Efficacy of Mapping Grassland Vegetation for Land Managers and Wildlife Researchers Using sUAS
Round 1
Reviewer 1 Report
This study used a consumer sUAS, semi-automated flight planning software, and graphical user interface GIS software to classify grassland vegetation with the aim of providing a user-friendly framework for managers and ecological researchers. It is interesting may be usable for land 2 managers and wildlife researchers. While there are some drawbacks as follows:
1. Introduction, the structure is disordered and illogical, please reorganize this section.
2. Materials and Methods, it is not clear and visualized, I mean you need provide some figures and tables to show the important information, rather than only some detail describing. Furthermore, no innovativeness was found
3. Results, it is too simple and hard to support your conclusions, it need to be dug deeper.
4. Same to 1, it need reorganized logically.
5. There are some mistakes on text formatting, e.g. Line 590-591, no title, and so on. Furthermore, it looks no enough innovativeness, please highlight it if there is some very novel method and idea.
Author Response
- We modified and slightly reorganized the introduction to make it clearer.
- We provided a new figure that represents a flow chart to assist ecologists and wildlife managers as to how our results could be used. This study was not intended to provide a new innovative way to conduct remote sensing. Instead, it was conducted to determine if a simplistic approach of using sUAS that could be used by the average ecologist or wildlife manager may provide adequate increased accuracy relative to freely available satellite and manned aerial imagery to justify the greater costs.
- We modified the reporting of our results in an attempt to make them clearer.
- We modified the discussion in an attempt to make them clearer.
- We have addressed the clerical error and attempted to better emphasize the goal wasn’t to develop a new innovative approach but determine if a simplistic approach of using sUAS imagery that is user-friendly to the average ecologist provide adequate improvement over satellite and manned aerial imagery to justify the greater costs.
Reviewer 2 Report
Very well prepared and presented study about the benefits from the use of UAVs in the vegetation mapping processes.
The manuscript is well organized and prepared. No major language issues were detected.
The study is supported by significant amount of statistical data and is well described, analysed and discussed.
My only recommendation is to check the formatting of the headings for the sections and subsections - in some cases there are points prior the text (Subsection 2.2, Section 3, etc.), as well as unequal spacing for the different headings from the same level.
My general recommendation is that the paper is to be accepted for publication in the journal.
Author Response
We corrected heading and spacing errors
Reviewer 3 Report
The authors sought to classify the vegetation classes based on drone-derived RGB remotely sensed data in comparison to the performance of airborne multispectral remotely sensed data. The methodology followed by the authors is not clear and repeatable. To this end, there is limited merit as far as the novelty of this work is concerned. for detailed comments and concerns check the annotations on the attached pdf.
Comments for author File: 
Comments.pdf
Author Response
We modified the introduction following most of the reviewer recommendations to reduce redundancy.
We added a study area map as well as a flow-chart to clarify our results recommendations.
We reiterate on page 5 that CHM stands for canopy height models
We justify the assumption on page 5 and point out that any lack of validity regarding this assumption should only decrease our accuracy, thus would not impact our inference.
In Section 2.1.3: we provide additional justification for this assumption regarding the 5 year old lidar data
Section 2.2.2. We used DroneDeploy derived CHM estimates for the manually digitized classification because we felt it better demonstrated the capabilities associated with using the drone and we were able to manually correct for any negative estimates. We were concerned the negative estimates from the DromeDeploy derived CHM estimates would confound the random forest classifications so we used CHM estimates derived from the Illinois Height Modernization Project dataset (ILHMP; I.G.D.C. 2015) for automatic random forest classifications. We have attempted to clarify this in the text.
Section 2.3.1: We rephrased this to clarify in the manuscript. What we were trying to communicate is that yes, we used field measures, but the person doing the manual digitization was not the same person that did the field measures. That ensured the person digitizing the layers could not rely on his/her memory from collecting the field data collection to increase accuracy.
Section 2.3.3: Yes generated 500 stratified random points for use during field validation.
Round 2
Reviewer 1 Report
This work looks nornal, though the authors tried to develop a repeatable and approachable workflow simplistic enough for the ecologist to effectively map for the ecologist. I mean more work need to do in the future.
Besides, some mistakes need to check, e.g. Line 97 what doyou mean "average"? etc.
Author Response
Changed "average ecologist" to "most ecologists"
Reviewer 3 Report
The authors have addressed the concerns raised in the previous version of the manuscript.
Author Response
No response needed,
