Next Article in Journal
The Influence of Climate Change on Forest Fires in Yunnan Province, Southwest China Detected by GRACE Satellites
Next Article in Special Issue
Time-Lag Effect of Vegetation Response to Volumetric Soil Water Content: A Case Study of Guangdong Province, Southern China
Previous Article in Journal
SRSe-Net: Super-Resolution-Based Semantic Segmentation Network for Green Tide Extraction
Previous Article in Special Issue
Assessment of Vegetation Dynamics and Ecosystem Resilience in the Context of Climate Change and Drought in the Horn of Africa
 
 
Article
Peer-Review Record

Spectral-Based Monitoring of Climate Effects on the Inter-Annual Variability of Different Plant Functional Types in Mediterranean Cork Oak Woodlands

Remote Sens. 2022, 14(3), 711; https://doi.org/10.3390/rs14030711
by Cristina Soares 1,*, João M. N. Silva 1, Joana Boavida-Portugal 2 and Sofia Cerasoli 1
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Remote Sens. 2022, 14(3), 711; https://doi.org/10.3390/rs14030711
Submission received: 31 October 2021 / Revised: 28 January 2022 / Accepted: 31 January 2022 / Published: 2 February 2022

Round 1

Reviewer 1 Report

The authors have put a lot of effort to revise the manuscript which has also improved the quality of the manuscript. There are several places in the introduction, methods and discussion sections where it feels a bit verbose and probably in those instances a summary (or deleting) is what is needed. For instance, in the introduction the authors devote a lot of space to definition of common terms without linking those definitions to the research gaps in climate-plant functional type interaction. Authors should also take care in how they interpret some of the results presented. For instance, attributing changes in leaf chlorophyll content to soil reflectance was very confusing and there is no clear justification anywhere in the manuscript how these inferences were drawn. Other section specific comments are presented below.

Abstract

L16 - space between “task” and “in”.

L37 - I think the authors should restructure this sentence/section to show the relevance of this assessment for formulation of climate policies for agro-silvopastoral systems.

 

Introduction

L59 - Not sure this phrase is needed “PFTs are expected to play different roles in ecosystems”. Authors can start sentence with “better knowledge …”

L79 - Replace “orbit period” with “revisit time”.

L88 – 123 Overall, this section could have done more to identify the gaps in research of climate-plant functional type interactions. In its present form there are a lot of definitions of terms and explanations of application of the concepts of interest to predicting this relationship. How is the time aggregation technique different to previous applications? Are there any benefits to retrospective and extant assessment of climate-PFT interactions?

 

Methods

L138 – Having explained multispectral sensors (LANDSAT and MODIS) in the introduction, I think the authors could also provide some background information to hyperspectral sensors. Otherwise, it feels strange to see the hyperspectral field measurements in the methods without any prior information.

Results

Figure 2 - It is very hard to read this graph due to interference between the curves. Is it possible to use colour to distinguish between the different functional type?

Moreover, it is unclear whether comparing the amplitude difference between wooded trees and herbaceous plants is relevant. It is expected that cork oak will have more biomass than herbs and shrubs and therefore experience less reduction in its biomass. This is the same for other life forms comparisons (authors already present these factual differences in Figure 5). What may be more interesting to know is the impact of seasonal variability on biophysical variables (e.g., tissue water content or chlorophyll content) and whether there is a lag in the response for each life form. Again, the authors have already presented the impact in Figures 5 & 6 and compared across seasons – it seems to me that this is a more interesting story to tell and probably should be what is first presented in the results section.

L258 and Figure 3 - what do the authors mean by “soil NDVI” and “soil MTCI”? In my opinion, this is a conceptually flawed terminology, both indices are derived from reflectance of live biomass (i.e., plant canopy/leaf), which makes any reference to soil very misleading. Could the authors clarify what these terms mean?

L267 – same comment as above.

Figure 5 - probably the most important result of this assessment. Yet, the figure is very blurry and hard to read.

L282 – 294 – What was the prior expectations of the authors? And how can these results be interpreted in the context of those expectations?

 Discussion

L388 – 405 Could it be that while cork oak trees are already experiencing water deficiencies (shown by the NDWI), but it is taking longer for the same effects to be seen in the biomass? I think the authors are repeating the results here without substantiating the implication of their findings in concepts and ecological theories.

L412 – 415 A similar query to discussion on cork oak. What does this mean? Is this adjustment to increasing temperature? Rainfall? Or both?

Climate influence – Very good. I think this is what is relevant in the discussion and probably the authors do not need the section presented in lines 381 – 434 and should start their discussion with this.

L475 – 476 Attributing changes in vegetation chlorophyll content to soil reflectance isn’t clearly explained in the methods and from a remote sensing perspective. If the authors are drawing inferences, this is not properly justified in the methods or anywhere in the manuscript.

L477 – 479 Probably a good justification to drop this index from the assessment.

L509 “IV” or “VI”.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

--> Many typos are there which can be corrected. For example lines 16, 381, etc.

--> Figure 1, Figure 5 are very blurred. Please update

--> The objectives in lines 120-123 can be presented in bullets points

--> In Section 2.1, line 128, put reference for the PFT classification

-->Line 16 can be modified as: "The objective of this study is to assess, ..."

 

 

Author Response

"Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

See attached comments

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

The template used is out-of-date, not this year version. Please use the latest template.
L 126-128: The geographic location of the study area should be presented as a new figure.
Figure 2 is not clear. It needs to draw it again in color as a kind of recognizable style. Also, there is a lack of explanation for Figure 2.
The reason why only NDVI and MTCI were compared in Figure 3 and Figure 4 should be explained.
The marks of A, B, C, and D in Figures -> (A), (B), (C), and (D)
The validity of the GLM presented in Table 2 should be verified.
What are the meaning and description regarding letters in boxes of Figure 5 such as ef, f, cd, de …?
As for vegetation changes according to climate factors using long-term field measurement data, it needs to add a new figure representing the overall workflow.
Readability is low. 
Please justify the practical implication for satellite remote sensing from this study.
NDVI is the most essential product from a spectral satellite image, but since on-site field measurement data were used without applying it to this paper, it is necessary to reconsider whether it is within the journal's scope.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors in general improved the paper according to my comments. I have just two remarks, as reported below. 

 

In my previous comment I wrote: 

REFEREE:

Resolution and number of significant digits should be better done in the models.
E.g. in the case of:
NDVICISTUS = 1.06 - 0.02 × TMAX_90 + 0.0001 × PP_HIDR
TMAX_90 ranges roughly between 20 and 30,
thus 0.02 × TMAX_90 might be
0. 15×(20÷30) ranges between 0.3 and 0.45
0.020×(20÷30) ranges between 0.4 and 0.6
0.024×(20÷30) ranges between 0.48 and 0.72
In general adapt the number of significant digits (in general I would recommend at least two significant digits). 

AUTHORS:

Answer: Thanks for the observation but it is not clear what is requested, so we did not proceed to the alteration.

REFEREE: 

Please use at least 2 signiicant digits. In other words, if a coefficient is 0.03451 do not report "0.03" but "0.034" or "0.0345" depending on the meaningfulness of the estimation.

***************************

REFEREE: 

I cannot explain a couple of models reported by the authors: NDVIGRASSES = 0.79 –0.002 × RAD_90 Why in the previous model thee NDVI get smaller as the radiation increases? It should be the opposite? MTCI in some cases in increased (oak, ulex) by precipitation, and in some other cases is reduced (grass) by precipitations. My impression is that the proposed models are best fitting models, that are working in some cases, in some other cases are just the result of a mathematical fitting, with no clear physical counterpart. Author should better discuss the actual meaningfulness of reported models.

AUTHORS:

Answer: We appreciate the question and we agree. What it means is that herbs are affected negatively with longer periods of high radiation. Does not mean that NDVI decreases with higher values of radiation. Here the period of climatic data aggregation, is the key for interpretation. Also, in this example of herbs, higher values of radiation are observed in summer, season in which herbaceous layer is absent.

REFEREE: 

Pleease add pertinent comments in thee paper.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

L 421-429: blank lines?

The reference guide should be checked out.

High-resolution figures need. 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

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

The manuscript examines the impact of climate change on different vegetation types in a Mediterranean woodland. Furthermore, the authors attempt to quantify biophysical and biochemical properties responses to climate change by proxy analysis of spectral indices using two models between 2011 and 2017. No doubt this is an important application of data science, remote sensing and statistical modelling to the field of ecology.

The combination of field measurements and hyperspectral imaging to determine spectral signatures is quite an interesting idea and one worth broad application in other ecosystems or forest landscapes. However, in its present form the manuscript fails to clearly elucidate the main assumptions that were made regarding the biophysical/biochemical properties and the modelling framework. As a result, it is unclear from reading the manuscript how influential the climatic factors were in determining temporal change to the biophysical/biochemical properties over the specified time period. This is more pronounced in the introduction where the flow between paragraphs is disjointed. Further, the interchangeable use of “analytical methods/techniques” with “statistical models” makes it hard for the reader to actively follow the description of the modelling that was implemented in the method section. For instance, basic description of common statistical models (like the GLM) seemed very convoluted and the meaning muddled. This is especially, so on how the statistical model was parameterised and would represent a challenge to anyone trying to replicate this study elsewhere. I would suggest that the authors read the articles listed below or any other related research articles on the subject area for examples on how best to properly integrate field data into statistical models and report the steps in a clear and succinct manner. Consequently, there is no clear description of how the climate variables that were measured over a 6-year period were integrated into the models. Overall, the methods lack clarity and several paragraphs inconsistent.

See the following papers for clear description on how to implement GLMs for ecological modelling.

Miller, J and Franklin, J (2002): Ecological Modelling. https://doi.org/10.1016/S0304-3800(02)00196-5

Schultz et al 2014; International Journal of Applied Earth Observation and Geoinformation. https://doi.org/10.1016/j.jag.2016.06.020

 

Below are section specific comments.

Abstract:

Line 18 – 21: For clarity of the objective, move the reference to time/monitoring period to the end of this sentence.

Line 30: How long is “long time period”? Best to give a number in days or months.

Line 38 – 41: Concluding sentence of the abstract lacks’ clarity. This is partly because of the choice of tense, in this case, evidenced instead of evidence and lack of reference to any of the biophysical properties of vegetation that were assessed. Makes it hard to decipher the overall implication of the study.

Introduction:

Line 65-66: Seems disconnected from the rest of the paragraph. The reference to PFTs appears to jump out of nowhere.

Overall the introduction needs to properly motivate the different aspect of analytical methods used in this assessment.

 

Materials and Methods:

Line 152 – 155: very confusing sentence. Perhaps try rewriting as two separate sentences.

Line 166-167: This sentence is poorly written and makes it harder to understand the entire paragraph. Are you trying to fit the model parameters and predictors using a function?  If so, what are these parameters? Is there any relationship between the model parameters and the correlation analysis implemented in the previous paragraph (lines 162 – 165). If not, then it raises more questions that needs to be addressed.

Line 171: Choice of the adjective “exhaustive” seems rather dramatic. Try using a more suitable adjective that conforms with standard academic writing.

Lines 172 – 175: Again, not clear what this sentence means. Are you splitting your data into training and validating datasets? It is strange that multicollinearity analysis is been carried out here, after a correlation test was implemented. Seems like there is too much going on here that needs simplification.

Results:

Line 191 – 192: Refers to a period when the annual precipitation was below annual averages. Looking at Figure 1, there is no way of telling when this period occurred. Perhaps, the authors need to reconsider the presentation of this figure and highlight this main finding as evidence of the influence of climate change in the ecosystems that was studied. Also, under what statistical test did the authors determine the non-significant results for the different climatic variables. Was these across the different PFTs? Or for each vegetation type?

Lines 271 – 301: This section is too long and the comparisons between the different indices and climate variable makes it easy for one to get confused. What could be helpful may be to highlight the key findings and refer the reader to table 2 if necessary.

Line 309: What does the phrase “overall, and per VI, in NDVI,” mean? From my point of view this did not make any sense, especially because the next word refers to the climatic variable temperature. My suggestion is a rewrite of this entire paragraph.

Line 327 and 328: This is a common occurrence throughout the manuscript. Authors should adhere to the correct presentation and scientific format of “degree Celsius”, that is, degree as a superscript.

Line 359: The first sentence is poorly written.

Line 367: Methods makes a reference to the fact that model performance was determined using Akaike Information Criterion (Line 168-169), surprised to see that there is no reference to this when evaluating model performance.

Discussion

Lines 375-377: Why is the emphasis on the evidence of differences between VI and vegetation type. This is expected considering that they are different PFTs, perhaps the focus should have been on the proxies of vegetation state (e.g., biomass) that was clearly outlined in the introduction and objectives and what impacts were picked up using the analytical framework proposed in this study.

Lines 380 – 383: what do you mean by “contrasting tendency”? Does this mean that there are times when the cork oak values become sensitive to climate variable and other times not so?

Line 384: Very clear and good message in this paragraph.

Lines 418 – 468: Seems to me like a repeat of the methods. I would have expected that the authors discuss in detail the implication of the quantified temporal patterns of changes in VIs, the relationship with climate variables in the context of biomass, tissue water content and chlorophyll as it relates to the different vegetation types responses to climate change.

Conclusion: This section fails to fully summarise the main findings, especially on specific impacts of climate change on properties – having a more focused discussion would have helped highlight the main results here.

 

Reviewer 2 Report

The authors proposed to investigate the influence of climate on Cork Oak woodland. To draw this connection, the authors conducted extensive field work, collecting both climate and hyperspectral data. Then, they built and compared preselected parametric and non parametric models, evaluating their ability to draw a relationship vegetation between indices (VI's) derived with field spectral data using a variety of climate metrics with different levels of aggregation.

Although I recognize the amount of work put into this paper, I am missing a red line. All the analysis conducted in this paper are informative, but they don't seem to fully work together to answer the initially proposed question. I believe the paper says little about the sensitivity of Cork Oak woodland to climate, and more about the sensitivity of VI's to climate in this particular ecosystem. Spectral indices are valuable proxies for vegetation health when aiming for spatial predictions, but using them is a weaker choice when working exclusively with field data. Unless, of course, previous literature clearly supports this relation (with field measures) between the chosen VI's and the target biophysical parameters, in which case the authors failed to clarify. Additionally, there should be a clear rationale on why to use VI's (e.g. cost-benefit, transferability). With this in mind, I recommend the redesign of the paper. Below, I provide highlighted some specific issues:

- Which kind of earth monitoring systems might readily make use of your results? This is important information to integrate in the discussion, since your results will most likely be used this way.
- Missing a clear statement of when the spectral data was collected, how this relates to the climate data and if these data were somehow aggregated. It would also be adequate to justify any choices made.
- Figure 2 is hard to read.
- Why use CART and GLM? And, more importantly, why make the comparison at all? If your objective is simply to evaluate the sensitivity of Cork Oak woodland to climate, this comparison seems unnecessary.

Back to TopTop