Evaluation of Satellite-Derived Bathymetry from High and Medium-Resolution Sensors Using Empirical Methods

Round 1

Reviewer 1 Report

My comments and suggestions were addressed. The present revised version of the manuscript seems better than the previous one.

Some typos to correct:

TEXT

L86: km2 is km²

L140: "2x2 or 3x3 10x 10 m" is "2x2 or 3x3 or 10x 10 m" ? Check it.

FIGURES

Figures 1, 2, 4, 6, and 7: to the right of each graphic scale, "Km" is "km", that is with the letter k lowercase.

 

Author Response

Point to point respond

Evaluation of Satellite-Derived-Bathymetry from High and Medium Resolution Sensors using Empirical Methods

Evagoras Evagorou ¹›²*, Athanasios Argyriou³, Nikos Papadopoulos³, Christodoulos Mettas¹›², George Alexandrakis ⁴ and Diofantos Hadjimitsis ¹›²

Reviewer 1

Comments

• Thank you for your comments, appreciated. Your suggestions were considered, and corrections have been made accordingly.

L86: km2 is km²

• Done.

L140: "2x2 or 3x3 10x 10 m" is "2x2 or 3x3 or 10x 10 m" ? Check it.

• Done.

FIGURES

Figures 1, 2, 4, 6, and 7: to the right of each graphic scale, "Km" is "km", that is with the letter k lowercase.

• Done.

Author Response File: Author Response.docx

Reviewer 2 Report

Line 86: 2 should be superscript

Line 130: pitch - roll & yaw -> pitch, roll and yaw

Line 203: -> where, ?????? and ???? represent the band values of the reflected red and green bands, respectively.

Some places of the word of "coefficient" should be "coefficients", e.g. Lines 261-288

Suggestion: using "images" instead of "imageries"?

 

Author Response

Point to point respond

Evaluation of Satellite-Derived-Bathymetry from High and Medium Resolution Sensors using Empirical Methods

Evagoras Evagorou ¹›²*, Athanasios Argyriou³, Nikos Papadopoulos³, Christodoulos Mettas¹›², George Alexandrakis ⁴ and Diofantos Hadjimitsis ¹›²

Reviewer 2

Comments and Suggestions for Authors

 

• Thank you for your comments, appreciated. Your suggestions were considered, and corrections have been made accordingly.

 

Line 86: 2 should be superscript

Line 130: pitch - roll & yaw -> pitch, roll and yaw

Line 203: -> where, ?????? and ???? represent the band values of the reflected red and green bands, respectively.

Some places of the word of "coefficient" should be "coefficients", e.g. Lines 261-288

Suggestion: using "images" instead of "imageries"?

• Your suggestion has been taken into account for the manuscript, and corrections have been made accordingly

Author Response File: Author Response.docx

Reviewer 3 Report

The authors have significantly improved their manuscript, especially in the section discussing their results. Please find below my last remarks concerning the manuscript.

Line 93: Please change “was estimated” to “were estimated”.

Line 98: Please add a space between “sensors” and “(Landsat…”.

Line 130: The authors did not make any corrections for wave heights. This could be mentioned at the end of line 130 after “yaw”.

Line 142: Please change “this” to “these” as you are using plural (“were”) for data.

Equation 6: Please provide a reference for this equation as it is different than Equation 9 provided by Stumpf and Holderied (2003) or correct it if necessary.

Line 328: “Reference not found”

Line 368: Please use a better word than “contributed” or use better syntax.

Line 400: Please delete “the” before “Figure 8”.

Line 409: Please change “d)” to “(d)”.

Lines 440-445: Please change the last paragraph of the conclusions section to reflect the changes made in the discussion.

Author Response

see comments below.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This manuscript describes the methodology, accuracy, and results of bathymetric maps generated from multispectral satellite datasets (SDB) acquired from three different sensors, the result of which are each other compared.

The manuscript is well structured, the methods are simply explained, math formulas seem to work well. Results, Discussion, and Conclusion sections are congruent. Figures and tables are fine, citations and references are sufficient.

Anyway, some more general information and related citations could be added in the Introduction section about the comparison between SDB and common on-field oceanographic techniques, i.e., I mean numbers of accuracy range of both methodologies, not only sentences on the advantages of the SDB related to the possibility of carrying on large-area surveys, in a short time, and at low-cost balance. This comparison should be shortly resumed also in the Discussion and Conclusions sections.

TEXT

In the whole text, the word "meters" following a number should be "m": change them. This is usual for any unit in Metrology.

2.1 Study area: in this section, a short geomorphological description of the studied bay (shape, coves, beaches, watercourses, etc.), located in the NW sector of the large Mirabella Gulf, could help not Greek readers to better understand the relation between emerged and submerged landscape.

2.2 In-situ depth Data: there is any tidal correction? Explain if you did it and how, e.g., evaluating tidal oscillations (even if in the Mediterranean Sea they are low) registered by a (local, national?) gauge station (which one and where?) in the hours of days (which ones and when?) of bathymetric surveys. This is a routine pre-processing operation, together with calibration of the speed of sound in water, important to reduce the errors before the depth map model construction.

The absence of tidal corrections could generate an error (even if little) in the correlation between observed depths (validation points) and extracted depths obtained by algorithms (see Results and Figure 5), which could explain the generally low R² values (0.37-0.76).

L290-291: eliminate the carriage return in the sentence.

FIGURES

Figure 1: add in the map a few toponyms, e.g., Elounda, Spinalonga Peninsula (that is Xερσòνησoς Kolokúθα), Aegean Sea, etc.

Figures 4, 6, and 7: In these three figures, add the graphic scale, as it is already in Figs. 1 and 2.

TABLES

The Tables work well.

Reviewer 2 Report

In this research, three most popular SDB retrieval methods were applied and compared. In my knowledge, these methods were extensively
studied already. Most the steps applied in this research are already in common practice, and there is nothing new in terms of how to apply those methods, the design of the experiments and the conclusions.

 

Some comments:

Single Band Linear Algorithm (SBLA), Multi-Band Linear Algorithm (MBLA) and Ratio Transform Algorithm (RTA) should be first mentioned in text, not in Table 2 caption.

The paper says that 80% of 5665 points were used to fit the models, which is a very big number to estimate only a few model coefficients. Small sets of data should be used and analyzed in my opinion. And when looking at the yellow colors in Figure 2, the majority of the validation and fitting points is very close each other, the selection of validation points should avoid such scenarios.

L373: "The results show that in shallow water from 0-2 meters depth, the accuracy of the estimated bathymetry depends significantly on turbidity." However, in my view it's very hard for the readers to establish such links and make the same conclusions when looking at the turbidity maps (Figure 6), error maps (Figure 7) and error diagrams (Figure 8).

 
from L281 to L291: The paragraphs are unreadable

L343: Figure 8's colors are not good for the printing paper.

When come to evaluate bathymetry models, the following recent paper is worth to read:
Review of near-shore satellite derived bathymetry: Classification and account of five decades of coastal bathymetry research
by Mohammad Ashphaq, Pankaj K Srivastava, D Mitra, Journal of Ocean Engineering and Science 6 (2021) 340–359

Reviewer 3 Report

The manuscript deals with the comparison of different methods for deriving depths from satellite imagery in a study area in the island of Crete. In-situ measurements are used for the derivation of the model parameters, while turbidity, derived from the satellite imagery is also examined together with the error estimates of the derived heights.

 

The manuscript deals with an application of known methodologies in the study area and comparisons are made in order to draw conclusions on the examined methods and the use of different imagery. Please find below my remarks and comments below. Based on them I propose a major revision of the manuscript.

 

Abstract

The concluding sentence should be consistent with what was mentioned earlier, i.e., state either WorldView 2, PlanetScope or Sentinel 2 instead of “high-resolution sensor” and one of the three already mentioned empirical models instead of “analogy algorithm”.

 

Keywords

Please capitalize the first letter for “Ratio transform algorithm” to be consistent with the other. Please also check the rest of the text.

 

Introduction

---------------

Line 39: Please change “for understanding of coastal systems evolution” to “for understanding the evolution of coastal systems” or delete “of”.

 

Lines 42-45: Please split the sentence in two at line 44, e.g., “These are commonly used…”.

 

Line 52: The inclusion of satellite-derived bathymetry in the production of navigation maps for coastal or low-depth areas is dangerous and should be avoided! For example, if we have an error of 3 m in estimating a depth of 8m then this is not a reliable result for navigation. Please delete this statement.

 

Line 85-94: Since you mentioned the results of the first research then you should do this also for the second one or, to make it shorter, summarize only the best results achieved.

 

Line 97: Please use the term section instead of “part”. The phrase “part of the dataset” is confusing.

 

Study area and Materials

------------------------------

Line 107: Please delete 35N as it does not apply to a geodetic reference system or state “UTM Zone 35N”.

 

Lines 107-108: Please use better English.

 

Line 113: Delete the second full stop.

 

Figures: Please make sure that the images are of adequate quality (resolution) as in the reviewer’s pdf this is not possible to validate.

 

Line 118: Probably it is meant: “with two TOPCON GNSS receivers using the Real Time Kinematic technique”.

 

Line 119: Please use “m” instead of “meters” and check also the rest of the manuscript for consistency.

 

Line 119: How did the authors validate the accuracy of their measurements? I understand that for the GNSS receiver the value provided is a measure of the internal accuracy of the measurements but for depth, how was the estimate of 0.1 m obtained?

 

Line 121: They are not exactly “random”. They follow the course of the vessel. Please delete “random”.

 

In-situ depth Data: What is the reference system for the depth values and how were the measurements referred to it?

 

In-situ depth Data: What corrections were applied to the measurements? Did the authors take into account tides?

 

Methodology

-----------------

Line 144: Please replace the verb “contribute” as the meaning is opposite.

 

Lines 166-167: Please rephrase to state correctly that cloud removal and shading masks weren’t required and therefore not created.

 

Line 178: Please change “linear interpolating” to “linearly interpolating”.

 

Lines 166-185: The authors mention initially the land/sea separation mask, then mention turbidity and then again the masking land/sea along with a filtering process. Please revise the text to show that filtering comes after the computation of NDWI and NDTI and not only after masking (in accordance with the workflow).

 

Equation 5: The symbol z is different then Zsdb mentioned next in the explanation of symbols.

 

Equations 6: How were the parameters of equation 6 estimated when the measured depth is on the right side of the equation?

 

Equation 7: Many papers in the literature use this absolute error parameter. The absolute value hides though the real distribution of errors as the sign is important (-/+). As this is my opinion, it is up to the authors to retain it.

 

Line 223: Please rephrase “The diagram below (Figure 3)” to “Figure 3”.

 

Results

----------

Lines 230: Incomplete sentence? Probably should be connected with the previous sentence.

 

Line 233: Please cite Vahtmäe and Kutsehave properly.

 

Lines 243-245: How can a correlation of 0.4 considered high?

 

Line 243: Please don’t use above/below but refer to the Table/Figure number instead. Please check this throughout the manuscript.

 

Line 281-283: Please rephrase it as it appears as figure caption. Same for lines 289-291.

 

Lines 289-294: This was already mentioned in lines 285-286 and does not add anything new. It may be deleted.

 

Lines 295-299: Linear interpolation

 

Line 298: Please correct the Error.

 

Figure 7: Linearly interpolating point values that do not cover the whole area does not lead to reliable plots. It would be preferable to use a color scale for the point distribution.

 

Discussion

--------------

Line 315-317: Even if there are “fine-grained sediments” at the bottom, errors of 1 or more meters would be a rare case. These kind of errors that might occur in specific point values can be screened and removed during the processing of echo-sounding data. Doubting about the accuracy of the echo-sounding data defeats the whole purpose of the authors research as all the models rely on these data. Unreliable in-situ data means unreliable results as all models are based on the in-situ data.

 

Line 323: How is turbidity related to the estimated errors? Looking at Figures 6 and 7, the high turbidity areas do not correspond to high error areas. Moreover, Figure 8 also does not justify this as for 2-4 m depth the turbidity is higher than 0-2 m depth range but the error estimates are lower. Especially for Planetscope, one can see that although turbidity is the same for ranges 4-6, 6-8 and 8-10 m, the error estimates for 8-10 m is very high, which means that it is not a matter of turbidity but of the algorithm followed for deriving the bathymetry model.

 

Line 331: Regarding the sensor resolution, how can it affect the results when comparing Worldview-2 and Sentinel 2? Moreover, the authors applied a median filter 3x3, which means that they changed the resolution of the processed values, thus having a lower resolution. Did this have any impact?

 

Line 352-353: In hydrography depths it is wrong to make error comparisons among different ranges of depth. This is the reason also that echo-sounding devices provide error estimates as a percentage of the measured depths. If the authors want to examine the errors, they must examine them as a percentage of the measured error and not in absolute values. For example, having a range of depth of 0-2 m and an error of 1 m, means that this can be from 50% to 100% error. These values are meaningless. On the other hand an error of 1 m for depths of 100-200m is obviously far less and corresponds to 0.5% to 1%.

 

Conclusions

---------------

Please adapt this section accordingly after taking into account the previous remarks.

 

References

-------------

Please correct references: 27, 36, 37, 40 and 49.