Large-Scale Mapping of Complex Forest Typologies Using Multispectral Imagery and Low-Density Airborne LiDAR: A Case Study in Pinsapo Fir Forests

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The aim of paper reviewed is to illustrate a methodology to map complex  Pinsapo Fir Forests typologies in southern Spain using multispectral and low-density airborne LiDAR data and machine learning algorithms. The approach showed considers species-area distribution and complex structure and aims to generate a map of existing Pinsapo fir forests and uuseful to identifying priority sites for conservation, restoration and legal protection that are currently lacking in southern Spain (Andalusia chain mountain). 

Overall, the study touches a topical subject in the context of forest science and ML classification criteria/methods using remote sensed product (laser scanner and spectral acquisitions). My general opinion of results showed in the paper is that there are some issues to be addressed before the manuscript can be suitable for publication. I suggest improving the text following my comments and suggestions in the PDF file attached.

My review response of this paper is minor revision.

Best wishes,

ZAP

Comments for author File: Comments.pdf

Comments on the Quality of English Language

none

Author Response

Reviewer #1 Comments to Authors

The aim of paper reviewed is to illustrate a methodology to map complex Pinsapo Fir Forests typologies in southern Spain using multispectral and low-density airborne LiDAR data and machine learning algorithms. The approach showed considers species-area distribution and complex structure and aims to generate a map of existing Pinsapo fir forests and useful to identifying priority sites for conservation, restoration and legal protection that are currently lacking in southern Spain (Andalusia chain mountain).

Overall, the study touches a topical subject in the context of forest science and ML classification criteria/methods using remote sensed product (laser scanner and spectral acquisitions). My general opinion of results showed in the paper is that there are some issues to be addressed before the manuscript can be suitable for publication. I suggest improving the text following my comments and suggestions in the PDF file attached.

My review response of this paper is minor revision.

The authors appreciate the reviewer’s suggestions, and his/her effort to correct the manuscript which has improved the scientific and written quality of the article. This work has not only served us for this work, but we are sure that it will guide the writing of other scientific works in the future.

1. where? in southern Spain

We appreciate the reviewer comment. However, no reference to the geographical area has been included in the title as the species only naturally exists in southern Spain (see L109-L111 in the manuscript file).

2. and biodiversity...stop!

We appreciate the reviewer comment. The text has been reviewed according to the reviewer suggestion:

L031-032.- Forest ecosystems provide critical and diverse productive and ecosystem services to human society [1] such as wood, carbon storage, and biodiversity [2,3].

3. please add also illegal logging

We appreciate the reviewer comment. The text has been reviewed according to the reviewer´s suggestion:

L032-036.- However, the future of forests is uncertain as a consequence of different factors such as illegal logging and, in particular, climate change [4] due to increasing temperatures and extreme droughts in many regions of the world [5], but especially in the Mediterranean regions [6].

4. from this procedure you obtain the response variables? ...if your reply is yes please specify it

We appreciate the reviewer comment. The text has been reviewed according to the reviewer´s suggestion. For a correct visualisation of the figure, the following sentence has been included in the figure caption:

L130-131.- Main workflow diagram for large-scale mapping of Pinsapo Fir Forest typologies. Response variables were extracted from the vegetation inventory data.

5. Nord symbol more larger in all maps done in the paper

Figures have been reviewed according to the reviewer´s suggestion.

6. this paragraph is more an introduction that a discussion of results obtained. Please consider it in your revision

We appreciate the reviewer comment. The text has been reviewed according to the reviewer suggestion:

L283-285.- In this study, a methodology for pinsapo fir forest typology is proposed as an example of applying multisource remote sensing data to classify in large spatial scale fragmented and uneven Mediterranean forests with critical conservation concerns.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

I have reviewed the article titled "Large-scale Mapping of Complex Forest Typologies Using Multispectral Imagery and Low-Density Airborne LiDAR: A Case Study in Pinsapo Fir Forests." This study addresses a critical need for improved forest typology mapping, particularly for the vulnerable Pinsapo fir forests. Integrating multispectral imagery with low-density LiDAR data to predict forest typologies is an exciting and innovative approach within the related literature.

However, the study's results show an overall accuracy of 62% and a Kappa index of 0.43, indicating moderate performance of the model. While the study acknowledges this, it needs to sufficiently discuss the potential reasons for this limitation or how future studies could address it. Also, the sampled plots did not represent four of the eleven proposed silvicultural types. This limits the generalizability of the findings and should be addressed in future research by incorporating more comprehensive sampling.

Apart from these, I have some minor suggestions for the authors to improve the article:

• Although the introduction section is detailed and explains the main purposes of the study, some paragraphs are quite long. For example, the paragraph starting with line number 49 can be divided into two. In the remaining sections, the authors should write paragraphs more concisely to help readers better understand the main ideas.
• The sentence starting with "Therefore, in this study the main objective..." on line 92 should be written as a new paragraph.
• What do SN, SG, and SB stand for? Although they are stated in Figure 1, abbreviations should be defined in the text where they are first used (line 106).
• When writing numerical data, attention should be paid to punctuation. It should be written in units of hundreds (lines 107, 109, and so on).
• After the flowchart in Figure 2, the headings can be considered subsections because they relate to the previous heading. Also, the section under "Application of the typology key to the sample plot" in this figure is unreadable. The resolution of the image should be increased to make it readable.
• The grid lines used in the maps need to be corrected. They should be updated to show coordinates, and only north-west or south-east grids should be shown. There is also no need to specify the scale in a map where a scale bar is used.

Author Response

Reviewer #2 Comments to Authors

I have reviewed the article titled "Large-scale Mapping of Complex Forest Typologies Using Multispectral Imagery and Low-Density Airborne LiDAR: A Case Study in Pinsapo Fir Forests." This study addresses a critical need for improved forest typology mapping, particularly for the vulnerable Pinsapo fir forests. Integrating multispectral imagery with low-density LiDAR data to predict forest typologies is an exciting and innovative approach within the related literature.

However, the study's results show an overall accuracy of 62% and a Kappa index of 0.43, indicating moderate performance of the model. While the study acknowledges this, it needs to sufficiently discuss the potential reasons for this limitation or how future studies could address it. Also, the sampled plots did not represent four of the eleven proposed silvicultural types. This limits the generalizability of the findings and should be addressed in future research by incorporating more comprehensive sampling.

Apart from these, I have some minor suggestions for the authors to improve the article:

The authors appreciate the reviewer’s suggestions, and his/her effort to correct the manuscript which has improved the scientific and written quality of the article. This work has not only served us for this work, but we are sure that it will guide the writing of other scientific works in the future.

1. Although the introduction section is detailed and explains the main purposes of the study, some paragraphs are quite long. For example, the paragraph starting with line number 49 can be divided into two. In the remaining sections, the authors should write paragraphs more concisely to help readers better understand the main ideas.

We appreciate the reviewer comment. The text has been reviewed according to the reviewer´s suggestion:

L57.- However, these forest inventories are often limited in temporal scope and spatial scale.

            In recent years, new approaches combining forest inventory field plots and remote sensing data have been emerged to improve forest characterization and mapping [21]……

L298.- These algorithms showed poorest results than RF in the accuracy tests for the validation dataset (Table 3) which is consistent with previous studies [49].

            The classification accuracy of forest types based on multisource remote sensing data was moderate, and the overall accuracy was 62%.......

L305.- On the other hand, inspection of the point cloud data revealed the presence of large trees and gaps with variable sizes in even-aged and two-aged stand types, which had a negative influence on the strength of the statistical relationship between ALS metrics and forest types [50].

            These results are partially explained by the selected predictors…..

L379.- Third, no forest inventory data are available for this study in SB. Although we strongly believe that pinsapo fir forests in SB have similar structure to those in some areas of SN, such data could increase the accuracy of classification models for this region.

            Thus, five important improvements should be addressed in future versions of the work….

2. The sentence starting with "Therefore, in this study the main objective..." on line 92 should be written as a new paragraph.

We appreciate the reviewer´s comment. The text has been reviewed according to the reviewer suggestion:

L092-097.-These gaps often result from lack of data with enough temporal recurrence, as observed in studies by [31].

            Therefore, in this study the main objective was to develop a forest typology for pinsapo fir forests based on forest composition and structural attributes and mapping this typology using a low-resolution LiDAR and multispectral data…

3. What do SN, SG, and SB stand for? Although they are stated in Figure 1, abbreviations should be defined in the text where they are first used (line 106).

We appreciate the reviewer´s comment. The abbreviations are defined in lines 45-46 of the revised manuscript.

4. When writing numerical data, attention should be paid to punctuation. It should be written in units of hundreds (lines 107, 109, and so on).

We appreciate the reviewer´s comment. The text has been reviewed according to the reviewer suggestion:

L110.- 4,973.9 ha

L112.- 600 and 1,600 mm

L269.- 3,241 ha

L270.- 1,075 ha

L274.- Table 5 (2,143.85 ha)

5. After the flowchart in Figure 2, the headings can be considered subsections because they relate to the previous heading. Also, the section under "Application of the typology key to the sample plot" in this figure is unreadable. The resolution of the image should be increased to make it readable.

We appreciate the reviewer´s comment. The subheadings have been organised in such a way that the highest level of subheadings is subsubsection (i.e. 2.1.1), as recommended by the journal. Figure resolution has been improved.

6. The grid lines used in the maps need to be corrected. They should be updated to show coordinates, and only north-west or south-east grids should be shown. There is also no need to specify the scale in a map where a scale bar is used.

Figures have been reviewed according to the reviewer suggestion.

Author Response File: Author Response.pdf