Estimating Rangeland Fine Fuel Biomass in Western Texas Using High-Resolution Aerial Imagery and Machine Learning
Round 1
Reviewer 1 Report
The manuscript investigated model performance at different scales and tried to identify significant factors that influence the model accuracy. Overall, the methodology seems reasonable. And experiments showed adequate evidence supporting what the authors claimed.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Comments to the Author
This paper presents a interesting issue about “Estimating rangeland fine fuel biomass in western Texas using 2 high‐resolution aerial imagery and machine learning”. I have the following comments.
1) In figure 1 I suggest add datum scale and a quadrante with coordinates.
2) Why e field samples were collected from October 23rd, 2018, to January 10th, 2019? Explain the main reason.
3) I suggest insert in the 2.2 section a table with name, description and image of classification for fuel types.
3) In 2.4 section, the author writes (Dillon et al., 2011), I suggest change to number in brackets.
4) The user's accuracy in section 3.1 is presented in the text and the author repeats it in table 2. I suggest the authors to leave only table 2 and discuss the result in the text.
5) In figure 2 I suggest to put a part of figure with zoom, the more significant result.
6) In figure 3 and 4 I suggest improving the resolution and size.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Estimating rangeland fine fuel biomass in western Texas using high‐resolution aerial imagery and machine learning
Li et al. (2022) have estimated rangeland fine fuel biomass in Texas using aerial images, a Landsat-derived image, and the Random Forest model. The thematic involved in this Article is within the scope of Remote Sensing and would be interesting to the readers of the journal. This is a solid manuscript, all methods are clearly explained, and all results are adequately shown and discussed. I also agree with the statistical procedure adopted to validate the models. Moreover, I congratulate the authors for discussing the limitations of the method adopted.
I have three main comments regarding the manuscript: (i) the Introduction section needs further development. It is too short and does not explore the novelty of the manuscript, (ii) the manuscript would benefit from an English style correction by a native-speaking colleague, (iii) a coordinate system and a scale bar must be added to all maps.
Please see a few minor comments below:
Line 31: Is support the most adequate term?
Line 50: Please use the sensors’ names instead of the satellites’ names.
Lines 62-68: Please rewrite these sentences in the present form.
Lines 68-69: Is this correct? Please verify.
Lines 80-82: Which model?
Line 102: Please change the color of the sample points in this figure. Moreover, please add coordinates and a scale to this map. Source? Which sensor/satellite? Date? Composition?
Lines 108-109: How did these corrections were performed?
Line 112: Which bands?
Lines 113-116: Please better explain these models.
Lines 118-122: Please explain how this image was used in the study.
Line 148: Please fix this citation style.
Line 195: Please see the comments for Figure 1. These changes must be applied to all figures of the manuscript.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 4 Report
I have several questions:
1.- The parameters used by the classifier must be indicated.
2.- 70% of the data was used for training and the remaining 30% for validation. Could you explain the validation method used?.
3.- It would have been very useful to test the robustness of the algorithm by performing a test with a dataset completely independent of training and validation.
4.- Additional parameters other than accuracy, e.g. sensitivity and specificity, should be given.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
