CH₄ isotopic signatures of emissions from oil and gas extraction sites in Romania

Large reductions of greenhouse gas (GHG) emissions to the atmosphere are required to mitigate ongoing climate change. The irreversible impacts of a global warming of more than 1.5°C above preindustrial levels were highlighted in a special report from the Intergovernmental Panel on Climate Change (IPCC, 2018). Limiting this increase to 1.5°C with a reasonably high probability requires not only net-zero emissions of carbon dioxide (CO₂) by 2055 but also a reduction of methane (CH₄) and black carbon emissions by 35% or more by 2050, relative to 2010 (IPCC, 2018). The European Union (EU) recently incorporated a net-zero emission objective for all GHGs by 2050 (European Green Deal; European Commission, 2019). In 2018, reported GHG emissions by EU countries (including the United Kingdom) were more than 4 billion tons CO₂ equivalent, with a contribution of 11% from CH₄ (European Environment Agency, 2020).

In 2019, the European Green Deal specially mentioned the necessity to “address the issue of energy-related methane emissions” (European Commission, 2019). Reported fugitive emissions from the extraction and handling of coal, oil, and natural gas were responsible for 12% of the EU’s total CH₄ emissions in 2018 (Crippa et al., 2019). In the United States, fugitive CH₄ from the natural gas supply chain was reported to be equivalent to total CO₂ emissions generated by its combustion (Alvarez et al., 2018). This is due to the global warming potential of CH₄ being 86 larger than the one of CO₂ over a 20 years-time horizon, including carbon cycle feedbacks (IPCC, 2013). An increased focus on CH₄ fugitive emissions is therefore particularly interesting for efficiently reducing GHG emissions.

GHG emissions are reported for each country using bottom-up approach (data-driven inventories), but top-down studies (atmospheric measurements and modeling) help verifying these estimates (Saunois et al., 2020). Romania is the fourth and third largest producer of crude oil and natural gas in the EU (including the United Kingdom), respectively (British Petroleum, 2020). It is especially relevant to investigate the CH₄ emissions in Romania as this region is not as well covered by atmospheric measurements than Western Europe (Integrated Carbon Observation System Research Infrastructure, 2019). The ROmanian Methane Emissions from Oil & gas (ROMEO) project (Röckmann, 2020) aims to provide experimental quantification of methane emissions from oil and gas extraction activities. In this framework, the present work reports methane isotopic data of atmospheric air samples collected mainly in the Romanian Plain, a major oil–gas production region in Romania.

As described in more detail below, the measurements of CH₄ isotopic composition (stable C and H isotope ratios) are widely used to characterize the sources of methane in atmospheric air samples (Levin et al., 1993; Tarasova et al., 2006; Röckmann et al., 2016; Townsend-Small et al., 2016; Zazzeri et al., 2016; Hoheisel et al., 2019; Lu et al., 2021), following genetic (thermogenic vs. microbial) CH₄ isotopic categorization (Schoell, 1980; Milkov and Etiope, 2018). The ethane (C₂H₆) to methane (C₂:C₁) ratio is also frequently used in atmospheric studies (Lopez et al., 2017; Mielke-Maday et al., 2019; Maazallahi et al., 2020) as an additional proxy of the gas origin, since thermogenic and microbial gas have different molecular compositions: Ethane (as well as higher hydrocarbons) is abundant in thermogenic reservoirs, with or without oil, and is practically absent in microbial gas. We note, however, that also high maturity thermogenic gas can be “dry,” ethane-free (Schoell, 1983; Milkov and Etiope, 2018), and this may confuse the genetic attribution of the gas based only on the lack of ethane. The CH₄ isotopic characterization can be used as tracer for the different sources in atmospheric transport models and inversions at different scales (Bousquet et al., 2006; Monteil et al., 2011; Houweling et al., 2017; Bergamaschi et al., 2018; Fujita et al., 2020) and therefore help improving the methane budget. More measurements of CH₄ source isotopic composition reduce the uncertainties in the signatures assigned to the different sources and to the resulting emission estimates.

The present study combined CH₄ isotopic and C₂:C₁ measurements in atmospheric air samples collected from aircraft and ground vehicles in different areas in Romania. From these measurements, we first determine the isotopic signature of CH₄ emissions from oil and gas production in Romania. Then, we investigate whether we can attribute CH₄ enhancements from aircraft measurements to the oil and gas extraction activities. Finally, we aim at providing further insight into formation processes of the fossil fuel reservoirs exploited in the Romanian Plain, based on both our results and previous literature.

The samples were taken during measurement surveys from an aircraft and from vehicles on the ground. The locations of all CH₄ enhancements that were characterized regarding isotopic compositions and/or C₂:C₁ ratio are presented in Figure 1 and summarized in Table 1. An enhancement describes a noticeable (usually from 100 ppb) increase in χ(CH₄) observed on a mobile analyzer. The target areas include important oil and gas basins and production sites in Romania.

We performed in situ measurements from a BN2 aircraft from the National Institute of Aerospace Research “ELIE CARAFOLI” (INCAS). Air samples were collected during 10 research flights on 9 days between October 2 and 17, 2019, over 7 regions of interest. The target areas were covered following a raster pattern (ca. 5 km width between the different legs), orientated perpendicular to the wind direction, and flying generally upwind. An example raster flight path is shown in Figure S1 of the supplementary material. The raster was generally flown at an altitude between 100 and 200 m above ground to be able to detect CH₄ enhancements from surface sources. We used a cavity ring-down spectroscopy (CRDS) analyzer onboard the aircraft (GasScouter G4302, Picarro, Santa Clara, CA, USA), to detect concentration elevations for determining when to collect samples. Unfortunately, due to an instrument malfunction, the ethane mode of the G4302 instrument was not operational during the campaign, therefore the instrument only measured CH₄ mole fractions (χ(CH₄)), at a measurement frequency of 1 and only operated at strongly reduced precision (instrument noise on the χ(CH₄) reading was approximately 200 ppb). Therefore, the CRDS analyzer was not used for quantification (another instrument was dedicated), but only to identify zones of CH₄ enhancements and sampling, which was still possible for significant enhancements, despite the reduced precision.

The instrument drew air through a 1/4” o.d. Teflon line that was connected to an air inlet at the outside, on the top-left of the cockpit. A 10-m 1/2” o.d. Dekabon tube was mounted on the outflow of the G4302 analyzer as a buffer volume. At a flow rate of 2.2 L/min, this volume is flushed in 15 s, so after a signal was detected by the instrument, we had about 15 s to decide whether it was significant enough to take a sample. If so, we diverted the air in the buffer volume toward the sample receptacle (2 L Supel™-Inert Multi-Layer Foil sample bags, Sigma-Aldrich Co. LLC, Saint Louis, USA). In some cases, the raster pattern of the flights allowed us to cross a zone of enhanced χ(CH₄) several times, and we collected several samples of the same suspected source. A total of 117 samples were taken from 31 CH₄ enhancements. The average background χ(CH₄) was 1975 ± 33 ppb, and CH₄ mole fraction enhancements ranged from 30 to 300 ppb above background. We only retained the samples with more than 60 ppb CH₄ above background and obtained CH₄ isotopic signals for 24 source locations.

Ground-based surveys were performed on 16 days between September 30 and October 19, 2019, targeted on 6 regions of interest in the Romanian Plain. The 3 ground teams that collected samples for this study traveled by car or truck, measuring CH₄ mole fractions using mobile analyzers (Ultraportable Greenhouse Gas Analyzer, Los Gatos Research, San Jose, CA, USA; GasScouter G4302 and G2201-i isotopic analysers, Picarro, Santa Clara, CA, USA; LI-7810 Trace Gas Analyzer, LI-COR, Lincoln, USA; Dual Laser Trace Gas Monitor, Aerodyne Research Inc., Billerica, USA). Once a CH₄ enhancement (ca. >200 ppb above background) was identified, the vehicle stopped and ambient air was collected in sample bags (2 L Supel™^_¢-Inert Multi-Layer Foil, Sigma-Aldrich Co. LLC, St. Louis, MO, USA; 3 L FlexFoil PLUS, SKC Inc., Eighty Four, PA, USA). One to three samples were taken for each χ(CH₄) anomaly, and 1 or 2 background samples were taken every survey day, in each region. A total of 161 samples were taken during ground surveys from 84 different sites.

Measurements of ethane and methane were performed in two vehicles during ground surveys. The surveys were performed on 16 days from October 1 to 18, 2019. In 1 car, a second G4302 instrument sampled air through a 1/2” o.d. teflon tube from the roof top of a car and continuously measured CH₄ and C₂H₆ at a frequency of about 1 Hz and a flow rate of 2.2 L/min. The noise was approximately 100 and 15 ppb around background level for CH₄ and C₂H₆, respectively. The instrument was installed on the back seat of the vehicle, running either on an internal battery or an external battery supply. The instrument connected to a tablet via Wi-Fi, on which the measured mole fractions were displayed with a delay of 5 s.

The second vehicle was equipped with a Dual Laser Trace Gas Monitor, based on a tunable infrared laser direct absorption spectroscopy using a combined quantum cascade laser and interband cascade laser (Aerodyne Research Inc., Billerica, MA, USA). Air was sampled through an RVS central inlet at the front of the trailer with a diameter of 60 mm at 3-m height, from which a 1/4” polyethylene tube goes to the instrument. CH₄ and C₂H₆ were measured continuously at a frequency of 1 Hz, with a flow rate of 6 L/min and a precision of 2.4 and 0.1 ppb, respectively. Precision is here reported as 3 times the standard deviation of 6-min constant concentration reading. The instrument was installed in the trailer, which was facilitated with infrastructure for electricity, a battery pack to run the mobile laboratory for 24 h, a router, and an air inlet. While driving, the measurements were displayed on a laptop, connected via Wi-Fi to the instrument with a delay of 5 s.

All samples were measured for χ(CH₄), δ¹³C, and δ²H in CH₄ at the Institute for Marine and Atmospheric Research Utrecht (IMAU) between October 8 and December 17, 2019. The analytical system is described in detail in Röckmann et al. (2016). The procedure is based on the extraction of CH₄ from the other air components, followed by its conversion into CO₂ or H₂ before isotope measurement using isotopic ratio mass spectrometry (IRMS). Measurements of δ¹³C or δ²H were performed separately, using 10–60 mL of sample air, depending on the CH₄ content. Each isotope measurement also returns a χ(CH₄) value, because the signal intensity is proportional to the amount of extracted CH₄. The system allows to process samples with χ(CH₄) in the same order of magnitude as ambient air. Some higher concentration samples were diluted with pure N₂ prior to measurement. Each sample was measured 2–4 times for both δ¹³C and δ²H, leading to an average uncertainty of 0.069 and 1.4 ‰, respectively, and of 5 ppb for χ(CH₄).

All reported isotopic values are relative to international reference materials: Vienna Pee Dee Belmnite (V-PDB) for δ¹³C and Vienna Standard Mean Ocean Water (V-SMOW) for δ²H. To do so, the sample measurements were alternated with a reference cylinder of ambient air, containing 1970.0 ppb CH₄ with δ¹³C = –48.072 (±0.026¹) ‰ V-PDB and δ²H = –88.311 (±0.3371) ‰ V-SMOW. The mole fractions were determined after measurements with a G2301 Gas Concentration Analyzer, Picarro, Santa Clara, CA, USA (precision at 5 min < 0.22 ppb). To calibrate the reference cylinder against the international scale for isotope ratios, it was measured repetitively (20 times) in the IMAU lab, against 2 other gases previously measured at the Max Planck Institute in Jena, Germany (Sperlich et al., 2016). The calibration of the reference gas was done in September 2019.

The Keeling (1958) plot method was applied to the isotopic data of each sampled χ(CH₄) anomaly. It is a mass balance approach used to derive the isotopic signature of an emission source, which is added to a stable background. This translates to the following expressions:

where c is the mole fraction and δ is the isotopic signature (δ¹³C or δ²H) of background (bg), source (S), or measured (m) CH₄. Thus, the source isotopic signature (δ_S) is given by the y-intercept of the regression line, when plotting δ_m against 1/c_m.

A prerequisite for this approach is that the background mole fractions and the isotopic values of background and source are stable over the sampling time period of each peak. To account for potential changes in the background CH₄ from 1 day and region to another, a sample of background air was taken for each measurement day and area. We assumed a stable background for each location, given that we used the background sampled on the same day and in the same region in each of the plots. In order to confirm that variations in the background do not cause significant biases, we also evaluated our results using the Miller–Tans method (Miller and Tans, 2003), which is suitable for calculating isotopic source signatures with a varying background, and found very similar source signatures.

For all linear regressions, the orthogonal distance regression (Boggs and Rogers, 1990) fitting method was used.

CH₄ and C₂H₆ data were extracted from the in situ measurements from two survey vehicles when we detected significant CH₄ enhancements in the vicinity of oil and gas facility locations reported by the regional operator. For each of these locations, we determined C₂:C₁ ratios from the slope of the linear regression between χ(C₂H₆) and χ(CH₄) measured values when a minimum excess threshold for CH₄ of 100 ppb above background was exceeded. C₂:C₁ ratios smaller than 0.0004 were set to 0.

Methane origin is generally biotic (Hunt, 1996; Clayton, 2005), including (1) thermal degradation of organic matter in sedimentary rocks (thermogenic gas), (2) metabolic reactions by certain microorganisms (microbial gas), and (3) biomass burning (pyrogenic gas). In some geological environments, methane can also be generated by chemical reactions in the absence of organic matter (abiotic gas; Etiope and Sherwood Lollar, 2013; Etiope, 2017). Microbial gas generated in sedimentary rocks in petroleum systems is fossil (radiocarbon-free), so as thermogenic gas, and can also be categorized as “geological methane.” Modern microbial gas is generated in surface ecosystems (wetlands, marshes, rice paddies, etc.) and by animal enteric fermentation; this category is mostly referred to as “biological methane” (Schoell, 1983). Microbial CH₄ generation may follow 2 metabolic pathways, methyl-type fermentation (anaerobic enteric fermentation, and to a larger extent in continental deposits and freshwater) and CO₂ reduction (preferred pathway in marine environments; Whiticar, 1999).

Geological hydrocarbon gases can be “wet” or “dry,” indicating the presence or absence of C2+ alkanes (Etiope, 2017). Alkanes are generally present in thermogenic systems, with or without oil (liquid phase). Geological gas composition and CH₄ isotope ratios are generally used to assess the origin of oil and gas associated methane. Milkov and Etiope (2018) compiled a global inventory of molecular and isotopic composition of natural gas from petroleum source rocks, reservoirs, and surface seeps and assessed the ranges of several variables corresponding to the various gas origins.

Natural gas is generally composed of a majority of CH₄ (>70%), with C₂H₆ (1%–10%), along with heavier hydrocarbons (until C₅–C₉) in smaller amounts and traces of inorganic gases (Hunt, 1996; Clayton, 2005; Etiope, 2017). The relative ratios of these compounds in the emitted gas provide information on its origin. The ratio of methane (C₁) over ethane (C₂) plus propane (C₃) is referred as Bernard ratio or C₁/(C₂+C₃) (Bernard et al., 1976). Although we did not measure C₃, the C₂:C₁ ratios already provide constraints on the Bernard ratio values. In addition, our new measurements were supplemented by data provided by the local oil and gas operator or the published literature. The operator provided hydrocarbon measurements at 14 sites, and Bernard ratios from another 13 sites were obtained from Filipescu and Huma (1979). Only 5 locations from these data sets are within our target regions and are compared with our measurements using figures 7 and 8 in Milkov and Etiope (2018). The operator also provided a classification of the oil density for the reservoirs we visited during our study.

It is important to consider that there are no published data on the CH₄ isotopic composition of natural gas from oil and gas production in Romania; the only available isotopic data are from gas seeps in the Transylvanian (microbial gas) and Carpathian (thermogenic gas) basins, reported by Baciu et al. (2008), Etiope et al. (2009), Etiope et al. (2009a), and Baciu et al. (2018). We will compare our results also with these published signatures.

In thermogenic deposits, an increase in δ¹³C and δ²H in CH₄ is correlated with the maturity of the source rocks (degree of degradation of organic matter; Schoell, 1980). Regarding the hydrocarbon composition, C₂₊ content initially increases from low maturity (wet gas, associated with oil), evolving into a drier gas, with less or no C₂₊ hydrocarbons at high maturity. Microbially formed CH₄ is also dry and contains little or no C₂₊ hydrocarbons, but is significantly more depleted in ¹³C, and generally occurs in shallower deposits (Rice and Claypool, 1981; Whiticar, 1999). While mixing of different gases can change the isotopic CH₄ composition of the original end members, gas migration among reservoirs and to the surface, being substantially advective, does not modify the isotopic CH₄ ratio. Migration can, however, modify the C₁:C₂₊ ratio, through molecular fractionation (Etiope et al., 2009). This hypothesis will be discussed in relation with our results and interpretation.

The isotopic source signatures at 83 locations related to oil and gas production are shown in Figure 2. One additional location was characterized: CH₄ enhancements from cattle grazing in an open field. The isotopic signals related to fossil fuel operations range between –67.8 and –22.4 ‰ V-PDB for δ¹³C and from –259 to –138 ‰ V-SMOW for δ²H. The distinction between geological gas emissions and modern microbial sources is possible through the δ²H signatures: While δ¹³C values are similar for modern and fossil microbial gas, the δ²H values are generally lower for modern gas, which derives from fermentation pathway (Figure 2a). Although sampled at different locations (Figure 1), there is a good agreement between our results and the literature values from Baciu et al. (2018), who mainly sampled natural gas seeps in north Buzu, Vrancea, Gorj, and in Transylvania (δ¹³C from –67.4 to –29.0 ‰ V-PDB and δ²H from –228 to –145 ‰ V-SMOW; Figure 2a and b).

C₂:C₁ ratios were obtained at 412 facilities, including 340 oil and 62 gas production facilities. Results for all sites are presented in Figures 2c and S2. Commonly used natural gas classification distinguishes between wet gas (C₂:C₁ > 0.1), thermogenic dry gas (C₂:C₁ < 0.1), and microbial dry gas (C₂:C₁ < 0.001; Clayton, 2005; Etiope, 2017). All C₂:C₁ ratios ranged from 0 to 0.91, therefore representing different types of natural gas, without a clear distinction between oil and gas production sites.

In Figure 3, we classified the ground surface δ¹³C isotopic signals depending on the exploitation of oil or gas, as reported by the operator for each site. We compare them with signatures observed from leaks in the natural gas distribution network and ruminants that we also sampled during our campaign. The δ¹³C of the natural gas from the network falls in the range of most sampled gas extraction facilities (in regions D, I, and P2). The δ¹³C from the ruminants (–61.6 ± 0.4 ‰) falls in the same range of values, which are typical of microbial CH₄ formation. The most depleted δ¹³C signals measured at oil wells, with values < –50 ‰ V-PDB, can occur in the case of very early mature thermogenic formations, when thermogenic gas is mixed with adjacent microbial formations or when secondary microbial CH₄ is produced. The distinction shown in Figure 3 is further discussed below, along with the analysis of the data from the different regions.

We can distinguish the CH₄ isotopic signatures from the different geographical clusters we covered; our results for each of them are shown in Figure 4. The C₂:C₁ ratios are not available for the exact same locations than the CH₄ isotopic signatures because they were obtained by different survey teams, but we can compare them on the cluster scale. The geographical distribution of the isotopic signatures and C₂:C₁ ratios is shown in the supplementary material (Figure S2). We evaluated the clusters with coinciding measurements of isotopic and C₂:C₁ ratios in Figure 5. Because we measured only ethane and methane (C₂ and C₁), we only report maximal values of Bernard ratios.

Additional measurements from the regional operator and the literature allowed to characterize the CH₄ formation processes in the different oil or gas reservoirs in more detail. All the complementary data are presented in the supplementary material (Table S1, Figures S1, S3, and S4).

Most samples were collected in the district of Prahova (region P). The ranges of isotopic values and C₂:C₁ ratios in the clusters P1, P2, and P4 are very wide, which corresponds to a heterogenous thermal maturity of the oil reservoirs that are being processed there (Figure 4). Similar heterogeneity is also found in the clusters P6 and P7, showing C₂:C₁ ratios from 0.0021 to 0.61. However, we identified four clusters with comparatively consistent results:

• P2: δ¹³C values < –59 ‰ were measured around two natural gas facilities of the area. Five χ(CH₄) enhancements near gas wells were measured in cluster P2, with C₂:C₁ < 0.002. Such values provide evidence for the presence of natural gas of microbial origin (Figure 5). Nearby oil extraction facilities in region P2 have higher C₂:C₁ ratios, and higher δ¹³C-CH₄ values, typical for associated gas of thermogenic origin (Figure 5). Our measurements support the presence of multiple origins of the fossil fuel deposits in this cluster.

• P5: The hilly region east of Târgovişte (labelled P5) hosts a cluster of oil extraction facilities with well-defined isotopic composition of the emitted gas: δ¹³C between –50.6 and –37.3 ‰ V-PDB, δ²H between –251 and –199 ‰ V-SMOW. The relatively high C₂:C₁, between 0.056 and 0.37, confirms the presence of only wet thermogenic gas in the exploited reservoir.

• P9: This cluster displays the highest CH₄ enhancements (χ(CH₄) > 10⁶ ppb) with particularly enriched and highly consistent isotopic signals from oil installations (n = 16): average δ¹³C = –36.8 ± 2.2 ‰ and δ²H = –160 ± 10 ‰. No other gas composition measurements was made in this region.

• P1.1: In this subregion of P1, the isotopic signatures suggest an intermediate maturity thermogenic or secondary microbial formation (δ¹³C from –51.0 to –41.4 ‰ and δ²H from –221 to –185 ‰; Figure S4). We obtained C₂:C₁ values > 0.02 for most locations in this cluster (41 of 45), indicating a relatively large ethane proportion (Bernard ratio < 50). These locations were nearby oil extraction and production facilities, which confirm the presence of wet gas of thermogenic origin (Figure 5). A Bernard ratio of 11.9 was reported by the operator for an installation 10 km away from this cluster. Subsurface measurements within the cluster (Filipescu and Huma, 1979) show a reservoir of oil with Bernard ratios from 9 to 18 at ca. 2,000-m depth, also consistent with our surface measurements. Therefore, most deposits in this region are likely from thermogenic CH₄ formation, but can also be of secondary microbial origin.

The CH₄ isotopic signatures we measured in the Dolj and Ialomita districts (regions D and I) are well defined (Figure 4), with relatively enriched δ²H and depleted δ¹³C, especially in Dolj. In this region, all facilities are associated with gas and condensed gas extraction plants. The same is true for the facilities with the lowest δ¹³C values sampled in Ialomita. The δ¹³C-CH₄ signatures we obtained were < –57 ‰ for 11 of the 12 sampled locations in these two regions. Values of δ¹³C < –60 ‰ suggest the presence of dry gas with microbial origin (CO₂ reduction pathway), as it is found in Transylvania (see the results of aircraft samples). Additional data support the hypothesis of the microbial origin of the extracted natural gas in regions in Dolj and Ialomita:

• Dolj (D): For a borehole southeast of Craiova, Filipescu and Huma (1979) mentioned shallow deposits of dry gas (ca. 30 km from the sampled sites, 200- to 300-m deep), with Bernard ratios > 3,000. This would confirm the presence of large microbial gas reservoirs, in this region of major natural gas production.

• Ialomita (I): Some facilities in Ialomita are related to oil extraction, from where we sampled CH₄ with isotopic signatures showing a thermogenic origin. The distinction with gas facilities is confirmed by the C₂:C₁ results of this cluster: The values range from 0 to 0.76, with C₂:C₁ from 0 to 0.002 only found around gas installations and the higher ratios found around oil installations (Figure 5). Filipescu and Huma (1979) reported a vertical gas composition profile in the vicinity of the cluster where we carried out measurements, nearby the sampled cluster showing the presence of dry gas overlying associated gas with increasing C₂ and C₃ content (Bernard ratios of ca. 376 and 76 at –1,200 and –2,000 m, respectively). The gas sampled by the operator likely corresponds to this deeper layer of associated gas (Bernard ratio of 47). Thus, we conclude on microbial dry gas overlying oil deposits from our measurements in Ialomita; and this hypothesis matches the gas composition data.

The Teleorman region (Te) can be divided into west and east subareas (respectively, Te1 and Te2), separated by the river Teleorman, based on the isotopic signatures:

• Te1: This subcluster is characterized by the heaviest isotopic signatures found in this study, sampled around both oil and gas extraction facilities (n = 4): average δ¹³C = –28.1 ± 4.2 and δ²H = –149 ± 11 ‰ (Figure 4). These values correspond to very late maturity thermogenic gas (Figure 2a). The Bernard ratio reported by the operator for a close-by facility is between the values reported in Filipescu and Huma (1979), from associated gas at two different depths. Filipescu and Huma (1979) also reported the presence of nonassociated gas in the deepest layers (>2,000 m below surface), with a Bernard ratio of 452. Therefore, the facilities we sampled are likely to reach the deepest formations of high maturity deposits, at the stage where dry gas formation starts to occur.

• Te2: CH₄ from oil wells (n = 5) was more isotopically depleted, especially in ¹³C: average δ¹³C = –54.6 ± 4.4 ‰ and δ²H = –199 ± 6 ‰ (Figure 4). This suggests an intermediate maturity thermogenic, secondary microbial, or mixed (with primary microbial gas) origin of the CH₄. The origin cannot be further constrained by the large range of measured C₂:C₁ ratios (Figure 5). Low δ¹³C values (–58 ‰ and lower) usually indicate microbial CH₄ and were measured at two oil wells. C₂:C₁ ratios from our surface measurements indicate relatively low ethane amounts (values from 0 to 0.08, n = 17), in agreement with Bernard ratios between 10 and 300 reported for one facility in the region by Filipescu and Huma (1979). However, the values reflect a certain heterogeneity, confirmed by the variations in δ¹³C by up to 10 ‰, and the presence of different densities in the oil deposits: from medium to heavy, compared to only light deposits in Te1. Here, the hypothesis of mixed gas origins is the most likely, including the presence of microbial dry CH₄ reservoirs alongside wet gas reservoirs.

Although the χ(CH₄) excess above background was much lower than what was observed by the surface vehicles, we could characterize the isotopic source signals of 24 locations. The Keeling plots were rejected if:

• (1) the excess χ(CH₄) above background was <60 ppb (based on IRMS measurements), and

• (2) the r² of the regression fit was <0.5 for δ¹³C and δ²H.

All accepted Keeling plots are available in the Supplementary Material (Figure S5), as well as the resulting source signals (Table S2).

The map in Figure 6 shows the resulting isotopic signatures of the χ(CH₄) anomalies sampled from the aircraft. They are compared with typical ranges of CH₄ isotopic source signatures in Figure 4. δ¹³C ranged from –64.6 to –35.8 ‰ V-PDB, and δ²H from –404 to –127 ‰ V-SMOW, and the majority of them correspond to the emissions from fossil fuel extraction (Figure 2b). Based on the wind direction for each flight day and of the recorded altitudes, we could link 18 CH₄ enhancements to underlying oil and gas extraction facilities and 6 to “unknown sources” (Figures 2 and 6). The CH₄ enhancements of unknown origin sampled in the aircraft had the most depleted isotopic values: δ¹³C between –64.6 and –54.9 ‰ and δ²H between –404 and –239 ‰ and were generally observed at higher altitudes (up to 2,000 m above ground, Table S2) and outside the oil and gas production clusters. These isotope ranges correspond to CH₄ from microbial fermentation processes (Figure 2), therefore likely coming from agriculture activities, waste management, or natural wetlands. The CH₄ enhancements could, for example, be advected from densely populated areas (urban) or large agriculture facilities. Waste and agriculture emissions of biogenic origin account for 33% and 40%, respectively, of the 2015 CH₄ anthropogenic emissions in Romania (Crippa et al., 2019). In comparison, fossil fuels fugitive emissions were estimated at 21%. Measurements carried out in the city of Bucharest (Fernandez et al., 2022) showed that biogenic emissions, especially from urban wastewater, were surprisingly widespread, much more than in other European cities.

Most other isotopic signals determined on samples collected from the aircraft were in the range of signatures sampled from the oil and gas extraction facilities on the ground (Figure 2). Without considering the CH₄ enhancements of unknown origins, the average δ²H signal was –196 ‰ V-SMOW. Most of the emissions from oil and gas facilities we characterized from ground sampling are in the Prahova region. It is also the area where most samples were taken from the aircraft, because of the location of the airfield. However, the wind often advected emissions from region O (Figure 7), where very few ground measurements were made. The relative enrichment in deuterium isotopes in the samples from the aircraft confirms the fossil fuel origin of the responsible CH₄ emissions.

The three χ(CH₄) anomalies observed above Transylvania show relatively low δ¹³C values ( –63.2 ± 0.1, –62.3 ± 1.3, and –58.2 ± 1.3 ‰), typical of microbial CH₄ gas (Figure 2a), and δ²H of –273 ± 26, –223 ± 20 and –188 ± 13 ‰. δ²H values >–250 ‰ are typical for the CO₂ reduction pathway rather than fermentation (Milkov and Etiope, 2018) and are consistent with the isotopic data reported for gas seeps in the same basin (Baciu et al., 2018). Therefore, we likely sampled CH₄ sources of microbially formed natural gas reservoirs at the two locations with δ²H > –250 ‰. This is confirmed by the relatively depleted δ¹³C values that we also found in natural gas leaks sampled on the ground. The presence of microbial gas reservoirs in Transylvania has been documented, for example, by Filipescu and Huma (1979), Pawlewicz (2005), and Baciu et al. (2018). The CH₄ source of the third enhancement intercepted by the aircraft is more likely to be from agriculture or waste from urban settlements in the area.

Figure 7 shows the CH₄ enhancements sampled from the aircraft that we could link to emissions sampled from the ground based on the wind directions. In Te1 and I regions, the isotopic signal from the aircraft compares well with signatures from the oil installations below, and we can confidently identify the oil and gas activities in the underlying clusters as the source of the CH₄ enhancements observed from the aircraft. The distinct isotopic signatures measured in region Te1 (relatively enriched in heavy isotopes; see results from ground surface samples) are also found in the aircraft samples (Figure 7). In Prahova, we show the results of 2 of 8 CH₄ enhancements observed from the aircraft, where the wind clearly came from clusters of oil and gas installations where we also sampled on the ground. Also here, the isotopic signals derived from the aircraft samples agree fall in the range of the ones observed at the ground, although this is a region with heterogenous isotopic signatures, which makes it difficult to precisely link a specific cluster to the isotopic signals from the aircraft (Figure 7).

The largest χ(CH₄) anomalies observed from the aircraft were when flying above Ialomita, with values of 300 and 400 ppb above background. The underlying oil and gas cluster was visited several times by ground vehicles. The two locations with the lowest δ¹³C (–60.9 ± 0.032 and –67.8 ± 1.18 ‰) correspond to gas installations and the others are oil wells or processing plants. The CH₄ enhancements observed from the aircraft likely originate from a specific processing plant (oil deposit, gas compression, and others) that was circled at low altitude. This facility was sampled two times from the ground with δ¹³C values of –48.5 ± 0.45 and –56.5 ± 0.13 ‰. The δ¹³C signals from the aircraft are in a similar range: –57.0 ± 0.22 and –51.4 ± 0.17 ‰. The range of isotopic values from this particular site suggests the presence of several CH₄ sources, reflecting the different activities (storage, extraction, and gas recovery) and potential temporal variability (discussed in Section 5).

CH₄ enhancements were observed during each raster flight above the target areas (Figure 1), which include dense clusters of oil and gas installations. The isotopic signatures confirmed that the sampled emissions originate from the underlying oil and gas extraction activities. The total emissions from fossil fuel extraction clusters can be estimated from aircraft measurements, for example, using a mass balance approach (Hiller et al., 2014; Karion et al., 2015; Peischl et al., 2016; Schwietzke et al., 2017; Fiehn et al., 2020). Top-down approaches often rely on the use of regional-scale transport models in order to evaluate reported emission inventories (Xiao et al., 2008; Henne et al., 2016). Information on the CH₄ isotopic signatures of the observed enhancements from the aircraft can be used for source attribution (Fisher et al., 2017), to constrain the measurement-based quantifications, and to verify model simulations on the origin of the emissions.

Previous studies used CH₄ isotopic source signatures to constrain the CH₄ budget globally (Schaefer et al., 2016; Schwietzke et al., 2016; Worden et al., 2017) and regionally (Röckmann et al., 2016; Bergamaschi et al., 2018; Menoud et al., 2020a). Mean isotopic values assigned to fossil fuel related emissions in the literature range between –37 and –44 ‰ for δ¹³C-CH₄ (Schaefer et al., 2016; Schwietzke et al., 2016; Worden et al., 2017; Menoud et al., 2020b), and around –175 ‰ for δ²H-CH₄ (Lu et al., 2021; Menoud et al., 2021), the latter being based on a substantially smaller sample size. For the specific case of the Romanian Plain, we suggest the use of average values from aircraft measurements made in this study: δ¹³C = –49.7 ± 6.4 ‰ and δ²H = –189 ± 38 ‰, from identified, mainly oil-related, fossil fuel activities. The average values from ground measurements around fossil fuel production sites in this study are δ¹³C = –46.0 ± 11.2 ‰ and δ²H = –188 ± 28 ‰, and within the range of uncertainty of the aircraft values.

The average isotopic signatures from our study do not take into account emissions from the production of natural gas in Transylvania and the distribution network. From the flight covering part of Transylvania, we found more depleted δ¹³C signatures than in the Romanian Plain, but we didn’t collect data on CH₄ emissions samples on the ground. Regarding the gas network, Fernandez et al. (2022) measured significant emissions in Bucharest and with also a relatively depleted δ¹³C isotopic signature (confirmed by our measurements of two gas leaks; Figure 3). Therefore, the δ¹³C value for CH₄ emissions from all fossil fuel activities in Romania is likely to be lower than the –49.7 ‰ suggested above and further different to the global values commonly assigned.

The χ(CH₄) anomalies we observed from the aircraft were not all related to emissions from the oil and gas sector. The samples of “unknown” origin we collected from the aircraft, and 1 sample above Transylvania, had a distinct isotopic signal, with δ²H < –250 ‰, that corresponds to CH₄ of microbial fermentation origin (Figure 2). Indeed, there are many other sources of CH₄ emissions in Romania. According to the EDGAR v5.0 inventory (Crippa et al., 2019), the main anthropogenic CH₄ sources in Romania in 2015 were the agriculture sector (40% of total emissions, including manure management) and waste management (33% of total emissions). Microbial fermentation can also occur in stagnant freshwater; therefore, lakes, swamps, and bogs are potential natural sources, in addition to the well-established natural gas seeps and mud volcanoes (Etiope et al., 2011). Future measurements could be targeted at constraining emissions from nonfossil fuel sources.

Schwietzke et al. (2017) emphasized the influence of episodic release from fossil fuel extraction or processing sites on atmospheric measurements. In this study, this was illustrated by the 2 flights performed over Ialomita (region I; on October 11 and 14, 2019), where the emissions very likely originated from the same facility that was circled by the aircraft (Figure 7). The isotopic signatures were δ¹³C = –57.0 ± 0.22 ‰ and δ²H = –131 ± 15 ‰ on the 11th, and δ¹³C = –51.4 ± 0.17 ‰ and δ²H = –176 ± 14 ‰ on the 14th. The results are statistically different, which reflects the diversity of releases, even from 1 localized facility. To evaluate the emissions from our target regions more precisely, it is necessary to combine data from several flights performed on several days and at different seasons.

The ground surface results show that we can determine the origin of emitted natural gas based on CH₄ isotopic measurements at the surface. Whereas the signatures in some clusters show consistent signatures, results from other regions were very heterogenous (P1, P2, and P4; Figure 4). The C₂:C₁ ratios we measured are generally in agreement with the CH₄ isotopic signatures. To support our conclusions, further measurements at the extraction and processing facilities would be beneficial, not only for gathering more data on the hydrocarbon composition but also to account for the different types of emissions that come from one facility. Cardoso-Saldaña et al. (2021) showed that the C₂:C₁ ratios are generally higher in storage tank emissions. There are also variations in the CH₄ emissions of a facility because of the presence of different sources, for example, storage tanks do not emit continuously. Our approach of measuring the CH₄ enhancements downwind of the facilities is sensitive to the variations in sources and their composition.

The analysis of other variables, especially total hydrocarbon contents, CO₂ contents, and the isotopic composition of CO₂ and C₂H₆ (Milkov, 2011; Milkov and Etiope, 2018), could help to further reduce the uncertainties in the origin of the gas emissions we investigated here. Nevertheless, we showed that atmospheric measurements allowed to draw conclusions on the geological origin of exploited gas reservoirs in the subsurface.

As part of the ROMEO project, we characterized CH₄ emissions from 83 oil and gas production sites for source isotopic signatures and 412 for C₂:C₁ ratios. The δ¹³C and δ²H isotopic signatures were also determined at 24 locations sampled from an aircraft. Our data show that over the target areas, most CH₄ significant enhancements that were encountered originated from the underlying oil and gas production activities. Thus, CH₄ emissions from oil and gas extraction activities are the main emission source for such large point sources within our target areas. The CH₄ isotopic composition over a certain area can sometimes be heterogeneous, but the distinction from other source categories than from oil and gas was still possible. This possibility of source attribution will support the quantitative interpretation of top-down emission estimates based on high-precision methane measurements that were also carried out on these and other flights in the same region.

Our results allow to characterize the origin of natural gas in several regions from atmospheric measurements. Both isotopic signatures and C₂:C₁ ratios indicated that most visited sites, related to oil production, emitted thermogenic gas. We also identified some microbial reservoirs, especially around gas production sites. New findings include the presence of microbial gas in at least four production areas of the Romanian Plain, located in the counties of Prahova, Ialomita, Dolj, and Teleorman. We also confirmed the presence of microbial, CH₄ emissions due to the production of natural gas in Transylvania.

The average isotopic signatures for CH₄ emissions from fossil fuel production over the Romanian Plain were –49.7 ± 6.4 ‰ V-PDB for δ¹³C and –189 ± 38 ‰ V-SMOW for δ²H, based on the measurements from the aircraft. The δ¹³C value possibly overestimates the average δ¹³C from fossil fuel activities in the whole country of Romania because it does not include emissions from natural gas production in Transylvania but is already lower than commonly used values for global fossil fuel emissions (Schwietzke et al., 2016). The global database of gas composition data made by Sherwood et al. (2017) did not include measurements made in Romania. Yet it is not the only place with relatively depleted δ¹³C-CH₄ values in geological deposits: The database as well as recently published studies (Zazzeri et al., 2016; Lu et al., 2021) reported δ¹³C values < –60 ‰ V-PDB in emitted CH₄ from coal or conventional gas exploitation in the United States, Australia, Poland, Japan, and other countries. Therefore, it is crucial to take into account the geographical variability of isotopic signatures when they are used to constrain the global or regional CH₄ budget.

All of the data used in this study are publicly available on a Zenodo repository (https://doi.org/10.5281/zenodo.6319952).

The supplemental files for this article can be found as follows:

• Table S1. Gas composition data measured at several extraction facilities by the operator in the regions where measurements were also carried out in this study.

• Table S2. Isotopic source signals of all χ(CH₄) anomalies sampled from the aircraft and considered significant (excess χ(CH₄) < 60 ppb and r² < 0.5).

• Figure S1. Example of flight pattern.

• Figure S2. Results from ground surface sampling around oil and gas facilities in regions P and Te2.

• Figure S3. Gas composition data in the Romanian Plain.

• Figure S4. Genetic diagrams of gas molecular and isotopic compositions, per cluster.

• Figure S5. Keeling plots of all sampled CH₄ enhancements from the aircraft.

We specially thank all other members of the ROmanian Methane Emissions from Oil & Gas team for their help during the campaign: Huilin Chen, Dominik Brunner, Oana Pîrvu, Mackenzie Smith, Niall Armstrong, Patryk Lakomiec, Sylvia Walter, Aurel Constantin, Sebastian Iancu, Alex Nica, Sorin Ghemulet, Constantin Visoiu, Cristian Pop, Alexandru Pana, Alexandru Tudor, Mihai Profir, Alin Scarlat, Lucian Cusa, Marius Corbu, Georgiana Grigoras, Sorin Vajaiac, Denisa Moaca, Vincent Edjabou, Julia Wietzel, Jaroslaw Necki, Pawel Jagoda, Jakub Baryzel, Andrei Radovici, Horatiu Stefanie, Alexandru Mereuta, Artur Ionescu, and Daniel Zavala-Araiza.

These surveys were funded by the ROmanian Methane Emissions from Oil & Gas project, funded by the Climate and Clean Air Coalition (CCAC) of UNEP under grant number PCA/CCAC/UU/DTIE19-EN652. This research project has also received funding from the European Union’s Horizon 2020 Research and Innovation Program under the Marie Sklodowska-Curie Grant Agreement 722479 (MEMO², https://h2020-memo2.eu).

The authors have no competing interests, as defined by Elementa, that might be perceived to influence the research presented in this article.

Contributed to conception and design: MM, TR.

Contributed to acquisition of data: MM, CV, HM, AH, IV, PB, AD, PK, MA, AC, CB, CS, MS, TR.

Contributed to analysis and interpretation of data: MM, CV, HM, GE, CB, TR.

Drafted and/or revised this article: MM, HM, GE, MS, TR.

Approved the submitted version for publication: MM, CV, HM, AH, IV, PB, AD, PK, SS, MA, AC, GE, CB, CS, MS, TR.

How to cite this article: Menoud, M, van der Veen, C, Maazallahi, H, Hensen, A, Velzeboer, I, van den Bulk, P, Delre, A, Korben, P, Schwietzke, S, Ardelean, M, Calcan, A, Etiope, G, Baciu, C, Scheutz, C, Schmidt, M, Röckmann, T. 2022. CH₄ isotopic signatures of emissions from oil and gas extraction sites in Romania. Elementa: Science of the Anthropocene 10(1). DOI: https://doi.org/10.1525/elementa.2021.000092

Domain Editor-in-Chief: Detlev Helmig, Boulder AIR LLC, Boulder, CO, USA

Guest Editor: Arvind P. Ravikumar, The University of Texas at Austin, Austin, TX, USA

Knowledge Domain: Atmospheric Science

Part of an Elementa Forum: Oil and Natural Gas Development: Air Quality, Climate Science, and Policy