A High Resolution Spatially Consistent Global Dataset for CO2 Monitoring
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsI recommend that the authors include a brief description of the dataset structure and, if possible, a sample of its contents when the link to the dataset is provided to make the paper more self-contained.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for Authors The article presents a super - resolution method for downscaling atmospheric COâ‚‚ concentration, enhancing the spatial resolution to 0.03 °× 0.04 °. However, as noted by the author, there is an abundance of domestic and international research on high - resolution reconstruction of COâ‚‚ concentration data, along with numerous available datasets. Moreover, there are legitimate concerns regarding the rationality of the method employed in the article. Training the COâ‚‚ dataset using ground temperature based solely on data distribution similarity is insufficient. Overall, while the article holds some research value, it lacks innovation and its methods are not entirely reasonable, thus requiring further modification and improvement.​ In addition, the article has the following specific issues:​- The coverage of the research status at home and abroad is insufficient. For instance, the research on COâ‚‚ monitoring satellites in Table 1 is incomplete, omitting China's GF and DQ - 1 satellites. The timeliness of some cited references, such as those in Table 2, needs enhancement. Given the rapid progress in this field, it is advisable to cite the latest research findings to better reflect current research trends and cutting - edge technologies.​
- Although the use of LST datasets for training models is theoretically mentioned, relying solely on data distribution similarity is not tenable. Furthermore, there is no explanation of how this similarity is defined. It is essential to conduct in - depth research on the intrinsic physical relationship between temperature and carbon dioxide concentration. This could be achieved through more comprehensive literature reviews, theoretical analyses, or experimental verifications. Such efforts would explore the reasons for choosing the LST dataset to train the model and strengthen the theoretical foundation of the research, thereby enhancing the rationality of using LST data to predict carbon dioxide concentration.​
- During the research process, the uncertainties in data and model results were not fully evaluated. Due to the inherent errors in carbon dioxide monitoring data and the introduction of uncertainties during model training and prediction, appropriate methods should be employed to quantitatively analyze these uncertainties.​
- Some research methods and processes are not clearly described. For example, it is not explained how the normalized XCO₂ data in Table 1 was obtained, nor is there any mention of the region or time period for the CO₂ data used.​
- In the results section, in addition to comparing with existing datasets, a comparison with the original OCO2 values should also be made. Both Table 4 and Figure 5 should include a comparative analysis with the original OCO - 2 data.​
- In the discussion section, a comprehensive comparison of the research results with other achievements in related fields is lacking. For example, by referring to other studies on the spatiotemporal distribution characteristics and concentration changes of carbon dioxide, the similarities and differences between the results of this study and existing research should be analyzed, and the reasons for these differences should be explored.​
- The labeling of some charts is not clear and detailed enough. In some charts, the units and meanings of the coordinate axes are not clearly labeled, as seen in Table 1. The red and blue annotations in Table 4 are inaccurate.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsRecommendation: Reject
Evaluation:
This paper introduces a downscaling method to enhance the spatial resolution of the OCO-2 dataset. The method mentioned in the paper is novel, using LST as training samples to learn the mapping relationship between high and low-resolution LSTs, and applying this mapping relationship to the OCO-2 dataset, thereby increasing its resolution by 16 times. However, I have doubts about the rationality of data selection and experimental design in the article, and there are also some issues in the narrative logic, specifically as follows:
Major comments:
- Based on the understanding of current research, there is a good spatial correlation between CO2 distribution and nighttime light. Here, how was it determined to use land surface temperature instead of other data, such as nighttime light, for model training?
- From the data normalization results in Figure 1, we can only see the similarity or dissimilarity of the four types of data in terms of numerical distribution. Will the difference in spatiotemporal distribution of these data affect model training?
- The original high-resolution LST images are resampled using bicubic interpolation to generate a low-resolution input dataset. The mapping relationship obtained by the final model is from low-resolution LST to high-resolution LST, and this mapping relationship is essentially brought about by the resampling effect of bicubic interpolation. Therefore, the most accurate mapping function that can be obtained from the model training is bicubic interpolation itself. Moreover, from the results and discussion sections, the method in this study is similar to the results of bicubic interpolation. So, what are the advantages of the model in this study?
- In lines 90-94 of the article, it is described that among the existing global XCOâ‚‚ datasets listed in the table, only the dataset from Wang et al. has a better resolution than NASA's dataset. However, in lines 88 and 89, the article only mentions that "XCOâ‚‚ retrievals from OCO-2 are integrated into a daily gapless gridded dataset using NASA’s modeling and data assimilation system," without providing detailed information on the resolution of the improved NASA dataset.
- In line 116 of the article, the conclusion "We can see that the LST distribution is the one that matches XCOâ‚‚ better" is drawn solely based on the numerical distributions of the data in Figure 1. However, LST and XCOâ‚‚ are, after all, two different variables, and whether their spatial distributions are also similar is a key factor affecting model performance. It is recommended to cite relevant literature or include experiments demonstrating the similarity in the spatial distributions of LST and XCOâ‚‚ to make the article more convincing.
Minor comments:
- In line 115 of the article, "DIV2K" and "DOTA" datasets are mentioned. It would be appropriate to provide a brief introduction to each of these datasets here.
- Figure 1 in the article lists the current satellites used for CO2 observation. The list could also include additional information such as the observation time, spatial resolution, and spatial coverage of each satellite to further emphasize the necessity of studying global high-resolution CO2.
- The third figure in Figure 6b is misnumbered.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsThis manuscript proposes a method to increase the resolution of a dataset that aims to improve the process of daily monitoring of atmospheric CO2 concentrations on a global scale with a greater degree of detail. To this end, an over-resolution model is created and reported to downscale the atmospheric CO2 data.
By design, the proposed method has great scientific potential and could be useful in the analysis of a particular type of data. Unfortunately for me, the authors have so far failed to present a sufficiently complete manuscript that would be interesting and highly appreciated by the scientific community.
The state-of-the-art review of the scientific problem under study is insufficiently complete and comprehensive and is not structured at a sufficiently good scientific level.
I have provided some, but not all, of my comments and recommendations below. I hope they will be of use to the authors.
Line 14: “… can achieve this by producing and sharing better data.” – I would recommend that this sentence be edited. We don't "produce" the data - it is what we get from our sensors and instrumentation. We can improve the tools we use to get that data. Furthermore, there is no such thing as "better data" - for some cases "better data" might be better spatial resolution, or better temporal resolution, or data with more parameters, e.g. -metadata, etc.
Line 14: “… a high resolution XCO2 dataset…” – High spatial or high temporal resolution? Please be more precise in the terminology used.
Line 15: „atmospheric CO2 pollution ... „ – This statement is categorically incorrect! CO2 is NOT an atmospheric pollutant per se. The problem with its role in global warming is the significant increase in its concentration over a relatively short period of time. Other components of the atmosphere are considered to be pollutants.
Line 16-19: What is this "super-resolution" essentially? OCO-2-derived dataset? There is no citation here to this dataset. I recommend that the authors rephrase this sentence for clarity and consistency of presentation.
Line 21-23: The phrase "area of global seamless fields" is not often used in academic or technical communities. Please be more precise in the terminology used.
Line 24: I think that more appropriate keywords could be used that are more relevant to the subject of this material.
Overall, I believe that it is advisable to revise the Abstract to more accurately and clearly reflect the content of the present material, as well as to correct some of the terms used.
Line 31: “anthropogenic warming”? There is no such term in scientific communities. Please correct.
Line 35-37: There is probably a lack of understanding by the authors about the type of data derived from ground-based measurements and remote sensing methods. Ground based sensors do NOT provide XCO2 data! Furthermore, the purpose of these data is not to provide information "on a regional or sub-regional scale". According to the authors - these data are "scarce". In what sense? "Scarce" spatially or temporally? Understanding the nature and characteristics of the types of data available to modern science is fundamental to solving a particular scientific problem.
Line 41: “… produce commonly used …” – The word produce is not the most appropriate in the context of the sentence. Please try to express yourself more precisely.
Line 43: The missions don't move, the satellites do. Please read carefully what you write and try to correct it. Also, missions do NOT map!
Line 44-45: Please revise this sentence. It is inaccurate and ambiguous ... „alleviate ... lack of completeness from satellite data...“ ... „seamless maps of ..” ? What kind of maps are these?
Line 45-66 The information in this section of the text should provide a snapshot of the state of the art of methods for mapping CO2 concentration levels and dynamics at regional and global scales, with corresponding advantages and disadvantages. Based on this review, the authors should justify what is new and why? What would be the advantages and differences compared to currently existing methods? This section of the text does not do this in a sufficiently clear and justified way. This text needs substantial revision..
Section 2. Datasets must be thoroughly revised. Here, all data used by the authors should be described clearly and in detail, well structured and presented with their specific characteristics - their spatial and temporal resolution, temporal range, etc., but not only. All these data have their own peculiarities and it is important for the reader to know exactly how they have been applied.
Line 105 “… structured as tiles of land surface temperature along swath of the satellite.” – Please edit this sentence. Such a sentence should not be found in a journal of the stature of Remote Sensing. Another important thing - the exact characteristics of the data used by the authors are missing.
Line 110: “high resolution CO2 seamless data…”? This is an incorrect expression. Please improve and clarify the terminology used.
Line 111-113: “low resolution to high resolution ...” – What kind of resolution are you talking about? Temporal, spatial? Furthermore, there is no justification for why the data needs to be downscaled. What is the purpose and what would be the benefit of this reduction?
The methodology applied by the authors to create their so-called "super resolution model" is described in sufficient detail and comprehensively. The sequence and schematic illustration of the workflow are correct and have a clear logical coherence.
Line 164 Nowhere in the manuscript, including here, is it clear what the exact dimensions of these "low-resolution global maps" are, and exactly what parameters are being referred to. If data from different sources were used, what are their characteristics, were these data combined with data from other sources, and if so, how was this done. A clear and precise definition of the parameters and characteristics of the input data is essential, as these are the main criteria for assessing the output data.
Line 173 What are these atmospheric components? Information about their characteristics is missing.
Line 178 What are these global daily datasets? By whom are they created? There is a lack of references, citation of sources, clear and precise definition of these data - temporal and spatial resolution, satellite and/or ground data, what their specific parameters are, etc.
Line 197 You claim that your data is from ground sources (Line 191), and you compare your data to the XCO2 variations. How exactly do you make this comparison? You need an excellent knowledge of exactly what is behind the XCO2 designation and exactly how this value is derived in practice.
Line 204 What are these over all metrics?
Line 207 How your data is validated with TCCON?
It is not permissible to insert figures in the middle of sentences. This has been done in several places in this manuscript.
Line 216-217 High or low concentrations of what? Above you claim your data is daily. On this figure which day do the pictures refer to?
The Discussions section should be in a separate section. The place of Section 5 should not be at the end of the article. Moreover, it needs a complete revision.
Please familiarize yourself with the requirements for the required structure:
https://www.mdpi.com/journal/remotesensing/instructions
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe author has carefully revised the article in response to the reviewers' comments, and the quality of the article has been greatly improved.
Reviewer 3 Report
Comments and Suggestions for AuthorsThank you for the author's detailed response to the issues I raised. However, the rationale of applying downscaling models constructed through LST to CO2 is worth discussing, and the model method constructed in the article has not significantly improved compared to existing methods. Therefore, I believe that this paper cannot be accepted.
Reviewer 4 Report
Comments and Suggestions for AuthorsI accept the changes made by the authors on the basis of the recommendations. However, I have minor comments that could be taken into account at the discretion of the editors.
/Line 31 in v.1/ Thanks to the authors for their response on the term “anthropogenic warming”. Adding a precedent/exception to the subhead of a statement does not automatically make the statement true. After all, we are talking about a scientific term here, not legalese.
I address this point for the sole purpose of making the statement and the terminology used as precise and correct as possible. I believe that the term ‘anthropogenic warming’ is not particularly appropriate for use in an article of this scientific level. However, let the editors also give their opinion on the matter.
I also note that in several places in the text there are figures/tables inserted in the middle of sentences. In my opinion this would be better avoided. Of course, the editors could make their recommendation on this.
I wish the authors success in their research work.