the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Constructing a measurement-based spatially explicit inventory of US oil and gas methane emissions
Abstract. Accurate and comprehensive quantification of oil and gas methane emissions is pivotal in informing effective methane mitigation policies, while also supporting the assessment and tracking of progress towards emissions reduction targets set by governments and industry. While national bottom-up source-level inventories are useful for understanding the sources of methane emissions, they are often unrepresentative across spatial scales, and their reliance on generic emission factors produces underestimations when compared with measurement-based inventories. Here, we compile and analyze previously reported ground-based facility-level methane emissions measurements in the major US oil and gas producing basins and develop representative methane emission profiles for key facility categories in the US oil and gas supply chain, including well sites, natural gas compressor stations, processing plants, crude oil refineries, and pipelines. We then integrate these emissions data with comprehensive spatial data on national oil and gas activity to estimate each facility’s mean total methane emissions and uncertainties, from which we develop a mean estimate of national methane emissions, resolved at 0.1° × 0.1° spatial scales (~10 km × 10 km). From this measurement-based methane emissions inventory (EI-ME), we estimate total US national oil/gas methane emissions of 15.7 Tg (95 % confidence interval of 14–18 Tg) in 2021 which is 2.5 times greater than the EPA Greenhouse Gas Inventory. Our estimate represents a mean gas production-normalized methane loss rate of 2.6 %, consistent with recent satellite-based estimates. We find significant variability in both the magnitude and spatial distribution of basin-level methane emissions, ranging from <1 % methane loss rates in the gas-dominant Appalachian and Haynesville regions to >3–6 % in oil-dominant basins, including the Permian, Bakken, and the Uinta. Additionally, we present and compare novel comprehensive wide-area airborne remote sensing data and results of total area methane emissions and the relative contributions of diffuse and concentrated methane point sources as quantified using MethaneAIR in sub-regions of the Permian and Uinta basins that together indicate diffuse area sources accounting for the majority of total regional oil and gas emissions. Our assessment offers key insights into plausible underlying drivers of basin-to-basin variabilities in oil and gas methane emissions, emphasizing the importance of integrating measurement-based data in developing high-resolution, spatially explicit methane inventories in support of accurate methane assessment, attribution, and mitigation. The high-resolution spatially explicit EI-ME inventory is publicly available at https://doi.org/10.5281/zenodo.10734300 (Omara et al. 2024).
- Preprint
(4689 KB) - Metadata XML
-
Supplement
(6526 KB) - BibTeX
- EndNote
Status: open (until 09 May 2024)
-
RC1: 'Comment on essd-2024-72', Anonymous Referee #1, 09 Apr 2024
reply
Omara et al compiled previously reported methane measurement data to study methane emissions from major US oil and gas producing basins, and developed a high spatial resolution emission inventory for 2021. I find the methods solid and the manuscript well-written. I only have some minor comments.
General comments:
- More information about spatial allocation from facility-level level to 10 km*10 km resolution is needed (section 2.4). As some basins have more data than others, how much uncertainty will spatial allocation introduce?
- Additional analysis: (1) As age of wells is important to methane emissions, would the authors add a plot to show the correlations between age of wells and methane emissions? (2) It would be good to add a map in Figure 7 to show the uncertainty from EI-ME emissions.
- It’s good to have a high spatial resolution emission inventory. Have the authors considered improving the temporal resolution of the inventory? If it is not possible, what are the challenges and how to make it possible?
Specific comments:
Title: should specify what year is the inventory for.
L21: should specify which year is the inventory for. And clarify what ‘mean emission’ represents (average of yearly emission, or average of all the uncertainty iterations).
L24: should add one decimal for ’14-18’ to be consistent with L23 ’15.7 Tg’
L42,59: ‘methane emissions’ to ‘methane emission’, please check throughout the manuscript
L84: how many wells do not have reported production days?
L104: need a little more information about how the previously published data are searched, such as keywords used for searching on google scholar (or somewhere else).
L166-168: I’m confused with potential bias accounting, can you provide more information? And what about the uncertainty associated?
L221-225: As EPA inventory underestimates emissions, how does using EPA emission factors impact your results, and how is 50% uncertainty assumed?
Figure 4: (1) why is the yellow bar for EDGAR hatched? (2) I suggest moving the year for each study from the bottom of the figure to the top.
Figure 5. The colors are similar and difficult to distinguish.
L373: Miller et al.,2023 is missing in the reference list.
Figure 7: consider moving the figure legends outside the maps (now they overlap with each other)
Citation: https://doi.org/10.5194/essd-2024-72-RC1 -
RC2: 'Comment on essd-2024-72', Anonymous Referee #2, 14 Apr 2024
reply
Omara et al. constructed a high-resolution inventory for methane emissions from the U.S. oil and gas industry based on reported site-level measurements. The work provides a baseline that incorporates the best information for future evaluation of oil & gas methane emissions in the U.S., thus an important contribution to the field. I appreciate that the statistical method applied in the study is carefully designed with adequate sophistication. I'd recommend publication of the manuscript in ESSD, after the following comments are addressed.
1. The title indicates the inventory is for "US oil and gas methane emissions". However, the work is actually for "contiguous US onshore up- and mid-stream oil and gas emissions". The language can be more precise in places like abstract, conclusion, and Section 3.1 (when national totals are compared). While the focus on "onshore up- and mid-stream" is explained in the main text, I do not find any explicit language about the spatial extent (can only be inferred based on Fig. 7 and 8). As Alaska is an important oil & gas production region, I am concerned about if the comparisons are "apple to apple" in e.g. Section 3.1 when varied "national" totals are compared and discussed.
2. The method for low-production wells is not described in the manuscript. Reference to Omara et al. (2022) is provided. However, given the importance of low-production sites found in this work, a brief description of the main idea (e.g., method and data source) of Omara et al. (2022) seems necessary.
In addition, there is inconsistency in the current description of the well-site measurements (Table 1, Line 145-146, and Fig. 1a). Table 1 shows that there are n=1153 samples for low-production and non-low production sites combined. But line 145-146 and the caption of Fig.1a indicate that the figure is for non-low production sites only and includes n=1153 samples.
3. Line 142: The fraction of methane in produced natural gas should vary greatly from basin to basin. Is there better information for this parameter? What's the impact of this assumption on the uncertainty?
4. Line 182-187: (1) Based on the description, it is unclear whether the distribution of fBDL or only the mean of fBDL is used in the "decrement total mean estimate by fBDL" step. (2) fBDL is defined below in L210 for mid-stream facilities, but the concept first appears here but fBDL is not defined.
5. Section 2.1 Non-SI units are used throughout the text. It'd better to provide a conversion for SI units.
6. Line 151: "as a function of"?
7. Table 2: Shen et al. (2022) results presented in Fig. 5. can also be shown here.
8. A recent publication by Sherwin et al. (2024) in Nature reported a large dataset of aerial site measurements over US oil & gas basins. A discussion, if possible, can provide interested readers with useful information. For instance, (1) How does this study compare with Sherwin et al. (2024) at the basin level? (2) What's the implication of this large measurement data to the national inventory compilation?
Sherwin, E.D., Rutherford, J.S., Zhang, Z. et al. US oil and gas system emissions from nearly one million aerial site measurements. Nature 627, 328–334 (2024).
Citation: https://doi.org/10.5194/essd-2024-72-RC2 -
RC3: 'Comment on essd-2024-72', Anonymous Referee #3, 29 Apr 2024
reply
The authors developed a detailed compilation of activity data and emissions measurements to estimate 2021 US oil and gas methane emissions. The measurements were drawn from published studies and used to develop emission rate distributions. These distributions were used to estimate methane emissions at the facility level and then, estimate methane emissions at the national level. The authors also compare their results to airborne measurements using MethaneAIR. Overall, their spatially explicitly inventory is a valuable contribution, especially for remote measurements using satellites. In addition, they find important spatial trends that can be used to improve emission estimates and inform mitigation. Below are some high-level and detailed comments that can improve clarity of the paper.
High-level comments:
- The authors use measurement data from various years (not 2021) to estimate methane emissions for 2021. However, due to regulations, technology advancements, and other factors, methane emissions distributions may be changing over time, as the authors acknowledge. It doesn’t appear that the authors try to correct for this temporal variability or address this in their discussion.
- The definition of the methane loss rate is unclear. Although the authors mention in the Results (though I would have expected this in the Methods) that an 80% methane content was assumed, it remains unclear how the conversion was done for oil facilities.
- The authors need to provide a clear definition for “measurement-based inventories”. The authors do not measure all sources but use methane emission rate distributions based on available measurements, which does not cover all sites but some subset. Therefore, the question is how many and which measurements are needed to have a representative sample that can be used to create “measurement-based inventories”.
Detailed Comments:
L16: Provide a clear definition of “measurement-based inventories”. It will be helpful to clarify the extent to which measurements are conducted for the inventory to be considered "measurement-based".
L18: How is "representativeness" assessed?
L20: How is the comprehensiveness of the spatial data assessed?
L25: Is production data at the annual level?
L28-32: Very long sentence. Break sentence in two.
L46: Countries submit national inventory reports to the UNFCCC but there is no UNFCCC Greenhouse Gas Inventory. They are required to follow IPCC guidelines.
L53: Define what is meant by “measurement-based inventories”. See previous and high-level comments.
L64: Reading on, it appears that the paper uses measurements not from 2021. How is the data corrected for temporal variability?
L65: Measurement data from which year(s)?
L67: relative to methane or natural gas production?
L81-83: Above, the authors mention that the loss rates are normalized by methane production. How is oil/gas production converted to methane production? Is the production data for 2021 used or is the production data corresponding to the month/year of measurement used?
Table 1: Are the estimated total methane emissions reported in column 6 done by the authors here in this paper or are these previous results? If they are estimated in this paper, they are better placed in the results.
L130-135: Many of these measurements were conducted before 2021. There needs to be a description as to how these measurements can be used to estimate emissions in 2021, and if some adjustments are needed.
L141: what are the units for the methane loss rate? If it’s unitless, it should say so. Is "CH4" methane lost or measured? What is "Gas"? All the variables here need to defined and their units clearly provided after the equation.
Figure 2. These distributions are better placed in the Results section.
Figure 3. The K-S test provides a measure of goodness of fit. How can it be used to assess the representativeness of the underly methane emissions measurements?
Figure 3. There are some arrows missing. I suggest an arrow be added to the black line from the dashed rounded box to the site-level CH4 emission rate histogram.
L240: How was 500 selected?
L260-261: How was data limitation determined? There are published studies on downstream natural gas, post-meter, and abandoned well emissions. Are the authors looking for some specific number of measurements?
L271: Remove the word "However"
L288-289: methane content in natural gas can be variable. Does the 95% CI include methane content variability or is it assume to be fixed at 80%?
Fig. 4: The legends should be moved outside of the plot. It would be helpful if the groupings of bars separated by dashed lines were annotated – e.g., green bars should just be labeled GOSAT.
L294-299: The caption describes the first three bars only but should describe the rest as well.
L345-346: If weakly correlated, should factors other than infrastructure be considered?
L501: How are production rates determined for midstream infrastructure?
L502-503: If this study uses the data for 2021, would it not be different from the 2018 gridded EPA GHGI inventory data?
L509: Figure 6b shows methane emissions for a sub-region of the Uintah Basin, for which the agreement was good. Therefore, I don't think it's the correct figure to be pointing to here.
Citation: https://doi.org/10.5194/essd-2024-72-RC3
Data sets
Measurement-Based Spatially Explicit Methane Emission Inventory (EI-ME) Mark Omara https://doi.org/10.5281/zenodo.10734300
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
288 | 101 | 21 | 410 | 20 | 12 | 18 |
- HTML: 288
- PDF: 101
- XML: 21
- Total: 410
- Supplement: 20
- BibTeX: 12
- EndNote: 18
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1