the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Jun 2023
Sensitivities of cloud radiative effects to large-scale meteorology and aerosols from global observations
Abstract. The radiative effects of clouds make a large contribution to the Earth's energy balance, and changes in clouds constitute the dominant source of uncertainty in the global warming response to carbon dioxide forcing. To characterize and constrain this uncertainty, cloud controlling factor (CCF) analyses have been suggested that estimate sensitivities of clouds to large-scale environmental changes, typically in cloud-regime specific multiple linear regression frameworks. Here, local sensitivities of cloud radiative effects to a large number of controlling factors are estimated in a regime-independent framework from 20 years of near-global satellite observations and reanalysis data using statistical learning. A regularized linear regression (ridge regression) is shown to skillfully predict anomalies of shortwave (R² = 0.63) and longwave CRE (R² = 0.72) in independent test data on the basis of 28 CCFs, including aerosol proxies. The sensitivity of CRE to selected CCFs is quantified and analyzed. CRE sensitivities to sea-surface temperature and estimated inversion strength are particularly pronounced in low-cloud regions and generally in agreement with previous studies. The analysis of CRE sensitivities to three-dimensional wind field anomalies reflects that CREs in tropical ascent regions are mainly driven by variability of large-scale vertical velocity in the upper troposphere. In the subtropics, CRE is sensitive to free-tropospheric zonal and meridional wind anomalies, which are likely to encapsulate information on synoptic variability that influences subtropical cloud systems by modifying wind shear and thus turbulence and dry-air entrainment in stratocumulus clouds, as well as variability related to midlatitude cyclones. Different proxies for aerosols are analyzed as CCFs, with satellite-derived aerosol proxies showing a larger CRE sensitivity than a proxy from an aerosol reanalysis, likely pointing to satellite aerosol retrieval biases close to clouds leading to overestimated aerosol sensitivities. Sensitivities of shortwave CRE to all aerosol proxies indicate a pronounced cooling effect from aerosols in stratocumulus regions that is counteracted to a varying degree by a longwave warming effect. The analysis may guide the selection of CCFs in future sensitivity analyses aimed at constraining cloud feedback and climate forcings from aerosol-cloud interactions, using both data from observations and global climate models.
-
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
(8881 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(8881 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1283', Anonymous Referee #1, 17 Jul 2023
General comments
This paper examines the sensitivity of cloud radiative effects (CREs) to a wide range of cloud-controlling factors and aerosol proxies using ridge regression models. The trained models are skillful in predicting both shortwave and longwave CREs, enabling further sensitivity analysis of CREs to different cloud-controlling factors, such as main low-cloud controls, three-dimensional wind fields, and various aerosol proxies.
Overall, I find this paper to be well-written and to present interesting results. The use of ridge regression and aerosol proxies expands the traditional framework of cloud-controlling factor analysis. Please see specific comments and technical corrections below.
Specific comments
Page 2, Introduction: Since this paper focuses on both the shortwave and longwave cloud radiative effect, it is recommended that the authors provide more background on the high-cloud feedback in addition to the low-cloud feedback.
Page 4, Line 15-20: To improve the paragraph's flow, please consider moving the sentence “Satellite observations from the polar-orbiting platform Terra are used” to line 20, before “Two commonly used proxies for CCNs …”
Page 4, Line 29: What is meant by "single-layer reanalysis"? Does it refer to data at a single level? Please provide clarification.
Page 4, Line 30: Please consider adding references for ERA5 reanalysis and Merra-2 reanalysis. For example:
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz-Sabater, J., et al. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999–2049. https://doi.org/10.1002/qj.3803
Randles, C. A., Silva, A. M. da, Buchard, V., Colarco, P. R., Darmenov, A., Govindaraju, R., et al. (2017). The MERRA-2 Aerosol Reanalysis, 1980 Onward. Part I: System Description and Data Assimilation Evaluation. Journal of Climate, 30(17), 6823–6850. https://doi.org/10.1175/JCLI-D-16-0609.1
Page 5, Section 2.2: A few suggestions for this subsection:
- The equation at line 10 is missing prime symbols in the parentheses?
- To help the readers to appreciate the connection and difference between ordinary least square(OLS) regression and ridge regression, the authors might consider rewriting the first equation in terms of coefficient β (and also including the loss function). This way, when λ=0, it reduces to OLS.
- The authors emphasized in the introduction that a benefit of ridge regression is providing robust estimation in the case of collinear predictors. It would be worthwhile to briefly explain why ridge regression can help with collinearity here.
Page 6, line 27: the authors state that the differences in predictive skill between the locally optimized and λ = 12 settings are negligible. And, it would be interesting to compare the predictive skill of the λ = 12 setting and that of the λ = 0.001 (-3 in log scale) setting. One way to demonstrate the superiority of ridge regression over OLS is to use a small penalty, such as λ = 0.001, and show that ridge regression is more skillful than OLS.
Page 8, line 1-5: what’s the average skill over the Sc regime? How does it compare with the previous studies?
Page 8, lines 5-9: The authors attribute the low skills in simulating the Southern Ocean (SO) to the low quality of reanalysis data. However, are there other possible explanations? Is it possible that the factors controlling SO clouds are not well represented by this set of CCFs?
Page 17, Line 9-10: It would be helpful to add references to discuss where sulfate would dominate CCN and where it would not.
Technical corrections
Page 1, Line 7: “CRE”, please gives full name since this is the first appearance.
Page 1, Abstract: It would be helpful for the readers to include the time span and latitude range covered in this study.
Page 2, Line 9: Please check the double quotation marks.
Page 4, Line 14: Please remove the s after CRE_LW.
Figure 1 (and other Figures): I noticed that some color bars in the figures have pointy ends while others are rectangular. If they indicate different meanings (e.g. some values have been clipped), please add a description in the figure caption (at least for Figure 1).
Figure 2: In the figure caption, please specify that λ=12 was chosen.
Figure 3 & Table 1: In Figure 3, “ML” can be changed to “Ml” to be consistent with the table and the texts. It might help to give the full name in addition to the abbreviation.
Page 8, Line 1-10: Please mention Fig.4, where the results are from. The same is also true for Fig.5 in section 3.2.
Page 8, Line 4: “ACI”, please gives full name since this is the first appearance.
Citation: https://doi.org/10.5194/egusphere-2023-1283-RC1 -
RC2: 'Comment on egusphere-2023-1283', Anonymous Referee #2, 18 Jul 2023
Review of the manuscript “Sensitivities of cloud radiative effects to large-scale meteorology and aerosols from global observations” by Hendrik Andersen, Jan Cermak, Alyson Douglas, Timothy A. Myers, Peer Nowack, Philip Stier, Casey J. Wall, and Sarah Wilson Kemsley
General
This study performed the sensitivity analyses with added cloud controlling factors with different aerosol proxies using a regularized linear regression (Ridge regression). Based on the high prediction skill of model, the authors have explained the dominant predictor (or proxy) for the CRE sensitivity.
The subject of this work is important to capture processes relevant to grasp CRE variability and cloud feedbacks. However, the manuscript needs some more explanations on the concepts for the variables and notable findings shown in the figures. Please see the details in the comments and questions below.
- On the aerosol proxy, before introducing the usage of aerosol proxy for CCF frameworks in the previous studies, specific explanation on its definition or concept could be more helpful for readers to understand that role and further results in this study.
- In Fig. 1, the resource of label bars is different. Please double-check.
- Please correct the ‘Res’ in the line 11 of page 6 section 2.2 (Ridge regression) as the Italic.
- What is the specific reason for the selection of four regions of clouds?
- What is ‘ACI’ in the line 4 of page 8?
- Lines 15-17 of page 8 is important interpretation on the role of regularization strength for ability of predictors to explain the CRE variability. They seemed to have more explanations.
- It would be better to describe what the positive/negative sensitivity of each CRE for certain variable indicates, such as explanations of SW reflectance or LW emission/trapping, before breaking down the explanations.
- A band of moderate positive CRELW-SST sensitivity shown in Fig. 5 seemed to be related to the tropical ascending (Ta) regimes. That is, in the mechanism of convection, vertical ascending in the convection may comparatively more relevant to the CRELW. Also, negative CRELW sensitivity to SST showed notable over trade cumulus (Tc) regimes. Is there any further interpretation or implication on this?
- ‘The sensitivity patterns of CRESW and CRELW (Fig. 7)’ in lines 6-8 should be corrected as ‘The sensitivity of CRESW and CRELW to 300’.
- For the section 3.3 Sensitivity of CRESW and CRELW to large-scale circulation, it is well understandable until the paragraph on the impact of 300. On the other hand, for the subtropics, too complicated contents are explained in a bulk. The conclusion is seemed to be that the decrease in low clouds due to the westerly leads the strong positive CRESW sensitivity to U700 and V700. If is correct, what authors intended to with the explanation of boundary layer humidity (RH925) in Fig. 9.
- For the lines 19-20 of page 17, the authors well explained the expected differences from the aerosol proxies, especially for the satellite-observed proxies. However, for the results written in the lines of 16-18 on the same page, they need more interpretations, exactly what differences have induced those results.
- Concluding findings on the zonal and meridional winds (4) and three aerosol proxies (5) on the page 19-20 are too comprehensive. Like other paragraphs, please encapsulate them more briefly.
Citation: https://doi.org/10.5194/egusphere-2023-1283-RC2 - AC1: 'Comment on egusphere-2023-1283', Hendrik Andersen, 15 Aug 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1283', Anonymous Referee #1, 17 Jul 2023
General comments
This paper examines the sensitivity of cloud radiative effects (CREs) to a wide range of cloud-controlling factors and aerosol proxies using ridge regression models. The trained models are skillful in predicting both shortwave and longwave CREs, enabling further sensitivity analysis of CREs to different cloud-controlling factors, such as main low-cloud controls, three-dimensional wind fields, and various aerosol proxies.
Overall, I find this paper to be well-written and to present interesting results. The use of ridge regression and aerosol proxies expands the traditional framework of cloud-controlling factor analysis. Please see specific comments and technical corrections below.
Specific comments
Page 2, Introduction: Since this paper focuses on both the shortwave and longwave cloud radiative effect, it is recommended that the authors provide more background on the high-cloud feedback in addition to the low-cloud feedback.
Page 4, Line 15-20: To improve the paragraph's flow, please consider moving the sentence “Satellite observations from the polar-orbiting platform Terra are used” to line 20, before “Two commonly used proxies for CCNs …”
Page 4, Line 29: What is meant by "single-layer reanalysis"? Does it refer to data at a single level? Please provide clarification.
Page 4, Line 30: Please consider adding references for ERA5 reanalysis and Merra-2 reanalysis. For example:
Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz-Sabater, J., et al. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999–2049. https://doi.org/10.1002/qj.3803
Randles, C. A., Silva, A. M. da, Buchard, V., Colarco, P. R., Darmenov, A., Govindaraju, R., et al. (2017). The MERRA-2 Aerosol Reanalysis, 1980 Onward. Part I: System Description and Data Assimilation Evaluation. Journal of Climate, 30(17), 6823–6850. https://doi.org/10.1175/JCLI-D-16-0609.1
Page 5, Section 2.2: A few suggestions for this subsection:
- The equation at line 10 is missing prime symbols in the parentheses?
- To help the readers to appreciate the connection and difference between ordinary least square(OLS) regression and ridge regression, the authors might consider rewriting the first equation in terms of coefficient β (and also including the loss function). This way, when λ=0, it reduces to OLS.
- The authors emphasized in the introduction that a benefit of ridge regression is providing robust estimation in the case of collinear predictors. It would be worthwhile to briefly explain why ridge regression can help with collinearity here.
Page 6, line 27: the authors state that the differences in predictive skill between the locally optimized and λ = 12 settings are negligible. And, it would be interesting to compare the predictive skill of the λ = 12 setting and that of the λ = 0.001 (-3 in log scale) setting. One way to demonstrate the superiority of ridge regression over OLS is to use a small penalty, such as λ = 0.001, and show that ridge regression is more skillful than OLS.
Page 8, line 1-5: what’s the average skill over the Sc regime? How does it compare with the previous studies?
Page 8, lines 5-9: The authors attribute the low skills in simulating the Southern Ocean (SO) to the low quality of reanalysis data. However, are there other possible explanations? Is it possible that the factors controlling SO clouds are not well represented by this set of CCFs?
Page 17, Line 9-10: It would be helpful to add references to discuss where sulfate would dominate CCN and where it would not.
Technical corrections
Page 1, Line 7: “CRE”, please gives full name since this is the first appearance.
Page 1, Abstract: It would be helpful for the readers to include the time span and latitude range covered in this study.
Page 2, Line 9: Please check the double quotation marks.
Page 4, Line 14: Please remove the s after CRE_LW.
Figure 1 (and other Figures): I noticed that some color bars in the figures have pointy ends while others are rectangular. If they indicate different meanings (e.g. some values have been clipped), please add a description in the figure caption (at least for Figure 1).
Figure 2: In the figure caption, please specify that λ=12 was chosen.
Figure 3 & Table 1: In Figure 3, “ML” can be changed to “Ml” to be consistent with the table and the texts. It might help to give the full name in addition to the abbreviation.
Page 8, Line 1-10: Please mention Fig.4, where the results are from. The same is also true for Fig.5 in section 3.2.
Page 8, Line 4: “ACI”, please gives full name since this is the first appearance.
Citation: https://doi.org/10.5194/egusphere-2023-1283-RC1 -
RC2: 'Comment on egusphere-2023-1283', Anonymous Referee #2, 18 Jul 2023
Review of the manuscript “Sensitivities of cloud radiative effects to large-scale meteorology and aerosols from global observations” by Hendrik Andersen, Jan Cermak, Alyson Douglas, Timothy A. Myers, Peer Nowack, Philip Stier, Casey J. Wall, and Sarah Wilson Kemsley
General
This study performed the sensitivity analyses with added cloud controlling factors with different aerosol proxies using a regularized linear regression (Ridge regression). Based on the high prediction skill of model, the authors have explained the dominant predictor (or proxy) for the CRE sensitivity.
The subject of this work is important to capture processes relevant to grasp CRE variability and cloud feedbacks. However, the manuscript needs some more explanations on the concepts for the variables and notable findings shown in the figures. Please see the details in the comments and questions below.
- On the aerosol proxy, before introducing the usage of aerosol proxy for CCF frameworks in the previous studies, specific explanation on its definition or concept could be more helpful for readers to understand that role and further results in this study.
- In Fig. 1, the resource of label bars is different. Please double-check.
- Please correct the ‘Res’ in the line 11 of page 6 section 2.2 (Ridge regression) as the Italic.
- What is the specific reason for the selection of four regions of clouds?
- What is ‘ACI’ in the line 4 of page 8?
- Lines 15-17 of page 8 is important interpretation on the role of regularization strength for ability of predictors to explain the CRE variability. They seemed to have more explanations.
- It would be better to describe what the positive/negative sensitivity of each CRE for certain variable indicates, such as explanations of SW reflectance or LW emission/trapping, before breaking down the explanations.
- A band of moderate positive CRELW-SST sensitivity shown in Fig. 5 seemed to be related to the tropical ascending (Ta) regimes. That is, in the mechanism of convection, vertical ascending in the convection may comparatively more relevant to the CRELW. Also, negative CRELW sensitivity to SST showed notable over trade cumulus (Tc) regimes. Is there any further interpretation or implication on this?
- ‘The sensitivity patterns of CRESW and CRELW (Fig. 7)’ in lines 6-8 should be corrected as ‘The sensitivity of CRESW and CRELW to 300’.
- For the section 3.3 Sensitivity of CRESW and CRELW to large-scale circulation, it is well understandable until the paragraph on the impact of 300. On the other hand, for the subtropics, too complicated contents are explained in a bulk. The conclusion is seemed to be that the decrease in low clouds due to the westerly leads the strong positive CRESW sensitivity to U700 and V700. If is correct, what authors intended to with the explanation of boundary layer humidity (RH925) in Fig. 9.
- For the lines 19-20 of page 17, the authors well explained the expected differences from the aerosol proxies, especially for the satellite-observed proxies. However, for the results written in the lines of 16-18 on the same page, they need more interpretations, exactly what differences have induced those results.
- Concluding findings on the zonal and meridional winds (4) and three aerosol proxies (5) on the page 19-20 are too comprehensive. Like other paragraphs, please encapsulate them more briefly.
Citation: https://doi.org/10.5194/egusphere-2023-1283-RC2 - AC1: 'Comment on egusphere-2023-1283', Hendrik Andersen, 15 Aug 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 397 | 182 | 17 | 596 | 10 | 9 |
- HTML: 397
- PDF: 182
- XML: 17
- Total: 596
- BibTeX: 10
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Hendrik Andersen
Jan Cermak
Alyson Douglas
Timothy A. Myers
Peer Nowack
Philip Stier
Casey J. Wall
Sarah Wilson Kemsley
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(8881 KB) - Metadata XML