Regional Monitoring of Fall Armyworm (FAW) Using Early Warning Systems

Round 1

Reviewer 1 Report

The present manuscript propose a regional monitoring framework of Fall Armyworm using image analysis at different spatial to temporal resolutions.

The study and the approach proposed is interesting and consider a large study area comprising 39 study sites in three countries: Zimbawe, Tanzania and Kenya. Different imagery were used comprising satellite (Sentinel2, Planet), UAV and ground sensors. This in my opinion, this is a juge dataset that deserves to be shared and analyzed.

Nonetheless, the manuscript has strong flaws mainly in the methods description and results presentation that cannot be ignored. The manuscript in the present form lacks of scientific soundness and rigor and readiness. It is really hard to keep the thread of the discussion throughout the overall manuscript. Not least, the manuscript requires a strong and accurate English proof. Many sentences are not pertinent (contributing to the lack of readiness) and could be omitted allowing a substantial shortening.

- The abstract is not focus, keep many copy/past formatting (carriage return), repetitions and must be rewritten.

- The introduction has some flaws:

i. missing relevant and updates references

ii. useless and not pertinent sentences

iii. confusing description

- Methods section must be rearranged reconstructing the sections names because in the present form are confusing and difficult to keep. Figure 3 is useless. A new Table can be developed by inserting all the information reported from lines 221 to 236. Phenological information is missing (must be inserted in Table 1 as well). In the same Table specific BBCH stages at which flights and monitoring were performed.  I have some concerns regarding the numbers of images collected and the number of study sites. As a sake of example, in line 227 it is reported that 7 images were collected for each study site resulting in 94 Sentinel-2 images for Zimbawe. But in line 184 it is reported that eight maize fields were sampled in Zimbawe. So I expected 7 images x 8 study sites = 56 images. Perhaps I am missing something (the same for the other imagery) but I am missing out 38 images.

It is possible to include any error bar around the mean NDVI value for each of the scatter plots of Figure 5 and 7?

For the above described reasons I do not recommend to consider the present manuscript as suitable to be published in Remote Sensing. I encourage the authors for a complete and substantial reformat of the manuscript to be re-submitted.

Minor comments:

L15 hung-er? Copy/paste? correct here and hereafter in the abstract.

L16 italic

L44 add (Zea mays L.) the scientific name the first time 

L46-47 add a reference here

L82 'de'?

L84: parasitoids are not natural enemies

L100 add the reference of the seminal publication by Rouse et al (1973) here

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper compares different indices and remote sensing platforms on monitoring Fall Armyworm.

The method description may be improved by adding a section to describe the comparative analysis experiments. A diagram could be useful to illustrate the information flow and the key models.

The result section stops short of details on the final data size to evaluate their statistic soundness. Figure 6 and Figure 7 show only 11 points, which lacks the base for statistical analysis.

Specific revision:
1.    Line 247 of Page #7: Replace “in the any of” with “in any of”.
2.    Line 215 of Page #6: Acronym “a.g.l.” may be fully spelled out as “above ground level” when it appears at the first place.
3.    Line 289-292 of Page #8: The sentence, starting with “The developed CAN_EYE software…”, may be re-written into two to avoid nested clauses.
4.    The title of Section 3.1 on Page #8: “short return interval”?
5.    Line 375 of Page #10: Sentence incomplete.
6.    Line 376 of Page #10: “assessed and investigate” – use the same past tense, or keeping one verb is enough.
7.    Line 379-380 of Page #10: “…one research study…” – redundant, keeping only one.
8.    Line 400 of Page #11: “the first analyses” to be “the first analyses”

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

   The manuscript by Buchaillot et al. is devoted to an important applied problem: development of new methods of revealing fall armyworm attack on basis of measurements of time series of measurements. Description of results seems to be confused (maybe, as a result of numerous measurements under different conditions); however, I suppose that this work is interesting and perspective. I have only several minor comments and questions.

   1. Introduction: Can other reflectance indices used for revealing of attacks of fall armyworm? For example, photochemical reflectance index (PRI) can be very sensitive to photosynthetic changes and, maybe, can be used for revealing of these attacks.

   2. Figure 5: Dependence of NDVI on the level infestation of FAW seems to have two phases (an increase of NDVI and decrease of NDVI). How potential reasons of these shape?

   3.  P. 10, lines 353-361; Figure 7: Only dependence of dNDVI on level infestation of FAW was described in this part of results. Comparison of dependence of dNDVI and similar dependence of NDVI (for same results) can be potentially interesting.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

This manuscript proposed a multi-platform system to monitor the Fall Armyworm (FAW), and I think it’s a very useful tool for agricultural management. My main comment is that the technical processes should be clearer and the results could be shown better, besides, writing could be more concise. Below are my suggestions.

 

Actually, it is not informative enough in Figure 2. I suggest adding maps of the sampling fields as a submap to the administrative map and directly adding labels of the location names on the map.

In lines 238, 240, and 283, the equations for NDVI are redundant and could be more concise.

In figure 9, the values of the X-axis are not days after sowing but dates, and please also check the X-axis and Y-axis described in the figure caption.

Many contents in the section “3 Results and Discussion” explaining how the methods work should be moved into “2 Materials and Methods”, e.g. Lines 310-326, 398-406, 449-454, 493-502.

I suggest adding a regional distribution map of FAW monitored by the method proposed in this manuscript to show the performance.

In figure 8, the significant sign should be described more clearer with the test type and P value.

 

Table 1 could be moved into the appendix of this manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors replies to all my concerns and the manuscript in the present form gained readability and scientific soundness and can now be accepted for publication in Remote Sensing.

Reviewer 4 Report

The authors have responded to all my comments and I have no more suggestions now.