Satellite ASTER Mineral Mapping the Provenance of the Loess Used by the Ming to Build their Earthen Great Wall
Round 1
Reviewer 1 Report
General remarks
I consider myself an expert in remote sensing (among other interests) with experience in geology. Therefore, I find the topic of the paper very interesting, and for the international readers, it might also deliver general and useful information. The author used ~ 250 ASTER L1T scenes from 25 overlapping paths with cloud cover (<10%) in summer. However, they applied some spectral mineral indices such us gypsum, silica, AlOH abundance, etc. for the area surrounding the Great Wall. Also, in situ measurements such as field portable spectrometers and Field-portable X-ray fluorescence (pXRF) were analyzed. Moreover, A high-resolution 3D surface models have been generated for each field site, using drone digital imagery. Besides, the authors collected field samples for laboratory works (e.g., XRD) for 1200 km of the Great Wall.
The manuscript is written in a readable, clear, and mixture (American or British) English, while the manuscript organization is no that well. However, several aspects look like a report (very good) but not as a scientific article. The key points are not highlighted but lost in the dense text. I still don’t fully understand who is the target of this article: a researcher in geochemistry? It feels that, as a remote sensing expert, I am not the target of this article, which is more oriented toward geochemistry. Nevertheless, I should be able to follow the main idea, which is not clear. Moreover, Part of the results is out of the scope of remote sensing. The main problems faced in this work that how can the authors used ASTER data (>15 m resolution) to determine minerals in the Great Wall, which has ~6 m width! Are the minerals concentrates enough to detect it using ASTER data! These two things must be discussed.
In summary, although the presumably huge work invested and the interesting topic of the paper, my suggestion is to major revised the manuscript. In the following, I present some more detailed arguments to support my decision. I think at least another round of careful revision is needed before it could be reconsidered.
Comments
Line #36: The introduction needs to be rewritten. It is so far from remote sensing topics.
Line #38: “forced” for what?
Line #46: the location of “Datong” should appear in Figure 1a. If "Datong" is out of the figure, please put an arrow to the eastward and write a label "Datong" there on the east side of the figure.
Line #47, Figure 1: North arrow is missing from Figure 1 A.
Line #48: the authors should put the RGB of the false-color composites of the ASTER data.
Line 48: the authors should not a mixture of American or British English, for example, lines #48, 53, 479, 536, 537, 727, 833, 932, 993, 1004, the author used the term “color” (American English); in lines #147, 150, they used the term “colour” and in line #168 they used the term “minimize” (British English).
Line #65: the authors should put a space between "," and "7".
Line #65: the authors should put a space between "8" and "m".
Line #68: the authors should put (Al2Si2O5(OH)4) between “kaolinite” and “and”.
Line #69: the authors should put “,” after “illite”, “wet”, and “cracking”.
Line 69: put “cracking, therefore, there are more” instead of “cracking and therefore is more”.
Line 74: the authors should put “(SiO2),” after “quartz”.
Line #75: delete “(SiO2)” and “(Al2Si2O5(OH)4)”.
Line #75: instead of “Ca.CO3” put “CaCO3”.
Line #76: instead of “Ca.O” put “CaO”.
Line #76: the authors should put “,” after “e.g.”.
Line #83: the authors should put “,” after “i.e.”.
Lines # 85-87: this sentence needs a reference "In contrast, the 20 to 70 μm fraction (i.e. coarse-silt) can remain in wind suspension for minutes to hours, travelling distances of meters to kilometers, and is termed short term wind suspension.".
Lines #87-88: this sentence needs a reference "Grains >70 μm in size (i.e. sand) are not carried by wind-suspension but instead bounce or saltate across the land surface, travelling only meters per wind event.".
Lines #88-90: this sentence needs a reference “Factors determining where this range of particle sizes can accumulate as deposits of loess include: prevailing wind direction; distance from source; and topographic and/or vegetation traps.”
Lines #126-127: instead of "ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer)" put "Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER )"
Line #130: "<100" could be ranged from 0.01 m to 99.99 m. I am asking the authors to give an exact pixel size, please give a specific resolution.
Line #132: The authors did not state the width of the Great Wall, which is very important in this work.
Line #166: The author should prepare a location map for the samples collected from the Great Wall. They can give a different symbol regarding the purpose of the sample.
Line #167: the authors should put “,” after “law”.
Line #168: the authors should put “,” after “thus” and “was”.
Line 174: The authors cited Figure 2 on page # 5 and they plotted the figure on page #10. I think the figure should come directly on page #6.
Lines #177-182: the subsection” 2.4. 3D model generation” should be explained well more detail information are missing. This paragraph needs to be declared "Camera calibration, photo aligning, dense point cloud building, mesh and texture building, DSM and ortho-mosaic building were processed with photogrammetric software called Agisoft hotoscan/Metashape.".
Line #180: the oblique angle should be declared.
Line #182: The put “Photoscan” instead of “hotoscan”.
Line #182: The software “Agisoft Photoscan/Metashape” should be cited and introduced.
Line #190: instead of “e.g. site 28, Figure 1d” put “e.g., site 28 and Figure 1d”.
Line #306: the authors should put “,” after “6b”.
Line #307: the authors should put “,” after “6c”.
Line #308: The subsection “2.14. Interpreting mineral transport pathways” should be rewritten and moved to the discussion section.
Lines #334-501: The key points are not highlighted but lost in the dense text. I still don’t fully understand who is the target of this article: researcher in geochemistry or remote sensing?
Line #889: The conclusions needs to be rewritten. It is very long (51 lines). Part of the conclusion should be removed (lines # 894-897). The others should be oriented to remote sensing (not to geochemistry) field.
Author Response
Overall comment.
The reviewer has hit on the concern that the authors have long shared and that is that the paper covers a lot ground. Note that we chose the Great Wall as a sampling target in part because little is known about its composition but mainly because it represented a 1200 km transect of similar material ideal for validating our regional satellite compositional maps. The primary objective of this project was to determine the provenance of the loess used by the Ming to build their earthen wall, which had two parts, (i) where did the Ming mine their loess for the earthen wall; and (ii) from where did this loess originate? Before submission, we had considered splitting the paper into at least two papers, namely one focused on the Great Wall results and a second on tracking the loess sediment transport pathways. However, the field data are essential validation data for the ASTER mineral maps and the ASTER mineral maps explain the compositional variation of the Ming earthen wall – a chicken and egg problem.
In the end, the earthen Great Wall is there because of the loess (pervasive, accessible and fertile) and the compositional variations along the earthen wall are the result of regional changes in composition of the loess. The ASTER mineral maps help unravel these compositional changes.
We have recast some of the Introduction, including being more specific about the aims, rewritten and shortened the Conclusions as well as changed a number of figures in the light of the reviewer’s comments. We have also corrected the typos/edits where we think appropriate. However, the paper remains long but we believe there is a lot of valuable information of interest to readers from many different professions.
Finally, this paper has been submitted to the 20th Anniversary of ASTER and its role in earth science mapping. We strongly believe our paper is well suited for this special issue and shows the value of the ASTER sensor in regional mapping surface composition for a complex but interesting application.
Reviewer 1
I consider myself an expert in remote sensing (among other interests) with experience in geology. Therefore, I find the topic of the paper very interesting, and for the international readers, it might also deliver general and useful information. The author used ~ 250 ASTER L1T scenes from 25 overlapping paths with cloud cover (<10%) in summer. However, they applied some spectral mineral indices such us gypsum, silica, AlOH abundance, etc. for the area surrounding the Great Wall. Also, in situ measurements such as field portable spectrometers and Field-portable X-ray fluorescence (pXRF) were analyzed. Moreover, A high-resolution 3D surface models have been generated for each field site, using drone digital imagery. Besides, the authors collected field samples for laboratory works (e.g., XRD) for 1200 km of the Great Wall.
The manuscript is written in a readable, clear, and mixture (American or British) English, while the manuscript organization is no that well. However, several aspects look like a report (very good) but not as a scientific article. The key points are not highlighted but lost in the dense text. I still don’t fully understand who is the target of this article: a researcher in geochemistry? It feels that, as a remote sensing expert, I am not the target of this article, which is more oriented toward geochemistry.
Apologies if I (the lead author) is mistaken but I am not clear if the reviewer is aware of the use of spectroscopy in remote sensing for measuring composition including the chemistry of stars to the mineralogy (which hosts the chemistry) of the planets and the Earth. The latter application is often called imaging spectroscopy or mineral mapping, hence the title of the paper.
The targeted audience is varied and includes (in priority): (1) ASTER geoscientists as this paper has been submitted to a special issue marking the 20th Anniversary since the launch of ASTER with the major focus on geoscience applications, which is the main purpose of this paper, i.e. can we use ASTER to help solve XX problem; (2) researchers working on the provenance of loess who traditionally have used point sample data and not compositional imaging as presented here for the first time; (3) people with a general interest in understanding the nature of the earthen Great Wall and how (from what/where) it was built and what factors might be promoting it’s degradation; and (4) those responsible for ensuring that the earthen Great Wall is preserved for future generations.
Nevertheless, I should be able to follow the main idea, which is not clear.
We appreciate this as a problem and have made some steps to improve this problem, especially I the Introduction section.
Moreover, Part of the results is out of the scope of remote sensing.
Could you provide more detail thanks?
The main problems faced in this work that how can the authors use ASTER data (>15 m resolution) to determine minerals in the Great Wall, which has ~6 m width! Are the minerals concentrates enough to detect it using ASTER data! These two things must be discussed.
We do not state or imply in the paper that ASTER can be used to directly map the mineralogy of the wall. As you suggest, we are faced with a spatial resolution issue. To tackle this problem, we first used a field sampling strategy that tested if there exist compositional relationships between the wall and the background (+/- 15 and 45 m) surface materials, i.e. a combined 90 m distance that spans one ASTER TIR pixel or 3 ASTER SWIR pixels or 6 VNIR pixels (all though were expended in the directional parallel to the wall by a factor of ~10). This 90 m wide area was targeted during the manual selection of ASTER ROIs. The end result of this process yielded similar patterns which both validated the ASTER pre-processing and revealed previously unknown compositional variations along the ~1200 km length of the wall.
Anyway, to avoid the possibility that a reader might make such as wrong assumption, we have added a sentence @ ~line 307 stating: “Note that we do not imply that the 15-90m pixel resolution of ASTER can be used to directly map the composition of the wall, which has a width ~8m. Instead our methodology relies on recognising similar compositional patterns between the ASTER and field data (which includes field sample points located 15 m and 45 m on either side of the wall) along the 1200 km length of the Ming earthen wall.”
In summary, although the presumably huge work invested and the interesting topic of the paper, my suggestion is to major revised the manuscript. In the following, I present some more detailed arguments to support my decision. I think at least another round of careful revision is needed before it could be reconsidered.
Comments
Line #36: The introduction needs to be rewritten. It is so far from remote sensing topics.
Done
Line #38: “forced” for what?
The Ming had had enough of Mongol raiding parties attacking their people across the Ordos region (and beyond).
Line #46: the location of “Datong” should appear in Figure 1a. If "Datong" is out of the figure, please put an arrow to the eastward and write a label "Datong" there on the east side of the figure.
Removed Datong and replaced with “Beijing” which is now shown in Figure 1a.
Line #47, Figure 1: North arrow is missing from Figure 1 A.
Lat long details are provided as is a small map of Asia so readers could easily work it out but we have added a N arrow anyway.
Line #48: the authors should put the RGB of the false-color composites of the ASTER data.
Done.
Line 48: the authors should not a mixture of American or British English, for example, lines #48, 53, 479, 536, 537, 727, 833, 932, 993, 1004, the author used the term “color” (American English); in lines #147, 150, they used the term “colour” and in line #168 they used the term “minimize” (British English).
All two replacements done.
Line #65: the authors should put a space between "," and "7".
Not sure but I guess not.
Line #65: the authors should put a space between "8" and "m".
Done.
Line #68: the authors should put (Al2Si2O5(OH)4) between “kaolinite” and “and”.
Done as well as all the other minerals listed
Line #69: the authors should put “,” after “illite”, “wet”, and “cracking”.
Done.
Line 69: put “cracking, therefore, there are more” instead of “cracking and therefore is more”.
Replaced “is” with “are”.
Line 74: the authors should put “(SiO2),” after “quartz”.
done
Line #75: delete “(SiO2)” and “(Al2Si2O5(OH)4)”.
done
Line #75: instead of “Ca.CO3” put “CaCO3”.
done
Line #76: instead of “Ca.O” put “CaO”.
done
Line #76: the authors should put “,” after “e.g.”.
Not done.
Line #83: the authors should put “,” after “i.e.”.
Not done.
Lines # 85-87: this sentence needs a reference "In contrast, the 20 to 70 μm fraction (i.e. coarse-silt) can remain in wind suspension for minutes to hours, travelling distances of meters to kilometers, and is termed short term wind suspension.".
done
Lines #87-88: this sentence needs a reference "Grains >70 μm in size (i.e. sand) are not carried by wind-suspension but instead bounce or saltate across the land surface, travelling only meters per wind event.".
done
Lines #88-90: this sentence needs a reference “Factors determining where this range of particle sizes can accumulate as deposits of loess include: prevailing wind direction; distance from source; and topographic and/or vegetation traps.”
done
Lines #126-127: instead of "ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer)" put "Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER )"
done
Line #130: "<100" could be ranged from 0.01 m to 99.99 m. I am asking the authors to give an exact pixel size, please give a specific resolution.
30 m now stated in the text. The problem is that we have resampled the 90m pixel resolution TIR bands up to 30 m and the 15 m pixel VNIR bands down to 30 m so that we could join together all VNIR, SWIR and TIR bands into a single 30 m file for image processing. I guess the point here at this introductory stage of the paper is to highlight to the reader that we are working with a very large region and that every pixel of this region has been measured for its surface composition using ASTER at a pixel resolution better than 100 m. Qualifying the <100 m with the details given above we believe would only confuse the reader and detract from this key point (of the paper).
Line #132: The authors did not state the width of the Great Wall, which is very important in this work.
We believe it is not appropriate here but is now given in Methods 2.13. Note the base width of the wall is typically ~6 m, not including any spoil/talus shed from the wall.
Line #166: The author should prepare a location map for the samples collected from the Great Wall. They can give a different symbol regarding the purpose of the sample.
Already given in Figure 1a.
Line #167: the authors should put “,” after “law”.
done
Line #168: the authors should put “,” after “thus” and “was”.
Not done.
Line 174: The authors cited Figure 2 on page # 5 and they plotted the figure on page #10. I think the figure should come directly on page #6.
done
Lines #177-182: the subsection” 2.4. 3D model generation” should be explained well more detail information are missing. This paragraph needs to be declared "Camera calibration, photo aligning, dense point cloud building, mesh and texture building, DSM and ortho-mosaic building were processed with photogrammetric software called Agisoft hotoscan/Metashape.".
We believe that this suggested expansion of drone data processing is not necessary as the derived DEM models simply provide the reader with a visual (qualitative) appreciation of the variety of wall conditions (Figure 1b-1e).
Line #180: the oblique angle should be declared.
See point above
Line #182: The put “Photoscan” instead of “hotoscan”.
done
Line #182: The software “Agisoft Photoscan/Metashape” should be cited and introduced.
done
Line #190: instead of “e.g. site 28, Figure 1d” put “e.g., site 28 and Figure 1d”.
Not done
Line #306: the authors should put “,” after “6b”.
Not done
Line #307: the authors should put “,” after “6c”.
Not done
Line #308: The subsection “2.14. Interpreting mineral transport pathways” should be rewritten and moved to the discussion section.
Not done. This section describes the methods/assumptions used in the image interpretation. Not the related interpretation which is described in the Results/Discussion.
Lines #334-501: The key points are not highlighted but lost in the dense text. I still don’t fully understand who is the target of this article: researcher in geochemistry or remote sensing?
Audience explained above. We had originally written summary points and their implications for each sub-section in the “Results” but later removed them as it largely replicated what we were presenting in the Discussion section.
Line #889: The conclusions needs to be rewritten. It is very long (51 lines). Part of the conclusion should be removed (lines # 894-897). The others should be oriented to remote sensing (not to geochemistry) field.
Conclusions are re-written and shortened.
Reviewer 2 Report
The authors used well known ASTER mineral indices to identify the distribution of the above-mentioned minerals in the study area. Only 25 sites were selected for field sampling and validation of ASTER results. The selected sites were selected from the remnants of the Ming earthen wall.
I would recommend rejection of the paper for the following reasons:
Sampling and validation sites are not enough and should be distributed over the identified pathways. Authors used ASD spectroradiometers for validation, but the manuscript did not include any figure of the ASD measurements. Accuracy assessment of the ASTER results should be added to the manuscript. There are some unnecessary field photographs, that can be replaced by ASD and XRD results. The introduction section does not include previous work of using remote sensing or ASTER for mineral identification. In Page (4), authors explained how they used extent of shadowing to measure the height of the wall? what is the reference of this method? How is it relevant to the objectives of this work?Author Response
Overall comment.
The reviewer has hit on the concern that the authors have long shared and that is that the paper covers a lot ground. Note that we chose the Great Wall as a sampling target in part because little is known about its composition but mainly because it represented a 1200 km transect of similar material ideal for validating our regional satellite compositional maps. The primary objective of this project was to determine the provenance of the loess used by the Ming to build their earthen wall, which had two parts, (i) where did the Ming mine their loess for the earthen wall; and (ii) from where did this loess originate? Before submission, we had considered splitting the paper into at least two papers, namely one focused on the Great Wall results and a second on tracking the loess sediment transport pathways. However, the field data are essential validation data for the ASTER mineral maps and the ASTER mineral maps explain the compositional variation of the Ming earthen wall – a chicken and egg problem. In the end, the earthen Great Wall is there because of the loess (pervasive, accessible and fertile) and the compositional variations along the earthen wall are the result of regional changes in composition of the loess. The ASTER mineral maps help unravel these compositional changes.
We have recast some of the Introduction, including being more specific about the aims, rewritten and shortened the Conclusions as well as changed a number of figures in the light of the reviewer’s comments. We have also corrected the typos/edits where we think appropriate. However, the paper remains long but we believe there is a lot of valuable information of interest to readers from many different professions.
Finally, this paper has been submitted to the 20th Anniversary of ASTER and its role in earth science mapping. We strongly believe our paper is well suited for this special issue and shows the value of the ASTER sensor in regional mapping surface composition for a complex but interesting application.
Other comments
The authors used well known ASTER mineral indices to identify the distribution of the above-mentioned minerals in the study area. Only 25 sites were selected for field sampling and validation of ASTER results. The selected sites were selected from the remnants of the Ming earthen wall.
I would recommend rejection of the paper for the following reasons:
Sampling and validation sites are not enough and should be distributed over the identified pathways.
This is a good idea but we suggest is not appropriate for two reasons: (i) Our sampling/validation strategy targets the primary aim of the paper namely: to determine the provenance of the loess used by the Ming to build their earthen Great Wall, which has two parts (where did the Ming mine the loess for their wall and where did this loess originate); and (ii) how do you go about sampling a region spanning >600,00 km2 and then meaningfully present the results (see paper by Cudahy et al, 2016 in Nature Science Reports). We suggest as a transect that spans the study region, which the Ming Wall proved to be excellent sampling target given that is made of loess.
Authors used ASD spectroradiometers for validation, but the manuscript did not include any figure of the ASD measurements.
We have now included selected ASD spectra as [part of Figure 3i.
Accuracy assessment of the ASTER results should be added to the manuscript.
What do you suggest is possible/practical other than what we have done using the comparison of wall transect results for both the ASTER ROI and field data? The similarity in patterns is a convincing, albeit qualitative, indication of the error.
There are some unnecessary field photographs, that can be replaced by ASD and XRD results.
We argue that the field photos shown in Figure 2 mosaic are important to the paper, for example: (i) recognition of the Ming’s “hangtu” construction method as we had to be clear we were sampling Ming wall and not earlier Dynasty walls; (ii) the Ming wall included rocks of the same type found locally bellow thin loess cover (addressed issue of where did the Ming source the loess); (iii) locating in situ pXRF sampling; and (iv) demonstrating how eroded layers of preferential groundwater attack can occur at different heights of the wall.
The introduction section does not include previous work of using remote sensing or ASTER for mineral identification.
See references 58-61 in the intro as well as others in the text [68, 69, 70, 71, 102].
In Page (4), authors explained how they used extent of shadowing to measure the height of the wall? what is the reference of this method? How is it relevant to the objectives of this work?
This method was developed by the lead author in the process of preparing for the fieldwork. That is, we had to establish where the Ming earthen wall was and what condition it was likely to be in (for our regional assessment of the wall’s erosional robustness we believed it best to compare the most preserved part for a given area which were spaced around every 50 km).
We have now deleted this section.
Reviewer 3 Report
Interesting and comprehensive study of both geomorphological/soil mapping and archaeological applications of ASTER imagery. Attached comments for improvements including explicit listing of objectives, explanation of use of L1T imagary, mosaicking algorithm strategy, labelling of y axis of some figures, and correction of some typos/equations. Worth publication following these adjustments
Comments for author File: 
Comments.pdf
Author Response
Overall comment.
The reviewer has hit on the concern that the authors have long shared and that is that the paper covers a lot ground. Note that we chose the Great Wall as a sampling target in part because little is known about its composition but mainly because it represented a 1200 km transect of similar material ideal for validating our regional satellite compositional maps. The primary objective of this project was to determine the provenance of the loess used by the Ming to build their earthen wall, which had two parts, (i) where did the Ming mine their loess for the earthen wall; and (ii) from where did this loess originate? Before submission, we had considered splitting the paper into at least two papers, namely one focused on the Great Wall results and a second on tracking the loess sediment transport pathways. However, the field data are essential validation data for the ASTER mineral maps and the ASTER mineral maps explain the compositional variation of the Ming earthen wall – a chicken and egg problem. In the end, the earthen Great Wall is there because of the loess (pervasive, accessible and fertile) and the compositional variations along the earthen wall are the result of regional changes in composition of the loess. The ASTER mineral maps help unravel these compositional changes.
We have recast some of the Introduction, including being more specific about the aims, rewritten and shortened the Conclusions as well as changed a number of figures in the light of the reviewer’s comments. We have also corrected the typos/edits where we think appropriate. However, the paper remains long but we believe there is a lot of valuable information of interest to readers from many different professions.
Finally, this paper has been submitted to the 20th Anniversary of ASTER and its role in earth science mapping. We strongly believe our paper is well suited for this special issue and shows the value of the ASTER sensor in regional mapping surface composition for a complex but interesting application.
Other comments
Interesting and comprehensive study of both geomorphological/soil mapping and archaeological applications of ASTER imagery. Attached comments for improvements including explicit listing of objectives, explanation of use of L1T imagary, mosaicking algorithm strategy, labelling of y axis of some figures, and correction of some typos/equations. Worth publication following these adjustments.
Ln 51. Difficult to read these sample site numbers.
Font size increased from 10 to 12.
Ln 58. An explicit explanation of “loess”, its context in the Chinese environment, and relevant references would be useful in this introduction. Your later interpretation for aeolian sources and paleo wind directions need reference(s) and background context
More detailed definition is already given in later sections though we have added here “Even though wind-generated loess deposits (dominated by angular, silt-size grains) are readily eroded [6,7],”
Ln 73. Where did this definition of sand come from? There are several version of texture – State which texture classification you use eg USDA or ISSS.
ISO 14688 (WG BBA from the reference you provided).
Ln 87. Grains >70 μm in size (i.e. sand) is a reference from Pye 1987.
Ln 122: Not clear, clarify. The formation of material includes the processes involved, indicated by the 3D soil profile, and related to climate, history etc. Comment on the depth issue and it relation to mappable unit classifications. Limitations in duplex soils, if present?
This is a direct quote from Smalley et al., 2019. The message we read from this reference is that understanding the nature of the grains of loess (such as its composition) is key to understanding the formation of the loess.
Ln 138: In summary, list the primary (& secondary?) aims of the study ? Correlation of field and ASTER spectroscopy to interpret the provenance of the Ming Great Wall? Testing the application of ASTER for geomorphological interpretation and archaeological investigations?
Expanded the sentence as follows: “From this multi-scale, spatially-comprehensive compositional framework, we obtain a more detailed understanding of the provenance of the loess used by the Ming to build their earthen wall. This overall aim though has two linked parts, namely, (i) where did the Ming mine their loess, i.e. was it from local sources (<1 km away) or from centralized quarries many kilometers away; and (ii) where did this loess originate. With regards to the second, does the loess mineralogy separate into certain particle size ranges, driven by different transport processes? Finally, has any compositional heterogeneity impacted on wall’s current erosional state as this could assist in wall’s future preservation.”
Ln: 230. Clarify the use of the radiance @ sense L1T instead of L2 surface reflectance and emissivity? Improved spatial accuracy? Use of ratios overcome seasonal/atmospheric conditions? How were the multi date ASTER TIR variable temperature issues overcome?
The ASTER L1T radiance@sensor data was the standard (only) orthorectifed product for all VNIR, SWIR and TIR bands at the time of data download (and presently). Orthorectification for all bands is critical given high amount of relief across the region and the need to have all bands processed together, especially for the green vegetation unmixing step. Besides, the L2 reflectance atmospheric correction is simply application of a gain and offset for each band per scene, which is what we achieved anyway with our invariant target ROI cross-calibration, which had the important benefit of ensuring that all offsets were properly accounted for prior to the gain corrections.
Ln 238: How were these image pairs and invariant targets identified? Manually? Software algorithm?
Manually. Done.
Ln 274: Check this. Don’t you mean (B5 + B7)/ B6 ? You have the same formula as your GI
Done
Ln 425. Enlarge figure or display as landscape in 2 pages?
Now as landscape.
Figures 4-7 y-axis now with numbers
Ln 899:
Conclusions have been rewritten.
ln 916.
These minerals are less discernible using ASTER in this study largely because of either low abundance (e.g. zeolite, carbonate, gypsum) ,ASTER’s limited spectral resolution (zeolite, halite, carbonate and gypsum).
Reviewer 4 Report
Comments on paper by Cudahy et al., “Satellite ASTER mineral mapping the provenance of the loess used by the Ming to build their earthen Great Wall”
General comments
This paper presents results from the combined use of ASTER global mineral maps and field spectral measurements as well as standard rock sample analysis techniques to track the origin of the loess used by the Ming to build their earthen Wall. The synoptic view provided by the ASTER mosaic and the derived mineral maps allowed for tracking of the sediment transport pathways leading to the formation of the loess. The combined observations also lead to a better understanding of the building techniques used by the Ming. This is a very extensive and comprehensive study on a very interesting (and original) subject.
However, it seems to me that there are almost two papers in one: one regarding the source of the materials used by the Ming to build the earthen wall, based on field observations and relationships with ASTER mineral indices for identification of the source materials; the other one using the synoptic view provided by the ASTER mineralogy maps to try and identify the source materials for the different Loess outcrops and derive a more general interpretation in terms of transport pathways, which is a major subject in itself.
It is easy to see how the synoptic view provided by satellite remote sensing images and more specifically ASTER here can help improving the understanding of transport pathways. And to that matter, this paper is really interesting and brings some new perspectives but I had a hard time figuring out if the main purpose is to understand sediment transport and sources of the loess based on ASTER maps or to determine the source of the material used to build the wall? I fail to understand how the techniques used by the Ming to build their earthen wall can directly contribute to the understanding of the dust transport pathways, even if this is also an interesting subject in itself. So I would suggest reorganizing or refocusing the paper in order to make it easier to read. A more logical structure / organization might be (1) field observations then (2) wall composition / building techniques and (3) sources of the material (vicinity or not) based on ASTER maps and finally, when “strange” materials are observed, remote sources and pathways.
I also have several questions regarding the ASTER mineral maps derived from various well known ASTER indices. It seems to me that, since most of the interpretation in terms of transport pathways relies on “spatial gradients”, it is important to have an idea of the spread of the values taken by the indices (histogram?). This is not possible here since the scattergrams in figures 4 to 7, which are then the base for the general interpretation in terms of transport pathways and source location, have no y-scale. The fact that the values derived from ASTER ROI are different from those obtained using the resampled field spectra is understandable (difference in observation conditions) but does not prevent from at least giving a feeling of the ranges. Moreover, using a color table does not indicate what the range of values is since it is easy enough to apply some contrast enhancement to a black and white image. Considering that most of the interpretation is based on spatial gradation of indices (which sometimes appear rather subtle as for kaolinite/white mica for example), I think it is necessary to add this information.
I also assume that the 250 ASTER scenes used here were acquired at different times of year (seasonal effects?) and possibly different years so I wonder what is the impact on the resulting indices, especially since there are also radiometric modifications consecutive to the various miscalibration correction issues.
Regarding the field observations (section 3.6 in particular), I would have appreciated a figure with some representative ASD reflectance spectra as well as a comparison between ASD resampled to ASTER and ASTER ROI spectra, illustrating the subsections (gypsum, Q-sand, kaolinite, white mica, montmorillonite, chlorite, carbonate, amphibole...). It would help understanding what is happening with the mineral maps.
Finally, although the use of ASTER global mineral maps makes it easier and faster to locate the various mineral combinations and thus assess the potential sources, it is hard to evaluate what is new with respect to already published papers based on more “traditional” approaches. I think a figure at the end synthesizing the different transport pathways would be useful and help avoiding repeating in section 4 (starting line 767 and onward) what has already been said in previous sections. It would also make it easier to compare with other sources of information regarding the same subject but using different approaches and hence emphasize what is new.
A few additional questions:
- The ASTER mineral maps reflecting the surface status at an instant t (present time), any idea of any potential changes in climate dynamics (wind strength and direction, drier, windier conditions, etc.) with respect to Ming times? Could some kind of time series be considered?
- Did the authors make any attempts to combine the indices in color composites to help understanding the various possible combinations?
- Any observations of sample mineralogy in a “traditional” manner, that is under a microscope / thin sections / SEM?
Despite all these remarks and questions, I still think this paper reflects a lot of work and is worth publishing. I would however recommend reworking the structure (or refocusing) and completing some of the information, in particular regarding the range of values for each ASTER mineral maps. This would, to my point of view, strengthen the arguments used to derive transport pathways and locating/confirming the sources of the loess sediments.
I therefore recommend that some (major) changes be made to this paper before publication.
Detailed comments:
* Line 75, 76: CaCO3 and not Ca.CO3
* Line 174: … measurements were taken at each of these sample points
* Line 182: Source/reference for Agisoft Photoscan/Metashape?
* Line 195: Particle size measurements of wall samples
* Line 274: Are the band numbers correct? They seem to correspond to bands in the TIR and not around 2.2 µm…
* Line 383: section title should be in italic
* Line 425: …field samples
* Line 547: Cenozoic?
* Line 553 and 555: SI3
* Line 647: … the Tibetan Plateau
* Line 705: …; and (iii) the importance…
* Line 783: These observations help understand why…
Author Response
Overall comment.
The reviewer has hit on the concern that the authors have long shared and that is that the paper covers a lot ground. Note that we chose the Great Wall as a sampling target in part because little is known about its composition but mainly because it represented a 1200 km transect of similar material ideal for validating our regional satellite compositional maps.
The primary objective of this project was to determine the provenance of the loess used by the Ming to build their earthen wall, which had two parts, (i) where did the Ming mine their loess for the earthen wall; and (ii) from where did this loess originate? Before submission, we had considered splitting the paper into at least two papers, namely one focused on the Great Wall results and a second on tracking the loess sediment transport pathways. However, the field data are essential validation data for the ASTER mineral maps and the ASTER mineral maps explain the compositional variation of the Ming earthen wall – a chicken and egg problem. In the end, the earthen Great Wall is there because of the loess (pervasive, accessible and fertile) and the compositional variations along the earthen wall are the result of regional changes in composition of the loess. The ASTER mineral maps help unravel these compositional changes.
We have recast some of the Introduction, including being more specific about the aims, rewritten and shortened the Conclusions as well as changed a number of figures in the light of the reviewer’s comments. We have also corrected the typos/edits where we think appropriate. However, the paper remains long but we believe there is a lot of valuable information of interest to readers from many different professions.
Finally, this paper has been submitted to the 20th Anniversary of ASTER and its role in earth science mapping. We strongly believe our paper is well suited for this special issue and shows the value of the ASTER sensor in regional mapping surface composition for a complex but interesting application.
Other comments
This paper presents results from the combined use of ASTER global mineral maps and field spectral measurements as well as standard rock sample analysis techniques to track the origin of the loess used by the Ming to build their earthen Wall. The synoptic view provided by the ASTER mosaic and the derived mineral maps allowed for tracking of the sediment transport pathways leading to the formation of the loess. The combined observations also lead to a better understanding of the building techniques used by the Ming. This is a very extensive and comprehensive study on a very interesting (and original) subject.
However, it seems to me that there are almost two papers in one: one regarding the source of the materials used by the Ming to build the earthen wall, based on field observations and relationships with ASTER mineral indices for identification of the source materials; the other one using the synoptic view provided by the ASTER mineralogy maps to try and identify the source materials for the different Loess outcrops and derive a more general interpretation in terms of transport pathways, which is a major subject in itself.
We thought long about splitting this paper into two or more shorter and possibly better focused papers, albeit likely submitted to different journals and hence audiences and so losing the common thread. In the end, we chose one paper as we only ever had one primary objective in this study and that was to determine if mapping mineral composition at multi-scales (satellite and field) would allow us to better track the provenance of the loess used by the Ming to build their earthen wall. This aim as you suggest having two parts, namely where did the Ming mine this loess and where did this loess originate.
Separating the paper accordingly is problematic as much of the underpinning data is shared. Note also that we did not present the bulk of the data collected as part of this study, only presenting that which we believed was key evidence for supporting conclusions. The ASD spectra you mention being one such example of omitting data as we believed that only spectral experts can read the compositional information contained in spectra. And how do you meaningfully portray hundreds of spectra? We chose to instead convolve the spect5ra to ASTER response so we could compare these with the ASTER ROI results.
It is easy to see how the synoptic view provided by satellite remote sensing images and more specifically ASTER here can help improving the understanding of transport pathways. And to that matter, this paper is really interesting and brings some new perspectives but I had a hard time figuring out if the main purpose is to understand sediment transport and sources of the loess based on ASTER maps or to determine the source of the material used to build the wall?
I fail to understand how the techniques used by the Ming to build their earthen wall can directly contribute to the understanding of the dust transport pathways, even if this is also an interesting subject in itself.
We chose the Ming earthen wall essentially as a validation medium for the ASTER mapping as it cross cuts the region as a useful transect and because we assumed, they mined this loess from the (near) surface and that they used the same construction methods throughout. The fact that the same wall shows different degrees of preservation along its length raises the question whether regional changes in the composition of the loess is the cause of its erosional state. Loess composition being key.
So I would suggest reorganizing or refocusing the paper in order to make it easier to read. A more logical structure / organization might be (1) field observations then (2) wall composition / building techniques and (3) sources of the material (vicinity or not) based on ASTER maps and finally, when “strange” materials are observed, remote sources and pathways.
I also have several questions regarding the ASTER mineral maps derived from various well known ASTER indices. It seems to me that, since most of the interpretation in terms of transport pathways relies on “spatial gradients”, it is important to have an idea of the spread of the values taken by the indices (histogram?). This is not possible here since the scattergrams in figures 4 to 7, which are then the base for the general interpretation in terms of transport pathways and source location, have no y-scale. The fact that the values derived from ASTER ROI are different from those obtained using the resampled field spectra is understandable (difference in observation conditions) but does not prevent from at least giving a feeling of the ranges. Moreover, using a color table does not indicate what the range of values is since it is easy enough to apply some contrast enhancement to a black and white image. Considering that most of the interpretation is based on spatial gradation of indices (which sometimes appear rather subtle as for kaolinite/white mica for example), I think it is necessary to add this information.
Ranges are now included but in the end, we were only looking for similar patterns in the field versus ASTER especially as the ASTER data values had been modified by a number of pre-processing steps.
I also assume that the 250 ASTER scenes used here were acquired at different times of year (seasonal effects?) and possibly different years so I wonder what is the impact on the resulting indices, especially since there are also radiometric modifications consecutive to the various miscalibration correction issues.
We stated in Section 2.10 (now 2.9) that: “Approximately 250 images from 25 overlapping paths were selected on the basis of cloud cover (<10%) and season (ideally late summer). However, for some areas, the only available images in the archive were compromised by cloud cover, snow and/or green vegetation, which potentially impacted on the accuracy of the subsequent image cross-calibration.”
In the same section we also state: “Even using this pre-processing strategy, residual calibration errors between images/paths remain in the final mosaic (highlighted by white block arrows in Figures 4a, 5a, 6a and 7a). Note though that these apparent calibration errors do not persist for the entire length of a given satellite path and less so for those areas close by where the invariant targets were selected. That is, poorly exposed areas with extensive snow or dynamic green vegetation cover and often associated with higher topographic elevation often show these overlapping image mismatches. There are also errors caused by local variations in atmospheric conditions, especially water vapour and aerosols. These localised errors could be reduced through additional cross-calibration of the aberrant images.”
Note that our manual cross-calibration method applies one gain and one offset for each band and for each scene which is similar to applying a MODTRAN climate model which also derives and allows the application of a gain (atmospheric absorption) and offset (atmospheric scattering) for each band and image.
Regarding the field observations (section 3.6 in particular), I would have appreciated a figure with some representative ASD reflectance spectra as well as a comparison between ASD resampled to ASTER and ASTER ROI spectra, illustrating the subsections (gypsum, Q-sand, kaolinite, white mica, montmorillonite, chlorite, carbonate, amphibole...). It would help understanding what is happening with the mineral maps.
Your first point is now included as Figure 3i.The second is much more problematic as the shapes are not readily comparable without first calculating a suite of gains and offsets to align the entire ASTER ROI data suite to that of the convolved ASD to ASTER suite. To the eye, they would look very similar without much if any variation. The way we have amplified this very subtle variation to gain up the spatial patterns provided by Ming Wall transect which gets back to the point you raise above regards why we chose the Ming Wall.
Finally, although the use of ASTER global mineral maps makes it easier and faster to locate the various mineral combinations and thus assess the potential sources, it is hard to evaluate what is new with respect to already published papers based on more “traditional” approaches.
This issue is raised both in the Introduction and then expanded in the Discussion section. In brief, there have been a plethora of studies trying to track the source of the loess but all to date have used point sample measurements of many different types. There is considerable disagreement. The problem has been the reliance on point sample data spanning a huge region. As discussed in the intro we have taken a “mineral system” approach developed by exploration geologists who use geophysical imagery to track relic fluid pathways in the hunt for economic deposits though we use here ASTER to map a loess transport system. There is no similar published work on the Great Wall results/interp.
I think a figure at the end synthesizing the different transport pathways would be useful and help avoiding repeating in section 4 (starting line 767 and onward) what has already been said in previous sections. It would also make it easier to compare with other sources of information regarding the same subject but using different approaches and hence emphasize what is new.
Good idea and one which we attempted but trying to get this all onto a single figure given the complexity of the system was beyond us. Needs more thinking.
A few additional questions:
- The ASTER mineral maps reflecting the surface status at an instant t (present time), any idea of any potential changes in climate dynamics (wind strength and direction, drier, windier conditions, etc.) with respect to Ming times? Could some kind of time series be considered?
Not that we could define.
- Did the authors make any attempts to combine the indices in color composites to help understanding the various possible combinations?
That would make interpretation a complicated spatial-composition pattern even more complex.
- Any observations of sample mineralogy in a “traditional” manner, that is under a microscope / thin sections / SEM?
The lead author has used such techniques many times for other studies but given that there have already been many other workers looking at this more detailed spatial scale for loess samples we thought it better to not replicate and provide even more data in this already busy paper.
Despite all these remarks and questions, I still think this paper reflects a lot of work and is worth publishing. I would however recommend reworking the structure (or refocusing) and completing some of the information, in particular regarding the range of values for each ASTER mineral maps. This would, to my point of view, strengthen the arguments used to derive transport pathways and locating/confirming the sources of the loess sediments.
I therefore recommend that some (major) changes be made to this paper before publication.
Round 2
Reviewer 1 Report
Reviewer’s report
on the manuscript entitled “Satellite ASTER mineral mapping the provenance of the loess used by the Ming to build their earthen Great Wall”
Submitted to Remote Sensing (Manuscript ID: remotesensing-642264)
Round 2
General remarks
I would like to thank the authors for their report. However, the manuscript still suffers from the same problems, the focus of the manuscript still shifted to be far from remote sensing topics. I cannot say it is without but I can say it is far from remote sensing topic. I said in the first report that, the manuscript should be oriented to remote sensing (not to geochemistry) field. Unfortunately, I got the following replay: “Apologies if I (the lead author) is mistaken but I am not clear if the reviewer is aware of the use of spectroscopy in remote sensing for measuring composition including the chemistry of stars to the mineralogy (which hosts the chemistry) of the planets and the Earth. The latter application is often called imaging spectroscopy or mineral mapping, hence the title of the paper.”
Although I have stated in the previous report that “The introduction needs to be rewritten. It is so far from remote sensing topics.”., the authors did not reorient the manuscript to be suitable for the Remote Sensing journal. The introduction chapter still far from remote sensing.
I have stated in the previous report that “Moreover, Part of the results is out of the scope of remote sensing.”. The authors stated, “Could you provide more detail thanks?”. Here I need to mention to subsections “3.1., 3.2., and 3.3.”, which “in my opinion” can be removed from the results and the manuscript. Am I wondering about the cohesion between these subsections and the remote sensing work in this manuscript? If there are related then it should be clearly highlighted.
In summary, although the presumably huge work invested and the interesting topic of the paper, my suggestion is to reject the manuscript in its current form with an opportunity to resubmit it to remote sensing after modifying the manuscript.
Comments for author File: Comments.pdf
Author Response
It is clear Reviewer 1 does not understand that “Remote Sensing” is a diverse range of technologies designed to distantly measure/map/monitor a diverse range of applications. This includes the spectroscopy of inorganic materials, like minerals - so called mineral mapping.
Our selected remote sensing tool/technology is ASTER and our target/application is to better understand the provenance of the loess used by the Ming to build their earthen Great Wall. Unfortunately, Reviewer 1 does not explain why they insist that our application of ASTER data, with related validation data, is somehow not “remote sensing”, especially given that we have submitted the paper to a Remote Sensing special issue on ASTER applications.
Unfortunately, Reviewer 1 has still not provided any suggestion on how to do a “complete” rewrite of the Introduction. We remain of the belief that each paragraph in the Introduction provides a reference/baseline/framework for the varied work covered in the Results and Discussion. We suspect that Reviewer 1 may be troubled by us opening the Introduction with a brief historical account of the Ming earthen wall construction. But we remain convinced that this is appropriate as it provides both a general interest lead into the paper but more importantly sets the foundation for understanding why we chose the Ming Wall for our investigation. That is, it was built very rapidly (months) using the same material (=loess) over a very large distances (>1200 km), thus making it ideal as a linear sampling target for validating our regional satellite ASTER mineral mosaics.
We thank Reviewer 1 for finally providing some guidance on what they consider as not appropriate for this paper, including Sections: 3.1 (Earthen Wall building materials); 3.2 (Earthen Wall Status) and 3.3 (Particle size). However, this is very confusing because at the same time they chose not to remove Section 3.4 (Field XRF Chemistry), especially given their very strong belief that geochemistry is not related to remote sensing. Very puzzling.
Instead, we strongly argue that the field data that we have presented in the paper, including Sections 3.1, 3.2 and 3.3, are essential for: (i) validating the satellite ASTER information; (ii) better understanding both the provenance of the loess (local and regional) used by the Ming to build their earthen wall; and (iii) whether changes in wall composition could affect its erosional durability. To this end, Section 3.1 (one paragraph only) is aimed at giving readers the confidence that authors understood what to look for when ensuring that we were sampling the Ming earthen wall and not one of the many other walls built by earlier Dynasties located nearby. If we got that wrong then we would not be able to compare the erosional status of the earthen wall sampled from different areas.
Section 3.2 (two paragraphs only) is focused on the remnant condition of the Ming earthen wall, which is a major theme of the paper. Without this information, especially the cited Figure 2j, how else could we assess/compare the role of composition in determining the wall’s erosional robustness (e.g. the well preserved Shandan Section).
The particle size information delivered in Section 3.3 is important for at least three reasons. First, particle size significantly affects the spectral behaviour of materials, especially the thermal infrared (TIR) (key reference cited in paper, Salisbury and Walter 1989) and so must be accounted for given our use of the ASTER TIR-based product called “Silica Index”. Secondly, loess is a well-sorted material, typically with multiple populations of discrete particle size ranges, with each related to different modes/origins and potentially compositions - an opportunity (and major theme) addressed in detail in this paper. From our particle size analysis, we concluded that the loess comprised four particle size populations, two of which we traced via the related mineral composition to its source using ASTER. Finally, we found that the proportion of fine to coarse size fractions is the key driver for determining the erosional robustness of the Ming earthen wall and that this particle size is measurable using ASTER TIR bands (silica index).
Thus, we can only conclude that Reviewer 1 does not understand the role of spectroscopy in remote sensing nor the value/importance of independent validation data.
Reviewer 2 Report
Title: Satellite ASTER mineral mapping the provenance of the loess used by the Ming to build their earthen Great Wall
Decision: Reject because the manuscript is scientifically incomplete and/or lacks a significant, novel contribution to the field.
The main objective of this paper is defining the sediment transport pathways of the loess used by the Ming to build their earthen Great Wall.
The authors successfully used well known ASTER mineral indices to identify the spatial distribution of the compositional minerals, but they did not validate their results. I highlighted this point in my previous comments, but they did not address it.
The conclusion of this research paper is built on assumptions, not facts.
This paper still needs more work to prove these assumptions and to validate the ASTER results. I would recommend that:
Collecting loess samples along the proposed pathways for minerals in figures 4, 5, 6 and 7. Perform XRD & ASD measurements for the collected samples. Use petrography or geochemistry to identify the mineralogical composition of the collected loess samples. Perform Isotopic analysis (age dating) for the collected samples from the Great Wall and from the proposed pathways.
Author Response
It is clear that Reviewer 2 does not acknowledge that the main aim of our paper is “ASTER mineral mapping the provenance of the loess used by the Ming to build their earthen wall”. It is both the title of the paper and stated more fully in the final paragraph of the Introduction together with the sub-aims, including mapping the loess sediment transport pathways, which Reviewer 2 has singled out as the only important one.
Reviewer 2 states that “the authors successfully used well known ASTER mineral indices to identify the spatial distribution of the compositional minerals, but they did not validate their results”. Inexplicably, Reviewer 2 has completely ignored Figures 2j, 3a, 3vb, 3c, 3d, 3e, 3f, 3g, 3h, 4b, 4c, 5b, 5c, 6b, 6d, 7b and 7c. Without reason, they have dismissed our sampling/validation strategy which targeted the 1200 km length of the Ming wall, providing both a valuable cross-section across the region and helped tackle the important issues of the compositional heterogeneity of the Ming wall and the provenance of the related loess. Instead, they want us to begin a new and “Perform XRD & ASD measurements for the collected samples. Use petrography or geochemistry to identify the mineralogical composition”. Did the Reviewer 2 even read the paper and see the XRD (Figure 3h), ASD (Figures 3i, 4c, 5c, 6c, 7c), geochemical (Figures 3e, 3f, 3g) and particle size (Figures 3a, 3b, 3c and 3d) data? Maybe Reviewer 2 is suggesting the close relationships we have shown between the ASTER and field data (Figures 4b vs 4c, 5b vs 5c, 6b vs 6c and 7b vs7cv) along the length of the Ming wall cannot be used as evidence to validate the regional integrity of the ASTER mineral mosaics. If this what the reviewer is suggesting then OUR remote sensing technology counts for nought as it can never be trusted because every pixel must be validated.
Not clear on why Reviewer 2 suggests “Perform Isotopic analysis (age dating) for the collected samples from the Great Wall and from the proposed pathways”. Apart from the fact that this would make an already too long paper even longer, what value would it bring especially given that each mineral type would need to be dated and tracked across thousands of kilometres from its source rocks to sites of loess deposition. Anyway, is this really a remote sensing issue? Note how this is a completely opposite view to Reviewer 1.
Anyway, it is no surprise that Reviewer 2 states “the conclusion of this research paper is built on assumptions, not facts”. Unfortunately, Reviewer 2 has NOT provided a single example of an assumption that they are concerned with in either of their reviews to the manuscript. They just ignore ALL the results.
