the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 30 Apr 2024
Atmospheric River Induced Precipitation in California as Simulated by the Regionally Refined Simple Convection Resolving E3SM Atmosphere Model (SCREAM) Version 0
Abstract. Using the Regionally Refined Mesh (RRM) configuration of the U.S. Department of Energy's Simple Cloud Resolving E3SM Atmosphere Model (SCREAM), we simulate and evaluate four meteorologically distinct atmospheric river events over California. We test five different RRM configurations, each differing in terms of the areal extent of the refined mesh and the resolution (ranging from 800 m to 3.25 km). We find that SCREAM-RRM generally has a good representation of the AR generated precipitation in CA, even for the control simulation which has a very small 3 km refined patch, and is able to capture the fine scale regional distributions that are controlled largely by the fine scale topography of the state. Although, it is found that SCREAM generally has a wet bias over topography, most prominently over the Sierra Nevada mountain range, with a corresponding dry bias on the lee side. We find that refining the resolution beyond 3 km (specifically 1.6 km and 800 m) has virtually no benefit towards reducing systematic precipitation biases, but that improvements can be found when increasing the areal extent of the upstream refined mesh. However, these improvements are relatively modest and only realized if the size of the refined mesh is expanded to the scale where employing RRM no longer achieves the substantial cost benefit it was intended for.
-
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
(11062 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(11062 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-839', Anonymous Referee #1, 18 May 2024
Summary:
Bogenshutz et al. present four detailed case studies chosen to highlight different types and aspects of ARs impacting the California coast using the ultra-high regionally refined E3SM atmosphere model, SCREAM. The purpose of the study is to show SCREAM’s potential utility simulating ARs for both climate and weather communities. Their main findings, aside from the realism of AR representation, include (1) refining resolution beyond 3km has little impact for improving precipitation biases, but (2) expanding the spatial extent of the refinement has modest benefits.
Overall Comments:
This is a very nice paper. It is clear, well-written and organized, and absolutely fits within the GMD scope. I have a few comments I would like to see addressed before full publication. Although these are relatively minor, they are important for clarity. Please see specific comments below, but the most important ones are regarding SCREAM model description and method/domain for computing max/mean IVT.
Specific Comments:
Line 18 and elsewhere: Typically the words “atmospheric rivers” are not capitalized.
Line 89: Out of curiosity, when you moved to using SHOC from a simplified PBL, was there anything that needed to be re-tuned?
Figure 2: Did you mean ne32 and not ne21?
SCREAM description and References: There are missing citations in the reference list, I definitely noticed both Caldwell et al 2021 and 2019 missing as I searched for these to get more information on SCREAM. Also, I recommend adding a bit more to the description part of the paper for ease rather than digging into the cited Caldwell papers, i.e. vertical resolution, features, biases, improvements not already mentioned, etc.
AR cases (these are more curiosity questions than comments on the paper itself): There are all great choices and span important and different types of ARs and dynamics. I am curious, however, on how SCREAM-RRR would perform for 1) a weak Cat 1 storm, or 2) different regions , or 3) wet vs windy “flavors”? Perhaps these are already planned activities, and I am not suggesting these be done for this paper. (I understand the intense computational and person-resources involved in these activities). Some questions to consider for future experiments (or answer if you have already done some of this): For a weak category 1 AR, or a “windy” AR, are there components of physics, dycore, and/or resolution that do better with an overwet environment vs not? Are biases different across the varied AR dynamics, i.e. Pineapple Express vs North Atlantic warm conveyor belt lift? (Has this been looked at)? Are there plans to look at moisture transports such as ARs within monsoonal flow which can be found in the E.Pacific?
Line 148 and elsewhere: The acronym “AR” only needs to be defined once.
Line 253: Why not add polygons to one of your maps with the locations of each of your evaluation points? This would be helpful visualizing where the skill is being evaluated. (Or maybe an inset with a zoomed map)?
Line 287 and Figure 5c: Pearson pattern correlation?
Line 327: Comparing SC2005 reanalysis and SCREAM, the character of the IVT plume and footprint are very different. One might argue that SCREAM really doesn’t capture the AR structure well here, especially in the ocean. Ideas as to why?
Lines 359-360: I understand the tension between the improvements via resolution and mesh area, but the other degree of freedom here is E3SM vs CESM when consulting with the results from the Rhoades paper. I think(?) that the vertical levels are different between these two modelling frameworks? How different is SCREAM vertical structure from the Rhoades study, and do you think this may matter regarding convection schemes?
Line 401: Did you mean Figure 10 (and not 3)?
Figure 10: I find it interesting that the EPAC 3km case seems to overdo the snow compared to the CA 3km in most cases, i.e. most notably at Virginia Lakes for NC2021 and everywhere SC2005. Thoughts?
Line 428, Figures 12/13, Lines 440-447: For the event maximum/mean IVT, did you use find, at each grid point in your RR domain, the maximum/mean values and plot this? Or just across the plotting extent in Fig 12/13? For SC2005 where SCREAM’s IVT core could arguably be over the ocean (based on Figure 4), this might be relevant. Are some of SCREAMs biases due to AR characterization and misplacement of the core, and not really just the fact that the plume hitting the coast is more narrow? For clarity, please add a sentence or two clarifying IVT max/mean methodology.
Citation: https://doi.org/10.5194/egusphere-2024-839-RC1 - AC1: 'Reply on RC1', Peter Bogenschutz, 31 Jul 2024
-
CC1: 'Referee comments on Bogenschutz et al. "Atmospheric River Induced Precipitation in California as Simulated by the Regionally Refined Simple Convective Resolving E3SM Atmosphere Model (SCREAM) Version 0"', James Benedict, 17 Jun 2024
Publisher’s note: this comment is a copy of RC2 and its content was therefore removed.
Citation: https://doi.org/10.5194/egusphere-2024-839-CC1 -
AC3: 'Reply on CC1', Peter Bogenschutz, 31 Jul 2024
Please see reply to reviewer 2 comment.
Citation: https://doi.org/10.5194/egusphere-2024-839-AC3
-
AC3: 'Reply on CC1', Peter Bogenschutz, 31 Jul 2024
-
RC2: 'Comment on egusphere-2024-839', James Benedict, 18 Jun 2024
General comments:
The study examines precipitation, snowfall, and surface temperature characteristics associated with 4 atmospheric river (AR) cases impacting California as simulated in the Simple Cloud Resolving E3SM (Energy Exascale Earth System Model) Atmosphere Model (SCREAM) with selected regionally refined grid mesh (RRM) configurations. Sensitivity of the results to the RRM horizontal resolution (3.25 km, 1.6 km, and 800 m) and upstream extent of the RRM domain is evaluated against both high-resolution gridded data sets and single-station time series. Generally, SCREAM is able to capture the fine-scale regional distributions of AR-related precipitation over California. However, the model systematically overestimates precipitation over upwind and higher-elevation areas, and underestimates precipitation on leeward sides of mountain ranges. SCREAM also tends to overestimate storm-total snowfall and mean surface temperatures during the AR events. Overall, it is found that precipitation and temperature distributions exhibit only modest sensitivity to RRM resolution. Although somewhat greater improvement is seen for the largest RRM domain extent due to better representation of large-scale meteorological patterns that guide the ARs, the authors assert that the required RRM domain expansion no longer achieves the intended computational cost savings.
I found the modeling approach and scientific assessment to be both clear and informative. The detail of the model evaluation was appropriate given the fairly large number of simulations conducted and AR cases examined.
A key finding of the paper, in my view, is the relative insensitivity of storm-total precipitation to model horizontal grid resolution in the range from 3.25 km to 800 m. This finding also highlights the current horizontal resolution limit of evaluation data sets (4 km for daily surface precipitation, and 25 km for ERA5 atmospheric fields). At such fine scales, the need would arise to start evaluating models against single meteorological stations, as the authors have done here.
Findings related to the RRM domain extent are also novel. The authors note that SCREAM is initialized with ERA5 but allowed to evolve freely thereafter. In my opinion, initializing the model 24 hours (or 18 or 30 hours, based on sensitivity tests) is perhaps a “low bar” to assess model performance, though I suppose longer lead times might result in a drift of the large-scale meteorological solution such that evaluation of AR impacts could become complicated. I view the paper’s findings mostly as an examination of AR impacts from simulations in which the large-scale meteorological patterns are quasi-prescribed, given the short lead time.
Specific comments:
L26-28: Rutz et al. (2014, https://doi.org/10.1175/MWR-D-13-00168.1) displays one of the earliest and nicest maps of AR contributions to total cool-season precipitation (their Fig. 8). Consider adding this reference here.
L49-51: Can a reference to these modeling improvement be added here?
L94-97: “SCREAMv0 used a prescribed-aerosol version of E3SM’s modal aerosol model, however in this work we use a version of SCREAM that prescribes both cloud-condensation nuclei number and aerosol radiative properties from an E3SMv2 simulation. This is known as Simple Prescribed Aerosol (SPA) and will be incorporated into SCREAMv1.” Do the authors think that this simplification, especially the prescribed CCN, could potentially be contributing to the “eagerness” of SCREAM to overestimate precipitation on the upwind side of mountain ranges?
Figs. 2 and 3: In future depictions of RRMs, it is recommended to recreate the plot formatting like the right panel in Fig. 1 so that land masses and coastlines are visible.
L159: “most locations” is too vague
L163-164: ERA5 and IVT should be defined here, as it's their first usage.
L169-170: D4 drought classification should be referenced.
L214-216: Three references on the large-scale meteorological patterns that precede west coast (of U.S.) AR landfalls:
——
Benedict, J. J., Clement, A. C., & Medeiros, B. (2019). Atmospheric blocking and other large‐scale precursor patterns of landfalling atmospheric rivers in the North Pacific: A CESM2 study. Journal of Geophysical Research: Atmospheres, 124, 11,330–11,353. https://doi.org/10.1029/2019JD030790Zhou, Y., & Kim, H. (2019). Impact of Distinct Origin Locations on the Life Cycles of Landfalling Atmospheric Rivers Over the U.S. West Coast. Journal of Geophysical Research: Atmospheres, 124, 11,897–11,909. https://doi.org/10.1029/2019JD031218
Carrera, M. L., Higgins, R. W., & Kousky, V. E. (2004). Downstream weather impacts associated with atmospheric blocking over the northeast Pacific. Journal of Climate, 17(24), 4823–4839. https://doi.org/10.1175/jcli‐3237.1
——L248-249: SNOTEL and Cooperative Observer Program should be referenced.
L273-274: It would be helpful to clarify how “bias” is calculated, as well as what spatial domain is used in the calculations for Fig. 5. The author’s use of “statewide” in L276 leads me to believe that the entire state of California is being used as the spatial domain for evaluation, but this should be more clearly stated.
A map containing locations (mountain ranges, counties) highlighted in the text should be added to help readers not familiar with California geography.
In Figs. 12 and 13, what temporal resolutions are used for SCREAM and ERA5 to evaluate maximum IVT? This should be noted in the L428 paragraph.
L486-494: It would greatly benefit the paper to add references to specific figures when summarizing the key findings.
L511: At the end of this paragraph, it might be good to add a sentence noting that model evaluation at sub-kilometer scales is challenged by the availability of observations needed to evaluate the simulations… though I suppose a comparison to a single station would be appropriate in some cases.
Technical Corrections:
L2: E3SM should be defined.
L42: “generation…have” —> “generation…has”
L62: “Winter Hydroclimate” need not be capitalized.
L73-74: “resolution is increased to 1.6 km”: It would be better to state something like “resolution is increased from 3 km to 1.6 km…”
L77: “5 km though, that work” —> “5 km, though that work”
The Fig. 1 caption seems to be missing one or more words.
Fig. 2 caption: Should be “ne32”, not “ne21”.
L135: Missing period at end of sentence.
L171: ; —> ,
L215: “set up” (verb) —> “setup” (noun)
L232: letting —> allowing
L281-283: Remove “Though”, add comma after “scores”, and change semicolon to comma. Also, “grids and there can be implications” —> “grids. There can be implications…”
L310: ; —> ,
L388: models —> model’s
L401: Should be Fig. 10, not Fig. 3.
L402: “though” —> “though they”
L403: Lakes —> Lake
L405: ; —> ,
L425: ; —> ,
L455: ; —> ,
L495: “The aforementioned positive precipitation bias seen in our simulations of 10 to 33%, is far less…” — please change to: “The aforementioned positive precipitation bias of 10-33% seen in our simulations is far less…”
L498: Remove semicolon
L505: “SCREAM’s resolution” —> “SCREAM’s muted resolution”…?Citation: https://doi.org/10.5194/egusphere-2024-839-RC2 - AC2: 'Reply on RC2', Peter Bogenschutz, 31 Jul 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-839', Anonymous Referee #1, 18 May 2024
Summary:
Bogenshutz et al. present four detailed case studies chosen to highlight different types and aspects of ARs impacting the California coast using the ultra-high regionally refined E3SM atmosphere model, SCREAM. The purpose of the study is to show SCREAM’s potential utility simulating ARs for both climate and weather communities. Their main findings, aside from the realism of AR representation, include (1) refining resolution beyond 3km has little impact for improving precipitation biases, but (2) expanding the spatial extent of the refinement has modest benefits.
Overall Comments:
This is a very nice paper. It is clear, well-written and organized, and absolutely fits within the GMD scope. I have a few comments I would like to see addressed before full publication. Although these are relatively minor, they are important for clarity. Please see specific comments below, but the most important ones are regarding SCREAM model description and method/domain for computing max/mean IVT.
Specific Comments:
Line 18 and elsewhere: Typically the words “atmospheric rivers” are not capitalized.
Line 89: Out of curiosity, when you moved to using SHOC from a simplified PBL, was there anything that needed to be re-tuned?
Figure 2: Did you mean ne32 and not ne21?
SCREAM description and References: There are missing citations in the reference list, I definitely noticed both Caldwell et al 2021 and 2019 missing as I searched for these to get more information on SCREAM. Also, I recommend adding a bit more to the description part of the paper for ease rather than digging into the cited Caldwell papers, i.e. vertical resolution, features, biases, improvements not already mentioned, etc.
AR cases (these are more curiosity questions than comments on the paper itself): There are all great choices and span important and different types of ARs and dynamics. I am curious, however, on how SCREAM-RRR would perform for 1) a weak Cat 1 storm, or 2) different regions , or 3) wet vs windy “flavors”? Perhaps these are already planned activities, and I am not suggesting these be done for this paper. (I understand the intense computational and person-resources involved in these activities). Some questions to consider for future experiments (or answer if you have already done some of this): For a weak category 1 AR, or a “windy” AR, are there components of physics, dycore, and/or resolution that do better with an overwet environment vs not? Are biases different across the varied AR dynamics, i.e. Pineapple Express vs North Atlantic warm conveyor belt lift? (Has this been looked at)? Are there plans to look at moisture transports such as ARs within monsoonal flow which can be found in the E.Pacific?
Line 148 and elsewhere: The acronym “AR” only needs to be defined once.
Line 253: Why not add polygons to one of your maps with the locations of each of your evaluation points? This would be helpful visualizing where the skill is being evaluated. (Or maybe an inset with a zoomed map)?
Line 287 and Figure 5c: Pearson pattern correlation?
Line 327: Comparing SC2005 reanalysis and SCREAM, the character of the IVT plume and footprint are very different. One might argue that SCREAM really doesn’t capture the AR structure well here, especially in the ocean. Ideas as to why?
Lines 359-360: I understand the tension between the improvements via resolution and mesh area, but the other degree of freedom here is E3SM vs CESM when consulting with the results from the Rhoades paper. I think(?) that the vertical levels are different between these two modelling frameworks? How different is SCREAM vertical structure from the Rhoades study, and do you think this may matter regarding convection schemes?
Line 401: Did you mean Figure 10 (and not 3)?
Figure 10: I find it interesting that the EPAC 3km case seems to overdo the snow compared to the CA 3km in most cases, i.e. most notably at Virginia Lakes for NC2021 and everywhere SC2005. Thoughts?
Line 428, Figures 12/13, Lines 440-447: For the event maximum/mean IVT, did you use find, at each grid point in your RR domain, the maximum/mean values and plot this? Or just across the plotting extent in Fig 12/13? For SC2005 where SCREAM’s IVT core could arguably be over the ocean (based on Figure 4), this might be relevant. Are some of SCREAMs biases due to AR characterization and misplacement of the core, and not really just the fact that the plume hitting the coast is more narrow? For clarity, please add a sentence or two clarifying IVT max/mean methodology.
Citation: https://doi.org/10.5194/egusphere-2024-839-RC1 - AC1: 'Reply on RC1', Peter Bogenschutz, 31 Jul 2024
-
CC1: 'Referee comments on Bogenschutz et al. "Atmospheric River Induced Precipitation in California as Simulated by the Regionally Refined Simple Convective Resolving E3SM Atmosphere Model (SCREAM) Version 0"', James Benedict, 17 Jun 2024
Publisher’s note: this comment is a copy of RC2 and its content was therefore removed.
Citation: https://doi.org/10.5194/egusphere-2024-839-CC1 -
AC3: 'Reply on CC1', Peter Bogenschutz, 31 Jul 2024
Please see reply to reviewer 2 comment.
Citation: https://doi.org/10.5194/egusphere-2024-839-AC3
-
AC3: 'Reply on CC1', Peter Bogenschutz, 31 Jul 2024
-
RC2: 'Comment on egusphere-2024-839', James Benedict, 18 Jun 2024
General comments:
The study examines precipitation, snowfall, and surface temperature characteristics associated with 4 atmospheric river (AR) cases impacting California as simulated in the Simple Cloud Resolving E3SM (Energy Exascale Earth System Model) Atmosphere Model (SCREAM) with selected regionally refined grid mesh (RRM) configurations. Sensitivity of the results to the RRM horizontal resolution (3.25 km, 1.6 km, and 800 m) and upstream extent of the RRM domain is evaluated against both high-resolution gridded data sets and single-station time series. Generally, SCREAM is able to capture the fine-scale regional distributions of AR-related precipitation over California. However, the model systematically overestimates precipitation over upwind and higher-elevation areas, and underestimates precipitation on leeward sides of mountain ranges. SCREAM also tends to overestimate storm-total snowfall and mean surface temperatures during the AR events. Overall, it is found that precipitation and temperature distributions exhibit only modest sensitivity to RRM resolution. Although somewhat greater improvement is seen for the largest RRM domain extent due to better representation of large-scale meteorological patterns that guide the ARs, the authors assert that the required RRM domain expansion no longer achieves the intended computational cost savings.
I found the modeling approach and scientific assessment to be both clear and informative. The detail of the model evaluation was appropriate given the fairly large number of simulations conducted and AR cases examined.
A key finding of the paper, in my view, is the relative insensitivity of storm-total precipitation to model horizontal grid resolution in the range from 3.25 km to 800 m. This finding also highlights the current horizontal resolution limit of evaluation data sets (4 km for daily surface precipitation, and 25 km for ERA5 atmospheric fields). At such fine scales, the need would arise to start evaluating models against single meteorological stations, as the authors have done here.
Findings related to the RRM domain extent are also novel. The authors note that SCREAM is initialized with ERA5 but allowed to evolve freely thereafter. In my opinion, initializing the model 24 hours (or 18 or 30 hours, based on sensitivity tests) is perhaps a “low bar” to assess model performance, though I suppose longer lead times might result in a drift of the large-scale meteorological solution such that evaluation of AR impacts could become complicated. I view the paper’s findings mostly as an examination of AR impacts from simulations in which the large-scale meteorological patterns are quasi-prescribed, given the short lead time.
Specific comments:
L26-28: Rutz et al. (2014, https://doi.org/10.1175/MWR-D-13-00168.1) displays one of the earliest and nicest maps of AR contributions to total cool-season precipitation (their Fig. 8). Consider adding this reference here.
L49-51: Can a reference to these modeling improvement be added here?
L94-97: “SCREAMv0 used a prescribed-aerosol version of E3SM’s modal aerosol model, however in this work we use a version of SCREAM that prescribes both cloud-condensation nuclei number and aerosol radiative properties from an E3SMv2 simulation. This is known as Simple Prescribed Aerosol (SPA) and will be incorporated into SCREAMv1.” Do the authors think that this simplification, especially the prescribed CCN, could potentially be contributing to the “eagerness” of SCREAM to overestimate precipitation on the upwind side of mountain ranges?
Figs. 2 and 3: In future depictions of RRMs, it is recommended to recreate the plot formatting like the right panel in Fig. 1 so that land masses and coastlines are visible.
L159: “most locations” is too vague
L163-164: ERA5 and IVT should be defined here, as it's their first usage.
L169-170: D4 drought classification should be referenced.
L214-216: Three references on the large-scale meteorological patterns that precede west coast (of U.S.) AR landfalls:
——
Benedict, J. J., Clement, A. C., & Medeiros, B. (2019). Atmospheric blocking and other large‐scale precursor patterns of landfalling atmospheric rivers in the North Pacific: A CESM2 study. Journal of Geophysical Research: Atmospheres, 124, 11,330–11,353. https://doi.org/10.1029/2019JD030790Zhou, Y., & Kim, H. (2019). Impact of Distinct Origin Locations on the Life Cycles of Landfalling Atmospheric Rivers Over the U.S. West Coast. Journal of Geophysical Research: Atmospheres, 124, 11,897–11,909. https://doi.org/10.1029/2019JD031218
Carrera, M. L., Higgins, R. W., & Kousky, V. E. (2004). Downstream weather impacts associated with atmospheric blocking over the northeast Pacific. Journal of Climate, 17(24), 4823–4839. https://doi.org/10.1175/jcli‐3237.1
——L248-249: SNOTEL and Cooperative Observer Program should be referenced.
L273-274: It would be helpful to clarify how “bias” is calculated, as well as what spatial domain is used in the calculations for Fig. 5. The author’s use of “statewide” in L276 leads me to believe that the entire state of California is being used as the spatial domain for evaluation, but this should be more clearly stated.
A map containing locations (mountain ranges, counties) highlighted in the text should be added to help readers not familiar with California geography.
In Figs. 12 and 13, what temporal resolutions are used for SCREAM and ERA5 to evaluate maximum IVT? This should be noted in the L428 paragraph.
L486-494: It would greatly benefit the paper to add references to specific figures when summarizing the key findings.
L511: At the end of this paragraph, it might be good to add a sentence noting that model evaluation at sub-kilometer scales is challenged by the availability of observations needed to evaluate the simulations… though I suppose a comparison to a single station would be appropriate in some cases.
Technical Corrections:
L2: E3SM should be defined.
L42: “generation…have” —> “generation…has”
L62: “Winter Hydroclimate” need not be capitalized.
L73-74: “resolution is increased to 1.6 km”: It would be better to state something like “resolution is increased from 3 km to 1.6 km…”
L77: “5 km though, that work” —> “5 km, though that work”
The Fig. 1 caption seems to be missing one or more words.
Fig. 2 caption: Should be “ne32”, not “ne21”.
L135: Missing period at end of sentence.
L171: ; —> ,
L215: “set up” (verb) —> “setup” (noun)
L232: letting —> allowing
L281-283: Remove “Though”, add comma after “scores”, and change semicolon to comma. Also, “grids and there can be implications” —> “grids. There can be implications…”
L310: ; —> ,
L388: models —> model’s
L401: Should be Fig. 10, not Fig. 3.
L402: “though” —> “though they”
L403: Lakes —> Lake
L405: ; —> ,
L425: ; —> ,
L455: ; —> ,
L495: “The aforementioned positive precipitation bias seen in our simulations of 10 to 33%, is far less…” — please change to: “The aforementioned positive precipitation bias of 10-33% seen in our simulations is far less…”
L498: Remove semicolon
L505: “SCREAM’s resolution” —> “SCREAM’s muted resolution”…?Citation: https://doi.org/10.5194/egusphere-2024-839-RC2 - AC2: 'Reply on RC2', Peter Bogenschutz, 31 Jul 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 283 | 100 | 32 | 415 | 19 | 18 |
- HTML: 283
- PDF: 100
- XML: 32
- Total: 415
- BibTeX: 19
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
Peter Bogenschutz
Jishi Zhang
Philip Cameron-Smith
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(11062 KB) - Metadata XML