Mapping Gaps in Sugarcane by UAV RGB Imagery: The Lower and Earlier the Flight, the More Accurate

Round 1

Reviewer 1 Report

The manuscript titled “Mapping Gaps in Sugarcane by UAV: The Lower and Earlier 2 the Flying on the Field, the More Accurate” aims to find the optimal spatial resolution and timing (reflected by the height of plant) of using UAV to map gaps in sugarcane. The experiment design and results look solid. However, the writing and the significance of the study needs major improvement.

Major comments

1. There are many long sentences in the Abstract and other places, which makes it hard to follow the logic of the manuscript. In addition, there are grammar and format issues need to be fixed. Many terms are just given abbreviations before given the full name (e.g., GNSS, CTC, DAH etc.). The description of R2 was not consistent in the text and equation.
2. The significance of mapping gaps in sugarcane needs to be stressed. Currently, little information is given here. What kind of decision-making depends on gaps in sugarcane, replanting, fertilizing?
3. Some unnecessary information should be removed such as the capacity of battery and speed of the UAV.
4. Have the authors thought about use object-based detection method to detect the sugarcane plants to decide the gaps?

 

Minor comments

1. Instead of using MAPE in table 1, use overall accuracy.
2. How about plant height less than 0.5 m? The seedlings of sugarcane should be smaller that that.
3. Figure 4, the R2 for pixel size of 6.0 cm and plant height of 0.5 m also look good to me. I guess the authors also need to compare the flight time to show if it is necessary to go for the highest possible spatial resolution.
4. L238-241, as I mentioned before, the performance of pixel size of 6.0 cm and plant height of 0.5 m look good to me. Beside the authors have not tested when the plants are less than 0.5 m. Therefore, the discussion here are not convincing enough.
5. How the method developed by the manuscript can be used at larger scale as not every farmer will have an UAV?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

* Overview and main merits:

The paper addressed this problem: Field-level variability and map gaps in sugarcane plantations. And the main methodology is image processing of visible (RGB) imagery data produced by on-board camera (UAV). The novel aspect in this work is mainly the optimal determination of the size of pixel (or, equivalently, the flight altitude) and when to fly based on the crop's phenology (e.g., height of plant). Note that this empirical study considers a fixed flight speed and no additional sensing information besides the RGB camera and GPS coordinates of each collected image.

The accuracy of the gap length's prediction is found to be impacted by both pixel size and the height of the plant. For the sugarcane case under investigation, the authors found that the UAV must fly no longer than 45 days for the planting, ideally 30 days (equivalent to a plant height of around 0.5m) to avoid unacceptable gap estimation errors.

In general, the studies seem to limit the setup/method to gaps not smaller than 1.0m.

 

* Main concerns (reviewer's perspective):

The lack of a more detailed Related Work section. Part of this one is already present at Introduction and Discussion sections. However, it is not clear if any additional sensing technology could be adopted to mitigate the issue of image quality due denser vegetation or plant height. For instance, no specific mention about how microwaves (e.g., radar), UV, NIR, MIR, and thermal images may (or may not) as a secondary sensing system.

Another point of concern is the lack of the "Visible (RGB)" term at both title and abstract. Since hyperspectral imaging has been recently introduced for such applications, the use of this term is a way to quickly highlight the type of sensing instrumentation under consideration.

Finally, there is no information in this work of how the flight speed impact the solution.

 

* Abstract:

• Clearly show what physical parameters are being observed with RS: Visible (RGB) images, in this case.
• The explanation about height of the plant can be improved on the lines of "how the height of the plant can impact the mapping of the gaps". Meaning that there is an ideal range of the plant height where the mapping accuracy is in its optimal zone. At the end, this analysis gives the moment when the UAV must flight. This sequence can be better introduced at the abstract.
• Specify the flight parameters that can be adjusted, besides the altitude (that will define the pixel size). I understood that the other parameter of consideration is "when to fly"
• The part related to "replace on-board sensors" does not fit in the Abstract and may be highlighted only when the UAV method is compared with ground-based instruments.

 

* Questions/Suggestions:

• L.30: Any research in this area? [1,2,7]? What methods were used? How do they compare with the ones in this study - with more details?
• L.53: A better description of the work in [6] is necessary. The text does not provide how the method works:
• How does the mentioned "on-board photoelectric sensor" operate?
• GNSS is introduced in a way that makes it difficult to understand if that technology plays any other role solely pinpointing the global coordinates.
• L.54: Acronym for GNSS is missing.
• L.53-75: There is an emphasis on the cross-comparison between sensors hosted by ground vehicles and UAVs that may not be necessary since this is not a survey paper. However, what is really missing is a deeper discussion about existing UAV-related works and this one, more on the lines of LL.75-81 with added references. Also, how other sensing technologies, in conjunction or not with RGB images have been used.
• L.84-85: The second objective is far from clear and may require at least one illustrative example.
• L.85: [TYPO] Ending "."
• L.109: Missing maximum horizontal distance in relation to the UAV operator.
• Fig.2: Can we assume that the plant heights 0.5m..1.7m correspond to the mentioned 30, 45, 60, 75 and 75 days of cultivation? This seems to be the case based on Fig.3. If positive, it would be important to add an explanation at LL.117.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

No further work to be done. the paper is well writing, the the research is well explained as well the experimetal work. The results are well presented and clear. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors had addressed my concerns.

Author Response

Thank you again for your thoughtful comments and suggestions to improve quality and strengthen the paper.