Construction of High Spatiotemporal Continuity Surface Water Bodies Dataset in the Haihe River Basin

Round 1

Reviewer 1 Report

Review of Construction of High Spatiotemporal Continuity Surface water  bodies Dataset in the Haihe River Basin

The paper is a very useful advance in technology and is well written and well documented.

The following five issues should be dealt with to make the paper acceptable for publication

(1)    In lines 416-420 the discussion describes what period was used for the analysis of the three methods (HRWD, JRC, GSWED).   There is confusion between the choice of period for analysis and the choice of what size of water bodies was included.  It appears that the period selected was chosen because GSWED data are not available after 2020.   The choice of what size of water body was included in the analysis is stated but no reason is given for this choice.   The selection should be rewritten to explain the reason for the size selection.

(2)    In Figure 8 results from HRWD are generally consistently higher than GSWED .  The only exceptions are in 2016 when the results for HRWD for January and June July and August are low and very similar to GSWED.   An explanation for these low values for HRWD in 2016 should be given.

(3)    In line 527 the explanation for JRC showing a higher water occurrence than HRWD is said to be due to the under detection phenomena in JRC    this is opposite to what is being explained and should be removed or modified to make sense.

(4)    Line 561  does not appear to make sense   what is an annual water occurrence of 25 ???

(5)    Lines 570-575 should be moved to section 8 conclusions  (or removed if already dealt with in Conclusions)

Minor editing corrections:

In the paper title the only words capitalized should be Construction and Haihe River Basin

Line 199               with a six-day  interval.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The study titled "Construction of a High Spatiotemporal Continuity Surface Water Bodies Dataset in the Haihe River Basin" was peer-reviewed. The manuscript focuses on a category reconstruction method that builds a spatiotemporal correlation model by integrating neighboring pixel labels and those at the same position in images acquired on other dates according to historical spatiotemporal water-distribution data driven by Sentinel-1 Synthetic Aperture Radar (SAR) data and surface water body evolution rules to reconstruct the class label (water or land) of Sentinel-2 cloudy pixels, achieving high spatiotemporal correlation. The methodologies appear more likely to be acceptable/reliable, and the research's originality is undeniable. The artwork is suitable and readable. However, I'd want to raise some minor issues about the methodologies, data quality, and overall structure of the work.

1. The abstract might be improved by mentioning the study's strong findings. In its existing form, the authors instantly decided that their built model outperformed the satellite products hands down. It is also worth noting the causes for JRC and JWCD's low performance. Some lines (18-23) are rather verbose and should be condensed. 

2. Why did the authors focus their efforts on building a combined product (S1 and S2) model using a large physical dataset rather than enhancing cloud removal techniques for JRC (especially during the rainy season)? The created dataset is more efficient, but its validated spatial-temporal coverage is significantly less and limited to a tiny Chinese region. What is your response to this question?

3. Why did JRC see a significant decline in the water body area for Miyun in September, whilst GSWD saw a drop in early May? Please clarify the ambiguity. The caption for Fig. 7 is reversed; please correct it. Please include the legend/unit of the color bar in Fig. 2.

4. The discussion section appears to be more of a results section, although it makes little sense. Please establish a separate discussion section based on your take-home message for the readers in comparison to previously observed outcomes. Please remove all discussion headings and relocate all results to the results section.

Please try to reduce the verbosity by avoiding unnecessary details in each section of the manuscript. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

1.      Line 317: “This mainly includes four steps. First, the entire study area is divided into 100 small blocks.” Are these blocks equal area? Is this each block based on watershed scale?

2.      Line 329: Please show the results of User Accuracy (UA) and Producer Accuracy (PA). In addition, the overall accuracy of HRWD is 0.93 which is high accuracy. Is there spatial difference of accuracy in this study? How to evaluate it?

3.      Line 369: In the statistical results, these differences are shown as omissions (yellow) or commissions (red) in Figure 6. However, I didn’t see the yellow legend. Maybe it is blue for omission error in Fig. 6, legend of c. Please check again.

4.      Line 399: In Figure 7 (b). Validation of water bodies in the Haihe River Basin in 2018 (b) Guanting. It is interesting that the water body surface area was lowest for JRC in Sept.. Why? What is happen in Sept. 2018?

5.      Line 401: In Table 4, the R2 values of HRWD for Guanting and Miyun are 0.87 and 0.89, respectively. These results are better than JRC and GSWED. However, if we consider the processing time and budget, is it suitable using HRWD comparing to JRC and GSWED?

6.      Line 544: Cloud cover during the rainy and phenological seasons causes JRC products to be unable to effectively identify water bodies with low inundation frequencies. Are there some suggestions to improve this limitation?

Author Response

Please see the attachment.

Author Response File: Author Response.docx