The Futility of Relative Methane Reduction Targets in the Absence of Measurement-Based Inventories Authors: Bradley M Conrad, David R Tyner, Matthew R Johnson Published: 2023 ABSTRACT AbstractUnder the Global Methane Pledge, Canada is developing oil and gas sector methane regulations targeting 75% reductions from 2012 levels by 2030. Without measured baselines and inventories, such policies are ultimately unverifiable and unenforceable. Using the major oil and gas producing province of Saskatchewan as a case study, we derive first-ever measurement-based methane inventories for the region and comprehensively model previous emissions back to the 2012 baseline. Although relative reductions of 23–69% have likely occurred, the dispersion of modeled possibilities and the high emissions from continuing production illustrate the limits of this approach as a meaningful policy metric. Moreover, nearly 90% of apparent reductions are explained by decreased production at heavy oil facilities, suggesting emissions have potential to rebound if production resumes. By contrast, derived measurement-based methane emissions intensities facilitate quantitative assessment and show that despite any past reductions, Saskatchewan’s 0.41 ± 0.03 g/MJ intensity remains among the highest in North America. This highlights how relative reduction targets absent measured baselines and inventories are inherently futile and risk rewarding high emitters while obscuring ongoing mitigation potential. Ultimately, required global methane reductions will only be achieved by adopting objectively and independently verifiable emission metrics while measuring and tracking progress toward a net zero future.Keywords: oil and gas, methane emissions, methane intensity, mitigation targets, methane inventory, inventory projections, measurement-based inventory, regulatory effectiveness Introduction Methane is a potent greenhouse gas that is more than 80 times stronger than carbon dioxide in the first 20 years after release1 and is implicated as causing more than one-quarter of currently experienced global temperature rise.2 However, methane has a short atmospheric lifetime such that elimination of emissions has near-immediate climate benefits, making it a critical target of international mitigation efforts. Under the Global Methane Pledge,3 more than 150 countries have now committed to a global effort to reduce methane emissions by at least 30% from 2020 levels by 2030, intended to eliminate 0.2 °C of warming by 2050. Success in this effort is imperative given that all Intergovernmental Panel on Climate Change (IPCC)-modeled pathways that successfully limit global heating to 1.5–2.0 °C or less require a 24–34% reduction in methane emissions from 2019 levels by 2030.4 Although significant methane is emitted across the oil and gas, agricultural, and waste sectors, sources in the oil and gas sector have the greatest near-term mitigation potential from both a technical engineering and economic perspective.5 Consequently, countries, including the United States and Canada, have announced new proposals to reduce oil and gas methane emissions by 2030, which are intended to achieve an 87% reduction from 2005 levels of covered sources in the U.S.6,7 and a notionally less-ambitious 75% reduction from 2012 levels for covered sources in Canada.8 Unfortunately, the framing of policy as a relative reduction target is inherently problematic, as it requires accurate and precise knowledge of methane emissions in the baseline year. This is especially true for the oil and gas sector, where bottom-up estimates and industry-reported data used in official inventories are widely recognized as inaccurate and underestimated.9−14 This means that baseline emissions are unknown and uncertain and that relative reductions stated in nationally determined contributions (NDCs) are ultimately unquantifiable, even if emissions are measured moving forward. Moreover, relative emissions reduction targets reward originally poor-performing jurisdictions and fail to consider widely varying emission intensities of different production types and regions. Uncertainty in emissions magnitudes and baselines, and the consequent ambiguity in achieved relative reductions, creates further policy challenges in regions where federal and state/provincial jurisdictions overlap. The province of Saskatchewan, Canada, is an especially notable example. Saskatchewan is Canada’s second-largest oil and gas producer15−17 and second-largest emitter of methane in the official federal inventory.18 In parallel with federal methane regulations implemented in 202019 that were intended to reduce methane by 40–45% from 2012 levels by 2025, Saskatchewan developed and revised its own independent regulations.20,21 These latter regulations were subsequently permitted to supersede the federal standard22 on the basis of modeled equivalent outcomes by 2025.23 Notwithstanding the uncertainty in this modeling related to poorly estimated inventories and source breakdowns, this deemed equivalency was especially notable given that the federal regulations were officially modeled to be significantly stronger in every year after 2022, with a 30% or greater cumulative reduction by 2030.23 Now, as the federal government is developing updated regulations24 structured to achieve 75% reductions from 2012 levels as committed under the Global Methane Pledge, the province of Saskatchewan is publicly resisting federal efforts25 while simultaneously claiming to have already achieved a remarkable 60% reduction in emissions from a 2015 baseline.25,26 This stated reduction is calculated superficially from industry-reported flaring and venting volumes,26,27 despite measurements showing these are severely underestimated28,29 and a general understanding that reported flare and vent sources only comprise a portion of total emissions.30−33 Moreover, since gas and oil production34 have dropped by approximately 32 and 21% during 2015–2021, it is unclear what is driving any achieved reductions and by how much the emissions burden of the produced resource might be improving. Together, uncertain baselines, lack of measurement data, and concurrent production declines raise the question of whether recent regulations are indeed effective, how emissions compare with other regions, and what mitigation potential might remain. Recognizing this important policy challenge, it is notable that two significant producing/consuming jurisdictions regulate or intend to regulate based on methane intensity. In the U.S. Inflation Reduction Act,35 fees for methane emissions at natural gas production facilities come into effect for emitted methane exceeding 0.2% of marketed gas from each facility; for facilities producing strictly oil, fees come into effect when methane emissions exceed 10 t per million barrels of oil. As part of the Fit for 55 package,36 the European Union has proposed setting a methane intensity standard for imported oil and gas at less than or equal to 0.2%.37 This study has three primary objectives. First, following published algorithms, source-resolved aerial survey data from 2020 and 2021 are used to derive the first measurement-based methane inventories for the upstream oil and gas sector in the Canadian province of Saskatchewan. Calculated with robustly characterized uncertainties using peer-reviewed methods, these inventories are contrasted with official estimates from Environment and Climate Change Canada to elucidate both the degree to which the federal inventory underestimates observations in the field, and how actual measured sources (e.g., tanks, engine sheds, unlit flares, etc.) differ from currently assumed breakdowns. These insights are critical for supporting optimal methane mitigation to meet 2030 targets. Next, following approaches consistent with official federal modeling,38 the second objective is to back-project the 2021 measurement-based inventory to judiciously and defensibly estimate upstream methane emissions in 2012, the federal baseline year for methane reduction targets. Using historical, industry-reported activity data from Petrinex34 and the Canada Energy Regulator,15,17 a range of models are analyzed to assess the impacts of production changes, regulations, and voluntary action in the intervening years. This analysis helps contextualize stated mitigation achievements, specifically illustrating that the lack of historical measurement data precludes precise assessment of relative reductions and potentially obscures the achievement of meaningful mitigation. Finally, the measured inventory is used to derive provincial methane emission intensities and trends for comparison with other jurisdictions. These results demonstrate the limits of relative targets in the absence of both current and historical measurement-based emissions inventories and that notwithstanding apparent and noteworthy reductions from previous levels, Saskatchewan methane intensities remain high relative to other jurisdictions and far above proposed international targets. Methods Aerial Survey Design Two comprehensive aerial surveys of upstream oil and gas infrastructure in Saskatchewan, Canada, were completed to enable derivation of separate 2020 and 2021 measurement-based upstream oil and gas methane inventories. Design and planning of both surveys followed the protocols detailed by Johnson et al.9 First, monthly data from Canada’s Petroleum Information Network (Petrinex)34 and the Government of Saskatchewan (public new, active, and suspended facility reports39 and vertical/nonvertical well geodatabases40) were used to identify oil and gas facilities and wells that were active or shut-in at the time of planning. As mapped in Figure 1, subsets of facilities and wells in each of the four main production regions of Saskatchewan were surveyed during August 31-September 11, 2020, and August 9-September 14, 2021. To maximize the sample size within funding constraints, subregions with high facility and well density were identified, and the surveys aimed to include all active and locatable facilities within each subregion. Polygons bounding facility/pad footprints were manually created to guide the plane using the latest satellite imagery in ESRI ArcGIS Pro and subsequently provided to Bridger Photonics for flight planning and implementation using their Gas-Mapping LiDAR (GML) technology.41 Following the surveys, updated Petrinex reporting data for the sampled sites were obtained and used to confirm active status at the time of the surveys and to calculate full-time equivalent populations of all facility and well types over all 12 months of 2021. As fully detailed in Tables S1 and S2 in the Supporting Information (SI), the completed surveys included 20 and 26% of all active facilities in the province in 2020 and 2021, respectively, and 5% of all active wells in each year. This significant sample size enabled robust assessment of uncertainties in the derived inventories following the method of Johnson et al.9 that separately considers measurement uncertainty, sample size and finite population effects, and source intermittency. Map of surveyed facilities and wells in the Province of Saskatchewan in 2020 and 2021 aerial survey. Samples are drawn from each of the four main production regions as defined by the Saskatchewan Government and represent the distribution of facility subtypes as detailed in Tables S1 and S2 of the Supporting Information. Inventory Development As in parallel studies to compute inventories for the provinces of British Columbia and Alberta,9,14 the present methane inventories were developed using a piecewise approach that considers two nonoverlapping components: an inventory contribution from GML-detected/measured sources (i.e., the measured source inventories; blue bars in Figure 2) and inventory contributions from sources below the detection threshold of GML (i.e., “unmeasured sources”), which are further parsed into nonpneumatics (e.g., tanks, compressors, unlit flares, etc.; orange bars), pneumatic instruments (light gray bars), and pneumatic pumps (dark gray bars). The measured source component of an inventory treats aerial survey data as a stratified sample of provincial oil and gas infrastructure, enabling independent analysis of subsets of facilities or wells. Referring to detailed methodology in Johnson et al.9 and additional discussion in Supporting Information Section S1.1, grouping of similar infrastructure into unique strata based on facility or well type permits the overall reduction of variance in calculated emissions while enabling robust bootstrap-based characterization of finite sample and population size uncertainties on the computed inventory.42,43 The unmeasured source component of an inventory estimates the contribution of nondetected/missed sources to the total inventory. Using the approach detailed in Johnson et al.,9 unmeasured source magnitudes are calculated by simulating the pass-by-pass altitude and windspeed conditions of the actual aerial survey over a feedstock data set of small sources near and below GML’s detection sensitivity limit (see in particular Section S2.2 and Figure S2 of their Supporting Information). Feedstock data employed in the present study are summarized in Supporting Information Section S1.2 and are composed of 2017 field measurements in Alberta,44 pneumatic counts from a 2018 field survey in Saskatchewan,45 and pneumatic emission factors from the Alberta Energy Regulator.46 This simulation ultimately yields average site-level unmeasured emissions (emission factors) by facility/well (see Tables S4 and S5) which, when multiplied by the facility/well populations, yield the contribution of unmeasured sources in each stratum to the inventory, as detailed in Tables S6 and S7. Similar to surveys of the neighboring provinces,9,14 the probability of detecting methane emissions from normally operating pneumatics was effectively zero as calculated using the probability of detection function for Bridger’s GML,47 such that the pneumatic portion of the unmeasured source inventory was derived using the standard bottom-up approach. Computed emission factors and unmeasured source inventory for nonpneumatic equipment and pneumatic equipment are summarized in Tables S4 and S5, respectively. Summary of upstream oil and gas methane inventories for Saskatchewan from Environment and Climate Change Canada’s (ECCC’s) 2012 baseline to recent measurement-based inventories. ECCC’s inventory for all years are shown in green from the 2021,55 2022,56 and 202318 national inventory reports (NIRs). These are compared with inventories derived from satellite data (2019) and the present aerial surveys (2020 and 2021). During 2019–2021, ECCC’s 2023 NIR underestimates measurement-derived inventories by a factor of 1.5–1.6. Crucially, slight methodological changes to the NIR during 2021–2023 show significant variation across reports, highlighting the inherent uncertainty when projecting inventories. *Shen et al.’s satellite-based inventory11 uses data spanning the 2019 calendar year, from 05/2018–02/2020. The total, aerially measured, provincial methane inventories for Saskatchewan, Canada, in 2020 and 2021 are obtained by summing the measured and unmeasured source components. Tables S6 and S7 show the inventories with 95% equal tail confidence intervals parsed by stratum and source “category” (measured, unmeasured nonpneumatic, and unmeasured pneumatic instruments and pumps). Tables S8 and S9 show the nominal inventory (excluding confidence intervals) by stratum and source type (e.g., engine sheds, tanks, separators, etc.) for each year. The 2021 breakdown of nominal results by stratum and source (Table S9) is used as a basis for projecting the derived 2021 inventory to previous years. Methane Intensities Methane intensities (total methane emissions normalized by volume or energy of produced resource) were calculated following the procedures employed in previous studies of the neighboring provinces9,14 as provided in Table S10. Also included in this table are intensity estimates derived from satellite estimates of Shen et al.11 based on measurements during May 2018–February 2020. Intensity calculations used values for energy content of hydrocarbons, methane fraction of natural gas, and standard density of methane consistent with Conrad et al.14 (see their Table S7) to enable direct comparisons. Back-Projection of Provincial Methane Emissions Mathematical Description Back-projections of the measured 2021 inventory to previous years considered combined effects of activity and policy changes consistent with the official inventory methodology of ECCC.38 Referring to the detailed derivation in SI Section S2, the total upstream oil and gas methane inventory for Saskatchewan in year y [kt/y], Q(y) was estimated via where Qij(yo) is the absolute emissions [kt/y] of source type i at stratum (i.e., facility/well type) j during the baseline year (yo) of 2021 and κij(y,yo) and λij(y,yo) represent unitless scaling factors that account for relative changes in activity and policies between the baseline and year of interest. The source- and stratum-resolved inventory for 2021 was used as a baseline, such that all Qij(yo) values are summarized in Table S9. A total of 144 unique back-projection scenarios to 2020–2012 were considered, each with different combinations of activity and policy scaling factors derived as described in the following sections and fully detailed in the Supporting Information. These 144 scenarios represent the range of possibilities from changes in production and reporting practices as well as both voluntary and regulatory operational changes in the intervening years. Activity Scaling Factors, κij Methane emissions from different facility types and sources may be expected to correlate with different measures of activity. For example, methane emissions from constant-bleed gas-driven pneumatics should scale directly with the number of instruments, whereas other sources such as venting at single-well heavy oil sites might reasonably be argued to scale with oil production, reported gas production, or total energy production at these specific site types or potentially the population of these specific sites in different years. Leveraging the available stratum-resolved measured inventory data, the derived inventory projections considered four activity scaling scenarios, in which the nonpneumatic emissions in each stratum could be scaled by primary production (i.e., produced gas or oil volume at a gas or oil facility type), produced gas volume, total energy production of gas plus oil, or active facility or well counts (see details in Table S11). In all cases, methane emissions from pneumatic equipment were assumed to scale with the number of active facilities or wells in each stratum. Policy Scaling Factors, λij Perhaps the most challenging aspect of back-projecting measured inventory data to previous years is objectively modeling impacts of past regulatory or policy changes on operating practices. In the absence of measurements, these potential impacts are ultimately unknown. More than anything, this inherent uncertainty exemplifies the futility of defining and evaluating policy on relative reductions from an uncertain baseline. Nevertheless, over the time frame of the present back-projections (2012–2021), the primary change was the introduction of Saskatchewan’s Directive PNG03620 which took effect in January 2020. This regulation introduced facility limits for venting and flaring as well as requirements to conduct twice per year leak detection and repair (LDAR) surveys at specific gas facility subtypes producing or receiving more than 60,000 m3/y of gas. In addition, prior to these regulations, during 2012–2019, some portion of companies may have begun voluntary LDAR measures as recommended in the Canadian Association of Petroleum Producers (CAPP) 2007 best management practice (BMP) guide. This guide suggested directed inspection and maintenance targeting components “most likely to result in big leaks” but left it to each company to “design a frequency monitoring program best suited for its operations”.48 Mirroring the approach used by ECCC in developing the national inventory,38 the net effect of the CAPP BMP is assumed equivalent to voluntary 0.5 times per year (every 24 months) surveys at all facilities during 2012–2014 increasing to 0.75 times per year (once every 16 months) during 2015–2019. Starting in 2020, full compliance with regulated twice per year LDAR surveys at applicable gas sites is assumed with continuing voluntary surveys of 0.75 times per year assumed for all other facilities. As shown in Tables S13 and S14, the present analysis extends this model to also consider scenarios with no voluntary early action, 50% compliance, and 100% compliance, noting that lack of assumed voluntary action will tend to overestimate the relative impact of regulations and hence overestimate any relative reductions in emissions from 2012. Additionally, the presented scenarios also perturbed whether different sources at applicable facility types were considered as fugitive sources (and therefore likely to be impacted by LDAR efforts), as summarized in Table S12. Beginning in 2020, section 5 of Saskatchewan’s Directive PNG03620 restricted “oil wells and oil facilities” to venting no more than 900 m3/d, requiring sites with greater volumes to conserve or flare all gas. Coincident with this policy change, industry-reported venting34 at both crude oil and heavy oil facilities fell drastically, including an apparent 57 and 44% venting decrease at subsets of crude oil and heavy oil sites operating during both 2018–2019 and 2020–2021. However, these reported decreases also coincided with large decreases in oil production, including a 21 and 22% drop at the same subsets of facilities operating in the two years before versus after January 1, 2020. Moreover, it is now well established that actual levels of venting in the Canadian oil and gas industry are consistently underestimated by reported data,12,13,30,32,33 especially at heavy oil sites28,30,49,50 which dominate emissions in Saskatchewan. Confidence in the reported data is further eroded considering that reported volumes of gas production, venting, flaring, and fuel use must only be measured if certain annually averaged volumetric thresholds are exceeded.21 Instead, produced gas volumes at oil sites are typically only estimated using infrequently measured or modeled gas-oil ratio (GOR) and gas-in-solution values,21 which themselves can be challenging to measure51 and notoriously uncertain at heavy oil sites.28,52−54 The fractions of produced gas that are flared, vented, or used for on-site fuel are then typically only estimated. Consequently, reported volumes cannot be expected to accurately characterize achieved methane mitigation via Directive PNG036. To overcome this challenge, six different scenarios were considered with the objective of bounding the true effect of Directive PNG036 on vent sources. As summarized in Table S15, stratum-by-stratum policy scaling factors were derived by closely analyzing reported data in Petrinex during the two-years prior and two-years following the introduction of the regulations. The first model acts as a baseline and assumes that all vented gas volumes are accurately reported within Petrinex. The second model assumes that pre-policy (2018–2019) vented gas volumes are accurate and venting per unit activity within each stratum (i.e., venting per site or venting per production for a given facility type) should remain constant. This assumes that Directive PNG036 had no effect on activity-normalized venting rates. A third and fourth model augment the first two by also allowing for the possibility that some postpolicy vented gas is mis-reported as fuel use, which recent field measurements suggest could be a significant factor in underreported vent volumes.28 Specifically, in cases where the reported fuel use-to-oil production ratio was noted to increase after January 1, 2020, the extra fuel use is instead assumed to be mis-reported venting. Similarly, a fifth and sixth model augments the third and first based on the common knowledge that oil well GOR tends to increase with well age, alongside decreasing oil production. Specifically, in cases where the as-reported stratum-level GOR is noted to unexpectedly decrease after the introduction of regulations, the prepolicy GOR is instead assumed and applied to the postpolicy oil production to recalculate the expected postpolicy gas production. Additional gas volumes under this scenario are then assumed to have been vented. Table S16 summarizes the calculated policy scaling factors for vent sources during 2012–2019 (from the 2021 baseline) for the 24 combinations of venting scenario and activity scaling scenario. In permutation with the considered range of LDAR models and source scenarios, this resulted in 144 unique inventory back-projection scenarios. Results and Discussion Provincial Upstream Oil and Gas Methane Inventory Aerial surveys of upstream oil and gas infrastructure were conducted in Saskatchewan, Canada in 2020 and 2021 using Bridger Photonics’ Gas-Mapping LiDAR (GML) technology.41 As detailed in the Methods Section, these source-level aerial measurements of methane emissions enabled the first measurement-based inventories for the province, following peer-reviewed protocols9 using a comprehensively characterized instrument.47Figure 2 plots the derived inventories (in blue, orange, and gray corresponding to contributions of measured, unmeasured nonpneumatic, and unmeasured pneumatic sources, respectively) with uncertainties that consider the effects of quantification error and sample size. The calculated 2020 inventory was 477.8 kt/y with an equal tail 95% confidence interval (black error bars) of −73.0/+90.0 kt/y; the 2021 inventory was 439.0 kt/y (−31.1/+34.4). The figure also shows upper-bound estimates that include additional uncertainty due to source intermittency (green error bars), which is negligibly small in this work (see Johnson et al.9 for further description of this analysis). While these results suggest a nominal 8.1% decrease in absolute methane emissions from the upstream sector between 2020 and 2021, this difference is within the plotted uncertainties. Figure 2 also plots the official upstream sector inventories from Environment and Climate Change Canada (ECCC) in green for 2012 and 2019–2021 as separately estimated in the 2021,55 2022,56 and 202318 national inventory reports (NIRs). The most recent NIR (2023) significantly underestimates methane emissions such that measurement-based inventories from the present aerial surveys are approximately 1.6× NIR values in 2020 and 2021 as indicated. A satellite-derived inventory11 using data spanning 2019 (purple) similarly suggests 1.5× higher methane emissions than the most recent official inventory for that year. This degree of underestimation by the NIR is consistent with recent measurement-based inventory studies in neighboring Canadian provinces.9,14 Moreover, there is also substantial variability in the annual emissions estimates in different versions of the NIR, ranging from ±20% in 2012 and as much as ±25% in 2019 over the past three reports. While ECCC is commendably working to improve their bottom-up estimates and modeling while also adapting to changes in provincially regulated reporting,57 an unavoidable consequence is a varying 2012 baseline with each update. Ultimately, this simply reflects the current uncertainty of the NIR and the limits of bottom-up estimates in the absence of measurements, especially in a time where emissions (and hence emission factors) can be expected to be continually changing. Taking the three most recent NIR estimates for 2012 (392.4–585.9 kt/y) at face value and comparing with the 2021 measurement-based inventory suggests that the change in Saskatchewan’s upstream sector methane emissions was likely between −25% (reduction) and +12% (increase). Alternatively, if the 2012 inventory reported in the 2023 NIR is inflated by a factor of 1.6 to be consistent with 2020 and 2021, then the present measurements suggest a 44% decrease in emissions since 2012, achieving targeted reductions outlined in the 2020 federal regulations four years early. The full range of possible relative reductions is analyzed in greater detail below. Breakdown of Inventory by Source Prior to attempting a more robust analysis of potential trends in emissions since 2012, it is important to examine the specific sources driving emissions. Figure 3 parses the measurement-based inventory by source type and compares with identifiable source categories in the most recent NIR. The limited granularity of the NIR breakdown (Figure 3a) reflects the shortcomings of the bottom-up estimate approach, where industry-reported venting is an unknown aggregation of industry-identified venting sources and fugitive equipment leaks are calculated using generic emission factors and extrapolations for assumed average site configurations. From a policy perspective, the unknown source detail in the ECCC inventory hinders regulatory development and makes it challenging to identify the root causes of its underestimation beyond the dearth of contemporary, measurement-based data. This lack of source detail is notably apparent for tank emissions, where the approximately 32% contribution to the 2021 measurement-based inventory is nearly an order of magnitude larger than the identifiable 3% contribution to the ECCC inventory. This discrepancy is largely due to the unidentified aggregation of some tank sources (flashing, breathing, and working losses and blanket gas venting58) into the industry-reported venting contribution. Without further breakdown of industry-reported venting, the degree to which tank emissions agree across the two inventories cannot be conclusively determined. However, in the extreme scenario that all reported venting is assumed to be tank-related, the official inventory would still underestimate measurements by more than 10%, suggesting substantial mismeasurement51 and/or underreporting of tank sources by industry. Breakdown of 2021 upstream methane inventories by source type. Pie charts identify (a) the relative contribution of source categories in Environment and Climate Change Canada’s (ECCC’s) 2023 national inventory report18 and (b) sources in the present measurement-based inventory. SCVF = Surface Casing Vent Flow; CHOP(S) = Cold Heavy Oil Production with or without Sand. Refer to Figure S3 for the equivalent results from 2020. The ECCC NIR (Figure 3a) also implies an outsized 23% contribution of surface casing vent flows (SCVFs, accidental or unintentional venting of gas collected in the well surface casing vents) in Saskatchewan. Since there is currently no regulated public reporting of SCVFs at producing wells in Saskatchewan, ECCC estimates SCVF emissions31 by scaling 2011 estimates59 on the basis of industry-reported60 “capable wells”. Specifically, ECCC uses historical measurement data from British Columbia and Alberta31 to calculate an emission factor of approximately 1.52 t/y of methane per capable wellhead, which is applied to all well types. By contrast, even if the entire 3% fraction of the measured inventory attributed “wellhead” emissions in Figure 3b is considered to be SCVFs, this implies a significantly lower emission factor of 0.154 and 0.413 tonnes/year per gas and crude oil wellhead, respectively. Conversely, the dominance of methane emissions from Cold Heavy Oil Production with or without Sand (CHOP(S)) is missed or obscured in the official inventory, especially CHOP(S) emissions via engine sheds and wellheads, which represent 38% of the measured inventory (167 kt). Not to be confused with unintended SCVFs, CHOP(S) wells intentionally produce gas that flows to the surface through the annular space between the production casing and well bore.28 This gas is typically vented, either at the wellhead or via an engine shed where a portion of the gas is used to power an engine driving the hydraulic well pump and nominally should be included in reported vent volumes. However, even if the entire 42% reported venting fraction in the ECCC inventory (114 kt) were assumed to be from CHOP(S), this would still significantly underestimate emissions. This is unsurprising given recent measurements of Festa-Bianchet et al.28 showing that actual venting at CHOP(S) sites in Saskatchewan was 3.9 times greater than industry-reported levels and underscores the challenge that ECCC faces when constrained to derive inventories using sparse and inaccurate industry-reported data. Correcting this potential reporting error alone would increase the ECCC inventory by almost 50% to 400.7 kt/y, nearly within the present study’s uncertainties. At a minimum, equipment-level emission factors derived from field-measurement data (see Tables S8 and S9 for source- and facility/well-specific emission factors from the 2020 and 2021 surveys, respectively) could significantly improve the accuracy of these official, bottom-up inventories. However, given that these new emission factors should themselves be expected to change if regulations are effective in driving rapid reductions of methane emissions, ongoing source-resolved measurement surveys will ultimately be critical to ensuring accurate equipment-level emission factors and inventories moving forward. Perhaps most importantly, from a regulatory development and mitigation perspective, the data in Figure 3b identify the relative contributions of vent, fugitive, and combustion sources to overall emissions. Fugitive sources in Saskatchewan (taken to include piping, non-CHOP(S) wellheads, other, and unknown in the present attributions), which are notionally detectable and fixable via LDAR surveys, account for just 8.6% of the measurement-based inventory. Combustion sources (primarily compressors and heaters) account for a further 3.4%, while remaining vent sources dominate the inventory at 88.0% of total methane emissions. By contrast, on a relative basis, the official ECCC inventory suggests that fugitive “equipment leaks” and SCVFs represent more than half of emissions. This gross overestimate risks steering regulatory efforts toward mandated LDAR, while the present results suggest that vent mitigation efforts would be much more effective. Inventory Projection to Estimate Relative Reductions The 2021 measurement-based inventory—representing the current best-available upstream oil and gas inventory for Saskatchewan—was used to predict historical methane emissions and objectively estimate relative reductions. Following the same general approach used by ECCC31 when estimating emissions over different years, the 2021 inventory was parsed at the source and facility/well type level (see Table S9) and scaled to previous years using a range of relevant activity metrics (oil, gas, and energy production and facility/well population) derived using publicly available, industry-reported data in Petrinex.34 In total, 144 unique projection scenarios were considered as described in the Methods Section and Tables S11–S16. The change in methane emissions relative to the projected 2012 methane inventory for each projection scenario is shown in Figure 4a; see Figure S4 for the absolute projections in kt/y and Figure S5 for projections relative to 2015. As summarized in Table S11, line color represents scenarios scaled by the noted activity metric, including primary production (i.e., produced volume of oil at oil facilities/wells or gas at gas facilities/wells), gas production, total energy production, or facility/well population. Plotted results represent 144 independent scenarios considering a range of added effects from voluntary LDAR measures since 2012, regulations implemented in 2020 under Directive PNG036, and potentially mis-categorized volumes in Petrinex (e.g., some vent volumes incorrectly reported as fuel gas). Relative reductions from a 2012 baseline via these simulations ranged from 23 to 69% with a projection-averaged reduction of 54%. From a 2015 baseline as in Saskatchewan’s regulations, relative reductions were 28 to 61% (average of 46%). While the latter simple average supports the Government of Saskatchewan’s claim that they have surpassed their Methane Action Plan61 objective of 40–45% methane reductions in the upstream sector since 2015, the range of possibilities precludes a definitive conclusion. Moreover, as any one of these 144 scenarios has the potential to best reflect the unknown reality, the simple average is not necessarily any more meaningful than any other scenario. Projections of the 2021 measurement-based inventory to estimate relative reductions in methane emissions from the ECCC baseline year of 2012. Relative reduction of methane emissions since 2012 are plotted for all projection scenarios (a) alongside simultaneous relative changes in production (b), including a notable 70% drop in production at single-well CHOP(S) (SW-CHOP(S)) sites. Absolute provincial methane intensities are also plotted for all projections, with an average reduction of 53% since 2012 (c). Subplot (d) plots the projection-averaged absolute change in methane emissions since 2012 alongside the change at SW-CHOP(S) alone. These data, coupled with the negligible change in methane intensity at SW-CHOP(S) sites (d), suggest that nearly all methane reductions in Saskatchewan can be attributed to reduced SW-CHOP(S) production. Any apparent changes in methane emissions must also be placed in the context of production changes over the same time period and the methane intensity of the products currently being produced. Figure 4b plots changes in oil and gas production from the Canada Energy Regulator15,17 (CER) and further segregates production at single-well CHOP(S) sites (SW-CHOP(S), calculated based on portion of total reported heavy oil production at subtypes 313 or 325 in Petrinex34). Overall, CER data suggest that oil and gas production dropped 8.4% on an energy-basis during 2012–2021, while production at SW-CHOP(S) sites decreased by more than 70%. Given that SW-CHOP(S) facilities emitted half (49%) of all methane in the 2021 measurement-based inventory, this shifting production trend is a primary driver of overall changes in emissions. Indeed, notwithstanding significant dispersion, the average of the projected provincial upstream industry methane intensities plotted in Figure 4c suggests that as a whole intensity has dropped by more than a factor of 2 from 0.799 to 0.378 g/MJ—or, following the Natural Gas Sustainability Institute’s (NGSI) protocol,62 from 5.15 to 2.44%. However, it is important to note that this substantial effect is driven by changes in overall production volumes and the proportion of production types, rather than changes in operating practices within specific facility types. Specifically, Figure 4d shows (red lines) that SW-CHOP(S) accounted for a projection-averaged decrease in methane emissions of 509.4 kt/y from 2012 to 2021, almost 90% of the total decrease across the province. Importantly, during this same period, the methane intensity at SW-CHOP(S) sites decreased by just 1.9% (Figure 4d; green line). Thus, nearly all methane reductions at SW-CHOP(S)—and, indeed, the province—during 2012–2021 can be attributed to a >70% drop in SW-CHOP(S) production with all other changes netting a comparatively small effect. Because SW-CHOP(S) remains the dominant source of methane emissions even after this decrease in production, addressing SW-CHOP(S) emissions continues to be crucial if Saskatchewan is to reduce emissions toward an intensity level in-line with leading jurisdictions. During 2018–2021—where available public Petrinex data is resolved by operator—almost three-quarters of industry-reported production decline at SW-CHOPS was attributable to four companies who also operated thermal in situ facilities. An overall transition in production from SW-CHOP(S) toward thermal in situ production could indicate a market-driven shift toward lower methane-intensity production (based on comparatively lower measured methane emissions per oil production from thermal facilities and notwithstanding significantly increased CO2 emissions from associated steam production). However, the trends in SW-CHOP(S) and thermal in situ production are somewhat independent. Less than two-thirds of the decline in SW-CHOP(S) production by thermal facility operators was offset by increased thermal in situ production. This implies that at most one-half of recent reductions at SW-CHOP(S) might be attributed to abandonment in favor of thermal in situ production. Thus, even if the operators active in both subsectors preferentially expand thermal in situ production moving forward, the drop in SW-CHOP(S) production (and associated methane emissions) might be expected to return by one-half or more if oil prices rise toward previous levels. Indeed, under the “Canada Net Zero” scenario, the Canada Energy Regulator63 predicts that by 2025, when Saskatchewan must renegotiate its equivalency agreement with ECCC, heavy oil production in Saskatchewan will be 31% higher than 2021 levels. Without dramatic improvement to SW-CHOP(S) methane intensity or immediate action (whether market- or policy-driven) to eliminate SW-CHOP(S), Saskatchewan may enter negotiations with emissions trending upward. In summary, the activity-based projections enable the estimation of emissions reductions over time. However, the ability to confidently derive meaningful conclusions from projections to justify and/or evaluate past interventions and regulations is dubious. These projections are constrained by the accuracy of reported production and dispositions of oil and gas in myriad ways. The choice of different bases for emissions scaling and subtle assumptions regarding the (in)accuracy of reported data can therefore render the impact of policy interventions effectively unknown in the face of projection uncertainties; a limitation that can only be overcome through measurements in conjunction with policy implementation. Notwithstanding necessary assumptions and uncertainties in the present projections, the case study of Saskatchewan also demonstrates how a change in production can enable substantial relative reductions in methane emissions. Without this critical context, regulations in jurisdictions with high production-normalized emissions that simply reduce production may be unfairly viewed as exemplary. While such decreases in emissions surely contribute toward broadly defined emissions targets, it is crucial to note that production-driven emission reductions do not signify effective policies and are not sustainable if production rebounds. This highlights the need to contextualize and regulate emissions and compare jurisdictional performance via measurement-based methane intensities. Methane Intensity In contrast to relative reduction targets, methane intensity is a concise and directly comparable measure of methane emissions per unit of production. There are two commonly used intensities that are effectively equivalent but are reported in different units. First, methane emissions can be normalized by the total energy of associated production, similar to carbon intensity,64 and presented in units of gram of methane per MJ of produced energy (i.e., g/MJ or, equivalently, kt/PJ). A similar methane intensity derived by the Natural Gas Sustainability Institute62 (NGSI; hereinafter called “NGSI methane intensity”) presents production-normalized methane emissions of natural gas in units of percent by attributing methane emissions to natural gas production on an energy-weighted basis. These two intensities are interchangeable for a fixed methane content of the produced natural gas. Figure 5a plots calculated total oil and gas sector methane intensities for Canada’s three western provinces independently and in aggregate, and Figure 5b plots similar results for five American basins65 and the country as a whole.66 Methane intensities of conventional (non oil sands) oil and gas production across North American basins.9,11,14,65,66 (a) Methane intensities for Western Canadian provinces independently and in aggregate via the present study, other 2021 aerial surveys,9,14 and satellite observations during May 2018 to February 202011. Additional downstream contributions from the ECCC NIR18 are added in white to give total sector values. (b) Methane intensities for American basins from satellite observations during 2018 and 202065 and a bottom-up analysis for 2015 (which separately estimates a downstream contribution as similarly shown in white).66 *Excludes the province of Manitoba. Figure 5 identifies that, in aggregate, Canadian (0.224 g/MJ or 1.44% using 2021 aerial survey data and 0.221 g/MJ or 1.42% using Shen et al.’s inversion11 of satellite data from May 2018 to February 2020) and American (0.252 g/MJ or 1.69%) methane intensities are effectively equivalent given the degree of uncertainty, with the main difference being due to a larger downstream component in the U.S.A. Despite this similarity, there exists an order-of-magnitude variation in methane intensity at the subnational level, ranging from 0.065 g/MJ (0.42%) in British Columbia (BC), Canada, to 0.66 g/MJ (4.40%) in America’s Anadarko basin (located in western Oklahoma and neighboring states). Spatial variation in methane emissions is a confluence of production characteristics (relative proportion of oil, gas, and other products), standard operating procedures, and regulations. Methane intensities like those plotted in the figure permit an objective comparison of jurisdictions, which explains why they are becoming a primary metric for international mitigation efforts.37,62,67,68 The present case study of Saskatchewan offers a critical finding regarding the evaluation of methane mitigation performance. Despite all provinces having internal methane regulations declared equivalent with federal regulations, and despite notable relative reductions in methane emissions during 2012–2021, Figure 5 shows that Saskatchewan is the most methane intense oil and gas producer in Western Canada by a significant margin (1.6× over neighboring Alberta and >6× over BC). This result illustrates that a jurisdiction can perform poorly on an intensity basis even while apparently meeting aggressive relative reduction targets. Moreover, notwithstanding the differing challenges coming with oil- vs gas-dominated production across the provinces, BC clearly demonstrates that low-intensity production can be achieved. In the face of impending import restrictions by the European Union, the fee structure in the U.S.A.’s Inflation Reduction Act, and public perception of carbon/methane intense production, the present analysis reveals the futility of relative reduction targets in setting a useful standard for methane emissions. Policy and regulations designed to meet objectively and independently measurable emission intensities along the path to a net-zero future are essential for ensuring competitive access to global markets, and ultimately for achieving the required global methane emissions reductions that underpin the Global methane pledge. Acknowledgments We gratefully acknowledge valuable input and assistance from Steve Smyth (Environment and Climate Change Canada, ECCC) in parsing bottom-up federal inventory data to facilitate source-level comparisons. The leadership of Natural Resources Canada (NRCan, especially Nicole MacDonald, Michael Layer, and Amin Asadollahi) under tight timelines to launch aerial surveys was essential to the success of this project as was the collaborative cooperation of Bridger Photonics Inc. (especially Ben Losby, Kerry Neal, and Mike Thorpe). Finally, we appreciate the assistance of Zachary Milani and Alex Szekeres (EERL) during the initial source attribution processing. Supporting Information Available The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c07722. Tabulated 2020 and 2021 measurement-based inventory data by stratum and breakdown by sources; additional inventory development methodology; mathematical description of inventory back-projections; additional figures showing the breakdown of uncertainties in the measured source inventory, empirical distributions of measurement conditions, inventory projections in absolute units of kt/y, and projection results from Saskatchewan’s baseline year of 2015; additional tables summarizing aerial survey coverage, methane intensity calculations, and projection scenarios including policy and activity scaling factor derivation and numerical results; and additional refs (69–71) (PDF) Tabulated 2020 and 2021 measurement-based inventory data by stratum and breakdown by sources; additional inventory development methodology; mathematical description of inventory back-projections; additional figures showing the breakdown of uncertainties in the measured source inventory, empirical distributions of measurement conditions, inventory projections in absolute units of kt/y, and projection results from Saskatchewan’s baseline year of 2015; additional tables summarizing aerial survey coverage, methane intensity calculations, and projection scenarios including policy and activity scaling factor derivation and numerical results; and additional refs (69–71) (PDF) Data to replicate Figure 2 through Figure 5 in the main text and additional figures and tables in the Supporting Information can be accessed via the Carleton University dataverse at 10.5683/SP3/ISA8G0. Anonymized detected emissions for each detected source in the aerial survey are also included. Data to replicate Figure 2 through Figure 5 in the main text and additional figures and tables in the Supporting Information can be accessed via the Carleton University dataverse at 10.5683/SP3/ISA8G0. Anonymized detected emissions for each detected source in the aerial survey are also included. Author Contributions Conceptualization: M.R.J., B.M.C., and D.R.T.; Software, validation, and data curation: B.M.C. and D.R.T.; Formal analysis: B.M.C., M.R.J., and D.R.T.; Writing—Original Draft: B.M.C. and M.R.J; Writing—Review and Editing: All authors. Project management and funding acquisition: M.R.J. This work was generously supported by Natural Resources Canada (NRCan, grant number EIP2-MET-001), Natural Sciences and Engineering Research Council of Canada (NSERC, grant numbers 06632 and 522658), and Environment and Climate Change Canada (ECCC, contract number 3000714200). The authors declare no competing financial interest. Supplementary Material es3c07722_si_001.pdf References Associated Data Supplementary Materials es3c07722_si_001.pdf