Geolocation Accuracy Assessment of Himawari-8/AHI Imagery for Application to Terrestrial Monitoring
Round 1
Reviewer 1 Report
- Please provide more global references with high impact factor in order to support the necessity of your methodology
- Please provide more relative studies in terms of geolocation accuracy.
- Figure 1 should be right after 2.1 subsection and not after two pages.
- You have used a lot of acronyms within the text and thus reading is becoming destructing.
- In subsection 2.3 you have to explain why you used data for the specific time zones you mentioned.
- Please simplify Figure 2
- Due to fact that Figure 5 is very crucial please analyze more within the text the results.
- Did you observed specific shifts within your study area. If yes, please provide a map.
- You have to extend the conclusions section
- You have to provide a discussion section as well comparing your work with relative studies.
Author Response
- Please provide more global references with high impact factor in order to support the necessity of your methodology
Response:
Thank you for your comments. We checked further references on geometric correction of geostationary satellites, particularty for high impact journals. With our best efforts, we could not find any relevant papers in high IF journals such as Remote Sensing of Environment, IEEE TGARS and Remote Sensing. However, we found a few papers in SCI journals.
Based on our literature survey, we added two more papers in the introduction section.
- Please provide more relative studies in terms of geolocation accuracy.
Response:
As described above, we searched similar papers on geolocation accuracies based on geostationary satellites, and we found a few papers. The analyses of these papers are geolocation accuracies of GOES-R (Tan et al. 2020) and COMS satellites (Jeong et al., 2020). We added discussion on the geolocation accuracy of GOES-R and comparison with this study since GOES-R is a similar generation geostationary satellites.
Reference:
Tan, B.; Dellomo, J.J.; Folley, C.N.; Grycewicz, T.J.; Houchin, S.; Isaacson, P.J.; Johnson, P.D.; Porter, B.C.; Reth, A.D.; Thiyanaratnam, P.; et al. GOES-R series image navigation and registration performance assessment tool set. J. Appl. Remote Sens. 2020, 14, 032405.
Jeong, J.; Han, H.; Park, Y. Geometric accuracy analysis of the Geostationary Ocean Color Imager (GOCI) Level 1B (L1B) product. Opt. Express 2020, 28, 7634.
- Figure 1 should be right after 2.1 subsection and not after two pages.
Response:
Revised as suggested.
- You have used a lot of acronyms within the text and thus reading is becoming destructing.
Response:
Thank you for your suggestion. We revised whole manuscript to reduce the number of acronyms used in this paper. Details are as follows:
1) We used JMA data instead of HSD (Himawari Standard Data) to make two datasets name consistent (named after data providers).
2) We avoided to use COFF/LOFF, COFF_ceres, COFF_hsd and only dCOFF/dLOFF were remained.
3) The acronyms of refRref, stdRref, refRobs_CEReS, stdRobs_CEReS, refRobs_JMA, and stdRobs_JMA were simplified to Rref, σRref, Robs_CEReS, Robs_CEReS, Robs_JMA, and Robs_JMA.
- In subsection 2.3 you have to explain why you used data for the specific time zones you mentioned.
Response:
The specific time-zone of 23:00UTC to 7:00UTC was selected based on the ranges of daytime at the nadir of Himawari satellite and corresponds to 8:00 JST (Japan Standard Time) to 16:00 JST. The selection is wider than those of other studies (e.g. 9:00-15:00 for Miura et al. (2019), 8:00-16:00 for Yan et al. (2019). Therefore, our selection of time-zone can be justified as an evaluation of geolocation accuracy toward terrestrial application. We added it as “we targeted the time from 23:00 UTC to 07:00 UTC based on the ranges of daytime at nadir of Himawari-8 and selection of previous studies (Miura et al. 2019 [3], Yan et al. 2019 [31]).” (see Line 172–174)
- Please simplify Figure 2
Response:
Based on the suggestion, we simplified the figure. See Figure 2 in the revised manuscript.
- Due to fact that Figure 5 is very crucial please analyze more within the text the results.
Response:
As you pointed out, the spatial pattern of the relative geolocation error is important. We added discussion about Figure 5 (modified to Figure 6) in the discussion section (see Line 338–361).
- Did you observed specific shifts within your study area. If yes, please provide a map.
Response:
Based on the suggestion, we added two cases of geometrically shifted case, large dCOFFi and dLOFFi, in the manuscript (Figure 2), and added its explanation.
- You have to extend the conclusions section
Response:
We re-read results and discussion section and conclusion section, and reorganized into three sections. We added and extended discussion section. Conclusion section was shorten based on the extended discussion section.
- You have to provide a discussion section as well comparing your work with relative studies.
Response:
Based on the suggestion, we split the result and discussion section into two separate sections. Please see these sections.
Reviewer 2 Report
The paper is well structured and the images, graphs and diagrams analyzed are clear and adequate to the scientific level of the Journal. However, I do not find elements of originality in the manuscript. Consequently, the manuscript, at present, cannot be considered a "research paper". The abastract, like the rest of the paper, can be better written. For example, it is necessary to better clarify the objectives of the paper and the potential of this approach in terrestrial applications.
Author Response
The paper is well structured and the images, graphs and diagrams analyzed are clear and adequate to the scientific level of the Journal. However, I do not find elements of originality in the manuscript. Consequently, the manuscript, at present, cannot be considered a "research paper". The abstract, like the rest of the paper, can be better written. For example, it is necessary to better clarify the objectives of the paper and the potential of this approach in terrestrial applications.
Response:
We appreciate your comments on the novelty and objectives of the study. Originality and objectives of our study is (1) to characterize temporal changes in the geolocation accuracy of original Himawari-8 AHI data, and (2) to demonstrate improvement of geolocation accuracy after applying further geometric correction using a visible band. We added this into abstract in order to clarify our objectives.
Reviewer 3 Report
This paper compares the geometric accuracy of the JMA HSD Himawari product and a post-processed CEReS product. The quality of the HSD product directly from JMA is disappointing, but it is fortunate that CEReS has undertaken to improve the Himawari product. The authors are correct that the CEReS product is better for land applications. It could also be an improvement for atmospheric motion vectors (AMVs) as well. There are no major problems with this paper. A couple small improvement suggestions:
- Discussion of potential benefits for AMV applications.
- The analysis of dLOFF and dCOFF represents geometric differences in two axes. Is there any rotation that can be detected? It would be harder to observe. Maybe a short rationale for the dLOFF and dCOFF parameterization could be included.
- The authors might consider, for the future, trying a Normalized Difference Vegetation Index (NDVI) approach to POC when geometrically correcting the Himawari product. NDVI can be less susceptible to shadows. The NDVI resolution would necessarily need to be 1km instead of 500m, but with subpixel interpolation there still could be some benefit.
- Supplementing Figure 3 with a histogram representation of the differences would be good. The density of points is far too high to get a get a sense of the statistics. Perhaps two sets of histograms each showing differences before/after JMA improvements could be included?
Author Response
This paper compares the geometric accuracy of the JMA HSD Himawari product and a post-processed CEReS product. The quality of the HSD product directly from JMA is disappointing, but it is fortunate that CEReS has undertaken to improve the Himawari product. The authors are correct that the CEReS product is better for land applications. It could also be an improvement for atmospheric motion vectors (AMVs) as well. There are no major problems with this paper. A couple small improvement suggestions:
Response:
We thank the reviewer for thorough and constructive feedback, which has greatly helped strengthen our manuscript.
Discussion of potential benefits for AMV applications.
Response:
Based on the suggestion, we added one paragraph on potential application to atmospheric motion vectors. Please see the last paragraph in the conclusion section.
The analysis of dLOFF and dCOFF represents geometric differences in two axes. Is there any rotation that can be detected? It would be harder to observe. Maybe a short rationale for the dLOFF and dCOFF parameterization could be included.
Response:
In principle, POC based geometric correction can estimate geometric shift for two axes. One axe is determined as east-west direction since AHI sensor scans east-west direction. Therefore, it is not possible to estimate rotation directly. We added a short explanation related to axes considered by POC (see Line 78–81).
We found dCOFF is much larger than dLOFF, which means east-to-west shift is a dominant factor.
The authors might consider, for the future, trying a Normalized Difference Vegetation Index (NDVI) approach to POC when geometrically correcting the Himawari product. NDVI can be less susceptible to shadows. The NDVI resolution would necessarily need to be 1km instead of 500m, but with subpixel interpolation there still could be some benefit.
Response:
Use of NDVI has a potential to improve geometric correction accuracy, however, we would like to use orignal 500m spatial resolution (red band) data. First, we found anomalous NDVI value is reported near the edge of cloud (Miura et al. 2019). Second, we would like to use better resolution, such as 500m. Furthermore, purpose of our paper is to show geolocation accuracy, not to show the methods of geolocation.
We have added a description of suggestion of NDVI approach (see Line 333–337).
Supplementing Figure 3 with a histogram representation of the differences would be good. The density of points is far too high to get a sense of the statistics. Perhaps two sets of histograms each showing differences before/after JMA improvements could be included?
Response:
Thank you very much for valuable comments. In the revised manuscript, we modified Figure 3 to include histograms to clarify it quantitatively and modified the corresponding descriptions in the result section.
Round 2
Reviewer 1 Report
Thank you for improving your manuscript based on my suggestions.
Author Response
Thank you for your comments.
After your review of English language and style, our manuscript has been proofread.
Reviewer 2 Report
The paper has improved since its first version. Consequently, I have no further comments.
Author Response
Thank you for your comments.
After your review of English language and style, our manuscript has been proofread.