Next Article in Journal
Machine Learning Model Optimization for Antarctic Blowing Snow Height and Optical Depth Diagnosis
Next Article in Special Issue
Unveiling PM2.5 Transport Pathways: A Trajectory-Channel Model Framework for Spatiotemporally Quantitative Source Apportionment
Previous Article in Journal
Pressure and Temperature Observations in Venice by Bernardino Zendrini from 1738 to 1743
Previous Article in Special Issue
Optimizing Vertical Unidirectional Airflow in Cleanrooms: An Integrated Approach to Floor Perforation, Plenum, and Fan Filter Unit Configurations
 
 
Article
Peer-Review Record

Construction of a High-Resolution Temperature Dataset at 40–110 KM over China Utilizing TIMED/SABER and FY-4A Satellite Data

Atmosphere 2025, 16(7), 758; https://doi.org/10.3390/atmos16070758
by Qian Ye 1,2, Mohan Liu 1,2,*, Dan Du 1,2 and Xiaoxin Zhang 1,2
Reviewer 1: Anonymous
Reviewer 2:
Atmosphere 2025, 16(7), 758; https://doi.org/10.3390/atmos16070758
Submission received: 30 April 2025 / Revised: 9 June 2025 / Accepted: 10 June 2025 / Published: 20 June 2025
(This article belongs to the Special Issue Feature Papers in Atmospheric Techniques, Instruments, and Modeling)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

A brief summary

The article “Construction of a High-resolution Temperature Dataset at 40-110 km Over China Utilizing TIMED/SABER and FY-4A Satellite Data” explores a new method developed to construct a high-resolution (with horizontal resolution of 0.5°×0.5° and vertical resolution of 1 km) temperature dataset over China in the middle atmosphere based on the XGBoost technique. The authors used measurements from SABER onboard the Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics (TIMED) and Fengyun 4A (FY-4A) satellite to construct accurate temperature profiles. The study resulted in a model that captures the characteristics of the vertical and seasonal variations of temperature. These findings contribute to the improvement of vertical climate models covering the mesosphere and lower thermosphere.

General comments

The manuscript presents a valuable study that attempts to provide a high-resolution temperature dataset in the MLT region over China from 2019 to 2023 based on observations of SABER/TIMED and Fengyun 4A (FY-4A) satellite and ERA5 reanalysis data. The study’s findings address the existing gaps in temperature data and provide insights for further research to improve the accuracy of the middle and upper atmosphere models. 

The article is logically structured, with a sufficient number of graphs representing the results. However, the manuscript would benefit from the detailed description of the model parametres, adding limitations of the model to the study, and comparison with other studies in this field.

Specific comments

Abstract

Line 20. I suggest adding the information and specifying the altitudes at which the model provides the most accurate results. 

Introduction

Line 31. The reference 1 is dedicated to the fires. It is doubtful that it is relevant to the introduction section.

2. Data and method

Lines 67–72. An extraneous paragraph has appeared. Please check.

Line 95-102. It is not quite clear how the authors implemented the Fengyun-4A (FY-4A) satellite data into the model. It would be helpful to clarify this here and somewhere in the results

  1. Temporal spatial coverage

Line 145-149. Figure 2. It is not clear at what altitude the plane is given or whether it is the total number at all altitude levels. Please clarify.

Line 151 and Line 154. Figure 3 and Figure 4. It is also not clear at what longitude the plane is given, as well as for Figure 4 at what latitude. Please specify.

  1. Results

Lines 168–169. Figure 5. Please indicate which plot is training and which is testing.

Lines 177. Figure 6. It seems like the plots show different planes but not hours. Please check.

  1. Summary and Discussion

The section would benefit from discussing the limitations of the current models and the data applied.

The comparison with related studies in the field would add significance for the reader.

Line 255-263. The findings would sound stronger if supported with the statistical data.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

In this study,  authors have constructed a high-resolution temperature dataset within  the mesosphere and lower thermosphere over China, utilizing data from SABER/TIMED, FY-4A satellite and ERA5 reanalysis. Authors have created a model based on the XGBOOST algorithm to generate reliable temperature profiles across altitudes of 40 to 110 km.

- Of course, the altitude range, that interests the authors, actually causes certain difficulties, since there is insufficient measurement data obtained for these altitudes. At the same time, a deeper understanding of the physics of atmospheric processes in this altitude range is an important and relevant scientific task. The authors consider vertical temperature profiles starting from an altitude of 40 km or even from the underlying surface (Figure 8), and pay attention to the relationship between atmospheric layers. When reading the manuscript, a recommendation arises to expand the understanding of how different atmospheric layers are related, in the opinion of the authors. First of all, how, in the opinion of the authors, the troposphere and lower stratosphere affect the overlying atmospheric layers. This is a key aspect of this work since the authors use information on atmospheric characteristics corresponding to different altitudes, altitude ranges. Please add some information in introduction.

- To create a model generating reliable temperature profiles for altitudes from 40 to 110 km, the authors used the XGBOOST algorithm.However, it is unclear how the reliability and representativeness of these profiles were determined.How did you combat overfitting of the XGBOOST algorithm?This is also a key question.

- To evaluate the performance of the XGBOOST method, the authors used the following statistical characteristics: R, RMSE, MAE and MRE.I recommend that the authors provide the formulas in the text for clarity.

- In this study, the authors used a dataset comprising a total of 6,608,473 samples to evaluate the performance of the model. The dataset was divided into training and testing subsets with an 80%/20% ratio. This is a standard approach. At the same time, very little is said about the input variables. What variables were used for training and what time averaging was used (1 hour average data?). Based on what physical concepts were the original variables selected? Also provide the characteristic altitude levels that you used. Perhaps it would be most desirable to present the input variables in the table.

 - Lines from to 26 to 35. This fragment of text contains the word «Thus», which is repeated twice.

It is possible to slightly change the text and remove one word «Thus» for a better text structure.

-Line 53. "relatively high special and temporal resolutions".

Please replace the word special with spatial. - Perhaps it is worth expanding the introduction. Tell us more about the source datasets used. In particular, the ERA-5 reanalysis is a widely known and used dataset. What errors does it have and how do they affect the data in the overlying layers? The ERA-5 data provide us with information about meteorological fields that reflect well the physics of atmospheric processes (Please see: https://doi.org/10.3390/atmos15010038, https://doi.org/10.3390/rs14246221). The work requires a review of the latest literature

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Overall, the manuscript is quite clear and shows interesting results. I encourage the authors to review the text carefully and add units for MAE and RMSE. After this, the manuscript can be published.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Back to TopTop