the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Dominant influence of Pacific climate modes on global observed and reanalysis cloud cover fields
Abstract. Global cloud cover represents a critical component of the climate system, with a considerable impact on the Earth's radiation budget. Small changes in clouds properties have a significant climatological impact because of the feedbacks that they generate, thus it is difficult to simulate the global cloud cover evolution in general circulation models. Observational investigations of cloud processes are constrained either by limited temporal and spatial extension of ground-based measurements or by imperfections in satellite data, like changes in geostationary satellite zenith angle, equatorial crossing time, or calibration. In this study, we used the Empirical Orthogonal Functions method to separate global patterns of total cloud cover variability in two satellite datasets from the International Satellite Cloud Climatology Project and the Pathfinder Atmospheres–Extended projects, each corrected for specific errors, and in the ERA5 Reanalysis. The first two modes explain most of the variance from what could be considered “signal” in both satellite data. Through Canonical Correlation Analysis, they are associated in a physically consistent manner with two different types of El Niño-Southern Oscillation (ENSO), namely the canonical ENSO which manifests itself in the eastern tropical Pacific and the El-Niño Modiki which manifest itself in the central Pacific. This work provides a comprehensive picture of the relationship between global total cloud cover and the tropical Pacific processes and indicates that cloud cover in the Indo-Pacific sector plays a significant role in the Earth radiative budget at interannual to decadal time scales. The similarity of the results across satellite and reanalysis data indicate that the both the observed and reanalysis cloud data sets contain consistent and valuable information related to global climate variability.
- Preprint
(1706 KB) - Metadata XML
-
Supplement
(527 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on esd-2021-75', Anonymous Referee #1, 08 Dec 2021
Review of “Dominant influence of Pacific climate modes on global observed and reanalysis cloud cover fields” by Petru Vaideanu et al.
The authors use Empirical Orthogonal Functions (EOFs) and Canonical Correlation Analysis (CCA) to explore patterns of variability and covariability mainly between sea surface temperatures and total cloud coverage. The results show coherent patterns of co-variability between SST and cloud patterns typical for Central and Eastern Pacific ENSO events, respectively. The manuscript also includes relationships of ENSO variability with other variables such as precipitation, mean sea level pressure, and winds. Lastly, CCA patterns obtained from temporally smoothed data are shown, which show some resemblance of the impact of Pacific Decadal Oscillation on clouds.
There are some unclarities concerning the methodology (Are all involved fields deseasonalized and detrended? What kind of temporal smoothing is applied for the results presented in Figs 2-4? Is the CCA applied to the full spatial fields or to a set of PCs derived from the EOF analysis?), but the results look reasonable. However, I have two more major difficulties with this study: first, it presents a very limited set of diagnostics (3 out of 6 figures show the same diagnostic using different data sets). Second, and this is the biggest problem, the results are not novel at all. There is a huge body of literature documenting the response of various atmospheric fields (including clouds) to ENSO. While probably not many studies have applied CCA to this very specific question (how do SSTs and clouds co-vary?), the method does not reveal anything new.
One example for ENSO-related cloud variability is Wang et al (2015), which in fact includes more up-to-date obs-data (e.g. CERES) than the present manuscript. An example for PDO-related variability of atmospheric quantities is Chen et al. (2019) (using obs and models). A quick online search brought many more papers with similar topics.
The authors do not claim that their results show much new beyond the state of the science, but scientific novelty is nevertheless a criterion for publication in ESD (see criterion 2 here: https://www.earth-system-dynamics.net/peer_review/review_criteria.html). I hence have to recommend to reject this manuscript.
I would like to mention that there are journals that only require soundness of methods for publication (e.g. Scientific Reports), which the authors may consider as an option.
References:
Chen, Y. J., Hwang, Y. T., Zelinka, M. D., & Zhou, C. (2019). Distinct patterns of cloud changes associated with decadal variability and their contribution to observed cloud cover trends. Journal of Climate, 32(21), 7281-7301.
Wang, H., and W. Su (2015), The ENSO effects on tropical clouds and top-of-atmosphere cloud radiative effects in CMIP5 models, J. Geophys. Res. Atmos., 120, 4443–4465, doi:10.1002/2014JD022337.
Citation: https://doi.org/10.5194/esd-2021-75-RC1 -
AC1: 'Reply on RC1', Petru Vaideanu, 13 Feb 2022
The comment was uploaded in the form of a supplement: https://esd.copernicus.org/preprints/esd-2021-75/esd-2021-75-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Petru Vaideanu, 13 Feb 2022
-
RC2: 'Comment on esd-2021-75', Anonymous Referee #2, 17 Jan 2022
Review of "Dominant influence of Pacific climate modes on global observed and reanalysis cloud cover fields" by Vaideanu et al.
This paper studies the dominant modes of variability in total cloud cover. The Principal Component Analysis (PCA) is used to decompose the observed total cloud cover into dominant modes and the Canonical Correlation Analysis (CCA) is used to identify the physical linkage with known atmospheric-ocean variabilities. The authors show that the dominant modes are the Central Pacific ENSO and the ENSO modoki in the tropical Pacific region.
The PCA and CCA analyses used in this study are well established in the scientific community and have been commonly used in climate studies. However, I am concerned about the physical interpretation of the results.
As the authors discussed in Section 1, cloud is a highly uncertain variable in climate prediction because of its spatial variability in both vertical and horizontal and because of the fact that cloud at different altitudes induces very different and sometimes opposite climate forcings. Therefore, I am a little hesitant to study the total cloud variability instead of studying high and low clouds separately. The authors should explain more clearly why studying the total cloud cover is important.
The principal modes in the total cloud cover, the CP ENSO and ENSO Modoki, have already been known in a previous study which studies high cloud fraction [Li et al. (2016), An Analysis of High Cloud Variability: Imprints from the El Niño-Southern Oscillation, Climate Dynamics, 10.1007/s00382-016-3086-7]. Therefore, the principal modes found in this study are not new and are primarily due to the high clouds instead of the total cloud cover, as is also explained by the authors in their Section 3.5. It also goes back to my comment above that why should we study the total cloud cover?
A critical issue with the authors' interpretation of the principal modes is that they ignore the second mode of ERA5R total cloud cover because "[t]he second EOF derived using the ERA5R TCC data (Supp. Fig. 1) is not of interest for our study due to its temporal characteristics." The PCA is a pure mathematical decomposition of any given matrix, random or not, where the principal modes are forced to be mutually orthogonal singular vectors and no physical constraints are applied in the construction of the singular vectors. Therefore, one must be extremely careful when trying to attribute physical meanings to the principal modes. The fact that the phenomenon of the authors' interests has shifted from the second mode to the third in ERA5R TCC means that the second and third modes are likely degenerated because the eigenvalues of these modes are statistically indistinguishable. A serious problem associated with degenerated modes is mode-mixing, which makes the physical interpretation of the degnerated modes difficult. As the authors are trying to compare ERA5R modes with those obtained from ISCCP and PATMOS-x, simply ignoring the second mode without considering possible mode-mixing in ERA5R TTC could potentially lead to inaccurate conclusion about the quality of the ERA5R assimilations. The authors may check whether there are also mode-mixing in the second and third modes in ISCCP and PATMOS-x TCC. For more details on mode-mixing, see Quadrelli et al. (2005), On Sampling Errors in Empirical Orthogonal Functions, Journal of Climate, 10.1175/JCLI3500.1.
A minor comment is that the authors mentioned in abstract and in the text a few times the ISCCP and PATMOS-x "each corrected for specific errors". I first thought that the authors corrected these data themselves but the authors actually downloaded the corrected data directly from the web. The discussion about the correction by Norris et al. in Section 2.1 is reasonable, but can the authors elaborate more on why they want to emphasize the correction in various places?
Citation: https://doi.org/10.5194/esd-2021-75-RC2 -
AC2: 'Reply on RC2', Petru Vaideanu, 13 Feb 2022
The comment was uploaded in the form of a supplement: https://esd.copernicus.org/preprints/esd-2021-75/esd-2021-75-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Petru Vaideanu, 13 Feb 2022
Status: closed
-
RC1: 'Comment on esd-2021-75', Anonymous Referee #1, 08 Dec 2021
Review of “Dominant influence of Pacific climate modes on global observed and reanalysis cloud cover fields” by Petru Vaideanu et al.
The authors use Empirical Orthogonal Functions (EOFs) and Canonical Correlation Analysis (CCA) to explore patterns of variability and covariability mainly between sea surface temperatures and total cloud coverage. The results show coherent patterns of co-variability between SST and cloud patterns typical for Central and Eastern Pacific ENSO events, respectively. The manuscript also includes relationships of ENSO variability with other variables such as precipitation, mean sea level pressure, and winds. Lastly, CCA patterns obtained from temporally smoothed data are shown, which show some resemblance of the impact of Pacific Decadal Oscillation on clouds.
There are some unclarities concerning the methodology (Are all involved fields deseasonalized and detrended? What kind of temporal smoothing is applied for the results presented in Figs 2-4? Is the CCA applied to the full spatial fields or to a set of PCs derived from the EOF analysis?), but the results look reasonable. However, I have two more major difficulties with this study: first, it presents a very limited set of diagnostics (3 out of 6 figures show the same diagnostic using different data sets). Second, and this is the biggest problem, the results are not novel at all. There is a huge body of literature documenting the response of various atmospheric fields (including clouds) to ENSO. While probably not many studies have applied CCA to this very specific question (how do SSTs and clouds co-vary?), the method does not reveal anything new.
One example for ENSO-related cloud variability is Wang et al (2015), which in fact includes more up-to-date obs-data (e.g. CERES) than the present manuscript. An example for PDO-related variability of atmospheric quantities is Chen et al. (2019) (using obs and models). A quick online search brought many more papers with similar topics.
The authors do not claim that their results show much new beyond the state of the science, but scientific novelty is nevertheless a criterion for publication in ESD (see criterion 2 here: https://www.earth-system-dynamics.net/peer_review/review_criteria.html). I hence have to recommend to reject this manuscript.
I would like to mention that there are journals that only require soundness of methods for publication (e.g. Scientific Reports), which the authors may consider as an option.
References:
Chen, Y. J., Hwang, Y. T., Zelinka, M. D., & Zhou, C. (2019). Distinct patterns of cloud changes associated with decadal variability and their contribution to observed cloud cover trends. Journal of Climate, 32(21), 7281-7301.
Wang, H., and W. Su (2015), The ENSO effects on tropical clouds and top-of-atmosphere cloud radiative effects in CMIP5 models, J. Geophys. Res. Atmos., 120, 4443–4465, doi:10.1002/2014JD022337.
Citation: https://doi.org/10.5194/esd-2021-75-RC1 -
AC1: 'Reply on RC1', Petru Vaideanu, 13 Feb 2022
The comment was uploaded in the form of a supplement: https://esd.copernicus.org/preprints/esd-2021-75/esd-2021-75-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Petru Vaideanu, 13 Feb 2022
-
RC2: 'Comment on esd-2021-75', Anonymous Referee #2, 17 Jan 2022
Review of "Dominant influence of Pacific climate modes on global observed and reanalysis cloud cover fields" by Vaideanu et al.
This paper studies the dominant modes of variability in total cloud cover. The Principal Component Analysis (PCA) is used to decompose the observed total cloud cover into dominant modes and the Canonical Correlation Analysis (CCA) is used to identify the physical linkage with known atmospheric-ocean variabilities. The authors show that the dominant modes are the Central Pacific ENSO and the ENSO modoki in the tropical Pacific region.
The PCA and CCA analyses used in this study are well established in the scientific community and have been commonly used in climate studies. However, I am concerned about the physical interpretation of the results.
As the authors discussed in Section 1, cloud is a highly uncertain variable in climate prediction because of its spatial variability in both vertical and horizontal and because of the fact that cloud at different altitudes induces very different and sometimes opposite climate forcings. Therefore, I am a little hesitant to study the total cloud variability instead of studying high and low clouds separately. The authors should explain more clearly why studying the total cloud cover is important.
The principal modes in the total cloud cover, the CP ENSO and ENSO Modoki, have already been known in a previous study which studies high cloud fraction [Li et al. (2016), An Analysis of High Cloud Variability: Imprints from the El Niño-Southern Oscillation, Climate Dynamics, 10.1007/s00382-016-3086-7]. Therefore, the principal modes found in this study are not new and are primarily due to the high clouds instead of the total cloud cover, as is also explained by the authors in their Section 3.5. It also goes back to my comment above that why should we study the total cloud cover?
A critical issue with the authors' interpretation of the principal modes is that they ignore the second mode of ERA5R total cloud cover because "[t]he second EOF derived using the ERA5R TCC data (Supp. Fig. 1) is not of interest for our study due to its temporal characteristics." The PCA is a pure mathematical decomposition of any given matrix, random or not, where the principal modes are forced to be mutually orthogonal singular vectors and no physical constraints are applied in the construction of the singular vectors. Therefore, one must be extremely careful when trying to attribute physical meanings to the principal modes. The fact that the phenomenon of the authors' interests has shifted from the second mode to the third in ERA5R TCC means that the second and third modes are likely degenerated because the eigenvalues of these modes are statistically indistinguishable. A serious problem associated with degenerated modes is mode-mixing, which makes the physical interpretation of the degnerated modes difficult. As the authors are trying to compare ERA5R modes with those obtained from ISCCP and PATMOS-x, simply ignoring the second mode without considering possible mode-mixing in ERA5R TTC could potentially lead to inaccurate conclusion about the quality of the ERA5R assimilations. The authors may check whether there are also mode-mixing in the second and third modes in ISCCP and PATMOS-x TCC. For more details on mode-mixing, see Quadrelli et al. (2005), On Sampling Errors in Empirical Orthogonal Functions, Journal of Climate, 10.1175/JCLI3500.1.
A minor comment is that the authors mentioned in abstract and in the text a few times the ISCCP and PATMOS-x "each corrected for specific errors". I first thought that the authors corrected these data themselves but the authors actually downloaded the corrected data directly from the web. The discussion about the correction by Norris et al. in Section 2.1 is reasonable, but can the authors elaborate more on why they want to emphasize the correction in various places?
Citation: https://doi.org/10.5194/esd-2021-75-RC2 -
AC2: 'Reply on RC2', Petru Vaideanu, 13 Feb 2022
The comment was uploaded in the form of a supplement: https://esd.copernicus.org/preprints/esd-2021-75/esd-2021-75-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Petru Vaideanu, 13 Feb 2022
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
776 | 191 | 45 | 1,012 | 72 | 39 | 45 |
- HTML: 776
- PDF: 191
- XML: 45
- Total: 1,012
- Supplement: 72
- BibTeX: 39
- EndNote: 45
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1