Estimation of Wheat Plant Height and Biomass by Combining UAV Imagery and Elevation Data

Round 1

Reviewer 1 Report

 The manuscript is soild in the introduction section and provides detailed state of the art of biomass estimation methods and the weaknesses of some methods.

The methodology could be enhanced with the arguments why the variables, varieties, planting densities, and N application rates were used to understand their contribution in discussions.

In the discussion (lines 413-415), plant density is a factor that could affect the extracted PH. However, it does not mention anything about it in the manuscript. The results and discussion could be improved with the inclusion of this variable.

In addition, in the discussion section, it is suggested to describe in more detail the reasons for incorporating plant height to improve the model (improve or extend Lines 479-482). Thus, it is not evident if the variation of wheat height in the last phenological stages is relevant for its incorporation in the models; a cost-benefit assessment would be interesting.

Specific observations are attached below: 

Table1. Does the term "number of samples" refer to plant sampling?

 L231: was any statistical method used for the validation of plant height?

L184: what does "image repetition rate" mean?

L231: does the number of GCP refer to the total during the two years of the experiment?

L334: Why did VARI not reach the same levels as other indices?

L347: What do you mean by two periods?

L 367-370: Why was the accuracy lower in the jointing phase?

L382-383: revise the wording

L413-416: In the methodology, three variables were considered in the two-year field experiment, including different varieties, planting densities, and N application rates, but in the results and discussion sections, these variables are not analyzed.

L441: Explain why the GPR model in the booting phase did not show an increase in the coefficient of determination.

L452: What do you mean by multiple periods?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

General comments:

This paper demonstrates that crop height can further improve the wheat biomass estimation that is only based on color indices (CIS). Specifically, the DEM and DSM are derived from UAV with RTK and the difference is used for estimating the crop height (CH). Both average and cumulative percentile are used for aggregating the crop height at the plot level. Then the paper compares models based on only CIS and combined CIS&CH. The results show the unique contribution of the crop height for the wheat biomass estimation. In general, this paper is well organized and lots of work have been done for this two-year field trial. However, there are some details missing in the paper and some concerns need to be addressed, which are as follows.

1. L74-76: DEM and DTM are slightly different, suggest using DEM. “(representing the vegetation)” is not clear, suggest changing to “(the altitude of the vegetation on top of the bare soil)”. For “(representing the terrain)”, suggest changing to “(the altitude of the bare soil)”.
2. The third objective is not well addressed in this paper, suggest deleting it. If it is necessary to keep this objective, more work needs to be done (a. Why the proposed method is different from other existing methods? b. For “over the multiple seasons”, how does the model trained in one year perform in another year?)
3. L150: how is the 216 calculated? 3 (replicates) x 3 (varieties) x 3 (densities) x 4 (N applications) = 108.
4. For “m2” and “R2”, use the superscript for “2” (m² and R²).
5. L156-161: In Fig 1, there are only 12 GCPs. What is the specification means in terms of “0.8m, 1m and 1.2m” (there should be a comma before ‘and’)? Suggest adding the precision of the RTK for the altitude in L161,
6. Table 1: Is that “Altitude” or “Flight Height Above Ground”?
7. L178: What are the precisions of the built-in RTK regarding longitude, latitude, and altitude?
8. L200: twenty plants or 108 plants?
9. L214: give the full name of RGRI.
10. Formula (1): why there are numbers 1 and 2?
11. L231: How are the plant heights validated by 648 samples? Are they aggregated at the plot level or at the sample locations?
12. L238: should the “lower plant boundary” be changed to “base plant boundary”?
13. Formula 2-4, delete the numbers
14. The 648 samples represent 648 plots, right?
15. Table 3: add units
16. L318: why only focus on the 90^th-100^th percentile? Use the superscript for the “th”.
17. Figure 8: the font is too small to read.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx