A Novel Remote Sensing-Based Modeling Approach for Maize Light Extinction Coefficient Determination

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study developed a novel semi-empirical model for the light extinction coefficient (kp) of maize, integrating data from various remote sensing platforms. Evaluation against existing models showed a 44% improvement in accuracy. The model demonstrated robustness across multiple sensors, with the normalized difference vegetation index (NDVI) playing a critical role. The research suggests practical applications and calls for further exploration under varied environmental conditions.

Comments

1.      How does the proposed semi-empirical model for maize's light extinction coefficient (kp) integrate data from multiple remote sensing platforms, and what advantages does this integration offer in comparison to single-platform models?

2.      In the context of the proposed kp model, how does the use of different remote sensing platforms, such as Landsat-8, Sentinel-2, Planet CubeSat, Multispectral Handheld Radiometer, and Unmanned Aerial Systems (UAS), contribute to the accuracy and reliability of the model?

3.      What specific statistical error metrics and Sobol global sensitivity indices were employed to evaluate the performance and sensitivity of the novel maize kp model, and how do these metrics contribute to the model's validation?

4.      Why are the Limited Irrigation Research Farm (LIRF) and the Irrigation Innovation Consortium (IIC) chosen as experimental sites for developing and evaluating the maize kp model, and how do these sites contribute to the model's real-world applicability?

5.      How does the study address the issue of surface heterogeneity in agricultural fields, and what implications does this have for the accuracy and generalizability of the proposed kp model?

6.      How does the proposed study address the limitations mentioned for existing kp models in cropland fields, and what innovative aspects are introduced in developing a semi-empirical spatial model for maize kp?

7.      NDVI is considered a strong predictor in the "canopy gap fraction" theory. Could you discuss why NDVI is particularly suitable, and are there alternative indices considered in the model?

8.      Eq. 12 outlines the linear interpolation model for 𝑑𝑑𝑓𝑐(𝑁𝐷𝑉𝐼). Could you provide more insights into the rationale behind this interpolation method and its impact on the accuracy of the model?

9.      Eq. 7, a unique semi-empirical and quadratic function of NDVI is used to relate fc. How is this function derived, and what factors influence its parameters?

10.  How is LAI measured in this research, and how is the temporal interpolation of LAI data performed?

11.  Explain the multispectral surface reflectance data sources used in this study (Landsat-8, Sentinel-2, Planet CubeSat, etc.).

12.  How are the CubeSat and Sentinel-2 data harmonized in this research?

13.  Describe the role of Unmanned Aerial Systems (UAS) and the handheld multispectral radiometer in data acquisition.

14.  Regarding Figures 5a and 5b, can you clarify the color-coding or labeling of points to distinguish between LIRF 2018 and LIRF 2022 datasets? This would help in understanding the temporal distribution of the data.

15.  For the scatter plots in Figures 6a and 6b, can you share specific examples of data points to illustrate the observed and estimated values of NDVIc and NDVI soil, along with their corresponding NRMSE values?

Comments on the Quality of English Language

Moderate editing of English language required

Author Response

Please find responses in the attached MS Word document. Thank you.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

The submitted manuscript for review primarily focuses on estimating the light extinction coefficient using information from various remote sensing sources. It is noteworthy that the authors utilized data obtained both from ground-based sources and from satellite and UAV platforms. Approaches that leverage data synergy from different sources are highly regarded by the scientific community.

Here are a few comments on the text:

The main concern regarding the study's assumptions is how the authors compare spatial data with different resolutions. In analyses involving information with varying resolutions, it would be beneficial to present methods for analyzing data at the same resolutions. I would like to draw attention to Table 1, where UAV data exhibit the highest accuracy. Could this be attributed to spatial resolution?

Another issue I see is the lack of information about the temporal synchronization between UAV, satellite, and ground-based data. How were the data compared? Did they constitute different independent datasets, or were they somehow compared with each other?

The description of satellite systems seems unnecessary, and it would suffice to refer to the literature regarding their characteristics. A similar situation arises with the conversion of DN values to reflectance in each of the systems.

I suggest condensing the article in the methodological section and adding a discussion section that analyzes factors that may influence the analysis results.

Author Response

Please find responses in the attached MS Word document. Thank you.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors,

 

Your study is very good, and the manuscript includes several small issues that require your attention before publication. There is one main issue, about your statistical calculations, that brings the reader to think your results look “too good to be true”.  I’m talking about your MBE and RMSE values, and their corresponding relative values (NMBE and NRMSE). For example: in Fig. 5, the R2 of the LAI analysis is much better than the R2 of the Fc. However, their relative MBE and NRMSE are alike. Can you please check your calculations to ensure they are correct?

In addition, please correct the following issues, according to specific line numbers:

Line 45 – you wrote “h_c” while I’m assuming it should be “f_c”

Line 110 – you wrote “to develop” when it should be “developed”

Line 129 – you wrote “RMSE of 12%”. Is it NRMSE, or the error is 12? Anyway, the k_p values are not % and the RMSE units are as the variable units.

Line 363 – you wrote that Landsat revisit time “was weekly”. It should be “every 8 days”.

Line 384 – you wrote that “If a given area of interest is located at mid-latitudes, Sentinel-2 temporal resolution is two to three days”. This is correct only for specific locations (that are in the overlap between the satellite strips) and for most locations, the revisit time is every 5 days.

Lines 478-485 – please indicate how many LAI measurements were used to calculate the site average.

Line 522 – you wrote “Eqs 30 and 32”, while it should be “Eqs. 29 and 31”.

Line 567 and elsewhere – you wrote the symbol + between the MBE and the RMSE values. Please remove it and write “0.02 and 0.07 values (MBE and RMSE, respectively).

Lines 618-619 – the larger relative errors of the NDVIsoil could be due to their very low average values, compared to NDVIc.  

Figure 6, lines 627-630 - Because you are showing the R2 values elsewhere, please add it also in this figure.

Line 677 – you wrote, “for any given f_c between 0 and 0.85.” yet according to Fig 5 it looks like you have measurements of 0.9. Please explain.

Comments on the Quality of English Language

No comments

Author Response

Please find responses in the attached MS Word document. Thank you.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors addressed the highlighted points. Therefore the article may consider for further process. 

Comments on the Quality of English Language

Minor grammar mistakes found