the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Dec 2024
Aircraft Observations of Continental Pollution In the Equatorial Lower Stratosphere over the Tropical Western Pacific During Boreal Winter
Abstract. Recent studies hypothesize that emissions from fires reaching the stratosphere can provide aerosols and aerosol precursors that initiate stratospheric ozone loss and lead to radiative heating of the stratosphere and cooling of the surface. Air from the troposphere enters the stratosphere primarily over the tropical western Pacific (TWP) during boreal winter. We report observations in the TWP of persistent, ubiquitous continental pollution in the tropical tropopause layer (TTL) and lower stratosphere (LS) during the Airborne Tropical TRopopause EXperiment (ATTREX) campaign in February–March 2014. We found concentrations of carbon monoxide (CO) enhanced up to 65 % over background levels in the deep tropics (5° S–15° N, 16 –17 km). Correlations of CO with hydrocarbon and halocarbon species indicated a biomass burning source, with the largest CO enhancements found in warmer, clear air. Satellite observations of CO did not detect the thin pollution layers observed by the aircraft, but did indicate Africa, Indonesia, and the western/central Pacific as geographical hot spots for CO in the TTL. Backward trajectories identified convective encounters in these areas as the dominant sources of polluted air in the TWP. Africa and Indonesia contributed about 60 % of the excess CO, transported to the TWP in two to four weeks. Our study confirms that air in the TTL over the TWP is affected by emissions from distant fires that can rapidly reach the LS in the principal source region for air entering the stratosphere, supporting the view that fires in tropical regions could impact stratospheric ozone and temperatures.
-
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
(4780 KB)
-
Supplement
(1217 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4780 KB) - Metadata XML
-
Supplement
(1217 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3832', Anonymous Referee #1, 13 Jan 2025
The manuscript "Aircraft Observations of Continental Pollution In the Equatorial Lower Stratosphere over the Tropical Western Pacific During Boreal Winter" by Pittmann et al. presents observations from the ATTREX campaign 2014 over the tropical western Pacific (TWP) region obtained with the Global Hawk.Â
The paper describes the dataset and comes to the conclusion that the UTLS composition in the TWP is dominated during the campaign time by continental pollution.
In my point of view the paper is in general well written, but some major and minor revisions should be conducted for a final publication in ACP.1) The pollution is termed "continental pollution". Even though the analysed pollution is of continental origin, the authors state that this is mostly from biomass burning pollution. Consequently, a re-naming of the manuscript should be considered, as the influence of anthropogenic pollution in the TWP UTLS is minor.
2) The authors make both in the abstract as well as in the conclusions statements about aerosol particles. However, the measurement data does not obtain any data point on aerosol particles. Even though aerosol particles are potentially included in the emission sources, their fate and transformation in the convective ascent as well as microphysical processing and / or sedimentation along the relatively long transit times is uncertain. I would expect substantial scavenging of primary emitted particles in the ascent, either by precipitation scavenging or by ice nucleation such that only a minor fraction of primary aerosol particles will be transported into the UTLS. Whether new particle formation also takes place in the TWP or in the transit to the measurement region is not yet certain. Therefore, the TWP cannot be named an ascent region of aerosols into the LS. And as this is not shown in this manuscript (and it is not its intention), these statements should be either weakened and discussed or should be removed from the manuscript.
Similarly, the role of VSLS are prominently mentioned in the beginning, but the analysis showed that they do not play a significant role in this context, i.e., as lofted species with a high ozone destruction potential.3) The back-trajectories are calculated for 40 days. This is a very long time period for air mass transport, without the consideration of diabatic processes (cloud processing), turbulence and small-scale mixing, such that the longest trajectories only represent a very crude approximation for the air mass history. Especially trajectories who travel over the Indian Ocean to Africa and from there back to the TWP region must be treated carefully. These limitations should be discussed in the manuscript.
The occurrence of convection associated with the trajectories also has to be discussed in a little more detail. As the trajectories are calculated from ERA5 dynamics, there is no guarantee that individual convective cells and elements on the order of <30 km are represented properly in the meteorological re-analysis data. Of course, the satellite data represent reality, but it should be analysed / discussed whether the convection in the re-analysis is co-located with the observed convective activity or whether this activity is sufficiently wide-spread such that air parcels in the convective region have a high probability to be lofted.
The analysis of air parcels in convection also reveals that it is not guaranteed that an air parcel actually has seen the convection, only because it has been in its vicinity: it could also happen that air parcels simply stream around the anvil of the convective towers and are not substantially modified by lofted near surface air.4) The selected / presented compounds often have quite enhanced life times, especially CH4 and CO2, such that convective signals of events more than a few days ago can hardly be distinguished from background conditions. Consequently, the profiles from Fig.2 for methane and carbon dioxide provide a measure for the substantial background variability. For species with shorter lifetime, like CO, the variability is smaller, even though the regimes of convectively and non-convectively influenced air masses cannot be easily distinguished (in Fig.2). In Fig.3 some of the flights show a distinct enhanced CO in the UTLS, which could indicate convective outflow; however, after travel times of more than a week, it is unlikely, that these structures remain that clearly visible, indicating more local or regional convection and further lofting, e.g., by convectively induced gravity waves and their mixing processes.
On the other hand, the shorter lived compounds (C3H8, C2H2) hardly show enhancements in the profiles. Given their shorter chemical lifetime, the air masses must have been lofted several days prior to the observations, if only in the Northern observations any signal can be identified. This rules out transport from Africa, but indicates more regional continental pollution. Even the CO signal in the Southern observations is quite low, and hardly shows UT enhancements. This does not fit well with the analysis from Fig. 14 that the corresponding trajectories could have seen convection only two weeks ago. A separation of continental convection or biomass burning in the vicinity of convection or anthropogenic activity near the convection might elucidate the origin or missing signals a bit better. This is most obvious for the Southern observations, but would also hold true for the rest of the data.5) The MLS data does not really sheds light onto the observations, as the signal cannot be properly distinguished from the background. Therefore, the manuscript could be shortened here, and the figures 12 and 13 could be moved to an appendix or supplement.
6) The pollution is mostly determined in the cloud-free air. As the authors state, that the pollution is mostly transported for several days, and that the trajectories are slowly lofted by radiative cooling, this is not very surprising. On the other hand this is a bit contradictionary to the finding of Fig. 11, which states that the pollution is usually associated with warmer air masses. As the typical temperature profile does not increase quickly in the low latitudes above the thermal tropopause, this needs an explanation.
7) I am missing an explanation, why the pollution is not found in cloudy or moist regions. Is this because the polluted air is at higher altitude than the clouds, i.e., because of the above mentioned lofting?
8) What information can be deduced from Fig. 9, which is not also included in Fig. 10? Can you (re)move Fig. 9 to the supplement, in order to slightly shorten the manuscript and reformulate the statements from the corresponding paragraph? Basically, the same conclusions, i.e., that the highest pollution levels are found in non-saturated regions, can be drawn from Fig. 10, which has better representation via the color coded visibility.
However, the color scale of Fig.10 (using rainbow colours) could be modified to be more perceptually uniform and color-blind / black-white friendly.9) Please discuss the variability in the vertical motion of the individual trajectories: especially trajectories from Africa travel more than 50 hPa upwards and downwards on timescales of a day or two (Fig. 14c, around day 13 to 21). What processes would you associate these relatively rapid air motions including downwelling, before the radiatively driven ascent prior to the measurements?
Citation: https://doi.org/10.5194/egusphere-2024-3832-RC1 - AC1: 'Reply on RC1', Jasna Pittman, 13 Mar 2025
-
RC2: 'Comment on egusphere-2024-3832', Anonymous Referee #2, 22 Jan 2025
Pittman et al. present in their manuscript in-situ measurements from ATTRAX campaigns, focusing on polluted air masses in connection to biomass burning. They analyze a number of pollution trace gases with different origin or atmospheric lifetimes. Further, they use MLS CO data for contextual information for the aircraft measurements, as well as backward trajectories in order to estimate the origin of measured air masses.
The manuscript is well-written and certainly within the scope of ACP and deserves publication after addressing my remarks below.General remarks:
- The introduction starts with the importance to observe pollution, which may enter the upper troposphere or even the lower stratosphere because of its potential to destroy ozone, and I fully agree that such kind of measurements are very important. However, the link to ozone is in the main part of the manuscript rather weak and is only briefly mentioned again in the conclusions.
- Section 3 starts with the presentation of the trace gases CO2, CH4 and CO, but in the instrument description section, it was mentioned that many more trace gases have been measured during the ATTREX campaigns. I would like to know why the authors chose to limit themselves only to these three trace gases.
- It is not clear to me why the UCATS instrument is introduced, but the O3 measurements are never used in this work.
- In general, the structure of section 3 should be better motivated (maybe in the introduction to this section?): Why are there subsections focusing on (3.1 & 3.3) CO2, CO, CH4, while others (3.2) focus on VSLS/HCs?
- A number of figures in this study use the "rainbow/jet" colormap, which is not suited to quantitatively represent data. I suggest to change to a colormap, which have color changes, which are perceptually constant (e.g. "turbo", "viridis" or "inferno")Specific points:
- Introduction: I think also studies showing stratospheric entry of polluted air masses in mid-latitudes should be briefly mentioned here. In particular about the Australian New Year fire (e.g. Khaykin et al., 2020 or Schwartz et al., 2020) and the Canadian wildfires (e.g. Pumphrey et al. 2020), which have been exceptional events.
- Section 2.3: The authors could mention if they have followed the recommended data screening for the MLS data (as explained in the MLS data quality document).
- Line 89: Here, three remote sensing instruments are mentioned, but it seems like they are not used for this study? I suggest to briefly mention why these measurements were not used in this study.
- Fig. 1: I understood the text that only a subset of the ATTREX-2 and ATTREX-3 flights are used in this study, while all of the flights are shown in this figure. I suggest to differentiate between the flights used in the study and the rest of them. E.g. the colors for the flights not used could be lighter, or the lines could be dotted.
- Line 132: Please use metric units (at least additionally to psia)
- Line 211: "Contribution from these processes could be evaluated by examining additional trace gases such as O3": So why is O3 not analyzed here, since it is available from the UCATS instrument?
- Figures 2&3: I suggest to combine these figures (and use a different colormap for latitudes and mixing ratios). I like the two perspectives on the data, but the figures are discussed together, so I would not separate them here. Further, label font sizes are very small for Fig. 3, please change it to the same style as Fig. 2.
- Figure 6: Even though it is mentioned in the caption, I suggest to label the rows accordingly directly in the figure to help the reader.
- Line 268 and Figure 6: I was wondering why the gases are always listed in this order. I would have ordered it according to the atmospheric lifetime.
- Line 303: "In all instances, HCs and CH3Cl were positively correlated ...": I think that one can only guess the correlation based on the plots provided, and only for the NH bin (where a better suited plot is provided in Fig. 7, which is not mentioned here). For the other bins, data points are sparse (in particular the pollution points), so I find it hard to see a correlation there.
- Figure 7: The grey open circles are hard to see at all, they are much too small. A legend explaining the different shapes of the points would also help (even though it is mentioned in the caption). Did I understand this correctly, that Fig. 7 only shows a subset of the data used in Fig. 6? Further, it would help to remind the reader on the approximate lifetimes of the substances shown in the panels, since they are discussed in the text.
- Figure 9: I'm not sure if I have understood the benefit of this figure. What arguments are based only on this figure, which cannot be explained by Figure 10? I think Figure 9 could be moved to a paper supplement.
- Line 383: "No correlation between temperature and CO was observed at these low H2O levels.": I think there is a weak (anti-)correlation with temperature in the plot, at least the points are getting darker blue for warmer temperatures?
- Line 384: "The second regime is air masses with H2O above 5 ppmv ...": Why is the threshold chosen to be 5 ppmv and not (e.g.) 4.5 ppmv?
- Line 392: "These warmer and polluted air masses were encountered 15 % of the time in the 3.5 to 5 ppmv H2O range and at altitudes between 16 and 17 km.": I do not understand this sentence, please rephrase.
- Line 513: "A ten-year MLS record showed...": Maybe I missed it, but there was no 10-year MLS record shown or mentioned in this manuscript yet? I think this thought should be discussed before the conclusions section.
- Line 550: Please also mention where to find ERA5 data (used for the trajectories). Further, is the trajectory model (a specific name is not mentioned in the manuscript) used in this work available? Do you have a website for that?Citation: https://doi.org/10.5194/egusphere-2024-3832-RC2 - AC2: 'Reply on RC2', Jasna Pittman, 13 Mar 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3832', Anonymous Referee #1, 13 Jan 2025
The manuscript "Aircraft Observations of Continental Pollution In the Equatorial Lower Stratosphere over the Tropical Western Pacific During Boreal Winter" by Pittmann et al. presents observations from the ATTREX campaign 2014 over the tropical western Pacific (TWP) region obtained with the Global Hawk.Â
The paper describes the dataset and comes to the conclusion that the UTLS composition in the TWP is dominated during the campaign time by continental pollution.
In my point of view the paper is in general well written, but some major and minor revisions should be conducted for a final publication in ACP.1) The pollution is termed "continental pollution". Even though the analysed pollution is of continental origin, the authors state that this is mostly from biomass burning pollution. Consequently, a re-naming of the manuscript should be considered, as the influence of anthropogenic pollution in the TWP UTLS is minor.
2) The authors make both in the abstract as well as in the conclusions statements about aerosol particles. However, the measurement data does not obtain any data point on aerosol particles. Even though aerosol particles are potentially included in the emission sources, their fate and transformation in the convective ascent as well as microphysical processing and / or sedimentation along the relatively long transit times is uncertain. I would expect substantial scavenging of primary emitted particles in the ascent, either by precipitation scavenging or by ice nucleation such that only a minor fraction of primary aerosol particles will be transported into the UTLS. Whether new particle formation also takes place in the TWP or in the transit to the measurement region is not yet certain. Therefore, the TWP cannot be named an ascent region of aerosols into the LS. And as this is not shown in this manuscript (and it is not its intention), these statements should be either weakened and discussed or should be removed from the manuscript.
Similarly, the role of VSLS are prominently mentioned in the beginning, but the analysis showed that they do not play a significant role in this context, i.e., as lofted species with a high ozone destruction potential.3) The back-trajectories are calculated for 40 days. This is a very long time period for air mass transport, without the consideration of diabatic processes (cloud processing), turbulence and small-scale mixing, such that the longest trajectories only represent a very crude approximation for the air mass history. Especially trajectories who travel over the Indian Ocean to Africa and from there back to the TWP region must be treated carefully. These limitations should be discussed in the manuscript.
The occurrence of convection associated with the trajectories also has to be discussed in a little more detail. As the trajectories are calculated from ERA5 dynamics, there is no guarantee that individual convective cells and elements on the order of <30 km are represented properly in the meteorological re-analysis data. Of course, the satellite data represent reality, but it should be analysed / discussed whether the convection in the re-analysis is co-located with the observed convective activity or whether this activity is sufficiently wide-spread such that air parcels in the convective region have a high probability to be lofted.
The analysis of air parcels in convection also reveals that it is not guaranteed that an air parcel actually has seen the convection, only because it has been in its vicinity: it could also happen that air parcels simply stream around the anvil of the convective towers and are not substantially modified by lofted near surface air.4) The selected / presented compounds often have quite enhanced life times, especially CH4 and CO2, such that convective signals of events more than a few days ago can hardly be distinguished from background conditions. Consequently, the profiles from Fig.2 for methane and carbon dioxide provide a measure for the substantial background variability. For species with shorter lifetime, like CO, the variability is smaller, even though the regimes of convectively and non-convectively influenced air masses cannot be easily distinguished (in Fig.2). In Fig.3 some of the flights show a distinct enhanced CO in the UTLS, which could indicate convective outflow; however, after travel times of more than a week, it is unlikely, that these structures remain that clearly visible, indicating more local or regional convection and further lofting, e.g., by convectively induced gravity waves and their mixing processes.
On the other hand, the shorter lived compounds (C3H8, C2H2) hardly show enhancements in the profiles. Given their shorter chemical lifetime, the air masses must have been lofted several days prior to the observations, if only in the Northern observations any signal can be identified. This rules out transport from Africa, but indicates more regional continental pollution. Even the CO signal in the Southern observations is quite low, and hardly shows UT enhancements. This does not fit well with the analysis from Fig. 14 that the corresponding trajectories could have seen convection only two weeks ago. A separation of continental convection or biomass burning in the vicinity of convection or anthropogenic activity near the convection might elucidate the origin or missing signals a bit better. This is most obvious for the Southern observations, but would also hold true for the rest of the data.5) The MLS data does not really sheds light onto the observations, as the signal cannot be properly distinguished from the background. Therefore, the manuscript could be shortened here, and the figures 12 and 13 could be moved to an appendix or supplement.
6) The pollution is mostly determined in the cloud-free air. As the authors state, that the pollution is mostly transported for several days, and that the trajectories are slowly lofted by radiative cooling, this is not very surprising. On the other hand this is a bit contradictionary to the finding of Fig. 11, which states that the pollution is usually associated with warmer air masses. As the typical temperature profile does not increase quickly in the low latitudes above the thermal tropopause, this needs an explanation.
7) I am missing an explanation, why the pollution is not found in cloudy or moist regions. Is this because the polluted air is at higher altitude than the clouds, i.e., because of the above mentioned lofting?
8) What information can be deduced from Fig. 9, which is not also included in Fig. 10? Can you (re)move Fig. 9 to the supplement, in order to slightly shorten the manuscript and reformulate the statements from the corresponding paragraph? Basically, the same conclusions, i.e., that the highest pollution levels are found in non-saturated regions, can be drawn from Fig. 10, which has better representation via the color coded visibility.
However, the color scale of Fig.10 (using rainbow colours) could be modified to be more perceptually uniform and color-blind / black-white friendly.9) Please discuss the variability in the vertical motion of the individual trajectories: especially trajectories from Africa travel more than 50 hPa upwards and downwards on timescales of a day or two (Fig. 14c, around day 13 to 21). What processes would you associate these relatively rapid air motions including downwelling, before the radiatively driven ascent prior to the measurements?
Citation: https://doi.org/10.5194/egusphere-2024-3832-RC1 - AC1: 'Reply on RC1', Jasna Pittman, 13 Mar 2025
-
RC2: 'Comment on egusphere-2024-3832', Anonymous Referee #2, 22 Jan 2025
Pittman et al. present in their manuscript in-situ measurements from ATTRAX campaigns, focusing on polluted air masses in connection to biomass burning. They analyze a number of pollution trace gases with different origin or atmospheric lifetimes. Further, they use MLS CO data for contextual information for the aircraft measurements, as well as backward trajectories in order to estimate the origin of measured air masses.
The manuscript is well-written and certainly within the scope of ACP and deserves publication after addressing my remarks below.General remarks:
- The introduction starts with the importance to observe pollution, which may enter the upper troposphere or even the lower stratosphere because of its potential to destroy ozone, and I fully agree that such kind of measurements are very important. However, the link to ozone is in the main part of the manuscript rather weak and is only briefly mentioned again in the conclusions.
- Section 3 starts with the presentation of the trace gases CO2, CH4 and CO, but in the instrument description section, it was mentioned that many more trace gases have been measured during the ATTREX campaigns. I would like to know why the authors chose to limit themselves only to these three trace gases.
- It is not clear to me why the UCATS instrument is introduced, but the O3 measurements are never used in this work.
- In general, the structure of section 3 should be better motivated (maybe in the introduction to this section?): Why are there subsections focusing on (3.1 & 3.3) CO2, CO, CH4, while others (3.2) focus on VSLS/HCs?
- A number of figures in this study use the "rainbow/jet" colormap, which is not suited to quantitatively represent data. I suggest to change to a colormap, which have color changes, which are perceptually constant (e.g. "turbo", "viridis" or "inferno")Specific points:
- Introduction: I think also studies showing stratospheric entry of polluted air masses in mid-latitudes should be briefly mentioned here. In particular about the Australian New Year fire (e.g. Khaykin et al., 2020 or Schwartz et al., 2020) and the Canadian wildfires (e.g. Pumphrey et al. 2020), which have been exceptional events.
- Section 2.3: The authors could mention if they have followed the recommended data screening for the MLS data (as explained in the MLS data quality document).
- Line 89: Here, three remote sensing instruments are mentioned, but it seems like they are not used for this study? I suggest to briefly mention why these measurements were not used in this study.
- Fig. 1: I understood the text that only a subset of the ATTREX-2 and ATTREX-3 flights are used in this study, while all of the flights are shown in this figure. I suggest to differentiate between the flights used in the study and the rest of them. E.g. the colors for the flights not used could be lighter, or the lines could be dotted.
- Line 132: Please use metric units (at least additionally to psia)
- Line 211: "Contribution from these processes could be evaluated by examining additional trace gases such as O3": So why is O3 not analyzed here, since it is available from the UCATS instrument?
- Figures 2&3: I suggest to combine these figures (and use a different colormap for latitudes and mixing ratios). I like the two perspectives on the data, but the figures are discussed together, so I would not separate them here. Further, label font sizes are very small for Fig. 3, please change it to the same style as Fig. 2.
- Figure 6: Even though it is mentioned in the caption, I suggest to label the rows accordingly directly in the figure to help the reader.
- Line 268 and Figure 6: I was wondering why the gases are always listed in this order. I would have ordered it according to the atmospheric lifetime.
- Line 303: "In all instances, HCs and CH3Cl were positively correlated ...": I think that one can only guess the correlation based on the plots provided, and only for the NH bin (where a better suited plot is provided in Fig. 7, which is not mentioned here). For the other bins, data points are sparse (in particular the pollution points), so I find it hard to see a correlation there.
- Figure 7: The grey open circles are hard to see at all, they are much too small. A legend explaining the different shapes of the points would also help (even though it is mentioned in the caption). Did I understand this correctly, that Fig. 7 only shows a subset of the data used in Fig. 6? Further, it would help to remind the reader on the approximate lifetimes of the substances shown in the panels, since they are discussed in the text.
- Figure 9: I'm not sure if I have understood the benefit of this figure. What arguments are based only on this figure, which cannot be explained by Figure 10? I think Figure 9 could be moved to a paper supplement.
- Line 383: "No correlation between temperature and CO was observed at these low H2O levels.": I think there is a weak (anti-)correlation with temperature in the plot, at least the points are getting darker blue for warmer temperatures?
- Line 384: "The second regime is air masses with H2O above 5 ppmv ...": Why is the threshold chosen to be 5 ppmv and not (e.g.) 4.5 ppmv?
- Line 392: "These warmer and polluted air masses were encountered 15 % of the time in the 3.5 to 5 ppmv H2O range and at altitudes between 16 and 17 km.": I do not understand this sentence, please rephrase.
- Line 513: "A ten-year MLS record showed...": Maybe I missed it, but there was no 10-year MLS record shown or mentioned in this manuscript yet? I think this thought should be discussed before the conclusions section.
- Line 550: Please also mention where to find ERA5 data (used for the trajectories). Further, is the trajectory model (a specific name is not mentioned in the manuscript) used in this work available? Do you have a website for that?Citation: https://doi.org/10.5194/egusphere-2024-3832-RC2 - AC2: 'Reply on RC2', Jasna Pittman, 13 Mar 2025
Peer review completion
Journal article(s) based on this preprint
Data sets
NASA ATTREX E. J. Jensen et al. https://espoarchive.nasa.gov/archive/browse/attrex/id4
MLS N. Livesey et al. https://disc.gsfc.nasa.gov/datasets?page=1&keywords=AURA%20MLS
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 255 | 65 | 21 | 341 | 39 | 19 | 37 |
- HTML: 255
- PDF: 65
- XML: 21
- Total: 341
- Supplement: 39
- BibTeX: 19
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Jasna V. Pittman
Bruce C. Daube
Steven C. Wofsy
Elliot L. Atlas
Maria A. Navarro
Eric J. Hintsa
Fred L. Moore
Geoff S. Dutton
James W. Elkins
Troy D. Thornberry
Andrew W. Rollins
Eric J. Jensen
Thaopaul Bui
Jonathan M. Dean-Day
Leonhard Pfister
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4780 KB) - Metadata XML
-
Supplement
(1217 KB) - BibTeX
- EndNote
- Final revised paper