Texture Features Derived from Sentinel-2 Vegetation Indices for Estimating and Mapping Forest Growing Stock Volume
Round 1
Reviewer 1 Report
The paper is well written and provides good insight into the employment of using Textural features for predicting forest growth stock volume. The authors have explained the methodology and discussed the findings thoroughly. I had some minor suggestions that will improve the current version of the manuscript. The introduction can be clearer and more concise for the readers to understand what the authors are trying to convey. I have provided detailed comments on each section further.
Abstract- English language inconsistencies. In the very first line, the authors mention " Forest growth volume was an indispensable variable". Forest GSV is always an important variable. Need to change all sentences to present tense. ("GSV is an indispensable variable"). These tense differences can be observed throughout the manuscript. I would suggest to carefully go through and make necessary changes throughout the manuscript.
Introduction: The research gap needs to be clearly specified. Current version is not providing clarity on the research gap and what the authors are trying to convey through this manuscript.
Line 73: "Values of pixel were less taken into account..." can be written more clearly and concisely as "Importance of pixel values in predicting forest growth parameters have not been studies before [][]" or "There is a need to evaluate the importance of pixel values derived from different metrics in predicting forest growth volume" .
Results:
Adding a graph showing the relationship of predicted and observed GSV will be useful for the readers to understand the effectiveness of the proposed model.
Discussion: Requires citation for sentences. In line 343, 385-386... Please add citations to claims made in the discussion section.
Also may provide quantitative values when suggesting variables had a moderate effect, higher effect in predicting GSV in discussion.
Author Response
We want to thank you for your valuable and helpful comments.
AR: Author’s Response
The line and section below about revised contents are based on the "Revision, changes marked" file
The paper is well written and provides good insight into the employment of using Textural features for predicting forest growth stock volume. The authors have explained the methodology and discussed the findings thoroughly. I had some minor suggestions that will improve the current version of the manuscript. The introduction can be clearer and more concise for the readers to understand what the authors are trying to convey. I have provided detailed comments on each section further.
Abstract- English language inconsistencies. In the very first line, the authors mention " Forest growth volume was an indispensable variable". Forest GSV is always an important variable. Need to change all sentences to present tense. ("GSV is an indispensable variable"). These tense differences can be observed throughout the manuscript. I would suggest to carefully go through and make necessary changes throughout the manuscript.
AR: Thank you for your detail consideration of our English errors. We are sorry that our carelessness, and the revision has been done.
Introduction: The research gap needs to be clearly specified. Current version is not providing clarity on the research gap and what the authors are trying to convey through this manuscript.
AR: The research gap has been added. Please see line 85-88 or below.
Generally speaking, (1) the existing research did not consider that the competence of the vegetation indices-based texture measures, (2) coupled with the performance of satellite imagery based on different pixel values, (3) and the combinations encompassing the above two points for estimating GSV.
Line 73: "Values of pixel were less taken into account..." can be written more clearly and concisely as "Importance of pixel values in predicting forest growth parameters have not been studies before [][]" or "There is a need to evaluate the importance of pixel values derived from different metrics in predicting forest growth volume" .
AR: Thank you for your detail and valuable consideration. The revision has been done, please see line 80.
Results:
Adding a graph showing the relationship of predicted and observed GSV will be useful for the readers to understand the effectiveness of the proposed model.
AR: Thank you for your valuable consideration. The scatterplots between observed GSV and estimated GSV have been added. See Fig 4.
Discussion: Requires citation for sentences. In line 343, 385-386... Please add citations to claims made in the discussion section. Also may provide quantitative values when suggesting variables had a moderate effect, higher effect in predicting GSV in discussion.
AR: Thank you for your detail suggestions. The citations and necessary description have been added. Please see section 5.2 line 408 and section 5.3 line 464.
Reviewer 2 Report
Manuscript focuses on the forest growing stock volume estimation and mapping. The subject is absolutely interesting and useful. Application of a new satellite data is accurate and brings new insights to the optical remote sensing. However the structure of the manuscript is its greatest weakness. There is a luck of methodology section. Authors presented simplified methods, but the section (2) is chaotic containing data, methods, and some bits of methodology. That is why the focus starts being unclear. The results section (3) is not consistent with the section 2. Then the discussion section (4) contains methodology and research results (e.g. lines 328-342). The inputs contained in the individual sections are intertwined.
In the subsection 3.3 (line 242) authors wrote that they presented three tree species but in fact authors presented only two species. This concept is unclear. The reference (13) is not available and quite old regarding quick outdated remote sensing information. Please improve style of writing because sentences are incomprehensible (e.g. line 422). Please reorganize manuscript according to the scientific writing standards.
Author Response
We want to thank you for your valuable and helpful comments.
AR: Author’s Response
The line and section below about revised contents are based on the "Revision, changes marked" file
Manuscript focuses on the forest growing stock volume estimation and mapping. The subject is absolutely interesting and useful. Application of a new satellite data is accurate and brings new insights to the optical remote sensing. However the structure of the manuscript is its greatest weakness. There is a luck of methodology section. Authors presented simplified methods, but the section (2) is chaotic containing data, methods, and some bits of methodology. That is why the focus starts being unclear. The results section (3) is not consistent with the section 2.
AR: Thank you for your detail and valuable consideration. We have added section (3) as methodology and added some necessary explanations. See section 3.1 or below.
Previous research has examined the sensitivity of the heterogeneity in vegetation and habitat taken by different pixel statistics using remotely sensed images, according to different variability of different statistical pixel results in the images. In this research, standard deviation values (STD) and mean values (MV) of pixels acquired from the univariate predictors were applied to predict the GSV of three kinds of plots, so as to explore the difference between both pixel values to GSV inversion. Besides, the textural metrics derived from bands and vegetation indices were also calculated to assess whether it yields different implications for different forest species. Eventually, ten feature sets were employed to detect the performance of different remotely sensed predictors.
In general, we firstly evaluated the relationships between in-situ GSV and univariate predictors which encompassed original spectral bands, vegetation indices, and the texture measures. All the predictors have been taken account into two kinds of pixel values. Then, we built ten groups of combinations that comprised a variety of remote sensing data or patterns derived from raw data and utilized the feature importance of Random Forest to select four ranked ahead features to investigate the estimated competence with different kinds of predictors.
Then the discussion section (4) contains methodology and research results (e.g. lines 328-342). The inputs contained in the individual sections are intertwined.
AR: The discussion has been improved. We have removed some inappropriate description. Please see below or line 390-406.
For all plots, an interesting detection was that many predicted results seemed to have a regular change (increase or decrease) along with the enlargement or minnish of window sizes for standard deviation and mean statistics. From all predicted GSV consequences, however, it showed that smaller windows have stronger competence to better fitting. This was analogous to what investigated the texture element of a small sliding window size from high resolution imagery showing a better association with horizontal vegetation structure. Besides, we also found that those textures achieving higher accuracies were based on red-edge, NIR, and SWIR, which have also been taken into account immense spectral predictors for vegetation attributes prediction. Thus, we draw that it is valuable to allow for texture measures derived from those spectral parameters which with great potential to predict forest characteristics.
In the subsection 3.3 (line 242) authors wrote that they presented three tree species but in fact authors presented only two species. This concept is unclear.
AR: Thank you for your detail consideration. We are sorry that our inappropriate expression. Actually, we wanted to convey three kinds of plots of our survey area. And we have revise the manuscript in section 4.3, line 294.
The reference (13) is not available and quite old regarding quick outdated remote sensing information.
AR: Thank you for your valuable advice. The reference (13) has been revised. See below or the reference in line73.
- Chen, J. Useya, and H. Mugiyo, “Decision-level fusion of Sentinel-1 SAR and Landsat 8 OLI texture features for crop discrimination and classification: case of Masvingo, Zimbabwe,” Heliyon, vol. 6, no. 11, p. e05358, Nov. 2020, doi: 10.1016/j.heliyon.2020.e05358.
Please improve style of writing because sentences are incomprehensible (e.g. line 422). Please reorganize manuscript according to the scientific writing standards.
AR: Thank you for your valuable advice. The sentence in line 422 has been improved. See below or section 6 line 500.
The survey plots were in Anji County consisting of Masson pine, theropencedrymion, and all plots which included a variety of dominant tree species.
Reviewer 3 Report
The manuscript has conducted extensive statistical analysis, which holds some value. However, there are certain areas that require improvement in terms of academic and theoretical rigor. One particular aspect is the manuscript's attempt to utilize texture information derived from vegetation indices for GSV inversion. Although this approach has potential, the background of relevant work is inadequately introduced, and the manuscript fails to emphasize its significance effectively. Furthermore, the results are not sufficiently summarized.
For instance, the ten indices utilized in the study should be appropriately classified and demonstrated to be suitable for GSV inversion based on previous research. Additionally, the background and rationale for employing standard deviation and mean values in the inversion process are not adequately explained. This omission hinders readers from fully comprehending the author's intentions. Moreover, the description of the inversion method itself is lacking in detail. It remains unclear how the "data range", "mean" and "variance" are incorporated as a predictor variable in the construction of the inversion model.
In conclusion, the current version of the manuscript lacks relevant background information and presents insufficient methodological descriptions. Therefore, it is not yet suitable for publication.
Author Response
We want to thank you for your valuable and helpful comments.
AR: Author’s Response
The line and section below about revised contents are based on the "Revision, changes marked" file
The manuscript has conducted extensive statistical analysis, which holds some value. However, there are certain areas that require improvement in terms of academic and theoretical rigor. One particular aspect is the manuscript's attempt to utilize texture information derived from vegetation indices for GSV inversion. Although this approach has potential, the background of relevant work is inadequately introduced, and the manuscript fails to emphasize its significance effectively. Furthermore, the results are not sufficiently summarized.
For instance, the ten indices utilized in the study should be appropriately classified and demonstrated to be suitable for GSV inversion based on previous research.
AR: The ten vegetation indices have been classified as red-edge bands and non-red-edge bands. And the related previous research about these vegetation indices were also described. See section 2.3 line 138-140 or below.
The vegetation indices have been utilized in tremendous of previous research [1]–[3], which demonstrated their great performance for GSV prediction, , especially for those based on red-edge bands.
[1] S. Sánchez-Ruiz, Á. Moreno-Martínez, E. Izquierdo-Verdiguier, M. Chiesi, F. Maselli, and M. A. Gilabert, “Growing stock volume from multi-temporal landsat imagery through google earth engine,” Int. J. Appl. Earth Obs. Geoinf., vol. 83, p. 101913, Nov. 2019, doi: 10.1016/j.jag.2019.101913.
[2] I. Chrysafis, G. Mallinis, S. Siachalou, and P. Patias, “Assessing the relationships between growing stock volume and Sentinel-2 imagery in a Mediterranean forest ecosystem,” Remote Sens. Lett., vol. 8, no. 6, pp. 508–517, Jun. 2017, doi: 10.1080/2150704X.2017.1295479.
[3] I. Chrysafis, G. Mallinis, M. Tsakiri, and P. Patias, “Evaluation of single-date and multi-seasonal spatial and spectral information of Sentinel-2 imagery to assess growing stock volume of a Mediterranean forest,” Int. J. Appl. Earth Obs. Geoinf., vol. 77, no. November 2018, pp. 1–14, 2019, doi: 10.1016/j.jag.2018.12.004.
Additionally, the background and rationale for employing standard deviation and mean values in the inversion process are not adequately explained. This omission hinders readers from fully comprehending the author's intentions.
AR: The details of inversion method have been added in section 3.1, line 169-184. See below.
Previous research has examined the sensitivity of the heterogeneity in vegetation and habitat taken by different pixel statistics using remotely sensed images, according to different variability of different statistical pixel results in the images.
In general, we firstly evaluated the relationships between in-situ GSV and univariate predictors which encompassed original spectral bands, vegetation indices, and the texture measures. All the predictors have been taken account into two kinds of pixel values. Then, we built ten groups of combinations that comprised a variety of remote sensing data or patterns derived from raw data and utilized the feature importance of Random Forest to select four ranked ahead features to investigate the estimated competence with different kinds of predictors.
Moreover, the description of the inversion method itself is lacking in detail. It remains unclear how the "data range", "mean" and "variance" are incorporated as a predictor variable in the construction of the inversion model.
AR: The details of three kinds of measures have been added in section 2.3, line 147-150. See below.
The texture measures were derived from ten spectral bands and ten vegetation indices, coupled with three kinds of texture metrics and three kinds of window sizes (e.g., NDII_ME17SD represents the predictors were derived from NDII which combined texture metrics ‘mean’ and 17×17 window size, and the pixel values are STD).
In conclusion, the current version of the manuscript lacks relevant background information and presents insufficient methodological descriptions. Therefore, it is not yet suitable for publication.
AR: Thank you for your valuable suggestions. We have revised the manuscript according to your advice.
Round 2
Reviewer 2 Report
Authors improved the manuscript according to the comments. However, I would suggest checking keywords. They are very general and do not guide a reader anywise.
Reviewer 3 Report
The manuscript is satisfactory in its current form, and I have no further comments.
