Next Article in Journal
Spatial Variability and Time Stability of Throughfall in a Moso Bamboo (Phyllostachys edulis) Forest in Jinyun Mountain, China
Previous Article in Journal
Some Observations of Waves at the Hong Kong International Airport and Their Numerical Simulations
Previous Article in Special Issue
Assessing Total and Tropospheric Ozone via IKFS-2 Infrared Measurements on Meteor-M No. 2
 
 
Article
Peer-Review Record

Estimation of High-Spatial-Resolution Near-Surface Ozone over Hubei Province

Atmosphere 2025, 16(7), 786; https://doi.org/10.3390/atmos16070786
by Pengfei Xu, Zhaoquan Xie, Yingyi Zhao, Yijia Wu and Yanbin Yuan *
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Atmosphere 2025, 16(7), 786; https://doi.org/10.3390/atmos16070786
Submission received: 13 April 2025 / Revised: 17 June 2025 / Accepted: 21 June 2025 / Published: 27 June 2025
(This article belongs to the Special Issue Ozone Evolution in the Past and Future (2nd Edition))

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors estimated surface O3 in Hubei province based on ground stations and satellite observations. They compared several machine learning models and found out that LightGBM is doing the best job. The paper is written in rather simple English and should be thoroughly checked and moved to more scientific language. In addition, in both the Introduction and Discussion (line 544 onwards) authors are omitting the drought effect of ozone. That should be incorporated into the discussion. In addition, discussion should be longer, as it is – it is not sufficient. If a part covering drought is added, it might properly prolong the discussion part.

 

line 48: acid rain likely not, it is SO2

line 53: Which ozone? I suggest defining abbreviation O3 as tropospheric ozone and then write only O3

line 68: concentration estimation model +- it would be bettter to reformulate

line 84,88, 109: year missing within the reference. Check throiughout the whole manuscript

line 178-179: the generation is regulated by UV radiation and presence of its precursors, not the other. Deposition is regulated by LAI, but not generation. You should write more clearly how do you mean that. Water vapour might affect O3 build-up by not allowing UV radiation to come through the clouds, however that would require some citation.

line 191: you mean O3 precursor emissions

line 195: that lead to an increase of NOx, which does not necessarily need to lead to the increase of O3

table 1. you should add here an abbreviation explanation

line 213: data are plural

line 216: is network density clearly linked to traffic?

figure 3, 6: "actual" value is not proper English. Should be "measured" value. Frequency has no units?

line 332: write more scientific English. "As can be seen from Figure" - is not proper advanced English

section 3.1.2: should go to the methods. Here you have Results section

line 373: you have already defined that abbreviation. Why do you define them again?

line 376-373: vertical diffusion? Nothing like this exists between the surface and PBL. It is vertical velocity, not diffusion

line 374: O3 does not accumulate

line 383: bad wording

figure 4,5: explain abbreviations within the figure caption

fugure 7: it is no value of ozone, but ozone concentration. Be specific

line 469: how do you know this? It is nowhere stated that you have VOC emissions from a database.

line 494: do not write that the figure is below. The editors might change the position of the figure

line 510: ozone is not just O

lines 560-575: this is not a discussion but more like a summary. You should have discussion here

line 575: where would the dispersion go? It suppresses rather its sinks

line 544: what is the effect of drought? It is not just the favourable conditions for O3 formation due to sunlight conditions, but at the same time restriction of its sinks - the vegetation, which during drought conditions and conditions of sharp light closes its stomata. Look at the paper 10.1016/j.envpol.2025.126081 dealing with drought and ozone. You might add that paper here.

 

 

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Overall Evaluation:
This paper presents a high-resolution ozone estimation for Hubei Province, China, using an ensemble machine learning approach that integrates ground-based observations, satellite remote sensing, meteorological data, and socio-economic parameters. By comparing six machine learning algorithms, the authors demonstrate that a stacked model combining XGBoost, LightGBM, and CatBoost achieves the highest performance. The study provides valuable insights into the spatiotemporal characteristics of ozone pollution and offers a replicable methodology for fine-scale air quality estimation. The topic is timely, and the approach is well-suited for Atmosphere. I have several major concerns regarding the presentation and novelty of this work, which must be addressed.

Major comments:

1) A major concern is that the spatial distribution maps of ground-level ozone concentrations estimated by the machine learning model exhibit very poor gradient clarity, making it extremely difficult to distinguish between high and low values. The colorbar and its corresponding scale are particularly misleading.

2) The captions of all figures and tables are incorrectly written. For example, “Figure 2. This figure illustrates the idea and process of the study” is not an appropriate or standard scientific caption.

3) The machine learning models employed by the authors have already been widely applied in mainland China. In terms of both the spatial scale of the data and the parameters used in the models, this study shows a high degree of overlap with existing research. It can be said that the study lacks innovation from a machine learning perspective. The authors should strengthen the discussion in the introduction regarding the insufficient research on ozone pollution, specifically in Hubei Province, and emphasize the advantages of their modeling approach over previous studies in the conclusion.

4) Based on the spatial distribution maps, the results of the machine learning models appear to be limited by the sparse availability of ground-based pollutant observations. Moreover, the authors have neglected to include key anthropogenic emission parameters—such as the intensity of anthropogenic NOx and VOCs—in their model training. The use of satellite-derived tropospheric column concentrations is known to be rather inaccurate, especially when used to estimate surface-level concentrations. Therefore, I question the reliability of the spatial distribution presented by the authors. A comprehensive comparison should be conducted with reanalysis datasets such as TAP developed by Tsinghua University and other Chinese atmospheric pollutant reanalysis products.

5) How did the authors prevent model overfitting?

6) Overall, the writing of this manuscript is quite poor, with many sections falling well below the standards of a scientific paper. A thorough and comprehensive revision is necessary.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I thank the authors for their changes and the updated Discussion part. It is much better now.

Author Response

We sincerely thank the reviewer for their positive feedback on our revisions. We are glad to know that the updated Discussion section is now considered much improved. We greatly appreciate the reviewer’s time and helpful comments, which have contributed to enhancing the quality of our manuscript.

Back to TopTop