Integration of Geospatial Tools and Multi-source Geospatial Data to Evaluate the Tropical Forest Cover Change in Central America and Its Methodological Replicability in Brazil and the DRC

Round 1

Reviewer 1 Report

Integration of geospatial tools and multi-source data for evaluate the forest cover change dynamics in Central America

-title should mention about tropical forest cover change and include scalability and replicability in Brazil and DRC instead of only central America

-Change Multi-source to multi-temporal because Landsat is the only satellite imagery source

 

Abner Jiménez, Alexander Hernández and Víctor Rodríguez-Espinosa

 

Abstract

-need to mention how significant is the study (integration of geospatial tools)

-need to simply mention at the end part of the abstract that tropical forests in both Brazil and DRC have decreased over time.

 

Introduction

-region instead of by countries: lines 36 – 42 need some references to support the statement perhaps previous related studies

-tropical forest shares similarities: lines 43 – 46 also need some references to previous related studies to support the statement.

-regarding the differences in forest definitions by each country (lines 52-57) maybe can add an example to give readers the exact idea of how does it differ.

 

Materials and Methods

-If possible, mention the size of study region coverage

-Line 207-212, maybe a figure or formula of Bayesian inference of the Maximum Likelihood Classifier could be included

-also, could include a reference to previous studies that using a similar classification method

-Table 1, should include the meaning of “ARB” abbreviation in the Nomenclature.

-Validation: Line 320-338, need to include the accuracy/validation RMSE & confusion matrix table results.

-Maybe results of the Reliability of forest cover mapping and forest cover (line 362-379) percentages can be combined into table format.

-Similar with Application of the methodology in other tropical zones (line 450-466)

 percentages and numbers can be presented in table format.

 

Conclusions

-Line 557-564, also can suggest that through the theoretical-methodological framework has been developed in this study, a regional classification should be given emphasize later adopted to country-level for multi-temporal forest cover changes study.

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

This study presents cost-efficient methods to generate national forest cover maps for the countries of Central America.

In the title change 'for evaluate' to 'for evaluating' or 'to evaluate'

Elaborate on the significance of the study in introduction. What is the importance of evaluating the forest cover change dynamics?

Abbreviate USGS in Pg.3, line 108.

In your study, you have used data from three years based on the availability. Discuss if you propose higher time intervals for future or related studies and challenges if any.

The text in Figure 1 is not clear.

Figure 6 and 7 can be improved, especially the labels.

Check the link in reference 32 - the link is not working.

Overall, there is no novel ideas or innovations in this study. Geoprocessing tools can also be done in ArcGIS Pro. Most of the figures can be improved. 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

It is important to map the vegetation types and their change over time in the present global climate change and deforestation scenario. In the present study, the authors use a hybrid classification technique to map the vegetation and its change in Central America and extrapolate their technique to two local sites, one in South America and another in Africa. The study is well conducted and has the potential to be used as a conservation and forest management tool. However, I have some major concerns with the manuscript in its present form.

Comments

-There is a grammatical error in the title. Please rephrase.

-The authors do not start the abstract with a research background and begin it with what they did.

-In the introduction, the authors go directly to the contextual framework without a research background. It is poorly organized with respect to language and the structure and it was very difficult to read. Sections 1.1 and 1.2 seems to be flipped. Please rework on the whole introduction to match your objective.

-L 31 – Some places you mention vegetation and in some forests. Be consistent. Vegetation might be the better term to use.

-L 33- Flooded what?

-Some examples of poor sentence formations are provided here. Please check the whole manuscript and correct it.(LL 33-35, LL 40-51, L 71, LL 78-79, LL 81-83, LL 85-86, LL 92-95, LL 97-100, L 120, LL 140-141

-LL 58-63- In your objectives you should state about extrapolating your data to local sites.

-references seem to be messed. Reference 1 comes after reference 2.

-Same sub-section numbering for 1.1 and 1.2.

-L 75- What cover?

-LL 205-206- Please provide the wavelengths of bands and why you considered only these bands for extracting spectral signatures. Can you also provide an image of spectral signatures for different vegetation types?

-L 292- The authors could provide the results of forest loss inside the forest masks and non-forest areas. This will be a good conservation tool to test the efficacy of defined forest areas versus non-defined areas. Also, they can then compare to previous studies which have only considered vegetated areas inside forest masks and whether their analysis match.

- Validation – the authors claim that their maps may be advantageous over other previously existing maps. In that case, they need to validate their maps against previously existing maps, especially high-resolution maps made from RapidEye. Also, the authors fail to validate the vegetation cover change which is important for such studies. The validation of different vegetation types is also missing which should include user and producer accuracies. Also, there is no validation for your classification in DRC and Para. How do you expect your extrapolations to work without validation?

-LL 352-354- Sentinel-2 based land cover maps at 20 m resolution for whole Africa is available. Why didn’t you use this?

-L 363- global?

Discussion- The authors should discuss the potential of improvement for the present study in the discussion. For eg: the Sentinel-2 data can improve spatial and spectral resolution and SAR data can improve classification in clouded areas.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 4 Report

Dear Authors,

 

Thank you for coosing your to be submitted to Remote Sensing. It represents an interesting approach for detecting the forest cover and forest cover changes in a semi-authomatic way. Hoever, despite the overall merit of the work I hav svral questions and comments to you to address: 1) the chosen methodology appears to be straightforard but the code (in Python) is applied in ArcPy. Is it possible to share it to the community to allo for a reproduction or it is proprietary? 2) the source data is the University of Marylend forest cover products. However, what is the percentage of the sampled points for training and for validation and ho do you make sure they are 100% correct to act as a ground truth? 3) The reported dynamics - the change detection approach is original - appears to be biased as you report the changes in absolute (ha) and relative (%) metrics though the overall accuracy is not very high. You kno that even if you have a very high accuracy you can state that these are area estimates which have a margin of error. This is apparent hen you start comparing ith other forest cover producs where the trends tend to deviate. Please, take this into account in your revised manuscript. 4) though iterated fe times, it is not clear what is the target spatial rsolution of the products? Is this based solely of Planet data? Please, make it distinctive in your manuscript. Thank you in advance for your kind consideration of my comments and suggestion on improvement.

 

Kind regards,

Reviewer

 

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Dear Authors,

Thank you very much for your revision. It would be nice if you have sent me files with a highlighted text about revision.

I just look into the reply to my comments and revised version. I think you have addressed my comments. I have no more comments.

Author Response

I attach the file in which We have highlighted in yellow color the changes made in the first version related to your comments.

Author Response File: Author Response.docx

Reviewer 2 Report

In Figure 8. what is 'mangle' deforestation, 'coniferas' deforestation, 'broadleaved' deforestation? Are those typos?

Author Response

The errors in the legend of Figure 8 were corrected.

Reviewer 3 Report

The present version of the manuscript is vastly improved based on the reviewer's comments. I hope the authors will do language editing before final publication. I am happy with all the changes that the authors made based on my comments. However, I felt that authors could have provided some references for future potential improvements (Eg: Sentinel).

Author Response

I am in the process of having a native-English speaker make the final corrections. Regarding your suggestion, in the last paragraph of the discussion (section 4) on potential improvements, references were added:

… These potential improvements are based on the findings of recent studies where it has been found that the combination of multi-source data, which were not used in this study, could provide significant improvements in the detection of forest types, as in the case of Sentinel- 2A (S2), Sentinel-1A (S1) in dual polarization, Shuttle Radar Topographic Mission Digital Elevation (DEM) combined with multi-temporal Landsat images [35].

Reviewer 4 Report

Dear Authors,

 

Thank you for investing your time to improve your manuscript following our recommendations. I have no further questions and comments but few minor comments:

Fig. 1 - correct for American to America in the title.

Fig. 3 - what does this figure represent - Landsat derived spectra? Please, specify.

Thank you once again for your kind co-operation.

 

Kind regards,

Reviewer

 

Author Response

Fig. 1 - correct for American to America in the title.

R// We use "Central America" when we refer to the name of the study area, and Central American when we use them as an adjective, for example "Central American region", Central "American countries" or "Central American Map".

Fig. 3 - what does this figure represent - Landsat derived spectra? Please, specify.

A // The following specification has been added to the image description: "derived from Landsat ETM+ images".