the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Feb 2023
High potential for CH4 emission mitigation from oil infrastructure in one of EU’s major production regions
Abstract. Ambitious methane (CH4) emissions mitigation represents one of the most effective opportunities to slow the rate of global warming over the next decades. The oil and gas (O&G) sector is a significant source of methane emissions, with technically feasible and cost-effective emission mitigation options. Romania, a key O&G producer within the EU, with one of the highest reported annual CH4 emissions from the energy sector (Greenhouse Gas Inventory Data - Comparison by Category, 2022), can play an important role towards the EU’s emission reduction targets. In this study, we quantify CH4 emissions from onshore oil production sites in Romania at source and facility level using a combination of ground-based measurement techniques. Measured emissions were characterised by heavily skewed distributions, with 10 % of the sites accounting for more than 70 % of total emissions. Integrating the results from all site-level quantifications with different approaches, we derive a central estimate of 5.4 kg h–1 site-1 of CH4 (3.6–8.4, 95 % confidence interval) for oil production sites. This estimate represents one of the highest when compared to measurement-based estimates of similar facilities from other production regions. Based on our results, we estimate a total of 120 ktons CH4 yr–1 (range: 79–180 ktons yr–1) from oil wells in our studied areas in Romania. This is approximately 2.5 times higher than the total reported emissions from the Romanian oil production sector for 2020. Based on the source level characterization, up to three quarters of the detected emissions from oil production sites are related to operational venting. Our results suggest that O&G production infrastructure in Romania holds a massive mitigation potential, specifically by implementing measures to capture the gas and minimize operational venting and leaks.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(1187 KB)
-
Supplement
(1219 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1187 KB) - Metadata XML
-
Supplement
(1219 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-247', Anonymous Referee #1, 24 Apr 2023
The authors combined ground-based measurements conducted by Delre et al and Korben et al with new UAV measurements to better estimate methane emissions from oil (and gas) production in Romania and to identify leaking components on well sites. They find that methane emissions from oil infrastructure in Romania exceed inventory estimates by a factor of two or much greater and that most of the emissions come from operational venting. These findings are timely and provide essential information from rapidly reducing methane emissions from oil infrastructure in Romania.
The authors have performed an in-depth statistical analysis of measurement distributions and the role of measurement method and the detection limit/proportion of non-detects. I particularly like their analysis of detection limits and non-detects, which unfortunately are not highlighted in the main paper.
A few sentences on the history and development oil production in Romania is needed. In addition, the authors should describe how the different datasets vary in terms of geology and operational practices, including comments about representativeness, and not just measurement methods. The measurements compared were made at different well sites and points in time. Although the role of temporal variability and site-by-site variations are acknowledged, the way that the results are presented, it appears that measurement method is the most important factor.
One of the final concluding points is that well sites with H2S emit less methane, with the argument being that they are more tightly managed due to safety reasons. However, as noted in the detailed comments, the literature show conflicting results, which may be due to geology, status of the well, or something else. Therefore, I suggest providing additional qualifiers and reference relevant studies (such as those noted below).
Most of the revisions that are needed are related to text and wording to clarify the messages. In general, it would be helpful to provide enough description so that readers can follow the topic without referring to the SI or another publication. Although a lot of the questions in the detailed comments below are answered in the SI, it would be better if a short description is provided in the main paper.
DETAILED COMMENTS ON MAIN PAPER
Line 44-46: Provide ranking of Romania (e.g., third or fourth) in terms of methane emissions from the energy sector in the EU.
Line 54: If possible, provide a more quantitative ranking (e.g., among the top 10).
Line 56: how representative are the studied areas of all areas in Romania?
Line 57: should it read "the entire Romanian oil production sector"? Based on the previous sentence, it appears that the 120 ktons/year is for a subset of the Romanian oil production sector.
Line 58: Is 2020 representative of methane emissions from other years? Because of the pandemic, there were changes in methane emissions in many oil and gas producing regions.
Line 58: was the source level characterization done as a part of the measurement campaign? Or was it based on other inventory estimates?
Line 75: Replace "Whereas" with "Although"
Line 95: add "the" in front of "largest"
Line 96: Delete "," after the first "EU"
Line 106: provide a list of measurement platforms in parenthesis. Are they all ground-based? It’s a bit confusing because UAV measurements are also included in this paper.
Line 110: "Ground-based" is mentioned numerous times in this paragraph. However, there is no description of what is meant by "ground-based" and what "non-ground-based" would represent. What are UAVs considered as?
Line 128: in the introduction, it sounded as though only ground-based data would be used. See previous comments.
Line 149: What counts as a cluster of high-density of production facilities? Provide a quantitative limit.
Line 153: specify whether the authors had site access or if they were screening from public roads only.
Line 155: define "site accessibility". Is this because operators did not give permission or because the roads are not in good condition or something else?
Line 156: is there data on farms and if so, was this data analyzed?
Line 170: how close was the emission point from the tracer gas release point?
Line 198-107: did Korben et al only use GPM?
Line 240: are the differences in MBA between UG and EMPA applications important or can they be viewed as minor? if they are important, they should be briefly (1 sentence) described here.
Line 284: what is the detection limit of the HFS?
Line 288: how many sites did the authors have access to (i.e., not from the road)?
Line 292: delete "at"
Line 295-299: provide a definition of "non-detects". Are these emissions detected by the OGI but not GPM, TDM, OTM, or UAV or vice versa? I think it's important to have one line here in the main text even if it's provided in the SI.
Table 1, Figure 2: Since these data do not represent the same sites, the differences may not be due to the method but due to the difference in emissions at sites. Plus, even at the same site, emission rates can change over time. There should be some description of the sites that are included in each box plot.
Line 357: remove extra "."
Line 369: inconsistent spacing between number of %.
Line 371: change to "35-83% higher"
Line 372: what are the ranges of uncertainty based on?
Line 382: it seems as though "wells" and "sites" are being used interchangeably. On a given well site, there are many components, including the well but also other infrastructure such as tanks. Provide a definition of both and stick with one term (e.g., "well site"). In addition, well sites can vary substantially in size and complexity. For example, Tyner and Johnson (2021) have a nice classification of well sites as on-site and off-site (https://doi.org/10.1021/acs.est.1c01572).
Line 391: what are "open-ended lines"? Since they represent such a large fraction, it is important to describe what they are.
Line 402: provide definition of an "oil park".
Line 406: replace "with" with "by"
Line 416: add "Methane" in front of "Emissions".
Line 447: replace with "we now confirm this discrepancy is large for Romania"
Line 454: this sentence suggests that gas flaring was practiced in Romania before. Why was it stopped? Or is this a typo?
Line 465-571: it is important to clearly define what is meant by site to make these comparisons.
Line 473: I would just say third or fourth highest. Clearly the Barnett and Fayetteville (where there's more unconventional development), the distributions are more heavy-tailed. Could we group the regions by production type or some other factor? It may be that Romania has the highest level of skewness among conventional production regions.
Line 485-488: provide more descriptions of the open-ended lines and what the non-identified components below the ground may be.
Line 489-497: It seems that this is an area for further research. One study based in Canada found that H2S emitting wells emit more methane: https://www.sciencedirect.com/science/article/abs/pii/S0048969722005836
However, another study found something similar to what was found here: https://www.sciencedirect.com/science/article/abs/pii/S0048969721059143
Line 501-505: Would it be safe to say that zero gas production wells are "abandoned wells"? How do the emission rates compare with those for abandoned wells in published literature? https://pubs.acs.org/doi/abs/10.1021/acs.est.0c04265
Line 522: "practices"
Line 531: as noted earlier, with the mention of "ground-based quantification methods", I expect some discussion or comments about the role of non-ground-based quantification methods. Also, UAVs are included in this analysis, which I would consider as aerial-based quantification methods. So, there's a need to change the terminology and discuss the limitations and benefits of "ground-based quantification methods" compared to others.
Line 533: replace with "oil well sites". I suggest using this term throughout the paper.
Line 538-540: This point may not be fully supported by literature. The role of H2S should be studied more systematically.
Line 539: H2S emissions is not a component. replace "this component" with "H2S emissions"
DETAILED COMMENTS ON SUPPLEMENTAL INFORMATION
Line 82: What about components other than the well borehole that may be leaking methane? It is discussed later on but an introductory paragraph on the range of potentially leaking components is needed.
Line 145 and section: What are the pros and cons of the UG and EMPA methods? It seems that the biggest difference is how wind is taken into account.
Line 153: How is the wind speed and direction determined? Is it measured directly in the field and if so, is it measured using the UAV or something else?
Line 165: How is the horizontal vector, u(y,z), determined?
Line 215: Reading later on, the role of Detection Limit in the likelihood function is described. However, there should be some information here or earlier on.
Line 221: Describe the direct search algorithm used here and provide a reference. Is this described in Zavala-Araiza et al (2015)? If so, it should be cited in this sentence.
Line 288-289: Is this due to the fact that detection limit emission rate is applied to non-detects? If non-detects were set to zero, the opposite would be true.
Table S3: Provide definition of S_0 in the caption again.
Line 331: What is the difference between east and west regions? In general, a better description of the basin and the general setting could be helpful in this paper.
Line 443-445: The importance of getting the low-end distribution right is something that is not widely discussed and is worth highlighting in the main paper.
Line 539: Why were only production and age evaluated? Was it because other data were not available or is there evidence to indicate that these are the most important characteristics to look at?
Line 561: Replace “for” with “of”
Line 568-569: I would argue that what is being shown here is that emitting wells are more likely to have higher gas production and associated gas, which is consistent with the finding that oil wells are more likely to be non-emitting. Therefore, I suggest revising the sentence: “no significant differences were found between emitting and non-emitting sites”.
Line 570: Replace “oi” with “oil”
Citation: https://doi.org/10.5194/egusphere-2023-247-RC1 -
AC2: 'Reply on RC1', Foteini Stavropoulou, 15 Jun 2023
We thank the anonymous referee for taking the time to thoroughly review our manuscript and for providing us with insightful and constructive comments that have greatly contributed to its improvement. Responses to individual referee comments are uploaded in the form of a supplement. Line numbers refer to the updated WORD manuscript with tracked changes.
-
AC2: 'Reply on RC1', Foteini Stavropoulou, 15 Jun 2023
-
RC2: 'Comment on egusphere-2023-247', Anonymous Referee #2, 03 May 2023
general comments
The work reported in this paper is from a co-ordinated field campaign conducted between many groups, measuring CH4 from oil production sites across Romania using four different approaches. The scientific question is firmly within ACP’s remit, and contains well established methods and statistical analyses to interrogate the results. The report is laid out clearly, with assumptions and methods presented well. Much of the detail of the methodologies is in the SI. I note that another reviewer has recommended to bring some of this into the main text. I would support this, although I think this is somewhat down to personal preference, as I did not see any guidance on this from the journal itself.
Substantial conclusions are reached, showing that methane emissions from these sources are over twice as high as currently reported, thus highlighting the inadequacy of the methods uses to report emissions in this sector and the potential for mitigation by stopping the leaks.
specific comments
L391 and elsewhere: open-ended lines is a term which is used, and is somewhat self explanatory, however it would help to define it for those not within this field. e.g. it is not clear whether these lines are designed like this, or if they should have a cap on the end but for some reason the cap has been left off.
L510 and 533: what are the Tier 1 EFs for developing and developed countries? Presumably the currently used EFs for reporting are quite different to what you have derived. This value would be useful to compare with the EF you have derived, eg in the conclusion as a comparison (and earlier when you discuss the IPCC methodology).
technical corrections
L67 and later: % symbol should not have a space before it, i.e. it should be 25% and not 25 %
L75 L80: Is Global Methane Tracker 2022, 2022 correctly referenced? Just checking, as it looks odd with 2022 appearing twice
Citation: https://doi.org/10.5194/egusphere-2023-247-RC2 -
AC1: 'Reply on RC2', Foteini Stavropoulou, 15 Jun 2023
We thank the anonymous referee for the valuable feedback and comments, which have helped to improve our manuscript. Responses to individual referee comments are uploaded in the form of a supplement. Line numbers refer to the updated WORD manuscript with tracked changes.
-
AC1: 'Reply on RC2', Foteini Stavropoulou, 15 Jun 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-247', Anonymous Referee #1, 24 Apr 2023
The authors combined ground-based measurements conducted by Delre et al and Korben et al with new UAV measurements to better estimate methane emissions from oil (and gas) production in Romania and to identify leaking components on well sites. They find that methane emissions from oil infrastructure in Romania exceed inventory estimates by a factor of two or much greater and that most of the emissions come from operational venting. These findings are timely and provide essential information from rapidly reducing methane emissions from oil infrastructure in Romania.
The authors have performed an in-depth statistical analysis of measurement distributions and the role of measurement method and the detection limit/proportion of non-detects. I particularly like their analysis of detection limits and non-detects, which unfortunately are not highlighted in the main paper.
A few sentences on the history and development oil production in Romania is needed. In addition, the authors should describe how the different datasets vary in terms of geology and operational practices, including comments about representativeness, and not just measurement methods. The measurements compared were made at different well sites and points in time. Although the role of temporal variability and site-by-site variations are acknowledged, the way that the results are presented, it appears that measurement method is the most important factor.
One of the final concluding points is that well sites with H2S emit less methane, with the argument being that they are more tightly managed due to safety reasons. However, as noted in the detailed comments, the literature show conflicting results, which may be due to geology, status of the well, or something else. Therefore, I suggest providing additional qualifiers and reference relevant studies (such as those noted below).
Most of the revisions that are needed are related to text and wording to clarify the messages. In general, it would be helpful to provide enough description so that readers can follow the topic without referring to the SI or another publication. Although a lot of the questions in the detailed comments below are answered in the SI, it would be better if a short description is provided in the main paper.
DETAILED COMMENTS ON MAIN PAPER
Line 44-46: Provide ranking of Romania (e.g., third or fourth) in terms of methane emissions from the energy sector in the EU.
Line 54: If possible, provide a more quantitative ranking (e.g., among the top 10).
Line 56: how representative are the studied areas of all areas in Romania?
Line 57: should it read "the entire Romanian oil production sector"? Based on the previous sentence, it appears that the 120 ktons/year is for a subset of the Romanian oil production sector.
Line 58: Is 2020 representative of methane emissions from other years? Because of the pandemic, there were changes in methane emissions in many oil and gas producing regions.
Line 58: was the source level characterization done as a part of the measurement campaign? Or was it based on other inventory estimates?
Line 75: Replace "Whereas" with "Although"
Line 95: add "the" in front of "largest"
Line 96: Delete "," after the first "EU"
Line 106: provide a list of measurement platforms in parenthesis. Are they all ground-based? It’s a bit confusing because UAV measurements are also included in this paper.
Line 110: "Ground-based" is mentioned numerous times in this paragraph. However, there is no description of what is meant by "ground-based" and what "non-ground-based" would represent. What are UAVs considered as?
Line 128: in the introduction, it sounded as though only ground-based data would be used. See previous comments.
Line 149: What counts as a cluster of high-density of production facilities? Provide a quantitative limit.
Line 153: specify whether the authors had site access or if they were screening from public roads only.
Line 155: define "site accessibility". Is this because operators did not give permission or because the roads are not in good condition or something else?
Line 156: is there data on farms and if so, was this data analyzed?
Line 170: how close was the emission point from the tracer gas release point?
Line 198-107: did Korben et al only use GPM?
Line 240: are the differences in MBA between UG and EMPA applications important or can they be viewed as minor? if they are important, they should be briefly (1 sentence) described here.
Line 284: what is the detection limit of the HFS?
Line 288: how many sites did the authors have access to (i.e., not from the road)?
Line 292: delete "at"
Line 295-299: provide a definition of "non-detects". Are these emissions detected by the OGI but not GPM, TDM, OTM, or UAV or vice versa? I think it's important to have one line here in the main text even if it's provided in the SI.
Table 1, Figure 2: Since these data do not represent the same sites, the differences may not be due to the method but due to the difference in emissions at sites. Plus, even at the same site, emission rates can change over time. There should be some description of the sites that are included in each box plot.
Line 357: remove extra "."
Line 369: inconsistent spacing between number of %.
Line 371: change to "35-83% higher"
Line 372: what are the ranges of uncertainty based on?
Line 382: it seems as though "wells" and "sites" are being used interchangeably. On a given well site, there are many components, including the well but also other infrastructure such as tanks. Provide a definition of both and stick with one term (e.g., "well site"). In addition, well sites can vary substantially in size and complexity. For example, Tyner and Johnson (2021) have a nice classification of well sites as on-site and off-site (https://doi.org/10.1021/acs.est.1c01572).
Line 391: what are "open-ended lines"? Since they represent such a large fraction, it is important to describe what they are.
Line 402: provide definition of an "oil park".
Line 406: replace "with" with "by"
Line 416: add "Methane" in front of "Emissions".
Line 447: replace with "we now confirm this discrepancy is large for Romania"
Line 454: this sentence suggests that gas flaring was practiced in Romania before. Why was it stopped? Or is this a typo?
Line 465-571: it is important to clearly define what is meant by site to make these comparisons.
Line 473: I would just say third or fourth highest. Clearly the Barnett and Fayetteville (where there's more unconventional development), the distributions are more heavy-tailed. Could we group the regions by production type or some other factor? It may be that Romania has the highest level of skewness among conventional production regions.
Line 485-488: provide more descriptions of the open-ended lines and what the non-identified components below the ground may be.
Line 489-497: It seems that this is an area for further research. One study based in Canada found that H2S emitting wells emit more methane: https://www.sciencedirect.com/science/article/abs/pii/S0048969722005836
However, another study found something similar to what was found here: https://www.sciencedirect.com/science/article/abs/pii/S0048969721059143
Line 501-505: Would it be safe to say that zero gas production wells are "abandoned wells"? How do the emission rates compare with those for abandoned wells in published literature? https://pubs.acs.org/doi/abs/10.1021/acs.est.0c04265
Line 522: "practices"
Line 531: as noted earlier, with the mention of "ground-based quantification methods", I expect some discussion or comments about the role of non-ground-based quantification methods. Also, UAVs are included in this analysis, which I would consider as aerial-based quantification methods. So, there's a need to change the terminology and discuss the limitations and benefits of "ground-based quantification methods" compared to others.
Line 533: replace with "oil well sites". I suggest using this term throughout the paper.
Line 538-540: This point may not be fully supported by literature. The role of H2S should be studied more systematically.
Line 539: H2S emissions is not a component. replace "this component" with "H2S emissions"
DETAILED COMMENTS ON SUPPLEMENTAL INFORMATION
Line 82: What about components other than the well borehole that may be leaking methane? It is discussed later on but an introductory paragraph on the range of potentially leaking components is needed.
Line 145 and section: What are the pros and cons of the UG and EMPA methods? It seems that the biggest difference is how wind is taken into account.
Line 153: How is the wind speed and direction determined? Is it measured directly in the field and if so, is it measured using the UAV or something else?
Line 165: How is the horizontal vector, u(y,z), determined?
Line 215: Reading later on, the role of Detection Limit in the likelihood function is described. However, there should be some information here or earlier on.
Line 221: Describe the direct search algorithm used here and provide a reference. Is this described in Zavala-Araiza et al (2015)? If so, it should be cited in this sentence.
Line 288-289: Is this due to the fact that detection limit emission rate is applied to non-detects? If non-detects were set to zero, the opposite would be true.
Table S3: Provide definition of S_0 in the caption again.
Line 331: What is the difference between east and west regions? In general, a better description of the basin and the general setting could be helpful in this paper.
Line 443-445: The importance of getting the low-end distribution right is something that is not widely discussed and is worth highlighting in the main paper.
Line 539: Why were only production and age evaluated? Was it because other data were not available or is there evidence to indicate that these are the most important characteristics to look at?
Line 561: Replace “for” with “of”
Line 568-569: I would argue that what is being shown here is that emitting wells are more likely to have higher gas production and associated gas, which is consistent with the finding that oil wells are more likely to be non-emitting. Therefore, I suggest revising the sentence: “no significant differences were found between emitting and non-emitting sites”.
Line 570: Replace “oi” with “oil”
Citation: https://doi.org/10.5194/egusphere-2023-247-RC1 -
AC2: 'Reply on RC1', Foteini Stavropoulou, 15 Jun 2023
We thank the anonymous referee for taking the time to thoroughly review our manuscript and for providing us with insightful and constructive comments that have greatly contributed to its improvement. Responses to individual referee comments are uploaded in the form of a supplement. Line numbers refer to the updated WORD manuscript with tracked changes.
-
AC2: 'Reply on RC1', Foteini Stavropoulou, 15 Jun 2023
-
RC2: 'Comment on egusphere-2023-247', Anonymous Referee #2, 03 May 2023
general comments
The work reported in this paper is from a co-ordinated field campaign conducted between many groups, measuring CH4 from oil production sites across Romania using four different approaches. The scientific question is firmly within ACP’s remit, and contains well established methods and statistical analyses to interrogate the results. The report is laid out clearly, with assumptions and methods presented well. Much of the detail of the methodologies is in the SI. I note that another reviewer has recommended to bring some of this into the main text. I would support this, although I think this is somewhat down to personal preference, as I did not see any guidance on this from the journal itself.
Substantial conclusions are reached, showing that methane emissions from these sources are over twice as high as currently reported, thus highlighting the inadequacy of the methods uses to report emissions in this sector and the potential for mitigation by stopping the leaks.
specific comments
L391 and elsewhere: open-ended lines is a term which is used, and is somewhat self explanatory, however it would help to define it for those not within this field. e.g. it is not clear whether these lines are designed like this, or if they should have a cap on the end but for some reason the cap has been left off.
L510 and 533: what are the Tier 1 EFs for developing and developed countries? Presumably the currently used EFs for reporting are quite different to what you have derived. This value would be useful to compare with the EF you have derived, eg in the conclusion as a comparison (and earlier when you discuss the IPCC methodology).
technical corrections
L67 and later: % symbol should not have a space before it, i.e. it should be 25% and not 25 %
L75 L80: Is Global Methane Tracker 2022, 2022 correctly referenced? Just checking, as it looks odd with 2022 appearing twice
Citation: https://doi.org/10.5194/egusphere-2023-247-RC2 -
AC1: 'Reply on RC2', Foteini Stavropoulou, 15 Jun 2023
We thank the anonymous referee for the valuable feedback and comments, which have helped to improve our manuscript. Responses to individual referee comments are uploaded in the form of a supplement. Line numbers refer to the updated WORD manuscript with tracked changes.
-
AC1: 'Reply on RC2', Foteini Stavropoulou, 15 Jun 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 924 | 264 | 20 | 1,208 | 56 | 13 | 12 |
- HTML: 924
- PDF: 264
- XML: 20
- Total: 1,208
- Supplement: 56
- BibTeX: 13
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Foteini Stavropoulou
Katarina Vinković
Bert Kers
Marcel de Vries
Steven van Heuven
Piotr Korbeń
Martina Schmidt
Julia Wietzel
Pawel Jagoda
Jaroslav M. Necki
Jakub Bartyzel
Hossein Maazallahi
Malika Menoud
Carina van der Veen
Sylvia Walter
Béla Tuzson
Jonas Ravelid
Randulph Paulo Morales
Lukas Emmenegger
Dominik Brunner
Michael Steiner
Arjan Hensen
Ilona Velzeboer
Pim van den Bulk
Hugo Denier van der Gon
Antonio Delre
Maklawe Essonanawe Edjabou
Charlotte Scheutz
Marius Corbu
Sebastian Iancu
Denisa Moaca
Alin Scarlat
Alexandru Tudor
Ioana Vizireanu
Andreea Calcan
Magdalena Ardelean
Sorin Ghemulet
Alexandru Pana
Aurel Constantinescu
Lucian Cusa
Alexandru Nica
Calin Baciu
Cristian Pop
Andrei Radovici
Alexandru Mereuta
Horatiu Stefanie
Bas Hermans
Stefan Schwietzke
Daniel Zavala-Araiza
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1187 KB) - Metadata XML
-
Supplement
(1219 KB) - BibTeX
- EndNote
- Final revised paper