Using UAV LiDAR to Extract Vegetation Parameters of Inner Mongolian Grassland

Round 1

Reviewer 1 Report

The manuscript "Using RIEGL VUX-1 UAV LiDAR to Extract the Vegetation Parameters of Inner Mongolian Grassland" deals with the evaluation of a LiDAR system installed on an RPA to retrieve grassland biophysical attributes such as canopy height, fractional coverage, and aboveground biomass. These parameters are of great interest to a variety of research fields ranging from ecology&biology to foddy&forage production. The measurements obtained by the RPA are compared with costly field data for validation purposes and comparisons. Although worth and scientific sound, the manuscript still lacks in some methodological aspects to allow reproducibility. Interestingly, some methodological aspects are explained in Results that required rephrasing sentences. Although several results are presented, the discussion and comparison with the state of the art of research is weak since only a few papers listed in references are from the last years. I recommend a resubmission to further accept it for publication in Remote Sensing. Although not being a native speaker, I feel some sentences could pass through a revision for clarity.

Abstract

L13: please rephrase the sentence adding some key-words such as "he retrieval of grasslands...", "the estimation of grasslands...";

L27: what are the implications and lessons learned?

Introduction

L35: where and in which environments?

L39: low or slow?

L49: define vegetation cover;

L58-59: please check the definition correctly;

L69-71: decreased vs. deceased?

L75: different grazing intensities are created? or evaluated on a site where there are good control and simulation of grazing intensities;

In general, the introduction shall reflect a better state of the art of research. There are some recent publications (even in Remote Sensing) dealing with the estimation of grasslands and pasture parameters from RPA that could be cited here and then the approach and advantages/disadvantages of LiDAR systems could be presented;

Material and Methods

L82: please explain shortly the goals of this experimental area;

L86: refer to hectares;

L91-92: how important are those species in the region?

Please consider adding a table with the parameters of the plots shown in Figure 1B;

Please also detail the topography of the experimental area;

L98-104: at the first moment I would ask criteria, software, parameters/settings and strategy, so, please explain that the steps shown on the methodological flowchart will be detailed in forthcoming sections;

Figure 2: please add the number of the sections/sub-sections on the flowchart;

L113-114: please consider adding two and four (rather than using numbers);

L120: please consider adding some pictures of the fieldwork or field sites when possible;

L141: add when possible some more info about the weight of the system on the platform;

Figure3B: if possible add a picture of the LiDAR system coupled with RPA;

L157: please explain how good LiDAR points capture ground features in very dense grass coverage; did you consider using some baren soil areas for calibration? or can you suggest it for future studies? or flight before and after the experiment?

L158: although being correct, I would suggest using DTM rather than DEM; DSM and DTM are specific terms and are both DEM and avoid misunderstandings;

L201: please write some more sentences explaining the strategy of the data analysis;

Results

L203: what is the percentage of points per sq meter that reached the ground surface?

L219-220: what does it mean in practical terms? please explain;

L259-260: explain it better and bring some more results for clarity;

L289-292: this is methodology;

Table2: add some previous sentences about data normality analysis before applying correlation analysis;

L314: only tree biomass? are there some with grasslands?

Figure 8: please consider applying a statistical test to evaluate changes and also a difference map to account for differences in estimation;

Discussion

Some more aspects of open research questions would be interesting. Some more comparisons with recent papers are still needed. Therefore consider adding the potential of RPA RGB measurements vs. RPA LiDAR measurements (citing advantages and disadvantages). Additionally, such measurements (as those demonstrated here) would also sound interesting when considering validation from both ICESAT-2 and GEDI as well as SAR systems;

L403: check typo of Rigel v. Riegl (in the entire manuscript);

Conclusions

L430: don't forget to add some mention about the autonomy of these systems for mapping large grasslands areas;

Thank you for the opportunity to evaluate this interesting manuscript.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

This paper examined the ability of a RIEGL VUX-1 UAV LiDAR system to detect to model AGB at a 0.1-m pixel resolution in under different grazing intensities. Four main parameters were used to estimate biomass and a difference in lidar-derived heights were highlighted  as a result of varying flight altitudes. The following revisions are suggested, but overall, this study should be improved. The significance of the RIEGL VUX-1 UAV LiDAR system should be emphasized at the very beginning, in the Abstract, and significance of the study should be further highlighted. Also, a statement on the modeling approach should be included here. Results, based on the R² were not very good (e.g. less than 0.6); authors should comment on this in the Abstract also.  The significance of the study should also be highlighted in the Introduction section (as indicated below) and more background information should be incorporated in the Introduction. As is, the Introduction is very short and reads more like a quick summary, and more information is highly recommended to support this study. As described in the Section 2.1 and from Figure 2, it seems all the data were used to train the model (vs splitting into training and test sets)? Please clarify. Please see more specific comments below.

L. 18: In the Abstract, state what is unique about RIEGL VUX-1 UAV LiDAR systems or at least describe why it warrants a study of the ability to detect grassland vegetation parameters. The importance of the study is not emphasized here but should be.

In the Abstract section, include a description of your modeling approach to estimating biomass.

L. 37 – 38: Add relevance references to support this statement (on FVC).

L. 39: Include how is grassland biomass traditionally measured from field sampling.

L. 75: AGB prediction maps were generated at a high resolution, 0.1 m – how does this compare to other studies?

L. 66: Authors mention “Wang et al.[19] used Velodyne’s HDL-32E UAV LiDAR system” for estimating biomass for the same area. Why is it important to understand the ability of RIEGLVUX-1 sensors to model similar parameters? Why is this study significant?

L. 98: What software was used to pre-process/process the lidar data?

L. 100: Why use a simple regression model? Is this the conventional approach to estimating grassland biomass? Some reference to the literature on modeling approaches could be integrated.

Figure 2: Were data randomly split into training and test sets? From the figure, it seems all the data were used to train the model.

L. 281: What is the regional scale? The study focuses on a local site. The inclusion of regional scale findings here is unclear.

L. 287: What “appropriate methods” were used to filter independent variables? Please clarify.

L. 289: Should be included in the Methods section (i.e. software used).

L. 295: Did authors check for multicollinearity? An example should be variance inflation factors (VIFs) below a certain threshold, based on the literature.

Figure 9: This seems more appropriate for the Results section

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The manuscript "Using RIEGLVUX-1 UAV LiDAR to Extract the Vegetation Parameters of Inner Mongolian Grassland" is resubmitted to Remote Sensing. Although the majority of the concerns were properly addressed, the current version of the manuscript still requires some changes for enhancing the manuscript. There was also a misuse of the "track changes" tool because the current changes did not substitute the former one (for example L149 and L150). The version must be final and up to date. Some methodological aspects are still not well explained and did not allow reproducibility. 

The discussion section did not include new references and the state of the art of research as requested. The addition of new references and comparison with international literature is necessary. Finally, the autonomy of the system is limited, and therefore, insights for the generation of maps can be provided in small regions. However, the system can be used to gather training and validation datasets for remote sensing systems at the orbital level and delivering data in a coarser spatial resolution system (such as SAR, optical and even GEDI and ICESAT-2 could be mentioned), and be therefore useful for estimating grassland attributes at larger regions. This concept could be explored more properly in the discussion and conclusion sections.

Some other minor comments:

L16, L49: add a space between sentences (these are only some examples, check this for the entire manuscript);

L17, L42, L53: add a dot between sentences (these are only some examples, check this for the entire manuscript);

L34: add the sentence "that could be useful for estimating grasslands in smaller areas and serving as references for other remote sensing datasets for estimating then in larger areas";

L50: check sentence meaning (change to time-consuming and costly);
L108: check the proper unit? It is also important to highlight that although the study area is very small, there is a high control on several "in-situ" parameters;

L114: are or is? please take some time to recheck the entire manuscript, there are still several typos;

L130-131: do you mean DSM minus DTM? please explain in more detail;

L132-143: I could not follow how you obtain such information since this IDL script is not provided or detailed. Impossible to reproduce it;

Please check the technical consistency of some key-sentences. For example:

L354-355: Analysis of LiDAR and ground-measured data, the LiDAR-derived canopy height lower than ground-measured.

do you mean?

Analysis of LiDAR and ground-measured data showed that the LiDAR-derived canopy height was slightly lower than ground-measured.

If this is correct, then the following sentence is wrong: "This shows that LiDAR data has the ability to penetrate the grassland canopy.". Please explain. I understood that if the ground coverage is high the penetration is difficult what is also related to the beam projection on the ground and also the discretization of the beam signal (if discrete or full-wave) to better characterize the ground component. Such a physical mechanism must be better explained.

Please improve the quality of the Figures. Please check that Figure 9 appears twice. 

Convert all units from g/m2 to g.m²;

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

Please check tracked changes throughout. Some sentences simply repeat information from the preceding sentence and overall, do not read well. Sentences need to be revised to make clearer, begin with a capital letter and completed with full stops. Currently, the revisions are confusing and unclear.

Line 14-16: Why is the first sentence repeated here?

Line 16-19: This sentence is not clear. Please revise to make concise and clear.  

Line 27:  remove “also”. This is the first result listed. State the results first, specifically the predictors and other quantitatively findings and then state as one of the overall conclusions that the sensor can improve modeling accuracy (given your findings).

Line 33-34: No need to re-state this.

Line 47: change to “includes”

Line 50: low to slow?

Line 51: People? Please revise to make clearer. Perhaps use “Biomass is traditionally measured..”

Line 69: Suggest using “lidar” instead of “LiDAR”

Line 89: Replace “In a word” with something more appropriate; perhaps “In summary”?

Line 90: How is it accurate? The R² reported doesn’t support this statement.

Line 94: This sentence seems repetitive.

Line 96: created or evaluated?

Line 99-101: Expand on this point further and integrate more references. Describe how the conclusions provide a theoretical reference. One sentence is inadequate.

Line 113: constructive or dominant?

Please proof-read thoroughly. There are many errors throughout. Line 124 states “The first step was LiDAR data pre-processing” and then immediately, authors state “Firstly, we used..” There are many similar mistakes throughout this paper. All are not listed here because they exist throughout the manuscript. Please revise accordingly.

Line 236: No mention of how multicollinearity was assessed; if variance inflation factors were examined, this should be included. What do you mean by “used simple regression model to filter the independent variables”? Note it should be variance inflation factor NOT variance expansion factor as authors indicate in their response to multicollinearity comment in the first round of revisions.

Author Response

Please see the attachment.

Author Response File: Author Response.docx