Estimating Rice Agronomic Traits Using Drone-Collected Multispectral Imagery

Round 1

Reviewer 1 Report

Manuscript is fine to be published

Reviewer 2 Report

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Manuscript: Estimating rice agronomic traits using drone-collected multispectral imagery

In this manuscript authors show their result about using a Sequoia sensor onboard UAV to derivate VI to be related with field measurements to explain agronomical threatments. In my opinion there are several lacks that have to be clarified.

Section 2.1

did you measure soil variability?m How did you prepare field experiment? Did you applied a clean irrigation? You have to be sure that filed have to any residuals of N, in other way your results can be "contaminated". Same explanation have to be performed related with water.

Incluye CRS in coordinates, line 160.

Line 223 Replace "aboveground" by "above ground level".

About VIs, did you calculate in manual way or did you use s batch process, scripts or something similar?

You do not explain anything about Ground Control Points?, GNSS sensor on Sequoia is not accurate enough as your experiment needs.

Section 2.3 

How did you identify samples on image?, did you use point value or an area?, what size of this area?

How did you measure samples coordinates? If you use a mean VIs for each area, did you measure variability?

Figure 2 is adequate but it is necessary to use Statistic to explain trends, correlations....

Are models from Table 6,7 and 8 general or site-specific?, if they are specific are not relevant and these table can be removed.

Reviewer 2 Report

This is a very interesting and well presented study using multispectral UAV to derive spatially refined estimates of canopy structural and chemical variables through vegetation indices. The test site and field design seems adequate. I propose minor comments, even if some of my comments may require some time. 

General Comments

A)     Introduction: Why not briefly talk about other nutrients (potassium , sulfur, and zinc), as well as stating the importance of N.

B)      Results: I think it’s worth doing a second analysis without the red-edge as many UAVs and multispectral instruments don’t include it. Red-edge is sensitive to smaller changes in leaf health, which may make it a valuable addition in this study to identify your canopy variables.

C)      Results: What about including results and discussing error associated with each fertilization – C0 to C3? Errors may be reduced by combining all plots as there may be plots with much lower or higher biomass, nitrogen concept, etc.

D)     Discussion: Missing a paragraph on potential spatial extrapolation of your method. How can this be done at larger scales, what is necessary in terms of time, effort, data, etc.

E)      Discussion: Further to my last point about spatial extrapolation, what about enriching your analysis with satellite imagery, e.g. from 10m Sentinel-2. Including an analysis like this would a) give clearer weight to UAVs with targeted information (or not) and b) provide evidence for extrapolating to a full region.

F)      Discussion: Consideration of climate in the discussion (as your images are over two years).

G)     Discussion: There is a recent canopy chemistry debate concerning the ability of remote sensing products to determine canopy chemistry concentrations, vs the need to consider scattering effects within canopies of different structures. E.g. Conifers and Deciduous trees have different Nitrogen, but this may be due to their very different canopy structure. A discussion on how your results may have been affected my this would be good. Please see:

Knyazikhin, Y., Schull, M. A., Stenberg, P., Mottus, M., Rautiainen, M., Yang, Y., … Myneni, R. B. (2013). Hyperspectral remote sensing of foliar nitrogen content. Proceedings of the National Academy of Sciences, 110(3), E185-E192. http://0-doi-org.brum.beds.ac.uk/10.1073/pnas.1210196109

Knyazikhin, Y., Lewis, P., & Stenberg, P. T. (2013). Reply toTownsend et al.: Decoupling contributions from canopy structure and leaf optics is critical for remote sensing leaf biochemistry Single-pixel imaging View project GEOLAND2 View project. Article in Proceedings of the National Academy of Sciences. http://0-doi-org.brum.beds.ac.uk/10.1073/pnas.1301247110

Specific Comments

1)      Line 18: Instead of saying 4 channel imager, why not say ‘green, red, red-edge, and NIR’.

2)      Line 20: State which Greek Prefecture ‘Nomos’.

3)      Line 20: ‘four different N treatments’ is vague.

4)      Line 21-22. Again, this sentence is vague.

5)      Line 50-51: Can you state the contribution of rice fields to the N2O GHG emissions globally?

6)      Line 61. Is rice in Greece mainly irrigated? What about your study site?

7)      Line 107-119: It would be worth mentioning Sentinel-2 in this discussion.

8)      Line 123-128. I feel this discussion on cameras would be important if there is temporal monitoring. What about for a snapshot (which is similar to UAV or other remote sensing instruments)?

9)      Line 47-48. Unclear and not very specific.

10)   Figure 1. Fertilizer treatment. Were the sites the same beforehand, same soil texture, depth, slope, etc.? I.e. were the sites fully controlled before for the initial soil nitrogen?

11)   Table 5. Would it be worth to show RMSEs as %, as each of the traits have different mean values? Otherwise it is difficult to compare.

Reviewer 3 Report

The topic of the paper is interesting, methods and practical tests are described very accurately.

The only doubts are on the simultaneous use of so many VIs derived from the same 4 radiometric bands; moreover the need of a two years survey to derive the regression models; this can be inconvenient to be performed in many cases.

Nevertheless, experimental results seem to show a certain soundness of the method.

The organization of the paper is good, bibliography is very wide, and English language is very good.

 One remark : in Table 1, and from line 181,  the acronym BBCH is never explained.