly from groundwater sources in the different re-gionstomeetdomesticandmunicipalneeds(Mm 3 yr − 1 ).Q abs ,calculat-ed in Tg yr − 1 , is obtained by converting the abstraction volumes in L (Mm 3 × 10 9 ) and the methane concentration in Tg ( μ g × 10 − 18 ). Herewe assume that all the methane contained in groundwater is degassedinto the atmosphere. Indeed, at the solubility equilibrium and at ambi-ent atmospheric conditions (ASW or Air Saturated Water at 20 °C),the methane concentration in water is 0.042 μ g L − 1 assuming an aver-age concentration in the atmosphere of 0.00018% (v/v) and solubilityconstants from Wilhelm et al. (1977). This concentration is 14 times smaller than the detection limit of our system (0.6 μ g L − 1 ; Moritz etal., 2015) and thus the residual pool of dissolved methane remainingin the water is not measurable and can be considered negligible.During the hydrological cycle, groundwater is discharged to surfacewater into streams, rivers or springs. In the St. Lawrence Lowlands wa-tersheds, the surface discharge reservoir is the St. Lawrence River, itstributariesandlakes(Larocqueetal.,2013).Asinthecaseofhumanab-straction,methanefromdischarged groundwaterisexpected to rapidlydegas to the atmosphere (e.g., Heilweil et al., 2015).Estimation of groundwater discharge to rivers from fl ow rate timeseries is dif fi cult when there is large number of ungauged rivers, as inthe St. Lawrence Lowlands. Scanlon et al. (2002) argue that rechargeand groundwater discharge to rivers can be equal if pumping, evapo-transpirationandunder fl owtodeepaquifersarelimited.Thishypothe-sisisconsideredreasonableforthestudyareawheremostriversareinagaining position, pumping and evapotranspiration are relatively small, Table 1 Summary of methane data from groundwater in the St. Lawrence Lowlands (this study), US and Canada shale gas exploration and exploitation areas.BEC NSF MONT St. LawrenceLowlandsSouthern Ontario NovaScotiaCentralNew YorkNortheastPennsylvaniaAppalachiansMarylandAlbertaBasin a MarcellusShale, PABarnettShale, TXThisstudyThisstudyThisstudyThis study [1] [2] [3] [4] [5] [6] [7] [7]Sample no. 52 22 44 118 1978 103 113 1701 78 186 111 59Surface (km 2 ) 2920 4585 9032 16,537 ≈ 90,100 ≈ 22,900 2315 ≈ 920 ≈ 1930 ≈ 277,000 26,700 n.d.Targeted shale Utica Utica Utica Utica Utica b , Marcellus,Keetle PointHortonBluff Marcellus Marcellus Marcellus – Marcellus BarnettMinimum ( μ g L − 1 ) 0.8 0.7 20 b 0.6 0 b 3 2 0.05 b 1 0.1 2 778Maximum ( μ g L − 1 ) 45,913 18,995 36,704 45,913 227,604 6000 8264 43,000 8550 42,900 111,000 51,025Median ( μ g L − 1 ) 152 134 86 111 9.13 b 3 7 0.76 1.5 5830 700 5996Average ( μ g L − 1 ) 3891 2693 5902 4417 4610 117 464 705 291 12,416 9155 14,38895% conf. ( μ g L − 1 ) 2582 2327 3180 1672 659 142 261 159 301 2158 3047 4190Referencesaslisted:[1]Hamilton(2011);[2]DrageandKennedy(2014);[3]McPhillipsetal.(2014);[4]Molofskyetal.(2013);[5]BoltomandPham(2013);[6]Humezetal.(2016);[7] Darrah et al. (2014). a Statistics reported on 62 wells (see Humez et al., 2016 for details). b Utica time-equivalent Collingwood-Blue Mountain Fm. Fig.3. Methanebaselineconcentrations(medianandaveragevalue)measuredingroundwaterfromshalegasdevelopmentareasofCanadaandUSAcomparedwithmethanefromareasofoilandshalegasexploitationfromAlberta(Canada),Marcellus(Pennsylvania,USA)andBartnett(Texas,USA).Bibliographicreferencesofexploiteddataarereportedindetailinlegendof Table 1.1333 D.L. Pinti et al. / Science of the Total Environment 566 – 567 (2016) 1329 – 1338
