Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2025 Apr 29;59(18):9008–9016. doi: 10.1021/acs.est.4c05602

Sevenfold Underestimation of Methane Emissions from Non-producing Oil and Gas Wells in Canada

Louise A Klotz 1, Liam Woolley 1, Bianca Lamarche 1, Jade Boutot 1, Mary Kang 1,*
PMCID: PMC12080243  PMID: 40299828

Abstract

graphic file with name es4c05602_0005.jpg

Millions of non-producing oil and gas wells around the world are leaking methane and other contaminants, contributing to increased greenhouse gas emissions and polluting our water, soil, and air. Quantifying methane emissions and understanding the attributes driving these emissions are important for evaluating the scale of the environmental risks and informing mitigation strategies. With our national-scale direct measurement database of 494 non-producing wells across Canada, we find total annual methane emissions from non-producing wells in Canada to be 230 kt/year (51–560 kt/year) for 2023, which is 7 (1.5–16) times higher than estimated in Canada’s National Inventory Report (34 kt/year) and accounts for 13% of total fugitive emissions from oil and natural gas systems in Canada. We show that the role of well attributes in methane emissions is best evaluated by considering the emitting component (wellhead/surface casing vent) and the spatial scale (e.g., national, provincial, subprovincial). Large uncertainties in methane emissions from non-producing wells can be reduced not only with additional measurements but also with detailed well attribute analysis using direct measurements. Identifying attributes linked to high emitters can also be used to prioritize mitigation, thereby reducing methane emissions and broader environmental risks.

Keywords: methane emissions, oil and gas wells, abandoned, well attributes, leakage

Short abstract

Methane emissions from nonproducing oil and gas wells are highly uncertain and are underestimated in current national inventories. Understanding well attributes governing methane emissions can help reduce uncertainties in emission estimates and inform the prioritization of mitigation efforts.

Introduction

Non-producing oil and gas wells can act as subsurface leakage pathways connecting oil and gas reservoirs to surrounding ecosystems, resulting in greenhouse gas emissions and other impacts on groundwater, soil, and air quality.1,2 There are more than 400,000 non-producing oil and gas wells in Canada, which represent more than 70% of all oil and gas wells in the country.1 There are also more than 3,000,000 non-producing oil and gas wells in the U.S., and millions more globally. Non-producing wells are one of the most uncertain sources of oil and gas methane emissions in Canada’s National Inventory Report (NIR),3 the U.S. Greenhouse Gas Inventory (GHGI),4 and other national inventories due to a lack of direct measurements and uncertainties in the number of wells.5,6 Non-producing well methane emissions have been shown to follow highly skewed distributions, with a small proportion of high-emitting wells dominating total emissions.5 Well attributes linked to these high methane-emitting non-producing wells remain unclear, and results from different studies are often conflicting.7 Understanding the factors driving high methane emissions is key to reducing uncertainties associated with non-producing well emissions, developing actionable mitigation strategies, and minimizing broader environmental impacts.

Many previous studies that have investigated the association between well attributes and methane emissions have found different results.7 The most common well attributes that are investigated are well status (active, plugged, unplugged), geographic location, fluid type (oil, gas, other), wellbore deviation, well depth, spud date, well density, operation mode (conventional or unconventional), and operator.7−12 The few attributes shown to consistently impact well integrity are geographic location,9,10,13 plugging status (unplugged wells emit more than plugged wells),10,11,14 and wellbore deviation (deviated wells emit more than vertical wells).9,10,12 Meanwhile, depth, age, fluid type, well density, operator, and other well attributes show inconsistent results.8−18 The large discrepancies among studies may be due to the use of different measurement methods, the fact that most well attribute studies focus on specific regions (Alberta, Pennsylvania, Colorado)6,11,14,18 or inconsistencies and gaps in operator-reported datasets.6

Methane emissions at non-producing well sites can originate from the above-ground wellhead infrastructure (referred to as wellhead emissions in this study), the surface casing vent (SCV) (referred to as SCV emissions in this study), or the surrounding soil due to gas migration. Here, we focus on the first two sources, as gas migration is a small portion of methane emissions from upstream oil and gas operations (about 0.1% in Alberta in 2021).19 A recent study conducted in Alberta and Saskatchewan addressed the importance of differentiating wellhead and SCV emissions when conducting measurements, as the underlying emissions processes and repair strategies differ substantially depending on the emitting component.6 In addition, these two sources are currently reported separately in Canada’s NIR: wellhead emissions are estimated using status-specific/type-specific (plugged/unplugged/unknown and gas/oil/other/unknown) emission factors developed based on direct measurements in British Columbia, Alberta, and Ontario5,6,20 and SCV emissions are estimated based on industry reporting for Alberta and British Columbia21,22 or extrapolated from two major studies on emissions from the upstream oil and gas sector.23,24 The U.S. Environmental Protection Agency (EPA) currently reports non-producing well emissions in the GHGI using status-specific emission factors from two main studies and does not specifically account for SCV emissions.11,25

Here, we develop a database of methane flow rate measurements at 494 non-producing wells across five provinces in Canada, which is the largest measurement database using a consistent methodology. This database also includes 105 previously unmeasured wells. We compile well attributes such as well status (plugged/unplugged), geographic location, fluid type (oil/gas/other/unknown), well depth, spud date, wellbore deviation, operator, and well density, and analyze them with respect to measured methane flow rates. We explore how various well attributes affect methane emissions at the national, provincial, and sub-provincial scales, and estimate national emissions using different emission factor categorization methods to better quantify uncertainties, which can inform methane emission reduction strategies, government inventories, and well integrity management.

Methods

Non-producing Oil and Gas Well Database

Using publicly available provincial and territorial well databases (Table S1), we develop a database of all reported nonproducing oil and gas wells in Canada. Although regulations differ slightly across provinces and territories in Canada, plugging is required when wells become permanently non-producing (Figure S1). For the purpose of this study, we define inactive, idle, shut-in, and suspended wells as “unplugged” and abandoned wells as “plugged” (Table S2). We also define a common terminology for the following well attributes: well type, wellbore deviation, well depth, and well age (SI-1 and Table S3). We define these categories in accordance with Kang et al.’s (2021) and ECCC3 definitions. Finally, we estimate the number of undocumented wells in Canada, which correspond to wells not reported in provincial/territorial well databases, often because they were drilled before modern oil and gas regulations came into place (SI-2).

Direct Measurements of Methane Flow Rates

We compile 678 methane flow rate measurements from 494 non-producing oil and gas wells across five Canadian provinces: British Columbia (41 measurements), Alberta (239 measurements), Saskatchewan (251 measurements), Ontario (106 measurements), and Quebec (41 measurements). All the measurements are collected using a static chamber methodology26,27 (SI-3 and Figure S2) during field campaigns that took place from 2018 to 2023,5,6,20,28,29 including 164 previously unpublished measurements at 105 previously unmeasured wells. In Alberta and Saskatchewan, we collect wellhead (261) and SCV (236) measurements separately. In cases where there is an SCV and we measured both the wellhead and SCV emissions (251 measurements), we define combined emissions as the sum of the wellhead and SCV emissions.

Well Attributes Analysis

We investigate the effect of different well attributes on emission estimates, defined as follows:

graphic file with name es4c05602_m001.jpg

with i the given kind (e.g., plugged, unplugged, and unknown) of an attribute (plugging status, fluid type, wellbore deviation, and operator), k the number of kinds for a given attribute (e.g., three for plugging status), well counti the number of wells of a given kind of attribute (e.g., well countplugged), percenti (e.g., percentplugged) the percentage of all non-producing wells that are of a given kind of attribute, well counttotal the total number of non-producing wells in the region, and emission factoreff the attribute-weighted emission factor.

Emission factors are defined as the mean flow rates (including positive and negative flow rates) and are governed by high emitters. For categorical well attributes (plugging status, fluid type, wellbore deviation, and operator), we define the effect size of a given attribute as the average relative difference between the effective emission factor (emission factoreff) and the national/provincial emission factor, defined as the mean of all measured flow rates regardless of well attribute (emission factortotal).

graphic file with name es4c05602_m002.jpg

We define effect sizes as large (effect size > 0.5), moderate (0.25 < effect size ≤ 0.5), small (0 < effect size ≤ 0.25), and none (effect size = 0). These values are defined arbitrarily, based on relative differences between the effect sizes we found. We investigate the effects of well attributes at national and provincial scales using the national or provincial mean flow rates as the emission factortotal.

For the continuous well attributes (age, depth, and well density), we perform a non-parametric Spearman correlation test, accounting for the non-normality of our flow rates dataset. We included negative and null flow rates in our analysis. We define correlation effect sizes based on the Spearman ρ and p-value as large (|ρ| > 0.5, p-value <0.1), moderate (0.3 < |ρ| ≤ 0.5, p-value <0.1), small (|ρ| ≤ 0.3, p-value <0.1),30 and none (p-value >0.1).

Emission Factors and National Emissions Estimate

We define emission factor categories at different geographical scales (national, west/east, and provincial) and define two emission factor categorization methods by well status (method 1) and well status and type (method 2). We define wellhead and SCV emission factors as the mean of the direct methane flow rate measurements, including null and negative flow rates, for each category/component/geographical region. We use nonparametric bootstrapping of the mean of measured emission rates to obtain 1,000,000 sets of emission factors (SI-6–1). Non-producing wells can exhibit negative flow rates (i.e., methane uptake), which are characteristic of microbial oxidation of methane. Methane uptake generally occurs in soils but can also occur in the wellhead and/or SCV, as they can be connected to the shallow soil surface, or there may be microbes inside the wellhead infrastructure.

We estimate total annual emissions for the different emission factor categorization methods following approach 2 of the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC, 2006). We use the distribution of bootstrapped mean emission rates (i.e., emission factors) (SI-6–1) and a uniform distribution of our activity data (i.e., well counts) (SI-6–2) to conduct a Monte Carlo analysis with 1,000,000 iterations (SI-6–3). We define the uncertainty as the 2.5th and 97.5th percentiles of the Monte Carlo iteration estimates (SI-6–3). We also compare our results to values presented in the Canadian NIR and the U.S. GHGI,3,4 as well as to another published study focusing on Canadian non-producing oil and gas well emissions.6

Results

Non-producing Well Distribution in Canada

We find 471,276 non-producing oil and gas wells reported in provincial and territorial databases as of 2023, with most wells located in Alberta (74%), Saskatchewan (16%), Ontario (5%), and British Columbia (4%) (Figure 1a and Table S5). Plugged wells represent 68% of the total, while 17% are unplugged, and 15% have an unknown plugging status. Oil wells represent 26%, gas wells 20%, other wells 9%, and 44% of wells have an unknown fluid type. Of the main producing provinces, Saskatchewan has the highest oil well proportion (61%, compared to 18% gas and 21% other), and British Columbia has the highest gas well proportion (37%, compared to 9% oil and 3% other). At the subprovincial scale, we see that in regions with more than 1,000 wells, the highest plugged well proportions are found in Edmonton (73%) and Fort McMurray, Alberta (89%); Fort Nelson, British Columbia (70%); and Swift Current, Saskatchewan (79%) (SI-7, Tables S5 and S6). We estimate the number of undocumented wells in Canada to be 49,446, with 61% of Canada’s undocumented wells located in Ontario (Figure 1b).

Figure 1.

Figure 1

Open in a new tab

(a) Non-producing well distribution across Canada with two pie charts for each province/territory (left pie chart shows well status, right pie chart shows well type). The size of the pie chart is indicative of well counts in each province/territory (Table S5). (b) Undocumented well count estimates for each province/territory in Canada (AB: Alberta, BC: British Columbia, MN: Manitoba, NB: New Brunswick, NL: Newfoundland and Labrador, NT: Northwest Territories, NS: Nova Scotia, NU: Nunavut, ON: Ontario, PE: Prince Edward Island, QC: Quebec, YT: Yukon).

Methane Flow Rate Measurements

From our 442 wellhead and 236 SCV emissions measurements, we find that Alberta has the highest mean wellhead (1.0 × 104 mg/h) and mean SCV (3.9 × 105 mg/h) flow rates in Canada (Figure 2). In Saskatchewan, we find the mean wellhead flow rate (6.7 × 103 mg/h) to be comparable to Alberta, but the mean SCV flow rate is 4 orders of magnitude lower (43 mg/h). Mean wellhead flow rates in British Columbia, Ontario, and Quebec are 1.2 × 103 mg/h, 3.8 × 103 mg/h, and 3.3 × 102 mg/h, respectively. The region with the highest mean wellhead flow rate from the five sampled provinces is Medicine Hat, Alberta (2.9 × 103 mg/h) in the southwestern portion of the province, which is 3 orders of magnitude higher than the region with the smallest mean wellhead flow rate (Dawson Creek in northeastern British Columbia, 3.4 × 101 mg/h) (Figure S5). The region with the highest mean SCV flow rate (8.1 × 105 mg/h) is Lloydminster, Alberta, a town bordering Saskatchewan. The region with the lowest mean SCV flow rate (2.3 × 10–1 mg/h) is Kindersley, Saskatchewan, approximately 250 km south of Lloydminster.

Figure 2.

Figure 2

Open in a new tab

Wellhead and surface casing vent (SCV) methane flow rates by well plugging status (plugged, unplugged, and unknown). The dotted line represents the mean methane flow rate in each province (including positive (emissions), negative (uptake) flow rates and flow rates below the detection limit, which are set to zero). n represents the number of wellhead and SCV flow rate measurements used to calculate mean methane flow rates for each province and component (wellhead vs SCV).

The highest wellhead (9.8 × 105 mg/h) and SCV (3.9 × 107 mg/h) flow rate measurements are from two separate unplugged gas wells in Alberta. This highest SCV measurement is 7.5 times larger than the highest measurement from a non-producing well in Canada in Bowman et al. (2023) (also an unplugged gas well) and is of the same order of magnitude as a recent study in Colorado, U.S., where the authors found an unplugged well emitting at 7.6 × 107 mg/h, the highest rate in published literature to our knowledge.13 Comparing mean flow rates, and without accounting for fluid type, we find similar national means (1.0 × 103 mg/h for plugged wells and 8.3 × 103 mg/h for unplugged wells) as in Riddick et al. (2024)13 (1.6 × 103 mg/h for plugged wells and 7.5 × 103 mg/h for unplugged wells).

The distribution of wellhead and SCV methane flow rates is highly skewed, with the top 12% of emitters accounting for 98% of wellhead emissions and the top 2.1% accounting for 98% of SCV emissions (Figure 2). These high emitters govern mean emissions and emission factors. However, the shape of the distributions varies across provinces, with Alberta and British Columbia showing a more distinct tail at the high end than other provinces. Most wellhead and SCV emission distributions at provincial and subprovincial scales are shown to follow log-normal distributions when considering a 1% significance level.

Well Attribute Analysis

Geographic Location

We find large effects of province on wellhead, SCV, and combined emissions (corresponding to well sites where both wellhead and SCV flow rates were measured) (Figure 3), with 1.8 and 1.9 times higher mean wellhead and SCV flow rates in Alberta, and a 17 times smaller mean wellhead flow rate in Quebec, compared to national wellhead and SCV mean flow rates. Looking at subprovincial differences (see “Geographic location” at the province level in Figure 3), we only see small differences in wellhead, SCV, and combined emissions estimates, indicating that estimates are best made at the provincial scale.

Figure 3.

Figure 3

Open in a new tab

(a) Effect size of different well attributes on wellhead, surface casing vent (SCV), and combined (cases where we measured both wellhead and SCV flow rates on a given well site) emissions estimates at the national scale and provincial scale. The effect size is based on the relative difference between the attribute-specific mean flow rate and the national mean flow rate for categorical variables (province, region, status, type, deviation and operator). N represents the number of wellhead and SCV flow rate measurements used to estimate these emission factors. Geographic location effects correspond to provincial effects at the national scale and subprovincial effects at the provincial scale. (b) Spearman rank-based correlation between continuous well attributes (well density (triangle), age (circle), and depth (square)) and wellhead, SCV, and combined emissions estimates at the national scale and provincial scale. Black points represent national scale correlations and colored points represent provincial correlations. Points in the orange area are considered statistically significant correlations (p-value <0.1). The black dotted line represents a Spearman ρ value of zero.

Well Status

We find large effects of well status for wellhead and SCV emissions in most provinces, except in Ontario and Quebec (small effect for wellhead emissions; Figure 3). At the national scale, we find moderate effects of well status on wellhead emissions and large effects on SCV emissions (Figure 3). We find the highest wellhead emissions from unknown status wells (9.6 × 103 mg/h), followed by unplugged wells (8.3 × 103 mg/h), and the lowest emissions from plugged wells (1 × 103 mg/h) at the national scale (Figure S6). At the provincial scale, we observe similar trends. As for SCV emissions, we find a larger difference between unplugged and plugged well emissions, with 25, 26, and 28 times higher mean SCV emissions from unplugged wells compared to plugged wells at the national scale, in Alberta, and in Saskatchewan, respectively. For combined emissions, we also observe large effects due to unplugged well emissions being much higher than plugged well emissions.

Fluid Type

We find that fluid type affects wellhead emissions moderately at the national scale (Figure 3), with three times higher mean wellhead emissions from gas wells (1.1 × 104 mg/h) compared to oil wells (3.5 × 103 mg/h) and the lowest wellhead emissions from other (2.7 × 103 mg/h) and unknown (5.5 × 102 mg/h) fluid type wells (Figure S7). As for SCV emissions, we see moderate effects of fluid type at the national scale and in Alberta and small effects in Saskatchewan. Gas wells show mean SCV flow rates 44 and 25 times higher than oil wells’ mean SCV flow rates, at the national scale and in Alberta, respectively. As for combined emissions, we find similar fluid type effects as on SCV emissions at the provincial scale and small effects at the national scale.

Wellbore Deviation

We find small to no effect of wellbore deviation on wellhead and SCV emissions everywhere, except in Alberta (moderate effect on wellhead emissions; Figure 3). In Alberta, we find that horizontal and deviated wells emit more than vertical wells, with 100 and 40 times higher mean flow rates for horizontal and deviated wells, respectively, compared to vertical wells. Looking at combined emissions, we find large effects at the national scale and moderate to small effects at the provincial scale.

Operator

We find small to no effect of operator on wellhead emissions everywhere except in Alberta (moderate) (Figure 3), where we find 19 and 40 times lower wellhead emissions from two operators (operators A and B) compared to other operators. At the national scale, we see the same trend in wellhead emissions between operator A, operator B, and other operator wells (Figure S9). As for SCV emissions, we find large effects of operator at the national scale and in Saskatchewan. For example, wells with unknown operators show the lowest mean SCV flow rates (6.5 mg/h), and wells with other operators (i.e., not operator A or B) exhibit the highest mean SCV flow rates (2.9 × 105 mg/h) at the national scale. Examining combined emissions, we find similar operator effects as observed for SCV emissions, with moderate effects in Saskatchewan and large effects at the national scale.

Age

We find only a small effect of well age on SCV (ρ = −0.14, p-value = 0.028) emissions at the national scale, with emissions decreasing as well age (Figure 3). At the provincial scale, we found no statistically significant and consistent effect, with opposite trends observed in Alberta and British Columbia. Nevertheless, we note that the Spearman correlations generally indicate that emissions decrease with well age.

Depth

We find no statistically significant Spearman correlation between well depth and wellhead or SCV at the national and provincial scale in most provinces, except in Quebec, where we find decreasing wellhead emissions with depth (ρ = −0.32, p-value = 0.045) (Figure 3). In general, the Spearman correlations appear to indicate that emissions can increase or decrease with depth.

Well Density

We find no effect of well density on national-scale wellhead, SCV, and combined emissions. However, at the provincial scale, we find moderate effects of well density on wellhead emissions in British Columbia (ρ = 0.32, p-value = 0.043) and small effects of well density on combined emissions in Saskatchewan (ρ = 0.17, p-value = 0.091), with wellhead emissions being higher in areas of high well density. Broadly, Spearman correlations appear to indicate that emissions can both increase or decrease with well density (Figure 3).

Emission Factors and National Emission Estimate

Comparing our wellhead emission factors (Tables S7 and S8) with emission factors defined in the ECCC NIR3 and the U.S. EPA GHGI,4,31 our emission factors are generally lower than ECCC values but higher than EPA values. At the national scale, we estimate lower wellhead emission factors for almost all well categories, except for plugged oil wells and plugged other wells, where we find 5 and 3.6 times higher emission factors, respectively. At the provincial scale, the ECCC defines emission factors only for plugged oil wells in Alberta, unplugged other wells in British Columbia, and plugged and unplugged gas wells in Ontario. We find 1.7 and 2.3 times lower emission factors for plugged oil wells in Alberta and plugged gas wells in Ontario, and we find a similar emission factor for unplugged gas wells in Ontario. Compared to U.S. EPA emission factors defined at the east/west scale, we find 39 and 460 times higher plugged emission factors for east and west regions, respectively, whereas unplugged well emission factors are on the same order of magnitude.

Combining status-type/specific provincial emission factors and well counts, we find annual wellhead and SCV methane emissions for Canada to be 11 kt/year (5.8–22 kt/year) and 220 kt/year (39–550 kt/year), respectively. Although our wellhead emission estimate is about half the ECCC 2024 national estimate for wellhead emissions (20 kt/year, 12–32 kt/year)3 (Figure 4a), our estimates of non-producing SCV emissions for Canada are 16 times larger than ECCC’s estimate for Alberta and British Columbia (14 kt/year, 13–15 kt/year) (Figure 4b). In our estimates, nonproducing SCV emissions in Alberta (210 kt/year, 34–550 kt/year) account for 97% of national non-producing SCV emissions. Our total (i.e., the sum of wellhead and SCV) emissions estimate (230 kt/year, 51–560 kt/year) is 7 times (1.5–16 times) higher than ECCC’s estimate (34 kt/year, 25–47 kt/year) (Figure 4c). Using our wellhead emission estimate and Alberta SCV emissions estimate in the Canadian NIR, non-producing oil and gas well emissions account for 13% of total fugitive emissions from oil and natural gas systems in Canada. Moreover, we estimate half the wellhead emissions and three times the SCV emissions estimated in Bowman et al. (2023) (93 kt/year, 13–440 kt/year).6

Figure 4.

Figure 4

Open in a new tab

Annual methane emissions estimates from non-producing wells in Canada, from (a) wellhead, (b) surface casing vent (SCV), and (c) total emissions (sum of wellhead and SCV emissions estimates). Using our flow rate measurements, we show emissions estimates derived from provincial status/type emission factors broken down by plugging status (plugged, unplugged, and unknown). n represents the number of wellhead and SCV flow rate measurements used to estimate these emissions. The estimates from the 2024 Environment and Climate Change Canada (ECCC) National Inventory Report3 and Bowman et al. (2023)6 are shown in red, representing all nonproducing wells. The black lines represent uncertainty ranges.

The results from our uncertainty analysis show higher uncertainty in SCV emissions estimates compared to wellhead emissions estimates, with lower and upper uncertainty ranges of 5.8–22 kt/year for wellhead emissions and 39–550 kt/year for SCV emissions. These uncertainties are dominated by emission factor uncertainties, ranging from −130% to +230%, whereas well count uncertainties range from −37% to +37%. In comparison to emissions uncertainties reported in Canada’s NIR (12–32 kt/year, −39% to +58%) for wellhead emissions and 13 to 15 kt/year (−7% to +10%) for SCV emissions in Alberta and British Columbia.3 Our lower and upper uncertainty limits for wellhead emissions are 2 and 1.5 times lower, respectively (Figure 4a). As for SCV emissions, our lower and upper uncertainty limits are 3 and 37 times higher than the lower and upper bounds reported in the NIR, respectively (Figure 4b). Moreover, our lower and upper uncertainty limits for total emissions (51–560 kt/year) are 4 and 1.3 times higher than the uncertainty bounds reported in Bowman et al. (2023) (13–440 kt/year)6 (Figure 4c).

Discussion

Uncertainties in Emission Estimates Dominated by Emission Factor Uncertainties, though Well Counts May Also Be Underestimated

Given the relatively small set of direct measurements of methane emissions from nonproducing oil and gas wells, emission estimates remain highly uncertain. New publications with higher measured methane flow rates than previously published are coming out regularly,6,13 indicating that emission factors for non-producing wells are likely to be revised upward in the future. However, future measurements may also indicate lower emission factors, as these datasets remain small and may not be representative of all wells. We find total annual methane emissions from non-producing wells in Canada (230 kt/year, 51–560 kt/year) to be seven times higher, on average, than estimated in Canada’s NIR (34 kt/year). The higher emission estimates are driven by higher SCV emission estimates, which, in turn, are driven by emission factor values. Compared to the uncertainty ranges reported in the NIR, our wellhead emission estimates have smaller uncertainty ranges, but our SCV emission estimates have larger uncertainty ranges. This discrepancy may be due to the difference in NIR estimation methods between wellhead emissions (based on emission factors) and SCV emissions (based on industry reporting and extrapolation), whereas our estimates rely on emission factors in both cases. Looking at total emissions (i.e., the sum of wellhead and SCV emissions), our lower uncertainty limit is double that of NIR’s lower uncertainty limit, and our upper uncertainty limit is 12 times that of NIR’s upper uncertainty limit.

Uncertainty in emission factors mainly arises from the large range of measured emission rate values and the way we categorize emission factors and wells. We find the lowest uncertainties when using status-specific emission factors (plugged/unplugged/unknown) aggregated in the west and east provinces. This is currently how the U.S. EPA estimates abandoned oil and gas well emissions in the GHGI.4 However, the emission factors they currently use are up to 460 times lower than our west- and east-status-specific emission factors. Performing additional measurements, focusing on regions and well categories currently lacking measurements, is necessary to increase our confidence in status- and type-specific provincial emission factors. There may also be opportunities to take advantage of other measurement methods that can rapidly assess large numbers of non-producing oil and gas wells (e.g., aerial surveys).28,32

Additional uncertainty in emission factors lies in the temporal variation of emissions. While microbial methane emissions/uptake have been shown to depend on temperature (i.e., higher methanogenic bacteria activity in warmer temperatures and lower methanotrophic bacteria activity in colder temperatures) and atmospheric pressure,33 thermogenic emissions variability is currently understudied. Recent studies have found daily variations in methane emissions from active wells in the U.S.34 and monthly, daily, and hourly variations in emissions from abandoned wells.33 In this study, we did not explicitly account for seasonal variation, as we lack temporally resolved methane flow rate data across the country for all seasons. However, repeat measurements at the same wells in Ontario show that winter (February) measurements were not statistically different from fall (October) measurements.20 For Alberta and Saskatchewan, we also have measurements from summer months (June and July) and fall months (October and November) but not at the same wells. Overall, given the order of magnitude of microbial methane uptake (<102) and the highest emitters governing total emissions, it is unlikely that the seasonality of microbial emissions/uptake will substantially affect total emissions estimates.

Our well count uncertainty range is highly subjective, as we did not account for the uncertainty in undocumented well count estimates. Undocumented well uncertainties are not sufficiently reflected in our uncertainty range and may lead to upward revisions in our methane emission estimates.2,32,35 Many wells that were drilled before modern oil and gas regulations came into place remain undocumented today, leading to high uncertainties in well counts in Canada and the U.S.36 In the U.S., the number of undocumented wells in 2020 was estimated to be between 310,000 and 800,000.36 In Canada, the only published estimate of undocumented wells in Ontario (30,000)37 increases the estimated well count in Ontario by a factor of 2.3 (Table S4).

The well count currently used in Canada’s NIR is 409,319 (ECCC, 2024), which is 15% lower than our reported well count (471,276, excluding undocumented wells). Our well count is unlikely to be an overestimate, as it corresponds to the wells in government databases with unique identifiers.

Wellhead and Surface Casing Vent Emissions Governed by Different Leakage Processes

Oil and gas wells are constructed as a series of barriers isolating the oil, gas, and other fluids from the surrounding environment (soil, groundwater, or atmosphere). Leaks occur when one or several of these barriers fail. Loose valves or failed gaskets on the wellhead infrastructure are one example of breakages that result in wellhead emissions, whereas subsurface cement or casing failures can result in SCV emissions.6,9,15,38,39 If a well does not have an SCV, subsurface leaks can migrate either through the wellbore or through the subsurface, potentially contaminating groundwater resources39 and/or migrating through permeable soil formations and reaching the surface as gas migration.16 Therefore, we find differences in how well attributes affect wellhead and SCV emissions. It is important to note that our well attribute analysis is specifically for identifying high emitters. Although our analysis may be used as a starting point for understanding broader leakage processes, it is not meant to provide causal relationships, and the roles of some factors may be confounded due to the multiple processes likely involved.

Performing well attribute studies on combined emissions of SCV and wellhead emissions leads to confounding well attribute effects and results in no clear relationships between well attributes and emissions. For example, in our study, combined emissions show large effects of wellbore deviation on emissions, even though we see only small effects of wellbore deviation on wellhead or SCV emissions when investigated separately. We also see a small effect of well density on combined emissions in Saskatchewan, even though Saskatchewan does not exhibit effects of well density on wellhead and SCV emissions. Previous well attribute studies using direct measurements of methane flow rates did not differentiate between wellhead and SCV emissions,11,13,14,18 which may contribute to inconsistencies among previous studies. For example, spud date and construction period effects have been inconclusive in many previous studies.10,13,18 However, by separating SCV and wellhead emissions, we find that wellhead emissions show no relationship to age, but SCV emissions are negatively correlated with well age in some regions.

The Effect of Well Attributes Vary with Geographical Scale and Location

The national- and provincial-scale effects of well attributes on emissions differ. Geographical location is the only attribute that is shown to consistently affect emissions in our results and across studies.8−10,13 However, geographical location encapsulates many underlying engineering, geological, operational, environmental, and policy factors that potentially affect well integrity, leakage, and emissions.40

Previously published well attribute studies focus on specific regions, states, and provinces, with none looking at national-scale effects, which may lead to inconclusive or biased results. We find notable differences between national and provincial scale effects for most attributes, with well depth showing opposite trends at the national and provincial scales, and other attributes (type, wellbore deviation, operator, age) showing either only national-scale effects but no provincial effects, or vice versa. Subprovincial effects replicate provincial effects in most cases, but some regions do not have enough measurements to get conclusive results. Overall, our results may explain why many studies assessing the relationship between abandoned oil and gas well leakage and well attributes show inconclusive or conflicting results.

In addition to the limitations in sample size and the geographical scope of these studies, it is important to ensure sample representativeness. Even though our dataset remains small, we targeted specific regions and well attributes to obtain better representativeness across well attributes and regions (SI-7, Tables S5 and S6).

Well Attribute Studies Can Help Focus Mitigation Efforts

We show that non-producing oil and gas well emissions are underestimated in the Canadian national inventory and are associated with large uncertainties, which may compromise national emission reduction goals. We find that more methane emission measurements are required to reduce the uncertainty in non-producing well methane emission estimates; however, this depends on the type of measurement and approach. Oil and gas regulators in Alberta and British Columbia require operators to conduct and report surface casing vent flows, but the use of this data has led to large underestimates in the NIR. Moreover, rather than simply measuring more wells, well attribute analysis may be used to strategically conduct measurements that reduce uncertainties the most, thereby minimizing measurement costs.

Acknowledgments

This research was supported by funding from the Clean Economy Fund to M.K., the Natural Sciences and Engineering Research Council of Canada (NSERC) and Fonds de Recherche du Québec - Nature et Technologies Établissement de la relève professorale (FRQNT) to J.B., and the McGill Summer Undergraduate Research in Engineering (SURE) program to B.L. and L.W. Special thanks to J.P. Williams, L. Bowman, S. Seymour, O. Barrigar, and S. Smyth for their assistance throughout the study, as well as G. Ding, A. Moali, C. Demouveaux, and the Saskatchewan Ministry of Energy and Resources for their support in field measurements.

Data Availability Statement

The well-by-well methane emission rate measurements are available via the McGill Dataverse: https://doi.org/10.5683/SP3/GDQNI4.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.4c05602.

  • Methods and results, including the well attribute classification, static chamber methodology, emission estimate uncertainty analysis, wellhead and surface casing vent emission factors, mean and median flow rates grouped by different well attributes (PDF)

The authors declare no competing financial interest.

Supplementary Material

es4c05602_si_001.pdf (8.1MB, pdf)

References

  1. Kang M.; Brandt A. R.; Zheng Z.; Boutot J.; Yung C.; Peltz A. S.; Jackson R. B. Orphaned oil and gas well stimulus—Maximizing economic and environmental benefits. Elementa: Sci. Anthropocene 2021, 9, 00161. 10.1525/elementa.2020.20.00161. [DOI] [Google Scholar]
  2. Kang M.; Boutot J.; McVay R. C.; Roberts K. A.; Jasechko S.; Perrone D.; Wen T.; Lackey G.; Raimi D.; Digiulio D. C.; Shonkoff S. B. C.; William Carey J.; Elliott E. G.; Vorhees D. J.; Peltz A. S. Environmental risks and opportunities of orphaned oil and gas wells in the United States. Environ. Res. Lett. 2023, 18, 074012. 10.1088/1748-9326/acdae7. [DOI] [Google Scholar]
  3. Environment and Climate Change Canada. National inventory report: Greenhouse gas sources and sinks in Canada, Environment and Climate Change Canada, 2024. [Google Scholar]
  4. Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990–2022, Environmental Protection Agency, 2024. [Google Scholar]
  5. Williams J. P.; Regehr A.; Kang M. Methane Emissions from Abandoned Oil and Gas Wells in Canada and the United States. Environ. Sci. Technol. 2021, 55, 563–570. 10.1021/acs.est.0c04265. [DOI] [PubMed] [Google Scholar]
  6. Bowman L. V.; El Hachem K.; Kang M. Methane Emissions from Abandoned Oil and Gas Wells in Alberta and Saskatchewan, Canada: The Role of Surface Casing Vent Flows. Environ. Sci. Technol. 2023, 57, 19594–19601. 10.1021/acs.est.3c06946. [DOI] [PubMed] [Google Scholar]
  7. El Hachem K.; Kang M. Reducing oil and gas well leakage: A review of leakage drivers, methane detection and repair options. Environ. Res.: Infrastruct. Sustainability 2023, 3, 012002. 10.1088/2634-4505/acbced. [DOI] [Google Scholar]
  8. Ingraffea A. R.; Wells M. T.; Santoro R. L.; Shonkoff S. B. C. Assessment and risk analysis of casing and cement impairment in oil and gas wells in Pennsylvania, 2000–2012. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 10955–10960. 10.1073/pnas.1323422111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Lackey G.; Rajaram H.; Sherwood O. A.; Burke T. L.; Ryan J. N. Surface Casing Pressure As an Indicator of Well Integrity Loss and Stray Gas Migration in the Wattenberg Field, Colorado. Environ. Sci. Technol. 2017, 51, 3567–3574. 10.1021/acs.est.6b06071. [DOI] [PubMed] [Google Scholar]
  10. Watson T. L.; Bachu S. Evaluation of the Potential for Gas and CO2 Leakage Along Wellbores. SPE Drill. Completion 2009, 24, 115–126. 10.2118/106817-PA. [DOI] [Google Scholar]
  11. Kang M.; Christian S.; Celia M. A.; Mauzerall D. L.; Bill M.; Miller A. R.; Chen Y.; Conrad M. E.; Darrah T. H.; Jackson R. B. Identification and characterization of high methane-emitting abandoned oil and gas wells. Proc. Natl. Acad. Sci. U. S. A. 2016, 113, 13636–13641. 10.1073/pnas.1605913113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Montague J. A.; Pinder G. F.; Watson T. L. Predicting gas migration through existing oil and gas wells. Environ. Geosci. 2018, 25, 121–132. 10.1306/eg.01241817008. [DOI] [Google Scholar]
  13. Riddick S. N.; Mbua M.; Santos A.; Emerson E. W.; Cheptonui F.; Houlihan C.; Hodshire A. L.; Anand A.; Hartzell W.; Zimmerle D. J. Methane emissions from abandoned oil and gas wells in Colorado. Sci. Total Environ. 2024, 922, 170990. 10.1016/j.scitotenv.2024.170990. [DOI] [PubMed] [Google Scholar]
  14. Riddick S. N.; Mauzerall D. L.; Celia M. A.; Kang M.; Bressler K.; Chu C.; Gum C. D. Measuring methane emissions from abandoned and active oil and gas wells in West Virginia. Sci. Total Environ. 2019, 651, 1849–1856. 10.1016/j.scitotenv.2018.10.082. [DOI] [PubMed] [Google Scholar]
  15. King G. E.; King D. E. Environmental Risk Arising From Well-Construction Failure—Differences Between Barrier and Well Failure, and Estimates of Failure Frequency Across Common Well Types, Locations, and Well Age. SPE Prod. Oper. 2013, 28, 323–344. 10.2118/166142-PA. [DOI] [Google Scholar]
  16. Wisen J.; Chesnaux R.; Werring J.; Wendling G.; Baudron P.; Barbecot F. A portrait of wellbore leakage in northeastern British Columbia, Canada. Proc. Natl. Acad. Sci. U. S. A. 2020, 117, 913–922. 10.1073/pnas.1817929116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Ingraffea A. R.; Wawrzynek P. A.; Santoro R.; Wells M. Reported Methane Emissions from Active Oil and Gas Wells in Pennsylvania, 2014–2018. Environ. Sci. Technol. 2020, 54, 5783–5789. 10.1021/acs.est.0c00863. [DOI] [PubMed] [Google Scholar]
  18. Saint-Vincent P. M. B.; Reeder M. D.; Sams J. I. III; Pekney N. J. An Analysis of Abandoned Oil Well Characteristics Affecting Methane Emissions Estimates in the Cherokee Platform in Eastern Oklahoma. Geophys. Res. Lett. 2020, 47, e2020GL089663 10.1029/2020GL089663. [DOI] [Google Scholar]
  19. Seymour S. P.; Xie D.; Li H. Z.; MacKay K. Sources and reliability of reported methane reductions from the oil and gas industry in Alberta, Canada. Elementa: Sci. Anthropocene 2022, 10, 00073. 10.1525/elementa.2022.00073. [DOI] [Google Scholar]
  20. El Hachem K.; Kang M. Methane and hydrogen sulfide emissions from abandoned, active, and marginally producing oil and gas wells in Ontario, Canada. Sci. Total Environ. 2022, 823, 153491. 10.1016/j.scitotenv.2022.153491. [DOI] [PubMed] [Google Scholar]
  21. Alberta Energy Regulator. Well Vent Flow/Gas Migration Report. 2022; https://www1.aer.ca/productcatalogue/365.html (Accessed 11 October 2023).
  22. British Columbia Oil and Gas Commission. Surface casing vent flow database. 2022; https://www.bc-er.ca/data-reports/data-centre/ (Accessed 11 October 2023).
  23. Canadian Association of Petroleum Producers. A national inventory of greenhouse gas (GHG), criteria air contaminant (CAC) and hydrogen sulphide (H2S) emissions by the upstream oil and gas industry; Canadian Association of Petroleum Producers, 2005. [Google Scholar]
  24. Environment Canada. Technical report on Canada’s upstream oil and gas industry; Environment Canada, 2014. [Google Scholar]
  25. Townsend-Small A.; Ferrara T. W.; Lyon D. R.; Fries A. E.; Lamb B. K. Emissions of coalbed and natural gas methane from abandoned oil and gas wells in the United States. Geophys. Res. Lett. 2016, 43, 2283–2290. 10.1002/2015GL067623. [DOI] [Google Scholar]
  26. Kang M.; Kanno C. M.; Reid M. C.; Zhang X.; Mauzerall D. L.; Celia M. A.; Chen Y.; Onstott T. C. Direct measurements of methane emissions from abandoned oil and gas wells in Pennsylvania. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 18173–18177. 10.1073/pnas.1408315111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Williams J. P.; El Hachem K.; Kang M. Controlled-release testing of the static chamber methodology for direct measurements of methane emissions. Atmos. Meas. Tech. 2023, 16, 3421–3435. 10.5194/amt-16-3421-2023. [DOI] [Google Scholar]
  28. Pozzobon C.; Liu Y.; Kirkpatrick J. D.; Chesnaux R.; Kang M. Methane Emissions from Non-producing Oil and Gas Wells and the Potential Role of Seismic Activity: A Case Study in Northeast British Columbia, Canada. Environ. Sci. Technol. 2023, 57, 21673–21680. 10.1021/acs.est.3c06062. [DOI] [PubMed] [Google Scholar]
  29. El Hachem K.; von Sperber C.; Allard C.; Heagle D.; Vyriotes D.; Staebler R. M.; Caron-Beaudoin E.; Kang M. Characterizing multifaceted environmental risks of oil and gas well leakage through soil and well methane and hydrogen sulfide emissions. Environ. Res. 2025, 264, 120254. 10.1016/j.envres.2024.120254. [DOI] [PubMed] [Google Scholar]
  30. Cohen J.Statistical Power Analysis for the Behavioral Sciences, 2nd ed.; Routledge: New York, 1988. [Google Scholar]
  31. Environmental Protection Agency. Inventory of U.S. Greenhouse Gas Emissions and Sinks 1990–2016: Abandoned Oil and Gas Wells; EnvironmentalProtection Agency, 2018. [Google Scholar]
  32. O’Malley D.; Delorey A. A.; Guiltinan E. J.; Ma Z.; Kadeethum T.; Lackey G.; Lee J. E.; Santos J.; Follansbee E.; Nair M. C.; et al. Unlocking Solutions: Innovative Approaches to Identifying and Mitigating the Environmental Impacts of Undocumented Orphan Wells in the United States. Environ. Sci. Technol. 2024, 58, 19584–19594. 10.1021/acs.est.4c02069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Riddick S. N.; Mauzerall D. L.; Celia M. A.; Kang M.; Bandilla K. Variability observed over time in methane emissions from abandoned oil and gas wells. Int. J. Greenhouse Gas Control 2020, 100, 103116. 10.1016/j.ijggc.2020.103116. [DOI] [Google Scholar]
  34. Lavoie T. N.; Shepson P. B.; Cambaliza M. O. L.; Stirm B. H.; Conley S.; Mehrotra S.; Faloona I. C.; Lyon D. Spatiotemporal Variability of Methane Emissions at Oil and Natural Gas Operations in the Eagle Ford Basin. Environ. Sci. Technol. 2017, 51, 8001–8009. 10.1021/acs.est.7b00814. [DOI] [PubMed] [Google Scholar]
  35. Boutot J.; Peltz A. S.; McVay R.; Kang M. Documented Orphaned Oil and Gas Wells Across the United States. Environ. Sci. Technol. 2022, 56, 14228–14236. 10.1021/acs.est.2c03268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Interstate Oil & Gas Compact Commission. Idle and orphaned oil and gas wells: State and provincial regulatory strategies; Interstate Oil & Gas Compact Commission, 2021. [Google Scholar]
  37. Manocha J.; Carter T.. Oil and gas exploration, production and legislation on Ontario farms. Ministry of Agriculture and Food, 1999. [Google Scholar]
  38. Gianoutsos N. J.; Haase K. B.; Birdwell J. E. Geologic sources and well integrity impact methane emissions from orphaned and abandoned oil and gas wells. Sci. Total Environ. 2024, 912, 169584. 10.1016/j.scitotenv.2023.169584. [DOI] [PubMed] [Google Scholar]
  39. Jackson R.; Gorody A.; Mayer B.; Roy J.; Ryan M.; Van Stempvoort D. Groundwater Protection and Unconventional Gas Extraction: The Critical Need for Field-Based Hydrogeological Research. Groundwater 2013, 51, 488–510. 10.1111/gwat.12074. [DOI] [PubMed] [Google Scholar]
  40. Klotz L. A.; Sonnentag O.; Wang Z.; Wang J. A.; Kang M. Oil and natural gas wells across the NASA ABoVE domain: Fugitive methane emissions and broader environmental impacts. Environ. Res. Lett. 2023, 18, 035008. 10.1088/1748-9326/acbe52. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es4c05602_si_001.pdf (8.1MB, pdf)

Data Availability Statement

The well-by-well methane emission rate measurements are available via the McGill Dataverse: https://doi.org/10.5683/SP3/GDQNI4.

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES