Evaluation of Future Simulations of the CMIP5 GCMs Concerning Boreal Wintertime Atmospheric Teleconnection Patterns
Round 1
Reviewer 1 Report
Comments to the authors
The study shows an analysis of teleconnections for the RCP4.5 and RCP8.5 scenarios for 7 CMIP5 models. The evaluation work shows that all the models capture part of these teleconnections with some variations. The study is interesting and can be accepted after addressing some issues. For example, would the application of other methods based on positive correlations generate higher significance values?
What happens with the aspect of temporal resolution, that is to say, when dealing with a seasonal behavior of the variation of geopotential values, for example using monthly values, as well as a temporal lag? Has the detection algorithm been tested under those conditions?
Author Response
The Author thanks the Reviewer for the useful comments on the manuscript (MS). Please find the answers below and in the attached pdf file.
Point 1: The study shows an analysis of teleconnections for the RCP4.5 and RCP8.5 scenarios for 7 CMIP5 models. The evaluation work shows that all the models capture part of these teleconnections with some variations. The study is interesting and can be accepted after addressing some issues. For example, would the application of other methods based on positive correlations generate higher significance values?
Response 1: Thank you for the suggestion. The pattern detection method used in this study is not applicable to detect positively correlated areas, consequently this question was not addressed previously. It would be noteworthy to try an algorithm which based on positive correlations, e.g. the method of empirical ortogonal teleconnections (van den Dool et al., 2000) because of the shortcomings of the presented detection method in the MS. Namely it can only identify each cluster/teleconnection which consists of only two action centers (ACs). It would corroborate the study, if teleconnections with more than two ACs could be captured, e.g. the Pacific-North American pattern with all of its four ACs. At this point of the study pattern detection methods based on empirical orthogonal function analysis – which also rely on the correlation matrix – are out of scope, mainly because it is difficult to associate physical meaning to the detected patterns (e.g. determine ACs) (Horel, 1981; Dommenguet & Latif, 2002).
Point 2: What happens with the aspect of temporal resolution, that is to say, when dealing with a seasonal behavior of the variation of geopotential values, for example using monthly values, as well as a temporal lag? Has the detection algorithm been tested under those conditions?
Response 2: The Author applied the detection algorithm on monthly data at an early stage of the study. The negative correlations obtained from gridded monthly time series were stronger than those presented in the MS because of the shorter time series, however the coarse temporal resolution led to fading ACs in the fields of strongest negative correlations (SNCs). For example the ACs of the Mediterranean Oscillation were undetectable in the SNC fields of the reanalyses/GCMs in some time periods. Because of this, monthly time series are not used yet.
A temporal lag was not applied in the detection algorithm because the Author intended to detect teleconnections which most intense regions vary simultaneously. Previously – as a test – time-lagged correlations were computed between teleconnection indices obtained from the SNC fields in the Euro-Atlantic region and atmospheric variables to identify connections between them.
References:
Dommenget, D.; Latif, M. A Cautionary Note on the Interpretation of EOFs. J. Clim. 2002, 15, 216–225. DOI: /10.1175/1520-0442(2002)015<0216:ACNOTI>2.0.CO;2
Horel, J.D. A Rotated Principal Component Analysis of the Interannual Variability of the Northern Hemisphere 500 mb Height Field. Mon. Weather Rev. 1981, 109, 2080–2092. DOI: 10.1175/1520-0493(1981)109<2080:ARPCAO>2.0.CO;2.
van den Dool, H.M.; Saha, S.; Johansson, Å. Empirical Orthogonal Teleconnections. J. Clim. 2000, 13, 1421–1435. DOI: 10.1175/1520-0442(2000)013<1421:EOT>2.0.CO;2.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The current paper was largely based upon the author's previous publications [24, 26], aimed to test any future changes in atmospheric teleconnection patterns in response to increasing future radiative forcings.
The manuscripts are well written to understand the main findings of the paper, though there are some rooms for improvement.
Major comments
1. Now the CMIP6 simulations are available, then why did the author use the previous one? I believe the current manuscript is not about the method, then the use of old CMIP5 is somewhat out of date.
2. The overall conclusion needs to be more elaborated. What does the general similarity of teleconnection patterns in future projections mean? Does that indicate the atmospheric intrinsic variability will be the largest factor controlling the atmospheric teleconnection patterns? Then what will be the impacts of future radiative forcing changes.
3. As a somewhat continued question of 2, how about the magnitude responses (not a correlation magnitude but a variance change) of atmospheric teleconnections to future radiative forcings? If the overall patterns remain the same but the magnitude changes significantly, does that indicate the amplification of intrinsic (naturally varying) atmospheric teleconnections due to future radiative forcing changes?
Minor changes
1. I understood the current manuscript was based upon the previous results. But it is much easier to read if you provide MCC and AUC values estimated from historical runs and ERA-20C (yrs. 1976-2005) in Table 1.
2. Line 79 ...leap days were removed for the consistency of GCMS, therefore ...
Author Response
The Author thanks the Reviewer for the useful comments on the manuscript (MS). Please find the answers below and in the attached pdf file.
Please note that line numbering refers to the clean version of the revised MS where changes made relative to the original MS are invisible.
Major comments
Point 1. Now the CMIP6 simulations are available, then why did the author use the previous one? I believe the current manuscript is not about the method, then the use of old CMIP5 is somewhat out of date.
Response 1. The scope of the study was the application of the pattern detection method on those GCMs, which historical outputs were validated by using the same detection algorithm in Kristóf et al. (2021) (denoted by [24] in the original MS). The CMIP5 GCMs are evaluated whether their RCP simulation outputs keep the teleconnection patterns, which were identified in their historical simulations. The results of this MS could be used to validate the RCP simulation outputs of the CMIP5 GCMs if sufficiently long time series will be available. The validation results may help the modelling community to improve climate models. This is emphasized in the revised MS as a concluding remark (in lines 357-361).
Point 2. The overall conclusion needs to be more elaborated. What does the general similarity of teleconnection patterns in future projections mean? Does that indicate the atmospheric intrinsic variability will be the largest factor controlling the atmospheric teleconnection patterns? Then what will be the impacts of future radiative forcing changes.
Response 2. The overall conclusion of this study is that teleconnections are detected in the vast majority of the GCM simulations through the 21st century, which indicates that natural variability is the major factor that controls the teleconnections. However, the increasing radiative forcing may affect the correlation magnitude between the most intense regions of the teleconnections and their spatial distribution. Those are implied by the larger variability of the AUC and the MCCmax values in case of the RCP8.5 scenario relative to the RCP4.5 scenario (see in Fig. 4). Thank you for the suggestion, in the revised MS the above-mentioned are emphasized in the Conclusion (in lines 350-356).
Point 3. As a somewhat continued question of 2, how about the magnitude responses (not a correlation magnitude but a variance change) of atmospheric teleconnections to future radiative forcings? If the overall patterns remain the same but the magnitude changes significantly, does that indicate the amplification of intrinsic (naturally varying) atmospheric teleconnections due to future radiative forcing changes?
Response 3. To answer this question, variances were determined in each grid cell at the geopotential height field at the 500 hPa pressure level (zg500) examined in the MS (the R codes are available in the script 07_function_-_computing_variances.R in the GitHub repository https://github.com/ekristof86/
R_codes_related_to_atmospheric_teleconnections). The majority of the seven GCMs analyzed in the MS produce larger variance values under the RCP4.5 scenario, compared to the variance values under the RCP8.5 scenario. Regarding the three analyzed time periods, the variances are decreasing if the consecutive periods are compared to each other. Based on these preliminary results, the increasing radiative forcing may lead to variance change concerning atmospheric teleconnections in the zg500, in the boreal wintertime in the Northern Hemisphere. However, to determine its significance, this topic should be analyzed more thoroughly. Thank you for the suggestion.
Minor changes
Point 1. I understood the current manuscript was based upon the previous results. But it is much easier to read if you provide MCC and AUC values estimated from historical runs and ERA-20C (yrs. 1976-2005) in Table 1.
Response 1. The MCC and AUC values are presented in Table A1 in the revised MS.
Point 2. Line 79 ...leap days were removed for the consistency of GCMS, therefore ...
Response 2. Thank you for the suggestion. It is inserted in line 95 in the revised MS.
Reference:
Kristóf, E.; Hollós, R.; Barcza, Z.; Pongrácz, R.; Bartholy, J. Receiver Operating Characteristic Curve Analysis-Based Evaluation of GCMs Concerning Atmospheric Teleconnections. Atmosphere, 2021, 12. DOI: 10.3390/atmos12101236.
Author Response File: Author Response.pdf
Reviewer 3 Report
It’s an intriguing research topic which perfectly aligns with Metrology/special issue scope. The methods seem more likely to be acceptable/reliable, and the originality of the research is unquestionable. All figures are appropriate and legible. However, I would like to mention a couple of suggestions as follows
1. Please mention why these particular GCMs were chosen for the study area.
2. Please make a stand-alone discussion section.
3. Please write the limitations of the study.
Author Response
The Author thanks the Reviewer for the useful comments on the manuscript (MS). Please find the answers below and in the attached pdf file.
Please note that line numbering refers to the clean version of the revised MS where changes made relative to the original MS are invisible.
Point 1. Please mention why these particular GCMs were chosen for the study area.
Response 1. The seven GCMs are selected based on the validation results of Kristóf et al. 2021 and 2020 (denoted by [35,36] in the revised MS). Those GCMs were considered as the most reliable relative to reanalyses datasets. In the revised MS, the reason of the selection is emphasized at the beginning of the Discussion in lines 276-278.
Point 2. Please make a stand-alone discussion section.
Response 2. A stand-alone discussion section was added to the revised MS which can be found in Chapter 4.
Point 3. Please write the limitations of the study.
Response 3. In the revised MS, limitations of the study, i.e., the pattern detection algorithm – henceforth: algorithm – are extended and emphasized in the discussion section. In summary:
- The resolution of the percentile values used as thresholds in the algorithm is too coarse.
- The algorithm cannot be used to identify teleconnections with positively correlated action centers.
- Nonlinear effects concerning teleconnections are not taken into account in the study.
- Area under curve (AUC) values – which are used as a metric – vary in a small interval.
References:
Kristóf, E.; Hollós, R.; Barcza, Z.; Pongrácz, R.; Bartholy, J. Receiver Operating Characteristic Curve Analysis-Based Evaluation of GCMs Concerning Atmospheric Teleconnections. Atmosphere, 2021, 12. DOI: 10.3390/atmos12101236.
Kristóf, E.; Barcza, Z.; Hollós, R.; Bartholy, J.; Pongrácz, R. Evaluation of Historical CMIP5 GCM Simulation Results Based on Detected Atmospheric Teleconnections. Atmosphere, 2020, 11, 723. DOI: 10.3390/atmos11070723.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
General Comment
The revised version of the manuscript “Evaluation of Future Simulations of the CMIP5 GCMs Concerning Boreal Wintertime Atmospheric Teleconnection Patterns” has substantially improved. The authors made an effort to consider all my suggestions and answered my previous concerns. I don’t have more comments for the revised version.
