Manual-Based Improvement Method for the ASTER Global Water Body Data Base

Round 1

Reviewer 1 Report

This manuscript seems to be providing a method to improve the original ASTER layer of water bodies. I think the resulting product is of interest to the research community, however the manuscript needs substantial work to make it suitable for publication. There are many grammatical errors throughout the manuscript which needs to be corrected (I started to label a few in the abstract, until I realised it needs to be thoroughly edited throughout). The manuscript is very brief and needs more information about the method used (see comment below), as well as how accurate this method is, and how to access the resulting product. If the method is based purely on manual threshold selection or on manually outlining every missed water body, with limited new research material, it may be more suitable as a research letter with the focus more about the final product quality and how it improves on the old water body layer.

Some grammatical errors

• Abstract – change ‘by-products’ to ‘by-product’
• Abstract - change ‘to public,’ to ‘to the public.’
• Abstract – missing word in ‘…separation of from land area’
• Plus there are lots more throughout the manuscript

Title – I don’t think ‘Ultimate’ is a suitable term. I think something like ‘A new method to improve the ASTER Global Water Body Data Base’ works better.

Page 3, Section 2.2, Step (2) – How is the ROI tool used to extract water bodies? Is it an automatic or manual process? Is there a single band threshold for all images, or a different threshold for each image? Or is it based on manually outlining each water body using the computer mouse? Can you please clarify this as not everyone uses ENVI.

Figure 4 - can you please make the red rectangular line thicker, as it is difficult to see.

Figure 5, Caption – can you also say what the red areas represent?

Table 2 might be more appropriate as an appendix, and I suggest adding a graph showing increased occupancy ratio with respect to latitude in the discussion.

Discussion – what is the accuracy of this method? How does it work across image tiles? If it is all done manually, how much time is involved in doing this? How does the accuracy change if a different person manually selects the water bodies? I think the discussion should also talk about how/why there are more lakes in the high-latitude areas using the new method.

Conclusion – is the new data available to the public?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors propose a study for enhanced detection of water bodies in satellite images. The results are promising; however, additional explanation and discussion are requested. I suggest to address the following comments in the main text:

- References provided in the introduction are mostly self-citations. Please improve this section presenting researches carried out by other authors in this framework.

- Please provide a description concerning figures 1a,b,c,d in the main text, highlighting the most important aspects. This comment applies to Figures 4a,b as well.

- Please provide scales for the images shown in Figures 1 and 4.

- What was the criteria used to select a given feature as a waterbody in the images? There are the attribute files in ASTER GWBD folder mentioned in Section I, but it is not clear if they were used somehow in the methodology for manual selection.

- Figures 2, 3 and 5 present relatively small areas compared to that shown in Figure 6. It is not clear how the proposed procedure was extended to such a wider area, if it must be done manually as stated in the main text.

- In Table 1, why is the lake occupancy reduced by less than a half in the improved image of N60W76 compared to the original image, if the improved image in Figure 4a shows more green areas than the original image?

- What was the process applied to the shaded-relief images shown in Figure 5 in order to provide such improvement? A description and a detailed explanation about the improved features which would be highlighted in benefit of the proposed approach are strongly suggested.

I also suggest the following corrections:

- In section I, first paragraph (page 1) please correct "ephemeris" to "ephemerids".

- In Figure 2b please correct the title from "Undeted waterbodies" to "Undetected waterbodies".

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper “Ultimate Improvement Method for ASTER Global Water Body Data Base” by Hiroyuki Fujisada et al provides a review about detection of waterbodies based on the digital elevation model (DEM) generation by using the data of Advanced Space-borne Thermal Emission and Reflection radiometer (ASTER) onboard Terra spacecraft. Unfortunately, using only reflectance and spectral index data that are provided by ASTER instrument it is not possible to detect perfectly all inland waterbodies. Method proposed by authors for improving the detection of waterbodies is using additional earth’s surface images: GeoCover2000 images from Landsat or Color-Land ASTER MosaicS (CLAMS) images if deficiency of GeoCover2000 images has to be covered. The paper describes how the deficient waterbody images are copied to the global waterbody dataset to make it more complete. However, the quality of figures should be improved making them more compact: the texts between images shold be removed as much as possible. I recommend accepting the paper by Hiroyuki Fujisada et al for publication in Remote Sensing after revision by authors. Editing of English language and style is also required.

Suggestions for authors.

1. Figure 1. No reference or explanation is given to (a) to (d) in text or in caption below the Figure.
2. Page 2, 11th row: … coordinates with 1 arcsecond or 1 as or 1”.
3. Fig 1 to 3. Description/explanation of the waterbody images should be given in the end of paragraph 2.1. No need to repeat them in captions.
4. Figure 2b. Undetected waterbodies…
5. Figure 4. Marking of the expanded sub-area needs to be improved.
6. Table 1. In the N60W76 tile, lake occupancy of the original image, likely, should be 6.6 %, and of the improved image 18.4 %, respectively. Number of significant places presented in table 1 seems to be too large and needs to be justified if not reduced.
7. Table 2. Number of significant places presented seems to be too large.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Most comments were replied satisfactorily, but I still have some suggestions:

Please, make it clear in the main text that shaded-relief images were achieved using a tool from ENVI, as done in lines 66-70.

Figure 6c, presented in this version of the manuscript, make clear that relevant improvements are restricted to the northern area of North Hemisphere due to the location of the analysed tiles. Therefore, there is no need to present the improvement for a worldwide map. Only the analysed areas could be displayed, thus the increase in occupancy ratio can be easily compared to the colorbar provided in this figure.

Additional in-text descriptions of figures 1a,b,c,d and figures 4a,b were requested in the previous review for highlighting the most important aspects seen by authors in each image. However, such additions were not provided in the new version of the manuscript. Why were such tiles chosen as significant examples of the proposed approach? It would be significant to mention at least the average size of waterbodies in each tile, thus the reader could have an additional parameter for comparison of the resolution capabilities.

Author Response

Reviewer 2 additional comment and response.

Please, make it clear in the main text that shaded-relief images were achieved using a tool from ENVI, as done in lines 66-70.

       Response: The sentence in lines 66-70 is not related to the creation of shaded-relief images but describe how to accomplish the waterbody improvements.

Figure 6c, presented in this version of the manuscript, make clear that relevant improvements are restricted to the northern area of North Hemisphere due to the location of the analysed tiles. Therefore, there is no need to present the improvement for a worldwide map. Only the analysed areas could be displayed, thus the increase in occupancy ratio can be easily compared to the colorbar provided in this figure.

       Response: Major object of this paper is how the improved ASTWBD was Created. Therefore, we believe that Figure 6c is important to show the increased occupancy ratio clearly.

Additional in-text descriptions of figures 1a,b,c,d and figures 4a,b were requested in the previous review for highlighting the most important aspects seen by authors in each image. However, such additions were not provided in the new version of the manuscript. Why were such tiles chosen as significant examples of the proposed approach? It would be significant to mention at least the average size of waterbodies in each tile, thus the reader could have an additional parameter for comparison of the resolution capabilities.

       Response: The image tiles with large increased occupancy ratios were selected as the typical examples. This sentence are added in section 2.3 ‘Typical Examples of Improvements’.