Combining Remote Sensing and a Geographic Information System to Map and Assess the Accessibility of Invasive Alien Species Forest Stands: Case of Acacia mearnsii on Reunion Island

Round 1

Reviewer 1 Report

In this paper, authors mapped the main forest stands using object-oriented remote sensing techniques, then the accessible forest stands taking into account favourable and unfavourable factors for their exploitation. Although the use of GIS together with remote sensing data may be new, the scientific aspect of interest is mainly the land cover classification the analysis of opportunity for forest sector derived from the exploitation of invasive aliene species. In my opinion the paper is well presented with well-structured methodology and discussion. In order to be published only some minor aspects has to be changed, as summaried below:

The very first statement needs to be better in context.

Lines 35-36. Should it also be considered a form of management, or just exploitation?

Line 41. Km2, use superscript.

Lines 64-67. Are there the alternatives that have been considered? Other fast-growing, perhaps non-invasive species?

Line 78. Are there other studies on the classification of invasive exotic species? A more in-depth review of the state of the art would be helpful for the paper.

Lines 81-82. Please, add some accuracy metrics to support the statement about the satisfactory results.

Lines 93-94. Please, provide some reference that used object-based classification.

Lines 113-114. Usually in procedures with remotely sensed data, GIS applications are exploited for visualization and sometimes processing functions.

Lines 161-164. Please, clarify the field survey. I misunderstood “recorded the geographic coordinates of the main forest stands”. It seems you acquire the border of each stands.

Line 164. Usually concerning position it is used the GNSS term. Since GPS refers on US satellite position system.

Lines 227. Do the sharp bands have spectral reflectance values in agreement with the initial values?

Lines 230-234. Consider to add Indices equation using bands name.

Lines 241-249. If available, provide at least one reference that used the same OTB function for image segmentation.

Lines 259. Although well known, add an introduction to random forests.

Line 275. Is the classification and accuracy evaluation Object-based or pixel-based?

So, how the class is assigned to each polygon?

If you consider the pixels of validation polygons, the size of polygons and the proportion between classes in training and validation should also be taken into account.

Add some information about segmented polygons. Such as mean and median area, average of pixels. These information are also useful for the results discussion.

Figure 6: please remove underlines in graph titles.

Due to the costs of Pleiades images, please, add pro e cons to the use of these images, compared to open access satellite data, such as Sentinel-2 images.

Lines 306. It is fine to emphasize the ArcMap module, but could the procedure herein presented , be processed following other programming approaches? This question is related to the relevance of GIS-based procedure, while having the vectors and rasters data all operational could be applied differently.

Add more details in the uncertainty map generation method. In remote sensing applications, the provision of uncertainty maps is now required by many researchers. In other cases, supporting the results with accuracy metrics may be enough. Give more emphasis to this map, which is certainly of interest.

English is ok for me.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This high quality article is clearly written, with good English, strong description of the motivation, aims, methodology and results. The presented methodology can be applied to other areas with minor adaptation. The discussion section provides a good overview of possible constraints of the presented methods, results and conclusions. I advised publication of the article after considering my remarks.

A few minor remarks:

133. It might be good to add the size of the study are in ha or km2

Figure 1C. The photographs of the forest types are small. It is difficult to see the differences. It would be good to increase their size. Maybe as a separate figure

Figure 1B. The white lines are missing from the legend.

Line 186 mentions a hydrographic network data set and refers to figure 1B.  I presume the hydrological network, (secondary) roads and ravines are all shown but mixed? This is unclear.

Section 2.3.3. + 4.1. Were other classification schemes considered? They may result in higher accuracies

line 324 “…and two broad types of felling techniques: mechanisable or non-mechanisable” Here something seems to be missing

The quality of Tables S2 and S3 can be improved. S2: blue symbols are small and vague, S3: the text is small and not “sharp”

Figure 4C is not analysed explicitly in the results chapter

Lines 459. Table 2 should probably be table 3

Figure 6. Some spelling correction notification remained in the figure

Lines 487-488, 496, 508, 510. It is not clear how the suitable area percentages are calculated/read from figure 6

see above

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This manuscript takes an interesting approach to how remote sensing and Geographic Information systems map and assess the accessibility of invasive alien species forest stands. The manuscript is well-structured and well-written. However, I have several questions and suggestions to the authors that in my opinion should be addressed before a possible publication. My questions mainly focus on some issues related to the methodology used. The statistical analysis is missing, I think the authors should also provide some statistical analysis to validate the results.

The references section is a bit chaotic. You have to check it carefully, especially the part with the journal's name and the year. Sometimes there is a space between the two, sometimes not, sometimes there is a comma, etc.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf