Taylor's statistical theory applied to the turbulence parameterization in the BAM-INPE global atmospheric model

Eras, Eduardo Rohde; de Campos Velho, Haroldo Fraga; Kubota, Paulo Yoshio

doi:https://doi.org/10.5194/gmd-2023-59

Preprints

https://doi.org/10.5194/gmd-2023-59

Preprints

Submitted as: development and technical paper

31 May 2023

Submitted as: development and technical paper |

| 31 May 2023

Status: this preprint was under review for the journal GMD. A final paper is not foreseen.

Taylor's statistical theory applied to the turbulence parameterization in the BAM-INPE global atmospheric model

Eduardo Rohde Eras, Haroldo Fraga de Campos Velho, and Paulo Yoshio Kubota

Abstract. Turbulence parameterization scheme for planetary boundary layer (PBL) based on Taylor's statistical formulation for turbulent flow is used in the Brazilian Global Atmospheric Model (BAM). Taylor's approach has been already applied to mesoscale and air pollution atmospheric models, but it is the first time that this approach is employed in a global model. The BAM model is operationally used to generate numerical weather forecasting by the INPE (National Institute for Space Research – Brazil). The simulation performed with BAM model using Taylor's parameterization is compared with the results obtained with the schemes presented by Holtslag-Boville, Sungsu-Park, and Mellor-Yamada trying to reproduce the ERA5 reanalysis data of two different initial conditions of two different seasons (dry season and wet season). The obtained comparison exhibit a positive result for the new parameterization simulating global precipitation, cloud coverage, and top atmospheric thermal radiation, specially in the dry season. Positive results for the Amazon basin were also obtained.

This preprint has been withdrawn.

Received: 20 Mar 2023 – Discussion started: 31 May 2023

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 4391 KB)

Withdrawal notice
This preprint has been withdrawn.
Preprint (4391 KB)

Download & links

This preprint has been withdrawn.

Eduardo Rohde Eras, Haroldo Fraga de Campos Velho, and Paulo Yoshio Kubota

Interactive discussion

Status: closed

RC1:
'Comment on gmd-2023-59', Anonymous Referee #1, 15 Aug 2023

Summary: In this study, the authors implemented a new boundary layer scheme into the BAM-INPE global atmospheric model. Then the authors conducted experiments to evaluate the performance of the new boundary layer scheme, using ERA5 reanalysis as the truth and in comparison to 3 other boundary layer schemes. The manuscript should be rejected in my opinion. I have 2 major objections to this manuscript. First, the entire paper reads much like a report, with no or little scientific discission. This is utterly unacceptable. I think the authors should publish this as a technical report on the website of the BAM-INPE model (if there is one), instead of being a scientific paper. Second, the paper did not really develop anything new. Consider both points, I would argue that the materials presented in this manuscript are not sufficient to justify its publication on GMD.
Minor comments:
1, equation 9: how is h defined and calculated?
2, algorithm 1: not necessary
3, line 154: how long is the 2nd simulation?
4, line 206-210: how do the authors see the daytime results and nighttime results? I don’t see any of these in the figure.
5, line 233-234: “The Amazon basin region was where Taylor parameterization got its best approximations comparing to the ERA5 reanalysis data”. This sentence is not wrong per se, but it should not be in the section “Planetary Boundary Layer Height” because the Taylor parameterization is NOT the best for “Planetary Boundary Layer Height”.
6, line 264-266: this sentence is unclear to me.
7, line 288-289: “at least consistent results for Mean Temperature at 2 meters”. This sounds a bit double standard. Just say the Taylor parameterization is the worst.
8, line 313-315: “The amount of radiation at the top of the atmosphere is a good indicator of the energy balance in the simulation, and the overall good result for this variable while using Taylor’s parameterization is a positive indication towards a good energy balance within the method.”. The authors never showed the net radiation at the top of the atmosphere, only the outgoing. If the authors comment on “energy balance”, shouldn’t they show the net radiation?

Citation: https://doi.org/10.5194/gmd-2023-59-RC1
- AC1: 'Reply on RC1', Eduardo Rohde Eras, 08 Sep 2023
  
  Before anything, We want to thank for all the corrections and comments. An acknowledgment was included in the manuscript nominated to the "Anonymous Referee".
  Answers for the Summary comments:
  "The manuscript should be rejected in my opinion."
  Answer: It's a good research for Brazil because it's the best parameterization for predicting rain, important for power generation. Also, almost all natural disasters are linked to meteorology, especially related to rain. It's vital for us in Brazil to know the best method to generate rain. Question of the Flying Rivers: Better in the Amazon is good for the Southeast (Zemp, Delphine, et al. "EXPLORING THE COMPLEX NETWORK OF THE AMAZON ‘S WATER PUMP AND FLYING RIVERS."). Whoever best describes the Amazon is good for other parts of Brazil.
  "First, the entire paper reads much like a report"
  Answer: Style is a matter of author property, the text might "feels" like a report, but it isn't. The paper has been written and it's honest, this is the first and most important step. Improve is now part of the process.
  "with no or little scientific discussion. This is utterly unacceptable."
  Answer: Key factor about Taylor: Representation of the Turbulence Spectrum. It's well represented, taking into account good physics. Three aspects:
  
  - Theoretical question: this spectrum follows Kolmogorov's law of -5/3 distribution of energy in different types of orders;
  
  - The ingress of energy is represented by the maximum of the spectrum that meets the observed. The inertial subdomain, different (experimental) spectrum shape for CLC, CLN and CLE suited by exponents;
  
  - The exponents will give the face of the function (shape) of the observed spectrum. Unlike curve fitting (zero at surface and zero at top) proportional to height: Z = 0, Z = h -> K=0. That's not curve fitting, Taylor is theory. The experiments generates good results, similar or even better than Yamada (second order!), being theory and first order.
  "I think the authors should publish this as a technical report on the website of the BAM-INPE model (if there is one), instead of being a scientific paper."
  Answer: Publish in a local paper is of no use, for us this is already known: One of the main developers of the model is one of the authors of the paper (Dr. Paulo) and the other author is one of the main researchers on Taylor (Dr. Haroldo). The scientific community has an obligation to disclose this result, who else used Taylor's theory to represent turbulence in a global model?
  "Second, the paper did not really develop anything new."
  Answer: The Result is Original: first use of Taylor in a Global OPERATIONAL model. Also, we overcome one of the most difficult variables to predict: precipitation.
  The "minor comments" are important contribution for the paper with valid corrections and new contents. Here are the answers:
  1, equation 9: how is h defined and calculated?
  Answer: The h is calculated using the Holtslag-Bovile implementation (Holtslag and Bovile: Local Versus Nonlocal Boundary-Layer Diffusion in a Global Climate Model, 1992 - Equation 3.11) that is based on Troen and Mahrt (1986).
  2, algorithm 1: not necessary
  Answer: Yes, it wasn't indeed. It's now off the paper.
  3, line 154: how long is the 2nd simulation?
  Answer: It was really missing this information, We fixed the text now.
  4, line 206-210: how do the authors see the daytime results and nighttime results? I don’t see any of these in the figure.
  Answer: The figure was wrong, it supposed to be at 72h of simulation and it was at 24h instead. Also, the text was clarified point were in the map was day and night.
  5, line 233-234: “The Amazon basin region was where Taylor parameterization got its best approximations comparing to the ERA5 reanalysis data”. This sentence is not wrong per se, but it should not be in the section “Planetary Boundary Layer Height” because the Taylor parameterization is NOT the best for “Planetary Boundary Layer Height”.
  Answer: Right, it was fixed in the text.
  6, line 264-266: this sentence is unclear to me.
  Answer: This sentence was a true mess. We completely rewrite the paragraph, it's better now.
  7, line 288-289: “at least consistent results for Mean Temperature at 2 meters”. This sounds a bit double standard. Just say the Taylor parameterization is the worst.
  Answer: The section 5.4 was rewritten in a more consistent way.
  8, line 313-315: “The amount of radiation at the top of the atmosphere is a good indicator of the energy balance in the simulation, and the overall good result for this variable while using Taylor’s parameterization is a positive indication towards a good energy balance within the method.”. The authors never showed the net radiation at the top of the atmosphere, only the outgoing. If the authors comment on “energy balance”, shouldn’t they show the net radiation?
  Answer: That is a valuable contribution, We did a new analysis in the data to generate the net radiation comparison, and in fact the implementation was deficient in this area. So the statement about model energy consistence wasn't correct. A new section (5.7) and two new tables (5 and 6) were included in the text with the new results and new explanations. Also, the conclusion were corrected and a suggestion of future work on model energy balance is included.
  
  Citation: https://doi.org/10.5194/gmd-2023-59-AC1
RC2:
'Comment on gmd-2023-59', Chiel van Heerwaarden, 12 Dec 2023
Dear authors,
I am writing this short review as the editor of the paper, because in all the months that have processed, and the endless list of people that I have contacted to review this paper, nobody accepted the task, except for one anonymous reviewer that already posted his review months ago. Therefore, I apologize for the delays that you have encountered.
After reading your manuscript in detail, I feel that I unfortunately have to reject this manuscript, because in my view it is not polished enough for a scientific manuscript. I like the idea of bringing an old, but established idea from Taylor to a global model. The validation and further analysis is however inadequate for a scientific publication and requires much more work. There is hardly any interpretation of the results and discussions of lessons learned from the findings.
I would like to give a few tips on how this can be achieved:
This paper discusses a boundary-layer parameterization, and compares this entirely to ERA5. In my view, a parameterization that should improve global models should be validated against observations. The boundary-layer height of ERA5 has many issues on itself, especially in complex terrain.

The Amazon region is a wonderful region for turbulence parametrization validation. I wonder why the authors did not use observations from the ATTO infrastructure to validate their model rather than ERA5 in Figures 6-9.

The chosen variables are not the best for a boundary-layer parametrization, and at least not in the most logical order. I recommend by starting with boundary layer basics: temperature, wind, humidity, and BL depth, and only then go to variables that derive from that, such as radiation fluxes, cloud cover and precipitation, as each of those are strongly influenced by other parameterizations in your model as well.

Discuss the figures not only in the degree of how well they match, but discuss what the findings mean for the parameterization that is the main topic of the paper: how do the choices of how the physics are described influence what we see in the picture?

I hope with these tips you can improve your work.
Citation: https://doi.org/10.5194/gmd-2023-59-RC2

Interactive discussion

Status: closed

RC1:
'Comment on gmd-2023-59', Anonymous Referee #1, 15 Aug 2023

Summary: In this study, the authors implemented a new boundary layer scheme into the BAM-INPE global atmospheric model. Then the authors conducted experiments to evaluate the performance of the new boundary layer scheme, using ERA5 reanalysis as the truth and in comparison to 3 other boundary layer schemes. The manuscript should be rejected in my opinion. I have 2 major objections to this manuscript. First, the entire paper reads much like a report, with no or little scientific discission. This is utterly unacceptable. I think the authors should publish this as a technical report on the website of the BAM-INPE model (if there is one), instead of being a scientific paper. Second, the paper did not really develop anything new. Consider both points, I would argue that the materials presented in this manuscript are not sufficient to justify its publication on GMD.
Minor comments:
1, equation 9: how is h defined and calculated?
2, algorithm 1: not necessary
3, line 154: how long is the 2nd simulation?
4, line 206-210: how do the authors see the daytime results and nighttime results? I don’t see any of these in the figure.
5, line 233-234: “The Amazon basin region was where Taylor parameterization got its best approximations comparing to the ERA5 reanalysis data”. This sentence is not wrong per se, but it should not be in the section “Planetary Boundary Layer Height” because the Taylor parameterization is NOT the best for “Planetary Boundary Layer Height”.
6, line 264-266: this sentence is unclear to me.
7, line 288-289: “at least consistent results for Mean Temperature at 2 meters”. This sounds a bit double standard. Just say the Taylor parameterization is the worst.
8, line 313-315: “The amount of radiation at the top of the atmosphere is a good indicator of the energy balance in the simulation, and the overall good result for this variable while using Taylor’s parameterization is a positive indication towards a good energy balance within the method.”. The authors never showed the net radiation at the top of the atmosphere, only the outgoing. If the authors comment on “energy balance”, shouldn’t they show the net radiation?

Citation: https://doi.org/10.5194/gmd-2023-59-RC1
- AC1: 'Reply on RC1', Eduardo Rohde Eras, 08 Sep 2023
  
  Before anything, We want to thank for all the corrections and comments. An acknowledgment was included in the manuscript nominated to the "Anonymous Referee".
  Answers for the Summary comments:
  "The manuscript should be rejected in my opinion."
  Answer: It's a good research for Brazil because it's the best parameterization for predicting rain, important for power generation. Also, almost all natural disasters are linked to meteorology, especially related to rain. It's vital for us in Brazil to know the best method to generate rain. Question of the Flying Rivers: Better in the Amazon is good for the Southeast (Zemp, Delphine, et al. "EXPLORING THE COMPLEX NETWORK OF THE AMAZON ‘S WATER PUMP AND FLYING RIVERS."). Whoever best describes the Amazon is good for other parts of Brazil.
  "First, the entire paper reads much like a report"
  Answer: Style is a matter of author property, the text might "feels" like a report, but it isn't. The paper has been written and it's honest, this is the first and most important step. Improve is now part of the process.
  "with no or little scientific discussion. This is utterly unacceptable."
  Answer: Key factor about Taylor: Representation of the Turbulence Spectrum. It's well represented, taking into account good physics. Three aspects:
  
  - Theoretical question: this spectrum follows Kolmogorov's law of -5/3 distribution of energy in different types of orders;
  
  - The ingress of energy is represented by the maximum of the spectrum that meets the observed. The inertial subdomain, different (experimental) spectrum shape for CLC, CLN and CLE suited by exponents;
  
  - The exponents will give the face of the function (shape) of the observed spectrum. Unlike curve fitting (zero at surface and zero at top) proportional to height: Z = 0, Z = h -> K=0. That's not curve fitting, Taylor is theory. The experiments generates good results, similar or even better than Yamada (second order!), being theory and first order.
  "I think the authors should publish this as a technical report on the website of the BAM-INPE model (if there is one), instead of being a scientific paper."
  Answer: Publish in a local paper is of no use, for us this is already known: One of the main developers of the model is one of the authors of the paper (Dr. Paulo) and the other author is one of the main researchers on Taylor (Dr. Haroldo). The scientific community has an obligation to disclose this result, who else used Taylor's theory to represent turbulence in a global model?
  "Second, the paper did not really develop anything new."
  Answer: The Result is Original: first use of Taylor in a Global OPERATIONAL model. Also, we overcome one of the most difficult variables to predict: precipitation.
  The "minor comments" are important contribution for the paper with valid corrections and new contents. Here are the answers:
  1, equation 9: how is h defined and calculated?
  Answer: The h is calculated using the Holtslag-Bovile implementation (Holtslag and Bovile: Local Versus Nonlocal Boundary-Layer Diffusion in a Global Climate Model, 1992 - Equation 3.11) that is based on Troen and Mahrt (1986).
  2, algorithm 1: not necessary
  Answer: Yes, it wasn't indeed. It's now off the paper.
  3, line 154: how long is the 2nd simulation?
  Answer: It was really missing this information, We fixed the text now.
  4, line 206-210: how do the authors see the daytime results and nighttime results? I don’t see any of these in the figure.
  Answer: The figure was wrong, it supposed to be at 72h of simulation and it was at 24h instead. Also, the text was clarified point were in the map was day and night.
  5, line 233-234: “The Amazon basin region was where Taylor parameterization got its best approximations comparing to the ERA5 reanalysis data”. This sentence is not wrong per se, but it should not be in the section “Planetary Boundary Layer Height” because the Taylor parameterization is NOT the best for “Planetary Boundary Layer Height”.
  Answer: Right, it was fixed in the text.
  6, line 264-266: this sentence is unclear to me.
  Answer: This sentence was a true mess. We completely rewrite the paragraph, it's better now.
  7, line 288-289: “at least consistent results for Mean Temperature at 2 meters”. This sounds a bit double standard. Just say the Taylor parameterization is the worst.
  Answer: The section 5.4 was rewritten in a more consistent way.
  8, line 313-315: “The amount of radiation at the top of the atmosphere is a good indicator of the energy balance in the simulation, and the overall good result for this variable while using Taylor’s parameterization is a positive indication towards a good energy balance within the method.”. The authors never showed the net radiation at the top of the atmosphere, only the outgoing. If the authors comment on “energy balance”, shouldn’t they show the net radiation?
  Answer: That is a valuable contribution, We did a new analysis in the data to generate the net radiation comparison, and in fact the implementation was deficient in this area. So the statement about model energy consistence wasn't correct. A new section (5.7) and two new tables (5 and 6) were included in the text with the new results and new explanations. Also, the conclusion were corrected and a suggestion of future work on model energy balance is included.
  
  Citation: https://doi.org/10.5194/gmd-2023-59-AC1
RC2:
'Comment on gmd-2023-59', Chiel van Heerwaarden, 12 Dec 2023
Dear authors,
I am writing this short review as the editor of the paper, because in all the months that have processed, and the endless list of people that I have contacted to review this paper, nobody accepted the task, except for one anonymous reviewer that already posted his review months ago. Therefore, I apologize for the delays that you have encountered.
After reading your manuscript in detail, I feel that I unfortunately have to reject this manuscript, because in my view it is not polished enough for a scientific manuscript. I like the idea of bringing an old, but established idea from Taylor to a global model. The validation and further analysis is however inadequate for a scientific publication and requires much more work. There is hardly any interpretation of the results and discussions of lessons learned from the findings.
I would like to give a few tips on how this can be achieved:
This paper discusses a boundary-layer parameterization, and compares this entirely to ERA5. In my view, a parameterization that should improve global models should be validated against observations. The boundary-layer height of ERA5 has many issues on itself, especially in complex terrain.

The Amazon region is a wonderful region for turbulence parametrization validation. I wonder why the authors did not use observations from the ATTO infrastructure to validate their model rather than ERA5 in Figures 6-9.

The chosen variables are not the best for a boundary-layer parametrization, and at least not in the most logical order. I recommend by starting with boundary layer basics: temperature, wind, humidity, and BL depth, and only then go to variables that derive from that, such as radiation fluxes, cloud cover and precipitation, as each of those are strongly influenced by other parameterizations in your model as well.

Discuss the figures not only in the degree of how well they match, but discuss what the findings mean for the parameterization that is the main topic of the paper: how do the choices of how the physics are described influence what we see in the picture?

I hope with these tips you can improve your work.
Citation: https://doi.org/10.5194/gmd-2023-59-RC2

Eduardo Rohde Eras, Haroldo Fraga de Campos Velho, and Paulo Yoshio Kubota

Viewed

Total article views: 435 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
308	97	30	435	26	21

HTML: 308
PDF: 97
XML: 30
Total: 435
BibTeX: 26
EndNote: 21

Views and downloads (calculated since 31 May 2023)

Month	HTML	PDF	XML	Total
May 2023	18	3	1	22
Jun 2023	99	19	3	121
Jul 2023	17	9	0	26
Aug 2023	25	9	2	36
Sep 2023	24	5	3	32
Oct 2023	15	6	1	22
Nov 2023	8	3	1	12
Dec 2023	24	12	3	39
Jan 2024	6	5	2	13
Feb 2024	15	6	2	23
Mar 2024	12	9	3	24
Apr 2024	17	3	6	26
May 2024	13	5	2	20
Jun 2024	15	3	1	19

Cumulative views and downloads (calculated since 31 May 2023)

Month	HTML	PDF	XML	Total
May 2023	18	3	1	22
Jun 2023	99	19	3	121
Jul 2023	17	9	0	26
Aug 2023	25	9	2	36
Sep 2023	24	5	3	32
Oct 2023	15	6	1	22
Nov 2023	8	3	1	12
Dec 2023	24	12	3	39
Jan 2024	6	5	2	13
Feb 2024	15	6	2	23
Mar 2024	12	9	3	24
Apr 2024	17	3	6	26
May 2024	13	5	2	20
Jun 2024	15	3	1	19

Viewed (geographical distribution)

Total article views: 393 (including HTML, PDF, and XML) Thereof 393 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 29 Jun 2024

Short summary

The portion of the earth atmosphere closer to the ground is responsible for heat, moisture and mechanical energy transportation between the surface and the air through turbulence, been very important for weather forecast. Between many solutions used to model this turbulence, this is the first attempt to use one based on Taylor's statistical theory in a global atmospheric model, achieving good results for precipitation and energy transportation, specially in the Amazon basin region.


Total:	0
HTML:	0
PDF:	0
XML:	0