Next Article in Journal
Simulating the Fate of Dimethyl Sulfide (DMS) in the Atmosphere: A Review of Emission and Chemical Parameterizations
Next Article in Special Issue
Perceptions of Climate Change and Health Risks Among Urban Older Adults in Mexico City: A Pilot Study
Previous Article in Journal
Some Recent Key Aspects of the DC Global Electric Circuit
Previous Article in Special Issue
Optimizing Microclimate Comfort in Macao: ENVI-Met Simulation of High-Density Urban Layouts Under the Climate in the Lingnan Area
 
 
Article
Peer-Review Record

A Comparative Study on the Methods of Predictor Extraction from Global Sea Surface Temperature Fields for Statistical Climate Forecast System

Atmosphere 2025, 16(3), 349; https://doi.org/10.3390/atmos16030349
by Yawei Cai and Xiangjun Shi *
Reviewer 1: Anonymous
Reviewer 2:
Atmosphere 2025, 16(3), 349; https://doi.org/10.3390/atmos16030349
Submission received: 15 January 2025 / Revised: 20 February 2025 / Accepted: 17 March 2025 / Published: 20 March 2025
(This article belongs to the Special Issue Extreme Climate Events: Causes, Risk and Adaptation)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper compared different predictor extracting methods under two climate forecast frameworks. But since it lacks scientific significance, I recommend rejecting this paper.

 

 

1.     Section 2.1

a.     Within the study period (1961–2022), the June Western Pacific Subtropical High Ridge Line (WPSHRL) was obtained from the National Climate Center of China for 1961 to 2016, and afterwards it was calculated based on ERA5. 

(1)  Are there differences in the latitudinal resolutions between the data from the National Climate Center and the EAR5?

(2)  Why not use the ERA5 in the whole period (1961–2022) for consistency?

The reason I suggest this is that in Sun et al. (2022), they also used the WPSHRL from the National Climate Center for 1961–2016, and reconstructed the WPSHRL from ERA5 for 2016–2022. But their WPSHRL (the black line in their Figure 4) and your WPSHRL (Figure 1b) is different for the period of 2016–2022.

2.     Section 2.2

a.     Line 105–110 describes the ZONE approach, but it misses one important information: what correlation criteria did the authors use to define “highly correlated”?

3.     Section 3

a.     In the ZONE and EOFg/EOFp extracting methods, the 62-year dataset (1961–2022) is used to select 20 representative zones and calculate 20 EOF modes, respectively. Then, these zones (zone-averaged SST) and modes (corresponding principle components) are used in the rolling forecast experiments.

Take the forward rolling experiment in 1992 for example. Although the forecast model parameters are updated only by data between 1961 and 1991, since the zones and modes are calculated based on data between 1961 and 2022, the forecast model in 1992 is not truly independent. Especially in the neural network framework, the authors selected the first six SST variables that have the highest correlation with WPSHRL as input.

4.     Section 4

a.     Line 279–285, the authors described the results for 2010, but the description is wrong. The two ZONE predictors (zone averaged SST) are positive, not negative.

5.     Section 5

a.     The discussion is pure speculation. The authors stated that: using meteorological elements as predictors seems to be contradictory to previous theoretical studies, but in fact not. The argument for this statement is without any solid proof from this work, since this work only focuses on using SSTs as predictors. Better to rephrase it, and put it in the conclusion section as possible future work.

6.     Section 6

     a.     Line 337–338 “The goal of this study is to gain experience in extracting SST predictors for the statistical climate forecast system”. As a scientific paper, the goal should be to have new findings, not gaining experience.

 

Grammers:

1.     Line 102 “In this initial stage of the research on extracting multiple SST predictors, we explore two simple and feasible approaches”. This sentence gives the impression that there is a second stage of the research, but in fact, these two methods are the only two used in this study. Better to rephrase it to avoid confusion.

2.     Line 117 “Table 111”. A typo?

3.     Line 249 “This paragraph introduces the comparison between the NN framework and the LR framework (Figure 5)”. It needs to be rephrased as something like “The forecast performance of four predictor extracting methods under two forecast frameworks are compared in Figure 5.” 

4.     Line 338–339 “the comparison between different SST predictor extracting methods is investigated”. The methods can be investigated, but no the comparison.

 

 

SUN, C., SHI, X., YAN, H., JIANG, Q. & ZENG, Y. 2022. Forecasting the June Ridge Line of the Western Pacific Subtropical High with a Machine Learning Method. Atmosphere [Online], 13. Available: https://mdpi-res.com/d_attachment/atmosphere/atmosphere-13-00660/article_deploy/atmosphere-13-00660.pdf?version=1650541239

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

I put all my recommendation inside the report

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This is an interesting study that presents a holostic approach to the correlations between sea surface temperature (SST) and climate forecasts in the context of climate change.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Back to TopTop