Next Article in Journal
Influence of Atmospheric Scattering on the Accuracy of Laser Altimetry of the GF-7 Satellite and Corrections
Next Article in Special Issue
Satellite–Derived Topography and Morphometry for VHR Coastal Habitat Mapping: The Pleiades–1 Tri–Stereo Enhancement
Previous Article in Journal
The Influence of Aerial Hyperspectral Image Processing Workflow on Nitrogen Uptake Prediction Accuracy in Maize
Previous Article in Special Issue
Easily Implemented Methods of Radiometric Corrections for Hyperspectral–UAV—Application to Guianese Equatorial Mudbanks Colonized by Pioneer Mangroves
 
 
Article
Peer-Review Record

Very High-Resolution Satellite-Derived Bathymetry and Habitat Mapping Using Pleiades-1 and ICESat-2

Remote Sens. 2022, 14(1), 133; https://doi.org/10.3390/rs14010133
by Alyson Le Quilleuc 1,*, Antoine Collin 2,3, Michael F. Jasinski 4 and Rodolphe Devillers 5
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Remote Sens. 2022, 14(1), 133; https://doi.org/10.3390/rs14010133
Submission received: 1 November 2021 / Revised: 15 December 2021 / Accepted: 22 December 2021 / Published: 29 December 2021

Round 1

Reviewer 1 Report

The main objective of the study is to integrate ICE-Sat-2 and high-resolution Pleiades-1 imagery to derive bathymetry and benthic habitat maps. This is an interesting study. However, several key issues need to be carefully addressed before reconsideration for a possible publication. More importantly, the state-of-the-art methods need to be discussed and considered. The accuracy assessment is also unclear. The training of the habitat classification is based on samples extracted visually. This would be a very challenging, uncertain, and objective task. This would pose a substantial limitation to the applicability of the employed methodology. Here are more detailed comments:

 

Line 20: the depth range should be reported. Otherwise, the accuracy (I assume it is RMSE) is not an appropriate indication.

Lines 46-47: this is not correct in general. What you are referring to is only empirical models that need in-situ depth samples to train/calibrate a regression model. However, there are fully physics-based models that rely on the inversion of the radiative transfer model and they do not need in-situ data. There is a need to clearly discuss the physics-based and empirical models in the Introduction. You can refer to, for instance, water color simulator (WASI) or BOMBER which are physics-based models:

Physics-based Bathymetry and Water Quality Retrieval Using planetscope Imagery: Impacts of 2020 COVID-19 Lockdown and 2019 Extreme Flood in the Venice Lagoon

WASI-2D: A software tool for regionally optimized analysis of imaging spectrometer data from deep and shallow waters

BOMBER: A tool for estimating water quality and bottom properties from remote sensing images

 

Line 64-65: this is quite unclear. The calibration needs to be clarified. I assume you refer to regression-based training. The Introduction is very shallow in terms of the literature review for the SDB method. It is not clear which type of regression (empirical) model is going to be employed. The literature methods like optimal band ratio analysis (OBRA), multiple optimal depth predictors analysis (MODPA), and Sample-specific multiple band ratio techniques for satellite-derived bathymetry (SMART-SDB) need to be discussed.

 

Also, water-column correction and substrate mapping literature need to be discussed.

 

Dataset: the research is based on only one cases study and could get benefit from others to enhance the reliability of findings and show its applicability in different sites.

 

Airborne In-Situ data sounds a bit strange as airborne data can not be considered as in-situ (field) measurements. It’s still remote sensing data that here relies on LiDAR sensor. Therefore, it is essential to assess the results based on in-situ data and not from another remote sensing source.

 

It seems that atmospheric correction is neglected in this study. The reason needs to be clarified and any possible impact to be discussed.

 

Could you specify the time gap between ICE-Sat2 and optical data acquisitions in terms of days (hours, etc.)?

 

DBSCAN method is proposed in this study or it is from the literature? If from literature, it needs citation.

Section 2.3: as mentioned above, state-of-the-art methods must be considered for depth retrieval. OBRA, MODPA, SMART-SDB, and other methods can be employed.

 

Section 2.3.3: other accuracy assessment metrics like bias also need to be added.

 

Line 290: what is the classification in ENVI? The descriptions of the methods need to be independent of specific software.

Section 2.4.2: selection of training samples for benthic habitats sounds a very challenging and uncertain task. This would be a key limitation on the applicability of the methodology considered in this study.

 

Fig.8: it is not clear which band ratio is considered and how it is selected.

 

Figure 9: the color bar needs to have more divisions to better understand the depth values represented by different colors.

 

Section 4.1: This title is unclear.

 

Line 490: how do you conclude that the RMSE in your study is lower than other studies? Her, only the impact of spatial resolution is studied. What about spectral and radiometric resolution?

 

Line 493: This is not a valid comparison to take the RMSE from other studies conducted in other case studies and compare it with your RMSE. the range of water depth and optical conditions are variable from a case study to another.

I couldn’t find the bathymetry retrieval accuracy assessment based on independent in-situ data.

The references are very shallow.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Given the absence of groundtruth benthic habitat data for the study area, the paper is extremely sound.

Page 10: "...brown and dark..." sounds unclear. Brown can be dark too. 
Page 12: It is not clear why the second degree polynomial was chosen
for comparison between Pleiades and ICESat-2 bathymetric measurements.
Page 14: The use of the support vector machine classification is not described with sufficient details. Were the authors using default ESRI parameters? Did they use the Gaussian kernel?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This manuscript uses two forms of remotely sensed data, lidar and multispectral imagery, to map bathymetry and bottom type in a coastal setting.  This is an important topic with a long history in the remote sensing community, but I don't see much, if anything, novel reported in this study.  Existing techniques in commercial software are applied and the results are not particularly impressive.  I have a number of concerns detailed in the attached PDF, along with many minor edits and requests for clarification.  The paper needs a significant revision but could be reconsidered if the authors can more clearly articulate what is new and distinct about their work and address all of the issues I've identified in the PDF.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I had provided detailed comments in the first round of revision. Most of them are addressed by the authors satisfactorily. I have only one concern now about the ratio model as the authors mention that the selection of bands is done by the software (ENVI) and they don't know how the method works! Implementation of a ratio model does not require being tied to specific software and can be easily implemented in any programming language. This is important to understand whether the optimal pair of bands are used for depth retrieval or not. Moreover, other techniques that they exploit all the spectral bands would bring more benefit to this study. Here, I mention the reply of the authors which is not clear:

"The software computed a ratio of two spectral bands, but the detail selection is not made explicit to the user."

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

I appreciate the thorough response document provided by the authors, who have satisfactorily addressed my concerns for the most part.  I have made a few more minor edits on the attached PDF that will need to be incorporated before the paper can proceed. In addition, the R2 values of the various regressions need to be reported, not just the RMSE.

Comments for author File: Comments.pdf

Author Response

The authors would like to thank you again for your corrections. We changed the minor edits in the manuscript, and we added the R² values of the various regressions in addition to the RMSE.

Back to TopTop