Oil and natural gas wells across the NASA ABoVE domain: fugitive methane emissions and broader environmental impacts

Arctic-boreal regions are experiencing major anthropogenic disturbances in addition to intensifying natural disturbance regimes as a consequence of climate change. Oil and natural gas (OG) activities are extensive in the Arctic-boreal region of western North America, a large portion of which is underlain by permafrost. The total number and distribution of OG wells and their potential fate remain unclear. Consequently, the collective impacts of OG wells on natural and cultural resources, human health and emissions of methane (CH4), are poorly understood. Using public OG well databases, we analysed the distribution of OG wells drilled between 1984 and 2018 across the Core Domain of the NASA Arctic-Boreal Vulnerability Experiment (‘ABoVE domain’). We identified 242 007 OG wells drilled as of 2018 in the ABoVE domain, of which almost two thirds are now inactive or abandoned OG wells. We found that annual drilling has increased from 269 to 8599 OG wells from 1984 to 2014 with around 1000, 700 and 1800 OG wells drilled annually in evergreen forest, deciduous forest and herbaceous land cover types, respectively. 65 588 OG well sites were underlain by permafrost in 2012. Fugitive CH4 emissions from active and abandoned OG wells drilled in the Canadian portion of the ABoVE domain accounted for approximately 13% of the total anthropogenic CH4 emissions in Canada in 2018. Our analysis identified OG wells as an anthropogenic disturbance in the ABoVE domain with potentially non-negligible consequences to local populations, ecosystems, and the climate system.


Introduction
Arctic-boreal regions are experiencing intensifying disturbance regimes through accelerating climate change and anthropogenic activities (Foster et al 2022). Large numbers of oil and natural gas (OG) wells have been drilled and are being drilled in these regions. The direct and indirect impacts of OG drilling include emissions of methane (CH 4 ), the primary constituent of natural gas, physical ecosystem disturbance due to the clearing of vegetation, soil compaction, air quality degradation from emissions of volatile organic compounds, nitrous and sulphur oxides, and soil and groundwater contamination due to leaks of saline water, various hydrocarbons and/or radionuclides (Dabros et Williams et al 2021). Here, we consider OG wells as anthropogenic disturbances themselves, but also as a potential source of fugitive CH 4 emissions to the atmosphere and a proxy for broader OG development related disturbances (e.g. access roads, waste handling and disposal, processing).
Focusing on western North America, the NASA Arctic-Boreal Vulnerability Experiment (ABoVE) aims to provide a mechanistic understanding of rapidly changing Arctic and boreal biomes, and their present and future ecosystem services (Fisher et al 2018, Miller et al 2019). A recent study found that 13.6% of the ABoVE Core Domain ('ABoVE domain') experienced a shift in land cover type between 1984 and 2014 (Wang et al 2020). Two dominating modes of land cover change have been identified, namely a notable loss in 'Evergreen Forest' and gain in 'Deciduous Forest' classes in the boreal biome, and gains in 'Shrubland' and 'Herbaceous' classes in the Arctic biome. Wildfire and climate change were identified as the main causes of these land cover changes, respectively, both with nonnegligible local, regional and global consequences. However, the role of OG wells, and of OG development more broadly, in these land cover changes remains largely understudied.
A large portion of the ABoVE domain is underlain by permafrost, i.e. perennially frozen ground (Gruber 2012, Biskaborn et al 2019. In northern Canada, permafrost temperatures are rising at rapid rates of 0.1 • C-0.5 • C per decade, leading to a deepening of the hydrologically and biogeochemically active layer and to changes in land cover types (Fraser et al 2018, Derksen et al 2019. Collectively, such drastic changes in landform and/or land cover affect many physical, ecological, and biogeochemical processes governing ecosystem composition, structure, functioning and services (Schuur et al 2015, Dabros et al 2018, Turestsky et al 2020. How these specific processes may be additionally influenced by anthropogenic activities and infrastructures, such as OG wells, remains largely unknown (Foster et al 2022).
Up to 26% of OG wells in Canada and the U.S. have well integrity failures that result in leakage (Wisen et al 2020, Abboud et al 2021, Lackey et al 2021. Permafrost thaw can lead to ground surface subsidence and collapse of soil structures ('thermokarst'), causing additional stress on the OG wells (Vaganova and Filimonov 2015, Lukyanov et al 2019, Mikhienkova et al 2020, potentially affecting the well integrity and creating additional pathways for fugitive CH 4 emissions and contaminants (figure 1). At the same time, OG wells may alter the thermal regime of their direct surroundings due to heat release from various well components, which may accelerate permafrost thaw (figure 1; Vaganova and Filimonov 2015, Lukyanov et al 2019, Mikhienkova et al 2020. This is especially true during the OG well drilling process, where warm mud is used to drill into the frozen ground, creating a thawed zone of up to 10 m radius in the direct vicinity of the borehole (Eppelbaum et al 2019, Chuvilin et al 2022. Overall, how OG wells and their direct surroundings may interact with permafrost thaw, exacerbate leakage and associated environmental and ecosystem impacts is currently understudied. OG wells represent a major source of CH 4 (Omara et al 2016, Zavala-Araiza et al 2018, MacKay et al 2021, Townsend-Small and Hoschouer 2021), a short-lived but potent greenhouse gas with a global warming potential of 82.5 over a 20 year time frame (Forster et al 2021). Several studies have quantified CH 4 emissions from the OG sector and have shown that current national greenhouse gas inventories are likely underestimating emissions from OG wells (Alvarez et al 2018, Johnson et al 2017, Tyner and Johnson 2021, Lyon et al 2021, MacKay et al 2021. Fugitive OG well CH 4 emissions arise from venting and flaring processes, as well as from leakage from OG well infrastructure (Omara et al 2016, Zavala-Araiza et al 2018, Kang et al 2021. When an OG well is no longer economical to operate, OG production stops and the well is abandoned. State, provincial, and territorial regulations require the plugging of abandoned OG wells, which involves isolating groundwater aquifers and OG reservoirs through cementing (King and Valencia 2014). Methane emissions from both active and inactive/abandoned OG wells remain highly uncertain, but were estimated to account for 32.6% and 19.3% of the national anthropogenic CH 4 emissions in Canada and the U.S., respectively (Environment and Climate Change Canada ECCC 2022, U.S. Environmental Protection Agency EPA 2022). These emission estimates include emissions from OG wells and supporting infrastructure (e.g. batteries, storage tanks, pneumatic devices). Given the importance of CH 4 emissions from OG wells, the availability of technical and economical mitigation options (Ocko et al 2021), and the potential for CH 4 emissions to serve as a proxy for broader environmental impacts, it is important to quantify CH 4 emissions from OG wells for the ABoVE domain.
The goal of this study was to identify OG well site locations in the ABoVE domain in the context of climate change-induced land cover change, permafrost thaw, and potentially non-negligible contributions to the atmospheric CH 4 burden. To meet this goal, we compiled publicly available information on active and abandoned OG well site locations and characteristics (i.e. production type, drilling date) to map OG well distribution across the ABoVE domain over time in relation to land cover change occurring between 1984 and 2014. We also evaluated OG well site locations with respect to permafrost distribution in 2012. In addition, we estimated annual CH 4 emissions from active and abandoned OG wells in the ABoVE domain in 2018 using a gridded national inventory of anthropogenic CH 4 emissions (Scarpelli et al 2021) and published emission factors for OG wells (Williams et al 2021).

OG well database for the ABoVE domain
We collected publicly available information on drilled OG wells in six Canadian Provinces and Territories, namely Alberta, British Columbia, Manitoba, Saskatchewan, the Northwest Territories, and the Yukon, as well as in one U.S. state, Alaska. (table S1). Each jurisdiction provides some information on well status (i.e. active, abandoned not plugged, abandoned and plugged) and well production type (i.e. gas producing, OG producing or unknown) (Kang et al 2021). However, the databases are often missing some attributes and employ differing terms to describe the OG wells. We carefully screened them to develop a harmonized OG well database with common terminology for well status, type and drilling date (SI-1).

OG well distribution across the ABoVE domain
We mapped OG wells according to their drilling date between 1984 and 2014 on the permafrost zonation index, an indicator of permafrost occurrence probability (Gruber 2012) and a recent land cover and land cover change data product  for the ABoVE domain (Wang et al 2020).

OG wells and permafrost
We classified the permafrost occurrence probability raster dataset into four permafrost zones (Helbig et al 2016), i.e. isolated (permafrost zonation index < 0.1), sporadic (0.1 ⩽ permafrost zonation index < 0.5), discontinuous (0.5 ⩽ permafrost zonation index < 0.9) and continuous (permafrost zonation index ⩾ 0.9), and derived OG well counts in each of them. Since the map was published in 2012, only OG wells drilled up to 2012 were included in this count. Wang et al (2020) used Landsat satellite imagery to detect breaks in land surface reflectance and assign land cover labels to statistical clusters of stable land cover over 1984-2014. The ABoVE domain was classified into 15 domain-specific land cover classes, further simplified into 10 classes (table S3, SI-4). For each year between 1984 and 2014, we mapped the annual number of drilled OG wells onto the simplified land cover map of the given year. We extracted the land cover class at each OG well site location to determine the land cover class in which OG wells were drilled and how annual OG well drilling changed over the years in each class. We included three North American Level II Ecoregions (CEC 2021) where most OG wells were located, namely the Boreal Plain, Taiga Plain and Alaska Tundra ecoregions. Ecoregions are regions showing similarities in ecosystem composition and environmental resources (Commission for Environmental Cooperation (CEC) 2021).

Fugitive methane emissions from OG wells
For comparison purposes with other Canadian emission estimates, we focused only on CH 4 emission from OG wells from the Canadian portion of the ABoVE domain (excluding Alaska). We extracted annual fugitive CH 4 emissions from active and inactive/abandoned OG wells in this domain from the Canadian gridded inventory (Scarpelli et al 2021). To provide context for OG well CH 4 emissions, we also extracted CH 4 emissions from high greenhouse gas emitting facilities reported in the Canadian Greenhouse Gas Reporting Program (ECCC 2019). This program compiles annual greenhouse gas emissions from facilities across various sectors (e.g. OG extraction facilities, oil sands mines, coal mines, pulp and paper mills, wastewater treatment facilities) emitting more than 10 kilotons CO 2 -equivalence, a unit accounting for the global warming potential of various greenhouse gases. For example, they used a global warming potential of 25 over a 100 year time frame for CH 4 , meaning that 1 ton of CH 4 amounted to Additionally, we estimated CH 4 emissions from abandoned OG wells in the ABoVE domain in Canada by multiplying the number of OG wells drilled up to 2018 with published emission factors developed for abandoned OG wells, which are reported in table S4 (SI-2, Williams et al 2021). Emission factors are reported for gas producing and OG producing wells, based on CH 4 flow rate measurements made on gas producing wells (n = 148) and OG producing wells (n = 196) with reported fluid production types. Additionally, Williams et al (2021) reported emission factors for wells with no reported fluid production type ('all plugged' and 'all unplugged'), which are averages of measurements performed on gas wells (n = 148), OG wells (n = 196) as well as measurements performed on wells with unknown fluid production type (n = 254). Additional details on emission factors derived by Williams et al (2021) are provided in the supplementary information (SI-2). Canada-wide estimates of abandoned OG well CH 4 emissions for the year 2018 were retrieved from Williams et al (2021) and the Canadian National Inventory Report (ECCC 2022).

OG well database
The total number of OG wells drilled in the ABoVE domain was 242 007 as of 2018. The majority were located in Alberta (n = 204 496) and British Columbia (n = 29 502) (table S5). 6778, 654, 501 and 76 OG wells were located in Alaska, the Northwest Territories, Saskatchewan and the Yukon, respectively. In Manitoba, no OG wells were located inside the ABoVE domain. OG wells in the ABoVE domain accounted for more than 30% of all OG wells in Canada (n = 787 553; Kang et al 2021). Sixty-three per cent of all reported OG wells in the domain were abandoned (n = 152 790) and 19% of abandoned OG wells in the domain were unplugged (n = 29 582). The highest proportion of unplugged OG wells was found in British Columbia (30% of all abandoned OG wells in the domain were unplugged) and all abandoned OG wells in the domain in Saskatchewan were plugged. Most active OG wells in the domain were producing gas (30%), compared to 20% of OG producing wells and 50% of wells with unknown fluid production type. As for abandoned OG wells in the domain, 24% were gas producing, 15% were OG producing and 61% were of unknown fluid type.

OG well distribution across the ABoVE domain
A substantial increase in drilled OG wells occurred during the 2000s, with OG wells drilled annually in the ABoVE domain in British Columbia and Alberta increasing from around 400-700 in the 1980s and 1990s to 2000-7000 in the 2000s (figure 2). As of 2014, 61% of the OG wells in the domain were abandoned, of which 20% were left unplugged.

OG and permafrost
In 2012, 37% of the ABoVE domain was underlain by continuous permafrost, while from north to south, 24%, 24% and 7% was underlain by discontinuous, isolated and sporadic permafrost, respectively (figure 3). However, most OG wells were located in the permafrost-free portion of the ABoVE domain (n = 146 242; 69%). Nevertheless, 65 588 OG wells were located in permafrost zones, of which 24% were located in the sporadic permafrost zone, followed by the isolated (4%) and continuous (3%) and discontinuous (<1%) permafrost zones (figure S1).

OG wells and land cover
Between 1984 and 2014, most OG wells were drilled in the 'Herbaceous' , 'Sparsely Vegetated' and 'Barren' land cover classes, accounting for 27%, 19% and 19% of total OG wells drilled in the ABoVE domain, respectively. The 'Fen' , 'Bog' and 'Shallows' land cover classes underwent the smallest amount of drilling during the study period (3%, 0.4% and 0.1% of total OG wells, respectively) ( figure 4(a)). Relating the OG well drilling to the percentage area of each land cover class in the ABoVE domain, we found that land cover classes occupying larger areas in the ABoVE domain did not consistently correspond to land cover classes with the highest proportion of drilled OG wells, indicating that the OG wells are over-represented in 'Herbaceous' , 'Sparsely Vegetated' and 'Barren' land cover classes, and under-represented in 'Evergeen Forest' , 'Shrubland' and 'Shallows' land cover classes. For example, the 'Barren' land cover class which underwent 19% of the total OG well drilling in the ABoVE domain only accounted for 10% of the total area of the ABoVE domain, whereas, the 'Evergreen Forest' land cover class, accounting for 26% of the total area of the ABoVE domain, contained only 12% of the total OG wells drilled between 1984 and 2014 in the ABoVE domain ( figure 4(a)).
Investigating how OG well drilling evolved throughout the study period, we found that most OG wells in the 1980s and 1990s were drilled in 'Herbaceous' , 'Sparsely Vegetated' and 'Barren' classes. A continued increase in the number of annually drilled OG wells started in the early 2000s, with around 1000-2000 OG wells drilled per year in the previously mentioned land cover classes. Starting in 2006, annual OG well drilling in the 'Evergreen Forest' class increased, overtaking 'Barren' and also underwent a significant increase in drilled OG wells from 2009 to 2014 (figure S3, SI-5). We found a large increase in OG wells drilled between 1984-1999 and 2000-2014 in 'Herbaceous' (+10 090), 'Sparsely Vegetated' (+12 347) and 'Barren' (+14 156) land cover classes ( figure 4(b)). The only land cover class undergoing a decrease in the number of OG wells drilled between 1984-1999 and 2000-2014 was the 'Shallows' land cover class (−77). Combining our well counts from the harmonized OG well database for the ABoVE domain and emission factors from Williams et al (2021), we estimated annual CH 4 emissions from abandoned OG wells in the Canadian portion of the ABoVE domain ranging from 0.003 Tg CH 4 per year (minimum emission factor, 'method 1') to 0.018 Tg CH 4 per year (maximum emission factors, 'method 2') (figure 5). Therefore, our estimate for abandoned OG wells in the Canadian portion of the ABoVE domain is 1.3-7.7 times larger than the estimate using the gridded inventory (Scarpelli et al 2021). Methane emission estimates from abandoned OG wells in the Canadian portion of the ABoVE domain corresponded to 32%-183% of the Canadian nationwide estimate for abandoned OG wells in 2018 (0.01 Tg CH 4 per year; Environment and Climate Change Canada-Greenhouse Gas Division 2022).

Fugitive methane emissions from OG wells
Even though most abandoned OG wells are plugged in the ABoVE domain (80%), CH 4 emissions from unplugged OG wells (which account for 20% of Figure 5. Methane emissions estimates from various sources in the Canadian portion of the ABoVE Core Domain: abandoned OG wells in the ABoVE Core Domain (grey dots represent our six estimates (further clarification in SI-2)), active and abandoned OG wells (Scarpelli et al 2021), facilities from various sectors (e.g. oil and natural gas, pulp and paper, coal and waste sector) emitting more than 10 kilotons of CO2-equivalence reported in the Canadian Greenhouse Gas Reporting Program (Environment and Climate Change Canada-Greenhouse Gas Reporting Program 2019) and wetlands (Bloom et al 2021). abandoned OG wells) represent 60%-65% of emission from abandoned OG wells, because unplugged wells generally emit more than plugged wells.
Abandoned OG wells in Alaska represent 1.8% of abandoned OG wells in the ABoVE domain and annual emissions amounted to 5.3 × 10 −5 Tg CH 4 per year to 8.9 × 10 −5 Tg CH 4 per year. Including abandoned OG well emissions occurring in Alaska only increases our estimate of annual CH 4 emissions from abandoned OG wells in the ABoVE domain by 0.4%-1.7%. To compare our estimate to other Canadian CH 4 emission estimates, we decided to exclude emissions occurring in Alaska.
For context, we compare CH 4 emissions from OG wells in the Canadian portion of the ABoVE domain to two other CH 4 emission estimates. We found methane emissions from all OG wells in the ABoVE domain were 213% higher than the Canadian Greenhouse Gas Reporting Program-reported CH 4 emissions occurring inside the ABoVE domain, amounting to 0.153 Tg CH 4 per year (figure 5). However, annual wetland CH 4 emissions (WetCHARTs) for the ABoVE domain were 61 Tg CH 4 per year in 2018 (figure 5).

OG well distribution in the ABoVE domain
We identified more than 242 007 OG wells drilled in the ABoVE domain as of 2018. The Boreal Plain and Taiga Plain ecoregions correspond approximately with the central and northern portions of the Western Canadian Sedimentary Basin, one of the world's largest OG-producing basins (MacKay and Pedersen 2022). OG well drilling has been ongoing in the Western Canadian Sedimentary Basin since the early 1900s (Gray 2005), and the rate of OG well drilling has increased each year, reaching up to 9000 annually drilled OG wells in the early 2010s. From 2017 to 2020, approximately 4000-5000 wells were drilled annually in the ABoVE domain. It is important to note that our OG well database was developed based on publicly available well databases published by each state, province and territory. The databases include geographic coordinates but do not specify the precision and accuracy of these coordinates, which can become problematic when comparing OG well site location datasets to other spatial data products. Therefore, there is a need for field-based studies to verify site locations and improve OG well databases.

Impacts of OG wells on ecosystems across Arctic-boreal regions are understudied
The Arctic-boreal region of western North America is especially vulnerable to disturbances and is experiencing fundamental land cover changes (Wang et al 2020, Carpino et al 2018. Most OG wells were drilled in the 'Evergreen Forest' , 'Sparsely Vegetated' and 'Barren' classes (figure 4). Wang et al (2020) reported a significant loss of the 'Evergreen Forest' class attributed mainly to the changing fire regimes, but the contribution of OG activities to this loss remains unclear. Moreover, 91% of the ABoVE domain is underlain by permafrost and 65 588 active and inactive/abandoned OG wells were located in permafrost zones in 2012. Permafrost is thawing at rapid rates (Biskaborn et al 2019). The resulting landform and/or land cover changes might affect active and abandoned OG well integrity, causing leakage of CH 4 and other contaminants to the surrounding soil, groundwater and the atmosphere (figure 1).

Methane emissions from OG wells are often underestimated
OG wells emit non-negligible> quantities of fugitive CH 4 emissions to the atmosphere (Omara et al 2016, Zavala-Araiza et al 2018, Kang et al 2021. Active and inactive/abandoned OG well CH 4 emissions in the Canadian portion of the ABoVE domain accounted for 13% of the total anthropogenic CH 4 emissions in Canada in 2018, based on the gridded methane inventory for Canada (Scarpelli et al 2021). However, this estimate is likely to be an underestimate. One source that we found to be underestimated is abandoned OG wells, which represent a large proportion of wells in the ABoVE domain. Moreover, current fugitive CH 4 emission estimates from OG wells in national inventories do not consider the potential role of enhanced emissions from permafrost thaw leading to preferential migration and additional fugitive emissions from OG wells.
In 2018, 63% of all drilled OG wells in the ABoVE domain were abandoned and the number of abandoned wells will continue to increase. Moreover, abandoned OG wells are generally not actively monitored/managed and their role on ecosystems in the Arctic-boreal region remain poorly understood. Therefore, there is a need to better characterize abandoned OG well attributes and accurately estimate CH 4 emissions from abandoned wells. Plugged OG wells generally emit less CH 4 than unplugged OG wells (Kang and Jackson 2016, Saint Vincent et al 2020); however, many abandoned OG wells remain unplugged, were plugged improperly or were abandoned before modern plugging technology and regulations became available (King and Valencia 2014). To derive our estimate of CH 4 emissions from abandoned OG wells, we characterized the wells in the public well databases by well production type, status, and drilling date. However, a large proportion of OG wells in the ABoVE domain had unknown production type (60%), which is consistent with datasets across the United States (Boutot et al 2022). Improved record-keeping of OG well characteristics, in addition to accurate location information, is necessary to better estimate OG well CH 4 emissions and to understand the broader environmental impacts of these wells.

Conclusion
The southern portion ABoVE domain is a major OG producing region, with 242 007 OG wells drilled as of 2018. Most of the wells were drilled in 'Evergreen Forest' , 'Sparsely Vegetated' and 'Barren' classes, suggesting that OG activities might be contributing to land cover changes in the ABoVE domain, particularly to 'Evergreen Forest' class loss. We found that about 65 588 OG wells were drilled in permafrost in 2012, which covers 91% of the ABoVE domain. Potential interactions between OG wells and permafrost, such as the presence of a thawed zone or thermokarst formation in the direct surrounding of the OG well, can release heat or additional stress on OG wells (figure 1). We found that fugitive CH 4 emissions from OG wells in the Canadian portion of the ABoVE domain accounted for 30% of the total anthropogenic CH 4 emissions in Canada and may be an important consideration in evaluating CH 4 emissions for all sources in the ABoVE region. This study provides evidence for an overlooked anthropogenic disturbance type that may exacerbate the rapid changes occurring across ecosystems in some portions of the Arctic-boreal region of western North America.

Data availability statement
The data cannot be made publicly available upon publication because they are not available in a format that is sufficiently accessible or reusable by other researchers. The data that support the findings of this study are available upon reasonable request from the authors.