Sentinel-2 Images Based Modeling of Grassland Above-Ground Biomass Using Random Forest Algorithm: A Case Study on the Tibetan Plateau
Round 1
Reviewer 1 Report
This paper proposed a new approach for grassland above-ground biomass in sentinel image based on feature extraction. However this work is not very clear. The whole process was more like a strategy of several general processes, although it can achieve better results than recently methods. The evaluation of the proposed method is simple. The core algorithm needed to be strengthened and detailed. Therefore, in the current state, the paper does not deserve to be published. Here (just for indication) some ways to improve the current version of the paper:
1. I don't think the mentioned main objectives in the manuscript were appropriate. Although the feature extraction is significant for RF model, the goal is to AGP estimation process. Therefore, the main contribution should be focused and exact.
2. The sample selection plots have been spread around the Alpine. Do the position have the relationship between the training set and test set? How to make sure the whole samples can represent the study area? Do the different grassland type have the effect to the AGB? I think there are many concerning points that should be introduced in the Data set section.
3. The author described the procedure of RFE algorithm in details. However there are several traditional algorithms can be implemented in this step. Since you think the feature extraction is significant for the whole procedure, how to prove this algorithm is the most effective than others?
4. Several abbreviations have showed in the section 4.2, please give your explanation.
5. The author have chosen the impurity importance and permutation importance to do the estimation? What is the meaning of the comparison of the performance with training data and test data.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
[Q1] Line 7, this sentence is problematic and incomplete. Please expands the research aim. The algorithm can be included in the research as well. Otherwise, readers will be missing here.
[Q2] line 21-28 can be moved backwards. You can start this paper with AGB, since the Tibet will be a case context for your research.
[Q3] From line 56 to line 57, it is not logic. Please rewrite or add some sentences here.
[Q4] Why RF? Why not other algorithms? Different models will generate diverse performance? Please justify this? The following papers can be referred.
"Soil exchangeable cations estimation using Vis-NIR spectroscopy in different depths: Effects of multiple calibration models and spiking." Computers and Electronics in Agriculture 182 (2021): 105990.
"Clay content mapping and uncertainty estimation using weighted model averaging." Catena 209 (2022): 105791.
[Q5] In the study area section. Please add more information to justify the Tibetan Plateau is a good case study area to support your research purpose.
[Q6] Section 2.2, please add more information about the sampling campaign.
[Q7] Section 2.5, Please add the date of your data in the supplementary materials.
[Q8] Please justify if other methods are also suitable and more capable of estimating. Moreover, please refer the documents in Q4, for adding the indicators for accuracy evaluation.
[Q9] According to Figure 6, the RF estimation accuracy is not high actually. Please discuss if other methods can be the alterative solutions. See papers in Q4.
[Q10] Please add a section of limitation and future work.
[Q11] Please rewrite the conclusion in an open view, instead of the results of this paper.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
Dear Authors
I have now completed the review of the manuscript titled “Sentinel-2 images-based modeling of grassland above-ground biomass using random forest algorithm: A case study on the Tibetan Plateau”. This study focused on the selection of S2 data and other auxiliary data. In this study, ten S2 bands, ten S2-derived vegetation indexes, 218 pieces of AGB field survey data, four types of meteorological data and three types of topographic data were used as the alternative input features for the AGB estimation model.
The topic is quite interesting and relevant. I have few comments to improve the quality and clarity of the manuscript.
- This manuscript needs clarity on images used, preprocessing details, seasonality, and detailed parameters while comparing two indices.
- Section 3.4, The number of regression trees was set to 1000. Why 1000?
- If we change the number of regression trees what will be the effect on the output? how we can optimize this number?
- Authors should include recent studies which used indices with ML[1-3].
- Authors should add the computational complexity of the RF model, see CDLSTM, SMOTEDNN, etc.
- Results and discussion are represented in a detailed manner, which is good.
[1] Planetscope Nanosatellites Image Classification Using Machine Learning
[2] Evaluating the Effectiveness of Machine Learning and Deep Learning Models Combined Time-Series Satellite Data for Multiple Crop Types Classification over a Large-Scale Region
[3] Sentinel-1 to NDVI for Agricultural Fields Using Hyperlocal Dynamic Machine Learning Approach
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Dear authors,
I have now completed the review of the revised manuscript titled " Sentinel-2 images based modeling of grassland above-ground biomass using random forest algorithm: A case study on the Tibetan Plateau”. I thinnk the authors have been made good efforts about the revised manuscript by giving point-to-point response. I suggest the editors accept it.
Reviewer 2 Report
Authors have well addressed all my concerns.
Reviewer 3 Report
Dear Authors
I have now completed the review of the revised manuscript titled " Sentinel-2 images based modeling of grassland above-ground biomass using random forest algorithm: A case study on the Tibetan Plateau”. I have observed that the authors put in good efforts to address all the comments satisfactorily.