the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Mar 2023
J-GAIN v1.0: A flexible tool to incorporate aerosol formation rates obtained by molecular models into large-scale models
Tinja Olenius
Abstract. New-particle formation from condensable vapors is a common atmospheric process that has significant but uncertain effects on aerosol particle number concentrations and impacts. Assessing the formation rates of nanometer-sized particles from different vapors is an active field of research within atmospheric sciences, with new data being constantly produced by molecular models and experimental studies. Such data can be implemented in large-scale climate and air quality models as parameterizations or look-up tables. Models benchmarked against measurement data provide a straight-forward means to assess formation rates over a wide range of atmospheric conditions for given chemical compounds. Ideally, the implementation of such formation rate data should be easy, efficient and flexible in the sense that same tools can be conveniently applied for different data sets in which the formation rate depends on different parameters. In this work, we present a tool to generate and interpolate look-up tables of formation rates for user-defined input parameters. The table generator routine applies a molecular cluster dynamics model with quantum chemistry input, but other types of particle formation models may be used as well. The interpolation routine uses a multivariate interpolation algorithm, which is applicable to different numbers of independent parameters, and gives fast and accurate results with typical interpolation errors of up to a few percent. These routines facilitate the implementation and testing of different aerosol formation rate predictions in large-scale models, allowing straight-forward inclusion of new or updated data without the need to apply separate parameterizations or routines for different data sets that involve different chemical compounds or other parameters.
-
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
(2641 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2641 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Daniel Yazgi and Tinja Olenius
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1464', Anonymous Referee #1, 03 May 2023
This manuscript presents a tool for efficiently incorporating formation rates from a molecular cluster dynamics model cluster dynamics model with quantum chemistry input into larger models with a look-up table approach. The tool includes a table generator and an interpolator. The table generator calls the molecular cluster dynamics model ACDC for a user-specified ranges of environmental parameters (vapor concentrations, temperature, relative humidity and so on) and generates n-dimensional table data for the particle formation rate. The interpolator can then be included in a host model and reads and interpolates the pre-calculated table data . They present an example case for sulphuric acid-ammonia new particle formation (NPF) and show that the typical errors are negligible to 10-20% for implementation into a global model. These errors seem highly acceptable compared to uncertainties in both theoretical and experimental rates.
The tool seems very useful, and the authors show well how it could be implemented in a global model and therefore seems like a good fit for GMD.
I have two main points I would like the authors to address.
- In general, it could be made clearer what is done with which parts of the tool.
- The tool is initially presented as being usable for both measurement data and ACDC data, as well as other models. However, it is not entirely clear upon the first read through whether the table generator is solely used for calling ACDC or whether it could also be used for other sources. On line 80 it is actually explained that it can only be used with ACDC, but because the tool was initially presented in the abstract with “The table generator routine applies a molecular cluster dynamics model with quantum chemistry input, but other types of particle formation models may be used as well.” I was still under the impression that the table generator was more generic.
- In my opinion, it should be clearer how other formation rate sources could be used with the tool and what pre-processing would need to be done if so. Alternatively, suggestions of such uses could be removed from the abstract (line 9-10).
- Related to this, it was unclear to me if the example presented in the result section for H2SO4–NH3 system is made using both the table generator and the interpolator, or if you simply interpolate a pre-calculated table from the cited Olenius et al. (2013) (see p6, L139-142). With careful reading, I think that the data from Olenius et al (2013) is the input data to ACDC, but this should be stated. On p3L87, the table generator is introduced, and is said to require “molecular cluster thermochemistry data for the given chemical compounds, and user-defined input for the ranges of the parameters that define the ambient conditions.” You could specify already here that “molecular cluster thermochemistry data for the given chemical compounds” is input to ACDC.
- In general, it would be helpful if input data and output data for each part were named a bit more consistently throughout the manuscript. Maybe this could be part of Fig. 1 for example?
- Since this tool is meant for implementation into host models, I have looked through the code repository and while it is in general tidy and includes informative readme files, it could be more helpful for users with some small tweaks:
- At first glance, it is not clear which steps would be needed to run or incorporate J-GAIN in a host model and clarifying this would improve reusability. A step-by-step instruction for use even in the base folder (e.g. which steps would be needed to reproduce some part of the example case in the manuscript, ideally including all bash commands) would make it clearer what and in which order the different procedures must be performed.
- Additionally, the fortran code does not seem to be commented. I especially think the implementation examples in the example folder (https://github.com/tolenius/J-GAIN/tree/main/examples/interp_dual) would benefit from some commenting to help potential implementers.
Other comments:
- The software prerequisites in the code repository do not link to anywhere where the software can be installed. The gfortran version needed was not the most recent and the build failed with the most recent version.
- P1,L1-2: “ New-particle formation from condensable vapors is a common atmospheric process that has significant but uncertain effects on aerosol particle number concentrations and impacts.” Vapors are by definition condensable, so maybe “condensable gasses” is better. Secondly, “impacts” is a bit of a loose end here, consider being more precise on what these impacts could be.
- P1,L4: “Such data can be implemented in large-scale climate…” I don’t think you can implement “data” in a model, the data can be used to calculate/predict formation rates in the models?
- P1,L5: “ Models benchmarked against measurement data provide a straight-forward means to assess formation rates over a wide range of atmospheric conditions for given chemical compounds. “ I don’t understand what this sentence means in this context.
- Fig1: I think this could be a great place to make it clear what parts of the tool does what, and what the user is expected to do. The ACDC input is also not mentioned here as far as I can see.
- P4,L94-95: “As the inclusion of charged species and hydrates in a molecular cluster data set requires a significant computational effort, these effects are not always available.” It’s not clear what “not always available” means here. From the host model?
- P5,L127: In this section it is not clear to me what the objective is. Are you simply providing some examples of what could be done in a host model? Are the examples related to code you provide? If this section is related to some code you provide, I would make this clear.
- P6,l144: “We also generate tables suitable for global applications” it is not clear what “also” means here, is it different to what you present in the sentence before?
- P7,L157-159: I assume the main point here is not the varying over time, but the fact that the interpolation is over all parameters? The way it reads now, it looks a bit like the varying over time should add something particular.
- P9,L184: “the run time exhibits a major increase when the table size increases beyond ca. 2^28” Just out of curiosity, why do you think this is?
- P9,L189: Would it not be more effective here to also state how much including the scheme would increase the total run time of the atmospheric component for example?
- P10,L201: The title reads: “Potential limitations in applying formation rates in a host model”, would it not be more precise to say “potential limitations for applying look-up tables in a host model”? There are already parameterisations for formation rates in most models.
- P11, L216-217: please double check this sentence.
Citation: https://doi.org/10.5194/egusphere-2022-1464-RC1 -
RC2: 'Comment on egusphere-2022-1464', Anonymous Referee #2, 06 May 2023
General comments:
The manuscript by Yazgi et al. presents a tool J-GAIN to generate and interpolate look-up tables of formation rates, allowing the implementation of theoretical particle formation rate data in atmospheric large-scale models. They conducted tests on the application and performance of J-GAIN using theoretical data for H2SO4-NH3 particle formation, which show that J-GAIN is efficient and accurate. The work is technically well performed and the chosen methodology is appropriate for the purpose of this study. The selected topic should be interesting across a range of atmospheric model development community. The manuscript is well written and easy to follow. Therefore, I recommend publication of this manuscript after a minor revision.
Specific comments:
- Line 105, it could be better to present how many points the higher-resolution reference table has. Although it is known that values determined from a table of sufficient resolution are guaranteed to be close to the original data, it is better to give specific relative errors between them in order to strongly convince readers.
- Line 211 and line 215, it seems it is inconsistent about whether different amines can be modeled as a lumped compound.
- Figure 3(a), it’s hard to see any other curves except the curve representing 29+1 points. It would be better to change the color or pattern of these curves, making the figure look more clear.
- Some minor mistakes are shown in the manuscript, e.g., Line58, “be close to the original data” instead of “be close the original data”. Please recheck and revise the whole manuscript.
- Please check the guidelines of GeoscientificModel Development for references, and all the references should be cited in the same style.
Citation: https://doi.org/10.5194/egusphere-2022-1464-RC2 -
AC1: 'Reply to reviewers', Tinja Olenius, 14 Jun 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1464/egusphere-2022-1464-AC1-supplement.pdf
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1464', Anonymous Referee #1, 03 May 2023
This manuscript presents a tool for efficiently incorporating formation rates from a molecular cluster dynamics model cluster dynamics model with quantum chemistry input into larger models with a look-up table approach. The tool includes a table generator and an interpolator. The table generator calls the molecular cluster dynamics model ACDC for a user-specified ranges of environmental parameters (vapor concentrations, temperature, relative humidity and so on) and generates n-dimensional table data for the particle formation rate. The interpolator can then be included in a host model and reads and interpolates the pre-calculated table data . They present an example case for sulphuric acid-ammonia new particle formation (NPF) and show that the typical errors are negligible to 10-20% for implementation into a global model. These errors seem highly acceptable compared to uncertainties in both theoretical and experimental rates.
The tool seems very useful, and the authors show well how it could be implemented in a global model and therefore seems like a good fit for GMD.
I have two main points I would like the authors to address.
- In general, it could be made clearer what is done with which parts of the tool.
- The tool is initially presented as being usable for both measurement data and ACDC data, as well as other models. However, it is not entirely clear upon the first read through whether the table generator is solely used for calling ACDC or whether it could also be used for other sources. On line 80 it is actually explained that it can only be used with ACDC, but because the tool was initially presented in the abstract with “The table generator routine applies a molecular cluster dynamics model with quantum chemistry input, but other types of particle formation models may be used as well.” I was still under the impression that the table generator was more generic.
- In my opinion, it should be clearer how other formation rate sources could be used with the tool and what pre-processing would need to be done if so. Alternatively, suggestions of such uses could be removed from the abstract (line 9-10).
- Related to this, it was unclear to me if the example presented in the result section for H2SO4–NH3 system is made using both the table generator and the interpolator, or if you simply interpolate a pre-calculated table from the cited Olenius et al. (2013) (see p6, L139-142). With careful reading, I think that the data from Olenius et al (2013) is the input data to ACDC, but this should be stated. On p3L87, the table generator is introduced, and is said to require “molecular cluster thermochemistry data for the given chemical compounds, and user-defined input for the ranges of the parameters that define the ambient conditions.” You could specify already here that “molecular cluster thermochemistry data for the given chemical compounds” is input to ACDC.
- In general, it would be helpful if input data and output data for each part were named a bit more consistently throughout the manuscript. Maybe this could be part of Fig. 1 for example?
- Since this tool is meant for implementation into host models, I have looked through the code repository and while it is in general tidy and includes informative readme files, it could be more helpful for users with some small tweaks:
- At first glance, it is not clear which steps would be needed to run or incorporate J-GAIN in a host model and clarifying this would improve reusability. A step-by-step instruction for use even in the base folder (e.g. which steps would be needed to reproduce some part of the example case in the manuscript, ideally including all bash commands) would make it clearer what and in which order the different procedures must be performed.
- Additionally, the fortran code does not seem to be commented. I especially think the implementation examples in the example folder (https://github.com/tolenius/J-GAIN/tree/main/examples/interp_dual) would benefit from some commenting to help potential implementers.
Other comments:
- The software prerequisites in the code repository do not link to anywhere where the software can be installed. The gfortran version needed was not the most recent and the build failed with the most recent version.
- P1,L1-2: “ New-particle formation from condensable vapors is a common atmospheric process that has significant but uncertain effects on aerosol particle number concentrations and impacts.” Vapors are by definition condensable, so maybe “condensable gasses” is better. Secondly, “impacts” is a bit of a loose end here, consider being more precise on what these impacts could be.
- P1,L4: “Such data can be implemented in large-scale climate…” I don’t think you can implement “data” in a model, the data can be used to calculate/predict formation rates in the models?
- P1,L5: “ Models benchmarked against measurement data provide a straight-forward means to assess formation rates over a wide range of atmospheric conditions for given chemical compounds. “ I don’t understand what this sentence means in this context.
- Fig1: I think this could be a great place to make it clear what parts of the tool does what, and what the user is expected to do. The ACDC input is also not mentioned here as far as I can see.
- P4,L94-95: “As the inclusion of charged species and hydrates in a molecular cluster data set requires a significant computational effort, these effects are not always available.” It’s not clear what “not always available” means here. From the host model?
- P5,L127: In this section it is not clear to me what the objective is. Are you simply providing some examples of what could be done in a host model? Are the examples related to code you provide? If this section is related to some code you provide, I would make this clear.
- P6,l144: “We also generate tables suitable for global applications” it is not clear what “also” means here, is it different to what you present in the sentence before?
- P7,L157-159: I assume the main point here is not the varying over time, but the fact that the interpolation is over all parameters? The way it reads now, it looks a bit like the varying over time should add something particular.
- P9,L184: “the run time exhibits a major increase when the table size increases beyond ca. 2^28” Just out of curiosity, why do you think this is?
- P9,L189: Would it not be more effective here to also state how much including the scheme would increase the total run time of the atmospheric component for example?
- P10,L201: The title reads: “Potential limitations in applying formation rates in a host model”, would it not be more precise to say “potential limitations for applying look-up tables in a host model”? There are already parameterisations for formation rates in most models.
- P11, L216-217: please double check this sentence.
Citation: https://doi.org/10.5194/egusphere-2022-1464-RC1 -
RC2: 'Comment on egusphere-2022-1464', Anonymous Referee #2, 06 May 2023
General comments:
The manuscript by Yazgi et al. presents a tool J-GAIN to generate and interpolate look-up tables of formation rates, allowing the implementation of theoretical particle formation rate data in atmospheric large-scale models. They conducted tests on the application and performance of J-GAIN using theoretical data for H2SO4-NH3 particle formation, which show that J-GAIN is efficient and accurate. The work is technically well performed and the chosen methodology is appropriate for the purpose of this study. The selected topic should be interesting across a range of atmospheric model development community. The manuscript is well written and easy to follow. Therefore, I recommend publication of this manuscript after a minor revision.
Specific comments:
- Line 105, it could be better to present how many points the higher-resolution reference table has. Although it is known that values determined from a table of sufficient resolution are guaranteed to be close to the original data, it is better to give specific relative errors between them in order to strongly convince readers.
- Line 211 and line 215, it seems it is inconsistent about whether different amines can be modeled as a lumped compound.
- Figure 3(a), it’s hard to see any other curves except the curve representing 29+1 points. It would be better to change the color or pattern of these curves, making the figure look more clear.
- Some minor mistakes are shown in the manuscript, e.g., Line58, “be close to the original data” instead of “be close the original data”. Please recheck and revise the whole manuscript.
- Please check the guidelines of GeoscientificModel Development for references, and all the references should be cited in the same style.
Citation: https://doi.org/10.5194/egusphere-2022-1464-RC2 -
AC1: 'Reply to reviewers', Tinja Olenius, 14 Jun 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1464/egusphere-2022-1464-AC1-supplement.pdf
Peer review completion
Journal article(s) based on this preprint
Daniel Yazgi and Tinja Olenius
Model code and software
J-GAIN, Software repository Daniel Yazgi and Tinja Olenius https://github.com/tolenius/J-GAIN
J-GAIN v1.0 Daniel Yazgi and Tinja Olenius https://doi.org/10.5281/zenodo.7457152
Daniel Yazgi and Tinja Olenius
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 271 | 90 | 14 | 375 | 8 | 7 |
- HTML: 271
- PDF: 90
- XML: 14
- Total: 375
- BibTeX: 8
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2641 KB) - Metadata XML