Assessing Forest Species Diversity in Ghana’s Tropical Forest Using PlanetScope Data

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In the introduction section, an explanation of the limitations of commonly used satellite images like Sentinel and Landsat can be beneficial to highlight research questions and novelties.

What sets this study apart in terms of novelty? Is it solely the utilization of Planet scope images? Emphasizing the novelty of your research is essential. Why do you believe this study warrants publication?

It would be beneficial to include vegetation indices based on red-edge bands, such as NDRE, to harness the full capabilities of the PS images. Why were only the most common VIs used? There are numerous VIs documented in the literature that could potentially exhibit strong correlations with diversity.

Line 184-186. present the formulas within a table for better organization and clarity.

How can the spatial mismatch between field data collected in a 20x50-meter plot and PlanetScope satellite images with a 3-meter resolution impact the accuracy and reliability of the analysis? I mean spatial scale discrepancy, heterogeneity, etc.

In Table 5, I believe the authors should consider incorporating additional features, such as PCA components, tasseled cap-based features, and other indices that can be calculated using the red-edge band. Moreover, environmental variables like elevation, slope, temperature, etc., could greatly enhance the modeling of species diversity. The current version utilizes a limited number of features, demonstrating limited correlation with field data, which were then used in the modeling process. Additionally, the low correlation observed between vegetation indices seems peculiar. It raises questions about how spectral bands exhibit a higher correlation than vegetation indices.

In the discution section, you mentioned that "The significant relationship between the spectral bands of PlanetScope (blue, green, green, yellow, red, red edge, and near-infrared) data and biodiversity variables (S, J', D2, and H') suggests that satellite images such as PlanetScope would be useful for predicting biodiversity in tropical forests". Did you observe a significant correlation? For example, an R2 value of 0.171 generally indicates a low level of variance in species diversity explained by the model.

In the discussion section, you explained why VIs exhibit a low correlation with species diversity: 'Vegetation indices did not perform well in terms of correlating with species diversity. This could result from the high heterogeneity of the BFR, with different species intermixed at small scales, making it challenging to distinguish between them accurately.' However, the question remains: how did spectral bands show higher correlation than VIs? The spatial resolution is 3m for all spectral bands and VIs, so if there is heterogeneity or mixed pixels, it could affect both, usually favoring spectral bands.

 

Comments on the Quality of English Language

 Generally, the language of the paper is adequate.

Author Response

Thank you for the review comments and suggestions. We have provided a point-by-point response to the comments in the documents uploaded.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The study titled "Assessing Forest Biodiversity in Ghana's Tropical Forest Using PlanetScope Dataset" examines the estimation of tree diversity in the Bobiri Forest Reserve (BFR) through the utilization of multispectral bands and vegetation indices extracted from PlanetScope images. This innovative approach represents a compelling use of an underexplored satellite product, thus proposing a valuable application for PlanetScope data. In summary, I find this paper interesting and look forward to its publication.

Comments for author File: Comments.pdf

Author Response

Thank you for the review comments and suggestions. We have provided a point-by-point response to the comments in the documents uploaded.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

General comments

The authors discussed the role of PanetScope imagery in mapping species diversity within a local study area of 30 sq. km in Ghana. Given the small study area, I would expect to see some advancements in image processing to map forest diversity. Unfortunately, I did not find this, so the research is a case study. Its results are hardly applicable over larger scales.

From this perspective, the practical role of the methodology is not clear to me.  First, the authors missed many important details in the methodology. For example, more information is needed to describe how they treated clouds and cloud shadows (observed for some images under <10% cloudiness threshold) in the image mosaicking process, which regression model they used, and how to address the significance of the correlation between remotely sensed data and diversity indices. Second, the authors utilized zonal statistics for spectral bands of PlanetScope imagery (I could only guess that they extracted a range of values, sd, etc. within sample plots perimeters because this was not described) and linked them to diversity indices. I have doubts that the regression models trained on such data then can be applied at a 3-m pixel level. Moreover, the correlation coefficients between diversity indices and selected spectral bands were > |0.2| only in a few cases. There is no discussion about multiple correlations at all.  

Text flow and wording must be carefully checked by a professional editor.

 My miscellaneous comments are provided below.

Specific comments

Title: Do not use “PlanetScope dataset”,  PlanetScope imagery or data would be better.

Abstract: Must be improved. Some sentences must be combined, and some information is hard to understand without reading the manuscript (e.g., standard deviation of band, range of band).  

Line 18: "remotely sensed dataset with a high spatial resolution" - be explicit here, use "3-m spatial resolution PlanetScope imagery".

Line 20-22: Correct following the edits of the first sentence.

Line 23: A List of indices should go after the reference data are described. Use the following logic in the sentence: satellite data were used to predict diversity indices.

Line 26: Why "some"? You have already specified them.

Line 31: What did you mean by range? After reading the manuscript I could understand that this refers to zonal statistics. Sorry, but it is not clear from the abstract.

Keywords: Most of the keywords are in the title.

Introduction

Line 42: "vital ecosystem service forest supports" - grammar check is needed.

Line 50: What is "low forest monitoring"?

Line 57: Hard to catch the idea.

Line 68: 730 trees? Is this the problem? Did you mean tree species? 

Line 91: Sentinel, be careful – there are more such typos in the text.

Line 99: Avoid starting sentences with []. It is a common problem for the manuscript style.

Line 110: Provide authors’ names before citing their research in brackets and avoid using this style; " discrimination" - mapping, classification. Do not describe what other authors did, provide only what they found.

Line 111: Did you mean ecosystem services?

Line 113: The Journal requires explanations of acronyms.

Line 118: I could agree if the information derived from high resolution imagery is aggregated at coarser resolution.

Line 125: PlanetScope.

Lines 129:130: What is the difference between 1 and 2?

Materials and Methods

Figure 1: Fix the legend - there are no three bands in the figure (just MS image), replace layer names with correct text, and add a coordinate grid.

Table 1: It is unnecessary to provide band specifications since it is available on the official site (fix green).

Lines 74-183: What did you do with 10% of clouds? Did you apply cloud and cloud shadow masks?

Line 184: How could the variability of diversity indices impact the way you extracted pixel statistics?

Line 186:187: You should have not described that you removed some pixel statistics since this is just a standard QGIS algorithm (I guess, zonal statistics). Leave only information on what you used. How are you going to apply zonal statistic variables at pixel level?

Lines 190:191: It was not clear how pixel statistics were calculated. I assume you used perimeters of sample plots. If that is the case, then you can link species diversity indices observed on the plot and corresponding spectral statistics. That is correct. I am wondering, how your model can work at 3-m pixel level.

Line 192: What model did you use in the regression?

Line 202: Fix the start of the sentence.

Line 231: What measure of "abundance" did you use in the study? Basal area? Notations for proportions are not uniform across indices.

Line 234: Still no information on the regression model. If you used only predictors from Table 1, why did you extract zonal statistics for those bands?

Results

Line 251: The sentence duplicates information from section 2.3.1. I believe this material suits more previous section.

Table 4: Species names normally go in italics.

Line 262: This sentence just steals valuable space from the manuscript.

Line 263: There is no correlation observed in most cases. At least you need to provide confidence intervals for R. How can you prove that relations between spectral variables and diversity indices would have the same trends (positive or negative) within other areas in Ghana? Now, these are just local results that have limited chances for replications in other areas or using other image mosaics.

Line 271: You predicted diversity indices, not biodiversity!

Line 272: Vis ---> Vis.

Line 281: What equation? Can you provide its form?

Figure 2: Very serious question: How can you use zonal attributes as predictors at 3-m pixel level?

Discussion

Line 312: I disagree. Everything shown in the paper had insignificant correlations with diversity indices.

Line 322: You did not justify that the correlation was negative. If you had provided confidence intervals for R, negative correlations could appear positive.

Line 384: You did not prove this statement - you have obtained a low level of correlation.

Conclusion

The first paragraph can be written without this study. The other text looks like a discussion. Be concise in your conclusions.

Line 395: You did not show that.

The manuscript must be substantially improved before re-submission. Text flow needs to be revised.

  Comments on the Quality of English Language

English must be improved by a native speaker / professional editor.

Author Response

Thank you for the review comments and suggestions. We have provided a point-by-point response to the comments in the documents uploaded.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors answered my comments, and I am satisfied. There are no additional concerns on my behalf. So, their paper can be accepted.

Author Response

 "Please see the attachment." 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The authors substantially revised the manuscript and improved the study design. In such a form the paper can be interesting to a certain group of readers.

 I have only minor suggestions;

Lines 27-32: Thanks for clarifying this, but leave it for the main text. Information on bands and indices is enough for the abstract. I suggest combining two sentences. Acronyms must be placed next to corresponding indices

Lines 43-44: Still, many keywords are in the title

Figure 1, Figure 3, Figure 4: Use Lon/Lat coordinates, or specify which projected CRS was used to build this map.

Line 210: In your response, you specified that all images were cloud-free. Write the same here, not less <10%.

Figure 3, Figure 4: increase font size.

Author Response

"Please see the attachment."

Author Response File: Author Response.pdf