Habitat Prediction of Northwest Pacific Saury Based on Multi-Source Heterogeneous Remote Sensing Data Fusion

Round 1

Reviewer 1 Report

The paper innovatively proposed a habitat prediction method based on the multi-source heterogeneous remote sensing data fusion. By fully using remote sensing data features with different data structure and size, the proposed method improved the habitat prediction accuracy of the Northwest Pacific Saury. It is significant to improve the habitat prediction accuracy of the Northwest Pacific Saury. 

Point 1:

Line 176: You should supplement the Env data distribution images to illustrate the characteristics of the Env data in terms of spatial distribution, as in the case of the L1B data. This will better explain the rationality for the different treatment method of Env data and L1B data.

Point 2:

Line 233: Why does the matching process of L1B data and fishery data use the mean processing to reduce the spatial resolution of L1B data? Why not choose to process a high-dimensional data format like Env_Fishery data? A more detailed explanation should be given.

Point 3:

Line 396: The description of the experimental approach of L1B regarding the relevant results in Table 3-3 is not clear enough. It is suggested to give a detailed description.

Point 4:

Line 428: The description of the trend of the experiment results obtained by different weights is not easy to understand. It is suggested to be modified.

Point 5:

Line 480: The experimental results of the traditional machine learning model RF in Table 3-4 outperform the BPNN and CNN. why not consider fusing the RF model with the neural network model? The reason for not choosing the RF model should be given more explicitly.

Point 6:

Line 589: The description of Figure 3-3 is not specific enough, and the experimental setting that produced the figure should be described. The analysis is too general to convince.

Point 7:

 

Line 598: The analysis of the effect of the month on the occurrence of high CPUE value samples should be added. This allows for a better correlation between the two effect factors.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper is very well written, and contributes multi-source heterogeneous remote sensing data fusion method for habitat prediction of Pacific Saury, which can solve the problem of the multi-source heterogeneous fusion and obtain more accurate habitat prediction results. And the proposed method is advanced and effective.

It is a topic of interest to the researchers in the related areas but the paper needs very significant improvement before acceptance for publication. My detailed comments are as follows:

 

a)    According to the exploratory data analysis, a censoring value is always calculated by censoring 5% to avoid the influence of extreme values, so generally, the front and back account for about 5% is removed. In this paper, the proportion of outliers deleted roughly accounts for 30% of the total amount of data, is it too much?

b)    Fig2-1 and Fig.2-2 showed process of deleting outliers. These 2 figures may be merged into one。

c)    Data expansion can undoubtedly expand the sample size of fishery data, but it may also give CPUE values to the location points that are not suitable for the survival of saury, and ultimately affect the objectivity of the research results. Please explain its rationality.

d)    Polynomial fitting algorithm in Table2-1 was unused. This line is suggested to be deleted

e)    Your manuscript needs careful editing and particular attention to sentence structure, such as errors in lines 105-107

f)     Line 273-276:How to map  1D factors, such as year, month, longitude and latitude , into a high-dimensional feature via 1D-CNN?

g)    Line 408: There exists a calculation error. The correct value should be 17.88%

h)    In the comparison experiment (3.3.1 & 3.3.2),the decision fusion model used both  Env_Fishery and L1B_Fishery data sources, while other models only used either Env_fishery or L1B_fishery. More data sources always provide more information available for prediction, which will naturally improve the accuracy of the model. So the comparison may not be reasonable.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The article is interesting and could make an important contribution to the field, but unfortunately in its current form the manuscript lacks research depth, visible by a focus on the case study rather than the research issue, proved by an insufficiently developed introduction and lack of discussions. Thus, the manuscript requires a strong development of these sections, which are also too descriptive, and lack the analytical touch required to emphasize the contribution of the study to the theoretical advancement of the field. Detailed comments are provided for each section of manuscript.
The introduction lacks a critical touch; normally, in a scientific article the introduction analyzes the existing literature in order to identify their shortcomings (ambiguities, controversies, misconceptions or lacks), justifying the need for research, and emphasizing the novel and original elements of the current study. However, in this manuscript the introduction only presents similar studies, but fails to analyze them critically. The authors should add a summary of criticisms in a paragraph preceding the one starting on line 81. The final paragraph of "Introduction" (lines 81-87) makes little sense; it gives the impression that the authors provide an absolutely unnecessary summary of their paper; this summary belongs to the "Abstract", which is meant to summarize the research, and therefore there is no need to duplicate it elsewhere. If this is the case, I suggest deleting the paragraph. Also, the introduction lacks the most important part, which is the declaration of the research goals. Perhaps they exist (lines 79-80), but the way they are written, using the past tense and without any clear statement (e.g., "this study aims to...") the presentation gives the impression that the authors provide a summary of their research.
The most important section of a research article, the Discussions, meant to emphasize the importance of research, justifying its publication, is missing. The discussions must include (A) the significance of results - what do they say, in scientific terms; (B) the inner validation of results, against the study goals or hypotheses; (C) the external validation of results, against those of similar studies from other countries, identified in the literature; (D) the importance of the results, meaning their contribution (conceptual or methodological) to the theoretical advancement of the field; (E) a summary of the study limitations and directions for overcoming them in the future research. Out of all these, only the significance of results is presented to some extent in the "Results" section; the other parts need to be developed too.
The abstract looks like a shopping list, focusing on the case study only, and not on the broader implications of research and only on what has been done. The abstract is supposed to deliver ideas, and not state the research steps in brief and provide useless figures instead of their significance. It needs to be rewritten entirely, and shift the focus from the case study to the research issue investigated in the study and to the contribution of research to developing the knowledge on the identified scientific issue.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors have fully and deeply addressed my previous comments, and as a result the manuscript increased its research depth and addresses a broader international audience. I do not have any additional comments and recommend its publication in the revised form.

Author Response

Please see the attachment.

Author Response File: Author Response.docx