Analysis and Evaluation of the Layered Precipitable Water Vapor Data from the FENGYUN-4A/AGRI over the Southeastern Tibetan Plateau

Round 1

Reviewer 1 Report (Previous Reviewer 1)

The review file is attached. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report (Previous Reviewer 2)

Review of “Evaluation and Analysis on the Precision of Layered Water Vapor Data from FY-4A/AGRI Satellite for the South-eastern Tibetan Plateau” by Song et al. The manuscript evaluates the FY-4A LPW product using the water vapor values from the radiosonde observations. The findings are valuable and I would like to recommend its acceptance for publication after necessary minor revisions.

 

1. In Section 3.3, before you show the bias, RMSE over 11 stations, the spatial distribution of FY4A/LPW should be presented (contour map).

2. L85, change -> 'changes'.

3. L188, why there is missing data with 0 value.

4. suggest to modify the sentence with 'And' as the initial, e.g., L308.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The comment file is attached. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Review of “Evaluation and Analysis on the Precision of Layered Water Vapor Data from FY-4A/AGRI Satellite for the South-eastern Tibetan Plateau” by Song et al. The paper validated the FY-4A LPW product using the water vapor values from the radiosonde observations of 11 radiosonde stations (RAOB TPW). The results indicated that FY-4A LPW had sound correlation with RAOB TPW in complex area on the South-eastern Tibetan Plateau, though underestimation exists as well.

 

Minor Comments,

L76-77, the resolution of VIS channels in FY-4A could reach 500m.

L85, With ‘the’ average elevation above 4000 meters.

L87, ‘too’ should be ‘to’.

L238-239, FIG 2b should be Fig 2c.

In Fig 2, check the data plotted in Fig 2c. It is same as the data points in Fig 2c.

 

Major comments,

1.       The spatial distribution of FY-4A LPW should be presented before comparison.

2.       The relatively larger bias in Summer is associated with the high LPW in this season. To normalized bias (bias / mean values) is more reasonable in evaluation of LPW.

The results barely describe the comparison of ground and satellite products. Authors should try to explain the cause of bias. For example, the overestimation at Yibin station and underestimation in other stations may be associated with the terrain. The role and influence of terrain in FY-4A retrieval algorithm should be illustrated

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

(1) Figure 2: Please elucidate the difference between Figure 2 (c) and Figure 2 (d). These two subfigures seem to be totally identical! However, they present different indices of RMSE, Bias, and R. This issue must be explained.

 

(2) The data in Figure 2 (a-d) represent the precipitable water vapor in the whole layer, high layer, middle layer, and low layer, respectively. While the text in Section 4.1 is not consistent with Figure 2, which is quite confusing. For example, “FIG 2 (b) sets forth the scatter diagram between the middle layer water vapor content inverted from FY-4A/AGRI LPW and RAOB TPW.” (Lines 238-240) and “The range of water vapor in the low layer contents is from 0 to 2.0cm.” (Lines 244-245) are contradict with the results in Figure 2. Meanwhile, Section 4.1 should be reorganized to contain more take-home messages. The current form is just a list of data, without any in-depth analysis.

 

(3) The process of the “three-dimensional matching between FY-4A/AGRI LPW and RAOB TPW in terms of time, space and vertical height” is not clearly clarified. First, “The time matching shall take the observation time of the RAOB data as the basis, and the time matching threshold value is set at ±60 min to search the most approximate FY-4A/AGRI LPW within the time interval.” (Lines 167-169): What is “the most approximate FY-4A/AGRI LPW”? Second, “Therefore, it is required to convert the RAOB data into Sigma layer in vertical height and conduct integral calculation to get the water vapor content in the corresponding height layer to guarantee the matching with the vertical space of FY-4A/AGRI LPW.” (Lines 177-180): The detailed information of the RAOB data should be added in the text. Third, “In consideration of the horizontal drift distance of air observation balloons and relatively low content of the atmospheric water vapor in the high layer, the threshold value of spatial matching is set at 20km. By taking the position of the RAOB station as the benchmark, search all the data grids of FY-4A/AGRI LPW in the radius of 20km, and calculate out the mean value of the FY-4A/AGRI LPW data grids searched.” (Lines 169-174): The threshold value was set to be 20 km. Is there any statistical basis for this conclusion? Additionally, is the mean value within the radius of 20 km really more accurate than the nearest neighbor? The aforementioned points are all required to be addressed in detail.

 

(4) The derivation of the Relief Degree of Land Surface (RDLS) is confusing. “The RDLS could be calculated out through date of the digital altitude model.” (Line 379): What is “the digital altitude model”? And how can the RDLS be calculated out through the “date” of the digital altitude model? Further, “Based on the established research achievement, this study took 30-meter data of DEM through through multiple experiments to select appropriate size of the window (39*39) and calculate out the RDLS of the research area” (Lines 382-384): Why the size of the window is set to be 39*39?

 

(5) Section 4.3.2: This section analyzes the correlation between the precision of FY-4A/AGRI LPW product with the RDLS, e.g., “Areas with relatively low RDLS like Wenjiang, Dazhou, Hongyuan and Ganzi had relatively small RMSE. Areas with relatively high RDLS like Jinchuan, Jiulong and Batang have relatively large RMSE.” (Lines 409-412), “The tend demonstrated that the RDLS have certain positive correlation with the Bias of FY-4A/AGRI LPW in different layers.” (Lines 405-406), “The tend demonstrated that RDLS have certain positive correlation with RMSE of FY-4A/AGRI LPW in different layers.” (Lines 423-424), etc. However, none of these conclusions are quantitatively corroborated. Can the “certain positive correlation” between the RDLS and the Bias in Figure 10 pass the confidence test? And the same is for the “certain positive correlation” between the RDLS and the RMSE in Figure 12.

 

(6) The English of this manuscript has to be heavily revised, where exists a massive number of mistakes, e.g., “root-mean-square of the stations” (Line 31), “generates significant influence too the regional and global atmospheric circulation” (Line 87), “provide basis for improve its inversion accuracy” (Line 103), “the clear shy atmospheric moisture profile” (Lines 124-125), and so on. In addition, the errors like “TERRA and AUQA” (Line 67), “Fengyun-4 Satellite (FY-A)” (Lines 71-72), “perceptible water” (Line 128), “temperature: 0.2°~0.3°” (Line 149) also must be corrected.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

(1) Abstract: What is the “root mean square of the stations”? This point has already been reminded in the comments of the previous round (Comment (6)). But the authors did not make a check at all.

 (2) Section 4.1: “The R of high layer was 0.90, which was the biggest among the three vertical layers.” (Lines 229-230); “The R of middle layer was 0.93, which was bigger than the R of high layer.” (Line 232): Both are the results of the Section 4.1. Which one should I believe in?

 (3) Figure 2: “the values of high layer were distributed from 0 cm to 2.5 cm, the values of middle layer were distributed from 0 cm to 3.2 cm, and the values of low layer were distributed from 0 cm to 2 cm.”: Why is the middle layer the highest?

 (4) Figure 3: The histograms of the Bias. (a) Whole layer; (b) high layer; (c) middle layer; (d) low layer. The data points of these four subfigures (5719, 5716, 5702, 5701) are not the same. Why? How can you retrieve 5719 samples for the whole layer based on the three layers with not enough quantity?

 (5) Introduction: “the resolution in the visible spectrum (VIS) is 0.5 km” (Line 75): This statement is inaccurate.

 (6) Section 4.3.2: “After multiple experiments, the size of the RDLS model were set to 39*30.”: In the revised manuscript, the size was set to 39*30. However, in the cover letter, the size was set to 39*39. Which one should I believe in?

 (7) Section 4.3.2: “The quantitative analysis indicated the correlation between Bias and RDLS of the whole layer, middle layer, and lower layer could pass the 0.05 confidence test.” From Figure 10, it is completely difficult to believe that the significance test can be passed.

 (8) Conclusion: This section is patchy and fragmented. It needs to be completely rewritten or reorganized.

 (9) The authors have not responded to the comments of the previous round carefully. Thus, in addition to the comments of the current round, the comments of the previous round which have not been well addressed are still required to be responded to point by point.