Introduction

Methane, a reduced form of carbon, is a minor component of the atmosphere at ~1840 parts per billion in the year 2016. Methane’s atmospheric residence time is 12–17 years, and methane has a global warming potential of 25 over 100 years (GWP100 of 25), compared with carbon dioxide, which has a GWP100 of 1.1 From 2000 to 2006, atmospheric concentrations of methane remained relatively constant, however, from 2007 to 2013, atmospheric methane increased by more than 5%, and in 2014 and 2015, concentrations increased by more than 10%,2 and are approaching high greenhouse-gas scenarios, which could exacerbate climate change. Proposed drivers of methane emissions include growth in fossil fuel use, increased agricultural production, thawing arctic permafrost, increased precipitation and tropical flooding, and the decrease in hydroxyl radicals in the atmosphere. However, confirmation is difficult because of uncertainties in the global methane budget. Identification and quantification of previously unidentified sources of methane emissions is crucial for improving methane budget estimates, and modeling climate change.3

Methane gas is predominantly derived from organic matter and produced by either thermogenic or microbial processes.4 Thermogenic methane forms when organic matter is buried and heated, usually at considerable depths over long periods of time, and includes methane associated with coal, gas, and oil formations. Microbial methane formation follows two pathways: via acetate fermentation or CO2 reduction, typically at shallower depths, by the actions of methanogenic organisms, members of the domain Archaea, on organic matter in anoxic environments, such as wetlands, landfills, and some aquifer sediments and coal deposits. Both thermogenic and microbial methane accumulate in groundwater as a result of generation, migration, and accumulation,5,6,7,8,9,10 with measured concentrations as high as 150 mg/L in the LAB.7 Groundwater pumping and the subsequent release of dissolved methane is potentially a significant source of methane emissions that has not been quantified in the global methane budget,11 although methane emissions from groundwater in the UK have been estimated at 2.2 × 10−6 Tg/a, about 0.05% of total UK methane emissions.12 Methane and other dissolved gases are typically removed from drinking-water distribution networks by several methods, included vented well caps, cascade, cone, and spray aerators, air stripping, or discharge into open or vented tanks.

While studying the sources and formation pathways for methane in LAB groundwater,7 the implications for the high methane concentrations and large volumes of annual groundwater pumping led to the estimation of methane emissions from the LAB, and a comparison with the Marcellus shale region of NE Pennsylvania. Further compilation of recent groundwater methane data,13 and annual groundwater pumping,14 enabled estimation of methane emissions from groundwater pumping in the Principal aquifers of the USA (Fig. 1).

Fig. 1
figure 1

Methane emissions resulting from pumping groundwater in the Principal aquifers of the USA. See Table 5 for aquifer names and values. Not shown but included in the total US groundwater pumping methane emissions are Alaska, Puerto Rico, Hawaii, and US Virgin Islands; *extrapolated estimate

Results

We estimate the emission of methane from the LAB at 2.9 × 10−3 Tg/a, by multiplying the annual groundwater abstraction for each aquifer (unit) by the average (arithmetic) methane concentration of each unit (Table 1), similar to approaches estimating the flux of nitrous oxide15,16 and methane from groundwater pumping.12 According to the Water Replenishment District of Southern California, in the year 2000, 3.1 × 1011 L of groundwater was pumped for municipal and industrial supply in the LAB.17,18 The distribution of pumping varies by chronostratigraphic unit (Table 2), with most groundwater pumped from the Pacific A, Pacific, and Harbor units.17,19 However, emissions from the Dominguez, Mesa, and Pacific units could not be estimated, since no wells in these units were sampled, and emission estimates were based only on the sampled units. This approach assumes all methane is completely degassed during pumping, treatment, and distribution of groundwater.12,15,16 Total ethane emissions also were calculated similarly, 3.5 × 10−6 Tg/a from the LAB aquifer systems.

Table 1 Los Angeles Basin groundwater sample ID, well depth, perforation intervals, dissolved methane concentration, and model layer and chronostratigraphic unit
Table 2 Groundwater-pumping methane and ethane emissions from each chronostratigraphic unit in the Los Angeles Basin, including the annual volume of groundwater pumped from each unit and percent of total pumping for each unit, average methane concentration of each unit, average ethane concentration of each unit, annual methane emissions and annual ethane emissions from each unit

Discussion

For comparison, groundwater-pumping methane emissions for the LAB, an area of 1088 km2, are ~0.7% of methane emitted annually from all sources in the South Coast Air Basin (SoCAB), an area of 17,469 km2, which includes urbanized portions of Los Angeles, Orange, Riverside, and San Bernardino Counties, and ~3% of the methane released from the Aliso Canyon storage facility leak in 2015–2016 (Table 2). The ethane released from groundwater in the LAB is only ~0.02% of annual SoCAB emissions, and ~0.05% the Aliso Canyon release. While a relatively brief event, the Aliso Canyon gas storage facility leak near Los Angeles from 2015 to 2016 released over 0.1 Tg methane and 7.7 × 10−3 Tg ethane to the atmosphere.20,21 Annual emissions of methane since 2007 in the SoCAB have been ~0.413 ± 0.086 Tg/a methane and ~0.023 Tg/a ethane.21,22 Methane emissions in the United States for various sectors including agriculture, energy, waste, industrial processes, land use change and forestry have been estimated as a total flux of ~29.03 Tg of methane23 in the year 2000. Global emissions of methane are estimated at ~275 Tg/a of methane in 2010.2

High methane concentrations, those in excess of values expected from water equilibrating with the atmospheric (~5 × 10−5 mg/L at 1 atm and 10 °C), have been identified in many aquifers in hydrocarbon-producing regions across the USA and the world e.g.24,25,26,27 The average methane concentration of 1701 Pennsylvania groundwater samples overlying the Marcellus is 0.7 mg/L24 (less than detection-level values were assigned a value of zero, so this is a lower limit), the average concentration in aquifers overlying the Bakken shale is 2.8 mg/L,28 Eagle Ford shale is 1.3 mg/L, Haynesville–Bossier shale is 2.6 mg/L, Fayetteville shale is 1.9 mg/L,29 and in the entire UK is 0.045 mg/L.30 The annual emissions of methane from the Pennsylvania groundwater overlying the Marcellus is 3.0 × 10−6 Tg/a, calculated using the average methane concentration (0.7 mg/L), and average annual withdrawal of 4.27 × 109 L/a in Susquehanna County https://water.usgs.gov/watuse/data/2005/index.html suggesting a relatively small contribution to the total US methane emissions from this area (Table 3). Future work is needed to quantify the withdrawal of groundwater in the Bakken, Eagle Ford, Haynesville–Bossier, and Fayetteville regions to calculate additional emissions of methane to the atmosphere from groundwater withdrawn from these hydrocarbon-producing regions. It should be noted that this is a coarse estimation of emissions from pumped groundwater to the atmosphere, based on a limited number of groundwater locations and samples that may be biased toward high-methane groundwater. Additional randomized sampling is needed for improved estimates.

Table 3 Comparison of methane concentrations in groundwater in different study areas in the USA, annual groundwater withdrawal rates, and calculated annual methane emissions from groundwater pumping in oil and gas regions

To constrain emissions of methane from groundwater pumping across the USA, the average methane concentrations in 15 Principal aquifers were calculated using data from 833 sampled groundwater wells;13 less than reporting-level values were assigned a value of zero (Table 4). By treating values reported as less than the detection level as zero, the methane emissions calculated in this work are a minimum estimate. If the non-detections are assigned the value of one-half the detection level, the average methane concentrations in nine of fifteen aquifers, and the average of all samples, were the same as using a value of zero, while the other six aquifers were within 45%. Public-supply wells were selected for sampling using a randomized spatially distributed grid–cell network to provide statistical representation for the public-supply aquifers.13 The methane data are representative of methane emissions associated with pumping from public-supply wells. The 15 Principal aquifers sampled13 correspond to 15 of the 66 Principal aquifers identified by Maupin and Barber,14 who estimated groundwater withdrawals from the Principal aquifers of the USA. Total withdrawals from the Principal aquifers in the USA was 1.06 × 1014 L/a in 2000.14

Table 4 Comparison of methane concentrations in groundwater, annual groundwater withdrawal rates, and calculated annual methane fluxes from groundwater pumping from 15 Principal aquifers and the six Principal aquifer systems of the USA

Annual methane emissions (Table 4) calculated for the 15 Principal aquifers discussed in McMahon et al.13, represent only ~50% of the groundwater withdrawal in the USA. To calculate the methane emissions from all 66 Principal aquifers, they were grouped into six Principal aquifer systems by lithology: carbonate rock, crystalline rock, sandstone, semi-consolidated sand, glacial unconsolidated sand and gravel, and nonglacial unconsolidated sand and gravel aquifer systems. Withdrawals from aquifers not included in one of the 66 Principal aquifers were reported in an “other” aquifers group.14 The emissions from all six Principal aquifer systems were calculated by assuming that the average methane concentration for the aquifers with data in a Principal aquifer system were representative of other aquifers in the same lithologic group without data. The methane concentration for each of the six aquifer systems were assumed to be the average of the aquifers within it. In some cases, such as the Sandstone Principal Aquifer System, there was only one aquifer13 methane concentration to base the value on. The “other” aquifers group was assigned the average methane concentration for all Principal aquifers systems (0.44 mg/L).

The annual emissions from the six aquifer systems and “other” aquifers were 0.041 Tg/a (Table 4). Subtracting the 2.3 × 10−6 Tg/a calculated from this method with the Maupin and Barber14 data for the LAB (included in the California Coastal Plain PAS), and adding the 2.9 × 10−3 Tg/a estimated from this study for the LAB, total annual methane emissions from the Principal aquifers in the USA are 0.044 Tg/a (Table 4), a small part (~0.2%) of the US total. However, groundwater-pumping methane emissions in the USA are similar to the emissions of methane from leaking oil and gas wells,31 or industrial processes (Table 3). Most of the groundwater methane in the USA is emitted from Semi-Consolidated Sand and Gravel Principal Aquifer Systems (Fig. 1). Better constraints on the methane concentrations and groundwater pumping in the three Principal aquifers, High Plains, Central Valley, and Mississippi River aquifers, which comprise 48% of the groundwater pumping in the USA, would greatly improve these estimates.

Uncertainties associated with the Tier 111 emission estimates are at least 5%, which reflects the measurement uncertainty for methane, but do not account for temporal methane variability, incomplete accounting of groundwater pumping, or the inaccurate estimation of an aquifer’s average methane concentration. Some temporal variability of methane resulting from sampling protocols32 or domestic wells33 may be dismissed in this study, as strict sampling protocols34,35 were used to sample public water-supply wells. However, significant variability in groundwater methane has been observed in samples (concentrations <0.3 mg/L) collected quarterly,30 and sample pairs, which varied by up to 67%,10 representing significant uncertainty. Annual groundwater pumping uncertainties reflects either inaccurate reporting or missed pumping. Most large water purveyors are known and track pumping closely, so the uncertainty on reported pumping from public-supply wells is small, <5%. When aggregating pumping volumes, some small operators may fail to respond or may be missed, such that their pumping was not included. These operators would represent a small portion of the groundwater pumping and as such, the uncertainty on unaccounted pumping is ~5%. In total, the uncertainty associated with annual groundwater pumping is estimated to be ~10%. The average methane concentrations estimated for each aquifer have the largest uncertainties, as calculations may be skewed depending on which wells were selected to sample. A randomized well selection process13,36 ensures that samples are representative of the concentrations in the aquifers being pumped. However, if studies do not collect enough samples to be representative, or their selection is not randomized, then the uncertainty could be up to 500% based on averaging sample concentrations in an aquifer. For the USGS Principal aquifer data,29 a randomization process was applied to available public-supply wells in the aquifers, and the uncertainty on the average methane concentrations in the public-supply aquifers was reduced to ~25%. The overall uncertainty on emissions estimates, based largely on possible temporal variability of methane is ~80%.

While small compared with total global emissions,11 methane emissions resulting from groundwater abstraction in the USA represent an important source to be quantified and should be included in the global methane budget. Emissions may be significant locally, where groundwater methane concentrations are high. Also, of local significance, methane, as a precursor to ozone formation, could represent an unaccounted for and potentially significant contribution to low-level ozone formation. More detailed analyses of correlations between methane concentrations and groundwater age, carbon system indicators, aquifer redox conditions, Eh potentials, as well as lithology and sediment age may help to better estimate methane occurrence and concentrations in groundwater.13,37 Within a given lithology group, methane concentrations are typically higher in younger sediments than in older sediments,13 and higher in older groundwater (pre 1950s) than recently recharged groundwater.7,13 Previous studies noted that sediment age can be important with respect to redox processes due to the preferential consumption of the most reactive components of sedimentary organic carbon by microbes over time.38,39,40 In support of this hypothesis, McMahon et al.13 found that for aquifers composed of similar lithologies, concentrations of methane, dissolved organic carbon, and ammonium were higher in the aquifers with younger sediments than in aquifers with older sediments.

If 78–99% of groundwater in the upper 2 km of the continental crust is older than 50 years,41,42 then groundwater methane may represent a significant source of methane, given increased rates of groundwater depletion (1500 km3/a).43 If 80% of the 1500 km3/a extracted is old groundwater with elevated methane concentrations (0.44 mg/L methane), then annual global groundwater-pumping methane emissions would be ~0.53 Tg/a, over three times the annual emissions from the metropolis of Los Angeles, California, but only 0.2% of global methane emissions. However, this estimate fails to capture the emissions of methane naturally/passively degassing from shallow methanogenic groundwater8 or from streams receiving methanogenic groundwater discharge,44 both of which could represent significant annual emissions. Further constraint of these processes is needed to better estimate extracted and natural methane groundwater emissions. Although expensive, given the number of groundwater wells, capturing the methane at the largest wells pumping groundwater with the highest methane concentrations could greatly reduce emissions. For example, in the Los Angeles Basin, capturing methane at 17 wells, which each pump more than 2.5 × 109 L annually, could reduce methane emissions by nearly a fifth.

Methods

The LAB is located within a northwest trending alluviated plain on the southern California coast, and contains deep sequences of petroliferous sediments and traps overlain by organic-rich marine and alluvial water-bearing sediments.19,45 Thirteen chronostratigraphic water-bearing units were identified using sequence stratigraphy to subdivide and link sedimentary deposits into unconformity-bound units.19 Wells were associated with chronostratigraphic unit based on well-screen depth and location (Table 1). Methane concentrations from 37 LAB groundwater samples ranged in concentration from 0.002 mg/L to 150 mg/L (Table 1), with an average value of 44.1 mg/L. Sample selection for the LAB was not randomized, and as a result, LAB methane emissions have a larger uncertainty (500%) than those of the Principal aquifers of the USA (80%), and thus may overestimate methane emissions. While thermogenic methane was detected locally in the shallow subsurface in conjunction with some of the historic explosions that occurred in the LAB,46 the methane distribution and isotopic systematics, and groundwater age results indicate that most excess methane in LA groundwater originates from microbial production in anoxic or suboxic conditions in the shallow aquifer systems, rather than the migration of deeper thermogenic methane in the hydrocarbon-bearing formations associated with oil fields in the basin.7

Groundwater samples were collected in accordance with the protocols established by the USGS National Water-Quality Assessment (NAWQA)34 and the USGS National Field Manual35 after purging the equivalent of three casing-volumes from each well with a submersible pump. Samples were analyzed for a suite of inorganic, organic, and isotopic constituents; for a complete description of the sampling, analytical methods, quality assurance, and data.47 Samples for analysis for concentration and isotopic composition of hydrocarbon gases were collected by attaching sampling tubing to a hose bib and flushing groundwater through the tubing before the flow was partially restricted with a control valve. An evacuated pouch containing a bactericide capsule was attached to the tubing and filled with 600–700 mL of sample water. The hydrocarbon gas concentrations and isotopic values were measured using gas chromatography and isotope ratio mass spectrometry, respectively. Methane concentration measurement uncertainty is ± 0.0001 mg/L7, and the statistics on the methane measurements have been previously presented.7,13