Airborne Hyperspectral Imaging for Submerged Archaeological Mapping in Shallow Water Environments

Round 1

Reviewer 1 Report

In this study, the authors bring new perspectives and provide successful case for documenting submerged archaeological stone structures using airborne hyper-spectral remote sensing data and multi-step processing method.

Airborne hyperspectral data were recorded in the VNIR spectral range (400-1000 nm) over the submerged megalithic site of Er Lannic (Morbihan, France). The method used to process these data included (i) visualization of submerged anomalous features using a minimum noise fraction transform, (ii) automatic detection of these features using Isolation Forest and the Reed–Xiaoli detector and (iii) morphological and spectral analysis of archaeological structures from water-depth and water-bottom reflectance derived from the inversion of a radiative transfer model of the water column.

The authors address the potential role of aerial hyperspectral imaging for the documentation, detection and conservation of submerged sites in coastal archeology. The structure and the rhythm of the manuscript are suitable, as well as the methodological approach and the discussion, and are suited to the objectives of the subject addressed. This article is clearly part of a broader study, because reference is made to the analysis of further acoustic data, such as MBES data and bathymetric LiDAR products.

The strong point is certainly the approach to the theme, that is the study of hyperspectral image as a source of possible archaeological data, in particular as a basic material for discovering the buried coastal monuments, for the conservation of cultural heritage. This is supported by LiDAR-based archaeological surveys, which prove how aerial remote sensing can be a source of information for reconstructing the archaeological landscape.

Furthermore, for the development of the research it could be an excellent archaeological remote sensing approach, in the submerged geographic conditions, the use of a high spectral and spatial resolution passive remote sensing data, for the identification of monuments and a comparison with the known data.

The article is well structured and the theme developed clearly and with impeccable methodology. In my opinion, we could improve the Section “1 Introduction” by moving the sub-sections “1.1 Dimensionality reduction and visualization, 1,2 Unsupervised anomaly detection and 1.3 Radiative transfer model over shallow water” to the Section 3 Methodology. Additionally, it will be highly appreciated if the authors can directly point out the research aims and goals in end of the Section Introduction.

Author Response

Dear reviewer,

Thank you very much for the thoughtful comments and suggestions.

We trust that all of your comments have been addressed accordingly in the revised manuscript.

In the following, we give a point-by-point answer to your comments :

In this study, the authors bring new perspectives and provide successful case for documenting submerged archaeological stone structures using airborne hyper-spectral remote sensing data and multi-step processing method.

Airborne hyperspectral data were recorded in the VNIR spectral range (400-1000 nm) over the submerged megalithic site of Er Lannic (Morbihan, France). The method used to process these data included (i) visualization of submerged anomalous features using a minimum noise fraction transform, (ii) automatic detection of these features using Isolation Forest and the Reed–Xiaoli detector and (iii) morphological and spectral analysis of archaeological structures from water-depth and water-bottom reflectance derived from the inversion of a radiative transfer model of the water column.

The authors address the potential role of aerial hyperspectral imaging for the documentation, detection and conservation of submerged sites in coastal archeology. The structure and the rhythm of the manuscript are suitable, as well as the methodological approach and the discussion, and are suited to the objectives of the subject addressed. This article is clearly part of a broader study, because reference is made to the analysis of further acoustic data, such as MBES data and bathymetric LiDAR products.

The strong point is certainly the approach to the theme, that is the study of hyperspectral image as a source of possible archaeological data, in particular as a basic material for discovering the buried coastal monuments, for the conservation of cultural heritage. This is supported by LiDAR-based archaeological surveys, which prove how aerial remote sensing can be a source of information for reconstructing the archaeological landscape.

Furthermore, for the development of the research it could be an excellent archaeological remote sensing approach, in the submerged geographic conditions, the use of a high spectral and spatial resolution passive remote sensing data, for the identification of monuments and a comparison with the known data.

The article is well structured and the theme developed clearly and with impeccable methodology.

Thank you very much for this comment.

In my opinion, we could improve the Section “1 Introduction” by moving the sub-sections “1.1 Dimensionality reduction and visualization, 1,2 Unsupervised anomaly detection and 1.3 Radiative transfer model over shallow water” to the Section 3 Methodology.

We modified the introduction section by moving 1.1, 1.2 and 1.3 to the “Materials and methods” section.

Additionally, it will be highly appreciated if the authors can directly point out the research aims and goals in end of the Section Introduction.

We directly pointed out the research aims and goals in the end of the introduction section (Lines 79-82) as follows: “Thus, the aim of this study was to evaluate the potential of AHI for detecting and documenting submerged sites in coastal archeology. The questions addressed are (a) can submerged archaeological structures be visible using AHI? (b) Can they be detected automatically? (c) Can they be characterized spatially and spectrally?”.

We really hope theses answers and related modifications can meet with your approval.

Kind regards,
Alexandre Guyot

Reviewer 2 Report

The research presented here is truly novel (and I might even say, inspired) and will be of great interest to the archaeological prospection community. Though the authors remark on certain limitations of the technique with respect to feature characterization (most specifically, the ability to discern between cultural or natural benthic features), I would add that there is significant value regarding the technique's ability to reveal patterns (which could be cultural in nature). This research will no doubt open new pathways for non-acoustic prospection in marine archaeology. 

A few very minor editorial suggestions:

You might consider changing "Imagery" in the title to "Imaging" (since the later refers more the technique discussed).

The term "water-bottom" is not common in the lexicon. (Note that it appears at least twice in the manuscript). I would suggest: seafloor or seabed. "Benthic" could be used as an adjective. 

Note that there is a formatting problem with the parenthetical references to Figure 6d, 6e, and 6f).

Author Response

Dear reviewer,

Thank you very much for the thoughtful comments and suggestions.

We trust that all of your comments have been addressed accordingly in the revised manuscript.

In the following, we give a point-by-point answer to your comments:

The research presented here is truly novel (and I might even say, inspired) and will be of great interest to the archaeological prospection community. Though the authors remark on certain limitations of the technique with respect to feature characterization (most specifically, the ability to discern between cultural or natural benthic features), I would add that there is significant value regarding the technique's ability to reveal patterns (which could be cultural in nature). This research will no doubt open new pathways for non-acoustic prospection in marine archaeology.

Thank you very much for your comment.

A few very minor editorial suggestions:

You might consider changing "Imagery" in the title to "Imaging" (since the later refers more the technique discussed).

We modified the title that is now “Airborne hyperspectral imaging for submerged archaeological mapping in shallow water environments”

The term "water-bottom" is not common in the lexicon. (Note that it appears at least twice in the manuscript). I would suggest: seafloor or seabed. "Benthic" could be used as an adjective.

“Water-bottom” is indeed less used than “seafloor” or “seabed” in coastal shallow-water studies (about 1 to 3 ratio), however it was chosen for its broader and more general usage (including inland water for example).

Note that there is a formatting problem with the parenthetical references to Figure 6d, 6e, and 6f).

Thank you for pointing out the formatting issue in figure 6. It was corrected.

We hope theses answers and related modifications can meet with your approval.

Kind regards,
Alexandre Guyot

Reviewer 3 Report

This technique has huge potential in remotely visualising already known archaeological sites. The results are really interesting and clear, and I am excited to read more about the potentials of AHI in exposing other coastal/submerge archaeological sites and paleo-landscapes in the near future.

However, if what I understood is correct, in order to develop an elevation model the analysis requires already know numerical data. In this case, past data of ALB/MBES were used to associate pixels with elevation values. From that the analysis works out the rest. If I am correct, this is then a major drawback as the propose technique is not fully non-intrusive as it requires some form of ground truthing and supplementary data.

Even though the technical information provide here, to me seems sound and the main argument is a valid one, perhaps a paragraph could be added, explaining the technique in a simpler, non-technical manner for readers lacking the technical knowledge/background, such as myself.

AHI is really interesting, but I have my doubts if it can work as a survey tool for discovering submerge archaeological sites and palaeo-landscapes in the absence elevation values. The models presented here are clear and I believe maritime archaeologists will find this technique quite appealing in mapping/viewing, submerge harbours in particular.

Author Response

Dear reviewer,

Thank you very much for the thoughtful comments and suggestions.

We trust that all of your comments have been addressed accordingly in the revised manuscript.

In the following, we give a point-by-point answer to your comments :

This technique has huge potential in remotely visualising already known archaeological sites. The results are really interesting and clear, and I am excited to read more about the potentials of AHI in exposing other coastal/submerge archaeological sites and paleo-landscapes in the near future.

Thank you very much for this comment.

However, if what I understood is correct, in order to develop an elevation model the analysis requires already know numerical data. In this case, past data of ALB/MBES were used to associate pixels with elevation values. From that the analysis works out the rest. If I am correct, this is then a major drawback as the propose technique is not fully non-intrusive as it requires some form of ground truthing and supplementary data.

Even though the technical information provide here, to me seems sound and the main argument is a valid one, perhaps a paragraph could be added, explaining the technique in a simpler, non-technical manner for readers lacking the technical knowledge/background, such as myself.

The methods used for the derivation of elevation / depth from multi or hyperspectral imagery are classically classified in two different families: empirical and radiative-transfer-based methods. While empirical methods require prior elevation data to infer water depth from remote-sensing imagery (usually by linear regression), radiative-transfer–based methods do not (see Dekker et al. 2011). In our study, a radiative-transfer–based method was used. Such a method is based on the radiative transfer theory, using a model that describes the relationship between the observed surface reflectance (remotely sensed spectral measurements) and the physical properties of the water column (including depth, bottom reflectance and optically active constituents in the water). The retrieval of these properties, through model inversion and optimization, is independent of external numerical data such as reference bathymetry data or in-situ depth calibration data. While in-situ data can be used for reducing bias or assessing estimated bathymetry, it is not required for deriving the elevation/depth. In the present study, existing bathymetric data (Litto3D product) was only used as reference for assessment and visual comparison (see Figure 9 and Figure 10).

For better clarity, we added some explanations in sections 2.7 and 3.6 as follows:

Line 216-218 : « This radiative transfer–based method, unlike empirical approaches, has the advantage of not requiring prior bathymetric data (an existing elevation model or in-situ measurements) for the inversion process and derivation of bottom depth [21]. »

Line 307-308 : « No external data (in-situ depth measurements or an existing elevation model) was used for refining the bottom-depth estimation. »

AHI is really interesting, but I have my doubts if it can work as a survey tool for discovering submerge archaeological sites and palaeo-landscapes in the absence elevation values. The models presented here are clear and I believe maritime archaeologists will find this technique quite appealing in mapping/viewing, submerge harbours in particular.

The above clarification may also partially answer the comment regarding the replicability of the method for discovering submerged archaeological sites and palaeo-landscapes. We do not have an answer to this replicability concern yet, but this first study opens some interesting perspectives towards shallow water archaeological prospection on different areas (with or without existing reference data) and for different types of sites (different typologies, different chronologies). Such perspectives were stated in section 5.4.

We hope theses answers and related modifications can meet with your approval.

Kind regards,
Alexandre Guyot

Reviewer 4 Report

General comments: This is an interesting paper that describes the employment of a well-known remote sensing method in a new environment. The test is thoroughly described and the overall impression is that this is a solid study. The results are more promising than convincing and ideally AHI should be tested on more submerged archaeological sites in different environments and geographical regions. This also include lakes. The characters of water are heterogeneous and of highly relevance in terms of the applicability of this approach. This is a first test, and as such it is an important first step in the direction of a further exploration of AHI in rugged shallow water environments.

Specific comments:
line 42: Add the word increasingly....is consequently increasingly becoming.... It has been so for quite a long time already.

line 58: I would say it is an exaggeration to say that ALB has gained great interest. Only few projects has been carried out with modest results. I suggest to replace great interest with some interest.....

line 60-62: Here you should mention more specifically the challenges related to water quality, water surface etc.

line 64: ...have so far focused......  Adding these words elucidate that you now bring in something new.

line 74-77:

line 96: What you mean with Unsupervised should be shortly defined in the beginning of this paragraph.

line 106: When referring to a specific paper I suggest that you mention names: ...developed by Reed and Yu [33].....

line 110: Same comment as regarding line 106.

line 149: Reference to G. de Closmadeuc is missing.

line 172: ....images were collectedd in clear sky and calm sea conditions. Please mention if this is a precondition.

line 193: Consider to add: Cf. Figure 1b. to the caption. It will increase readability.

line 210: I think that the word workflow better describes this than method. Also, it is unclear what you mean with ...axes sponding to the..... Please redraft this sentence in order to clarify what you mean.

line 210-215: These are the objectives you address in this paper and should appear at the beginning of the article, for instance after line 77.

line 218: Consider if it is possible to make a clearer connection between Figure 4 and the text in chap. 3. Methodology for instance by numbering the various steps in the diagram.

line 319-323: Please check if the numbers and letters are correct.

line. 327: ...many natural ground features. Here you should mention some of these.

line 337-338, 341, 348: Error!

line 370 - 374: Consider a table instead of text with all these numbers.

line 407-420: This paragraph should include and refer to existing research as a part of the discussion.

line 438-451: A comparison to ALB would be appropriate here.

line 458: ....heritage preservation plans. Perhaps replace preservation plans with the word management.

Author Response

Dear reviewer,

Thank you very much for the thoughtful comments and suggestions.

We trust that all of your comments have been addressed accordingly in the revised manuscript.

In the following, we give a point-by-point answer to your comments:

General comments: This is an interesting paper that describes the employment of a well-known remote sensing method in a new environment. The test is thoroughly described and the overall impression is that this is a solid study. The results are more promising than convincing and ideally AHI should be tested on more submerged archaeological sites in different environments and geographical regions. This also include lakes. The characters of water are heterogeneous and of highly relevance in terms of the applicability of this approach. This is a first test, and as such it is an important first step in the direction of a further exploration of AHI in rugged shallow water environments.

Thank you for this comment.

Specific comments:

line 42: Add the word increasingly....is consequently increasingly becoming.... It has been so for quite a long time already.

We modified the text as suggested.

line 58: I would say it is an exaggeration to say that ALB has gained great interest. Only few projects has been carried out with modest results. I suggest to replace great interest with some interest.....

We modified the text as suggested.

line 60-62: Here you should mention more specifically the challenges related to water quality, water surface etc.

We modified the text (lines 77-78)  as follows:

:” … underwater mapping requires addressing challenges related to the complexity of (i) the data (including high dimensionality and signal-to-noise ratio), (ii) the object of study (degraded and partially documented structures) and (iii) the environment, especially the complex light-matter interactions in water, affected by multiple environmental factors such as water constituents, surface conditions or benthic composition.”

line 64: ...have so far focused......  Adding these words elucidate that you now bring in something new.

We modified the text as suggested.

line 74-77:

line 96: What you mean with Unsupervised should be shortly defined in the beginning of this paragraph.

A short description of “unsupervised” was added at the beginning of the paragraph (lines 182-183) as follows: “In the machine learning field, unsupervised learning is the task of identifying structures or relationships in the input data without prior knowledge by mean of reference or labelled data.”

line 106: When referring to a specific paper I suggest that you mention names: ...developed by Reed and Yu [33].....

Names were added to specific papers (by Reed and Yu [33], by Liu et al. [34]).

line 110: Same comment as regarding line 106.

Names were added to specific papers (by Reed and Yu [33], by Liu et al. [34]).

line 149: Reference to G. de Closmadeuc is missing.

The following reference was added : de Closmadeuc, G. Découverte d’un cromlec’h dans l’île d’El Lanic (Morbihan). Bull. Société Polymath. Morbihan 1867, 28–30.

line 172: ....images were collectedd in clear sky and calm sea conditions. Please mention if this is a precondition.

This was clarified (lines 125-126) as following: “During the survey, images were collected in clear sky and calm sea conditions (preconditions to reduce sun-glint effects and solar irradiance variation)”.

line 193: Consider to add: Cf. Figure 1b. to the caption. It will increase readability.

(a) and (b) were already included in the caption of Figure 1.

line 210: I think that the word workflow better describes this than method. Also, it is unclear what you mean with ...axes sponding to the..... Please redraft this sentence in order to clarify what you mean.

This section was renamed to “Workflow” and the sentence redrafted as follows (Lines 223-224): “The methodology developed and the associated workflow (Figure 4) were organized in subsections corresponding to the research objectives of this study.”

line 210-215: These are the objectives you address in this paper and should appear at the beginning of the article, for instance after line 77.

Aims and objectives of the paper were added at the end of the introduction section (Lines 79-82 as follows: ”Thus, the aim of this study was to evaluate the potential of AHI for detecting and documenting submerged sites in coastal archeology. The questions addressed are (a) can submerged archaeological structures be visible using AHI? (b) Can they be detected automatically? (c) Can they be characterized spatially and spectrally?”.

line 218: Consider if it is possible to make a clearer connection between Figure 4 and the text in chap. 3. Methodology for instance by numbering the various steps in the diagram.

The subsections of chap. 3 Methodology were renamed so that a clear connection can now be made between the diagram workflow and the text.

line 319-323: Please check if the numbers and letters are correct.

The caption was corrected.

line. 327: ...many natural ground features. Here you should mention some of these.

The sentence was enriched as follows (Line 337) :  “… many natural ground features (e.g. sands, granitic rocks, dry algae, grass).”

line 337-338, 341, 348: Error!

The reference sources were corrected.

line 370 - 374: Consider a table instead of text with all these numbers.

The information provided in text is in Table 4, which contains complementary information.

line 407-420: This paragraph should include and refer to existing research as a part of the discussion.

The following references to research were added to the discussion section (lines 419, 421, 426).

Dekker, A.G.; Phinn, S.R.; Anstee, J.; Bissett, P.; Brando, V.E.; Casey, B.; Fearns, P.; Hedley, J.; Klonowski, W.; Lee, Z.P.; et al. Intercomparison of shallow water bathymetry, hydro-optics, and benthos mapping techniques in Australian and Caribbean coastal environments: Intercomparison of shallow water mapping methods. Limnol. Oceanogr. Methods 2011, 9, 396–425

Bajjouk, T.; Mouquet, P.; Ropert, M.; Quod, J.-P.; Hoarau, L.; Bigot, L.; Le Dantec, N.; Delacourt, C.; Populus, J. Detection of changes in shallow coral reefs status: Towards a spatial approach using hyperspectral and multispectral data. Ecol. Indic. 2019, 96, 174–191.

Gao, J. Bathymetric mapping by means of remote sensing: methods, accuracy and limitations. Prog. Phys. Geogr. Earth Environ. 2009, 33, 103–116.

Göritz, A.; Berger, S.A.; Gege, P.; Grossart, H.-P.; Nejstgaard, J.C.; Riedel, S.; Röttgers, R.; Utschig, C. Retrieval of Water Constituents from Hyperspectral In-Situ Measurements under Variable Cloud Cover—A Case Study at Lake Stechlin (Germany). Remote Sens. 2018, 10, 181.

line 438-451: A comparison to ALB would be appropriate here.

The following  sentence was added to highlight the capability of benthic cover types characterization for AHI compared to ALB (Lines 461-464): : “Nevertheless, a first characterization of the bottom spectral properties is a valuable information offered by AHI and an advantage over ALB for which the backscattering intensity of a single wavelength signal is less adapted for the description of benthic composition [55].”

[55]        Pan, Z.; Glennie, C.; Fernandez-Diaz, J.C.; Starek, M. Comparison of bathymetry and seagrass mapping with hyperspectral imagery and airborne bathymetric lidar in a shallow estuarine environment. Int. J. Remote Sens. 2016, 37, 516–536.

line 458: ....heritage preservation plans. Perhaps replace preservation plans with the word management.

We replaced “heritage preservation plans” by “cultural heritage management”.

We hope theses answers and related modifications can meet with your approval.

Kind regards,
Alexandre Guyot

Reviewer 5 Report

This is a very high quality research article. Spectral imagery are normally used in ground investigation. This is the only, or one of the few occasions, where spectral sensors have been applied to underwater archaeological domaine. This is a high novelty on the archaeological remote sensing research.
Summary is well organized; may be a little long.
Background, material and methods are clearly exposed.

Results are are really great.

I'm really glad to evaluate this article that i recommend for publishing in present form. 

Author Response

Dear reviewer,

Thank you very much for the positive and thoughtful comments.
It is highly appreciated.

Kind regards,
Alexandre Guyot