Mapping Benthic Habitats by Extending Non-Negative Matrix Factorization to Address the Water Column and Seabed Adjacency Effects

Round 1

Reviewer 1 Report

The present paper describes two original non-negative matrix factorization-based unmixing algorithms for mapping of sea weeds community under water column. Authors describe methods and results in detail and the algorisms are fairly accurate. The paper is acceptable after minor revision.

Lines 163-164: Definition of clear water condition should be corrected.

Lines 326, 467, 473, and legend to Figure 5: Species name ‘Taxifolia’ should start from lower case letter.

Legend to Figure 4: ‘2’ of ‘K2’ should be subscript.

Lines 461-462: Abbreviations ‘Chl’, ‘SM’ and ‘CDOM’ should be explained before Figure 1 (the first appearance). Does ‘Chl’ mean chlorophyll-a?

Legend to Figure 6: ‘sustratum’ should be ‘substratum’.

Figure 8 (a): There is a vacancy at Pixel = ca. 85. Explanation is necessary.

Lines 545 and 558: Does ‘photophilic algae’ mean ‘phytoplankton’ or simply ‘algae’?

Discussion: Authors are suggested to add more detailed explanation on the difference between WUM and WADJUM with reference to the ground truth data.

Author Response

Editor comments 

Decision: Accept after minor revision

Comments: Both reviewers found the manuscript to be of high quality. The article should be accepted after minor revisions from Reviewer #1 and #2 are made.

We thanks the editor for the selection of reviewers and very fast review.

We are very grateful to the two reviewers for the time and energy they devoted to reading and analyzing our work, and their relevant comments and suggestions.

We provide below a point-by-point response.

Reviewer#1 

The present paper describes two original non-negative matrix factorization-based unmixing algorithms for mapping of sea weeds community under water column. Authors describe methods and results in detail and the algorisms are fairly accurate. The paper is acceptable after minor revision.

We are honored by your appreciation.

• Lines 163-164: Definition of clear water condition should be corrected.

Yes, this is a mistake, the clear water conditions are defined in our work as [ Chl= 0.03  mg.m^-3, SM = 0.01 mg.L^-1, CDOM = 0.01 m^-1 ]; thank you for the correction (done).

• Lines 326, 467, 473, and legend to Figure 5: Species name ‘Taxifolia’ should start from lower case letter.

Done.

• Legend to Figure 4: ‘2’ of ‘K2’ should be subscript.

Done

• Lines 461-462: Abbreviations ‘Chl’, ‘SM’ and ‘CDOM’ should be explained before Figure 1 (the first appearance).

The abbreviations ‘Chl’, ‘SM’ and ‘CDOM’ have been introduced and defined in line 39 (section Introduction). For the convenience, we have changed the legend of Figure 1, by adding the following text (in bold italic font):

“(b) Environment vector d which represents the weight of the target pixel contributing to the diffuse radiance, for a target radius of $0.5 ~m$, for various seabed depths ranging from 1 m to 10 m, and moderately turbid water conditions, corresponding to a chlorophyll-a concentration Chl= 1 mg m^-3, a suspended matter concentration  SM = 1 mg L^-1 , and a colored dissolved organic matter concentration  CDOM = 0.1 m^-1.”

• Does ‘Chl’ mean chlorophyll-a?

Yes, it means chlorophyll-a. We have added this information in the legend of Figure 1, and also in line 162. Thank you for the note.

• Legend to Figure 6: ‘sustratum’ should be ‘substratum’.

Done.

• Figure 8 (a): There is a vacancy at Pixel = ca. 85. Explanation is necessary.

Figure 8-a shows the ground truth provided by the Ifremer Institute, after classification of the Vortex ROV color image of the seabed obtained at the bottom level. The high resolution classification is performed for three classes: sand, Posidonia, and Taxifolia, corresponding to a pixel size of 2.3 cm wide. The vacancy at Pixel = ca. 85 corresponds to a zone where the supervised classification is incomplete.

We added the following text in line 529:

“Note that the ground truth shows incomplete classification at Pixel ca. 85 (Figure 8-a).”

• Lines 545 and 558: Does ‘photophilic algae’ mean ‘phytoplankton’ or simply ‘algae’?

Photophilic algae means a variety of algae conditioned by light penetration. The rocky substrates of the infralittoral stage where sufficient light conditions prevail are covered by extremely rich and varied stands of photophilic algae. In the Porquerolles area, some species such as Cystoseiretum crinitae, Lithophyllo-arbacietum lixulae, and Dasycladetum vermicularis are present, being an environmental issue. We added the following text in line 548 of the revised manuscript:

“Photophilic algae consist of a rich variety of algae conditioned by light penetration, often located on rocky areas. They constitute an important environmental issue.”

• Discussion: Authors are suggested to add more detailed explanation on the difference between WUM and WADJUM with reference to the ground truth data.

The water quality around Porquerolles is quite clear. Simulations showed a very weak difference between WUM and WADJUM results for abundance estimation in clear water. Moreover, the difference between WUM and WADJUM performances gets smaller when the knowledge of the water quality is approximate. It is then coherent to find no notable difference between WUM and WADJUM on the Porquerolles area, with reference to the ground truth data. We completed the discussion with the following, in line 615 of revised manuscript:

“this is in accordance with the results obtained in the simulation using synthetic data, which showed that in clear water, the performances of WUM and WADJUM were similar up to 10 m depth, for abundance estimation. Furthermore, the unperfect knowledge on the water quality also reduces the difference between WUM and WADJUM performances.”

Reviewer 2 Report

The reviewed article concerning mapping benthic habitats by extending nonnegative matrix factorization to address the water column and seabed adjacency effects is very interesting and has high scientific advantages. In the introduction, the background and comprehensive review of the problem's literature were presented. Three main parts of the article: Models and method, Analysis of performance based on simulations and Application to real data contain theoretical essentials and implementation of the research. Results of the research have been presented in graphic form. Conclusions, on the basis of the research, are clear.

Line 171 : index of 1.3%, . should be corrected
Line 225: Table 1 - not Table I
Line 600: 2 ∗ 10⁴ should rather be written in the form 2·10⁴

Author Response

Editor comments 

Decision: Accept after minor revision

Comments: Both reviewers found the manuscript to be of high quality. The article should be accepted after minor revisions from Reviewer #1 and #2 are made.

We thanks the editor for the selection of reviewers and very fast review.

We are very grateful to the two reviewers for the time and energy they devoted to reading and analyzing our work, and their relevant comments and suggestions.

We provide below a point-by-point response.

Reviewer#2

The reviewed article concerning mapping benthic habitats by extending nonnegative matrix factorization to address the water column and seabed adjacency effects is very interesting and has high scientific advantages. In the introduction, the background and comprehensive review of the problem's literature were presented. Three main parts of the article: Models and method, Analysis of performance based on simulations and Application to real data contain theoretical essentials and implementation of the research. Results of the research have been presented in graphic form. Conclusions, on the basis of the research, are clear.

We are honored by your appreciation.

• Line 171 : index of 1.3%, . should be corrected

Done

• Line 225: Table 1 - not Table I

Done, and also everywhere in the text.

• Line 600: 2 ∗ 104 should rather be written in the form 2·104

Done