A Remote Sensing Method to Monitor Water, Aquatic Vegetation, and Invasive Water Hyacinth at National Extents
Round 1
Reviewer 1 Report
An interesting paper very (too) dense using both Landsat-8 and Sentinel-2 images. We appreciated it, and the relationships between field data and remote sensing of Invasive Aquatic Plants. Some pertinent results on Water Hyacinth Ecology in South Africa explain its distribution as presented with remote sensing tools. The particular focus in helping colleagues to use their method and the tools they offer are a very tremendous positive position, but the readers must believe their conclusions without sufficient explanations.
We agree about the interest not to de dependent on previous maps of water bodies (L. 441-442), as well as using Sentinel-2 for aquatic vegetation detection.
About Water Hyacinth colonisation of temporary waterbodies, it could be useful to follow its dynamics along the year, its resistance to dryness, its possibility of seed production and seeding. Taking into account its phenological changes (L. 477-482) is highly pertinent.
This study highlights original results in remote sensing:
- the importance of spectral bands in the thermal infrared. This information, correctly discussed, shows the importance of knowing the phenological cycle of the water hyacinth. Complements on this subject are necessary.
- The superior performance of TOA images compared to BOA images in detecting aquatic plants: This point deserves more detailed explanations.
The link between the spatial resolution of the images used, including thermal images, and the size of the objects detected is not discussed. What is the minimum size of hyacinth canopy that is detectable? And are the undetected parts important in the monitoring and management of the hyacinth? A scale bar should be added to the images in Figure 7.
Nevertheless, some improvements are suitable.
- For a better understanding, do precise that Water Hyacinth stands are monospecific and usually very large. For the 4 others IAAP, the work could be less easy!
- Most used indices should be presented not only with a reference! Specially the MNDWI, …
- Figure 3: maps should be numbered: the legend would be clearer.
- Figure 4: where is it? I understood later that it was Hartbeespoort dam (cf fig. 7): it is right? For me using A, B, C instead of top, bottom left and bottom right would be clearer.
- Figure 5 announced lines 181-182 is lacking. What is it, and is it useful? Thus, all the following figures must have changed numbers if the authors keep this lacking figure 5 (and a quick resubmission should be compulsory).
- Figure 6: where?
- Table 2: Validation loss 0.740 ± 0.0078: is it an error? If not, explain why (I have not found any explanation).
Minor comments:
- 77 it is not a bias unless a difference!
- 440: 4.2. Stage 2 instead of stage 1? (cf fig. 2)?
- 476: 4.3. Stage 3 instead of stage 2?
- For the reference list, ref. 22: why capitals? I didn’t understand where the paper/book of ref. 38 is published
Author Response
Thank you for these valuable comments. Especially the comments concerning the TOA data outperforming the BOA data for aquatic vegetation detection and the consequences of the minimum detectable object size. Addressing these comments has significantly improved the manuscript. Please find detailed responses to each of your comments in the attached document.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
I enjoyed reading this paper. It is well-written, well-organized and well-presented. The use of the Google Earth Engine (GEE) platform to track invasive species is an upcoming field and not a trivial problem hence this paper advances the field of Remote Sensing in a meaningful way. The objective of the paper seems 3-fold:
- delineate accurately the boundaries of water bodies
- map aquatic vegetation in general within these water bodies
- differentiate target species (in this case, water hyacinth - WH) from other aquatic vegetation
There are some changes that I would recommend to make the paper easier to understand and follow. I'll organize them by section but see the attached pdf for specific comments.
Introduction: The introduction does a good job of introducing the datasets, the GEE platform, and the need for the study. However, it does not discuss the methods used in the study (water detection, aquatic veg detection through methods similar to those used in this study) in the context of other studies and methods currently in use for the 3 objectives outlined above. The introduction should make the case for the use of the methods that the study ends up implementing.
Methods: More explanation is needed in methods. Even though the authors cite the paper for their index MNDWI, the formula is simple enough that it needs to be rewritten here. The reader might not have read the earlier publication. Also, the method for arriving at the 95th percentile is unclear. Clearly explain how you use these thresholds, are they per pixel, etc. Percentile index values are used later in the methodology too so I would suggest, introduce that idea in the introduction itself, especially if others have used it before. And what makes them powerful. I think the strength of the study is that it is a spatio-temporal study - it used temporal and spatial indices or hybrid ones. You should highlight that.
Equation 1 is also unclear and should be explained further. I don't mean the equation itself but why it works. Why does the inequality give a more reliable mapping of water than the index by itself? This is never quite explained. Even if you have explained this in some other publication, it needs a simple explanation here. What is the logic behind this equation?
Similarly the OTSU and canny techniques should also be simply explained (at least in 1-2 sentences). And the inputs those techniques need and the output they provide. Same for random forest. Basically the methods section probably needs the most clarification.
For the rest of the comments, refer to the comments within the attached pdf.
Comments for author File: Comments.pdf
Author Response
Thank you for these valuable comments. We found them helpful and believe that the addressed comments will assist in making the manuscript more accessible to a wider audience. Please find detailed responses to each of your comments in the attached document.
Author Response File: Author Response.pdf
Reviewer 3 Report
Interesting research that involves a new method to identify one invasive species (water hyacinth) in nationwide.The article is correctly written and understood.
Detailed comments
- In the "Introduction" and/or "Discussion" section, there is no information about the results of the previous publications on the classification of water hyacinth. Including the methods used and the accuracy achieved. The authors do not cite one in more important articles about the examined species- https://www.sciencedirect.com/science/article/abs/pii/S2352938517302148
- L115 - The authors write "They were combined with the ‘none’ class (i.e., other riparian or aquatic vegetation)" - What other types of vegetation were used? Do they present all the variability of water vegetation in the analyzed area?
- Water hyacinth - Please provide the full Latin name of the species.
- Fig 10 should be expanded with the classification results shown in a different scale. The current scale does not allow a visual evaluation of the quality of the result.
Author Response
Thank you for your comments. Readers will be able to easily get a visual idea of the results of the species discrimination without going to the interactive map. Please refer to the attached document for detailed responses to your comments.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments to authors, version 2
Abstract and keywords: harmonize the Latin name!
I am surprised of the difference between the 1st version and the 2nd one in the calculation of water hyacinth cover!
The presentation is much clearer
The authors have taken into account most of our previous remarks and recommendations, and I have particularly appreciated the additive explanations in sections 4.2, 4.3 and 4.5.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
Most of my review concerns have been addressed. The only remaining concern I have is regarding figures and figure captions, specifically for #6, #7 and #10. The authors tend not to have legends on the figures or to explain the colors in captions. For example, in Figure 6, what do the colors green, black, red and blue indicate? The caption is just as mystifying. In Figure 7, the first 2 rows seem to be the same area. Are they the results from the 2 different methods? If so, what about the last 2 rows. The whole figure is therefore unclear. Again, the caption is not sufficiently explanatory. Naming the rows will help. Clearly explain what each row is showing.
Figure 10, legend is missing again - for the blue and green colors. Also, blue and green blend into each other (the small figure panel to the left of the label B or C). Not sufficient contrast. Choose a contrast color - red WH and blue water, something like that.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
The text in its current form I accept. No additional comments.
Author Response
Reviewer 3 was satisfied with the manuscript. Many thanks for your earlier comments and suggestions.
