Cropland Suitability Assessment Using Satellite-Based Biophysical Vegetation Properties and Machine Learning
Round 1
Reviewer 1 Report
Dear authors and editor,
The manuscript “Cropland suitability assessment using satellite-based biophysical vegetation properties and machine learning” addresses the topic of machine learning for determining land suitability instead of traditional GIS-based approaches. The authors focus on the question: may land suitability for machine learning models based on LAI and FAPAR be more objective? To answer this question, the authors implemented a multicriteria method based on data fusion and machine learning.
The strong points of the manuscript are the scientific approach and the quality of writing whereas the weak points are the data fusion and the low spatial resolution of some data (why did they not use Landsat or Sentinel 2 for some key datasets?). For these reasons, I am a bit concerned that some errors may be included in the final classification. Maybe the authors could discuss this a bit. The topic is interesting for the scientific community and the agricultural communities.
Introduction: The Introduction is well written but is really a bit too long. I suggest lingering a bit less on the examples of the use of LAI and FAPAR, as they may be redundant.
Material and Methods and Results are well explained and clearly presented. The resolution of some figures (e.g. Figure 7) could be improved.
Discussion and Conclusions: adequate.
I have noticed some confusion throughout the manuscript on the kind of the analysis: sometimes the authors say this is a solely GIS-based multicriteria analysis (e.g. keywords and line 152), sometimes they say it’s not (e.g. line 464). This issue must be fixed.
The tables in the Appendix should be called A1 and A2.
Here are some specific comments:
Line 221: could you please report the resolution of the orthophoto?
Line 248: could you please give more information about the bioclimatic indices?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Cropland suitability assessment using satellite-based biophysical vegetation properties and machine learning
The paper is presenting a cropland suitability assessment approach based on machine learning. This paper is also providing a detailed information on the methodology, data processing required and the limitations of each approach.
This type of assessment is relevant due to the future climate change and rapid population growth, the paper needs some minor revisions.
- Table 1. Include the values of precipitation and temperature for covering days of the year.
- list characteristics of the soil for subset A and subset B in line 251 or 252
- During the study, were there any dry years or drought years. How does the results vary with dry and wet years? Compare and provide a brief description in this regard.
- I think this approach can also be used to assess suitable croplands for future climate change. Discuss briefly how this approach can be modified for future climate change studies
- Arrange covariates in Figure 7 for consistency between Subset A and Subset B.
- References: I see some inconsistencies in the format used for references cited. Please check the format
Author Response
Please see the attachment.
Author Response File: Author Response.pdf