Evaluation of Leaf Chlorophyll Content from Acousto-Optic Hyperspectral Data: A Multi-Crop Study

Round 1

Reviewer 1 Report (Previous Reviewer 1)

Thank you for the efforts made by the authors in revising the manuscript, particularly for conducting additional experiments and detailed information regarding imager specifications and processing procedures. One last issue that require attention as below:

1. On page 5 in line 150 to 168, it seems that the imager you used in this study employs a dual crystal design, which is relatively uncommon. Could you please provide a reference or a brief schematic outlining your design.

Author Response

We added a reference which describes a tandem acousto-optical tunable filter in detail.

Reviewer 2 Report (Previous Reviewer 2)

This article starts from a linear model for a single crop and gradually develops a linear model that uses a single index suitable for multiple crops to effectively measure chlorophyll content. This paper demonstrates the potential of using AOTF for CC assessment. The number and nature of the samples significantly affect the performance of the evaluated model. What are possible strategies for data collection? This article is more of a feasibility experiment and does not compare with other measurement methods or HSI equipment.

Author Response

1. The number and nature of the samples significantly affect the performance of the evaluated model. What are possible strategies for data collection?

Indeed, spectral-index-based concept of chlorophyll content evaluation is in principle dependent on the number and nature of the samples. As shown in Table 5, spectral indices are usually introduced and well-proven only for particular crops. That is why it is important and challenging to develop an approach to multi-crop chlorophyll content estimation. For high measurement accuracy and reliability, it makes sense to collect as complete variety of crops as possible that can be found in the field under study. The mathematical model has to be verified periodically and refined if necessary due to the change of the growth stage and other factors. We added a few statements on a possible strategy for data collection in the manuscript.

2. This article is more of a feasibility experiment and does not compare with other measurement methods or HSI equipment.

In this feasibility study, we compare various chlorophyll indices in single-crop and multi-crop modes (Tables A1) as well as different regression models (Table 6). For the presented concept, tunable imaging filter able to address any wavelength rapidly is most efficient base for HSI system. Therefore, we focused on acousto-optic imager and described it in detail. Multiple studies carried out using HSI systems of other types are included in the reference list.

Reviewer 3 Report (New Reviewer)

(1) Include some information of CC estimation using radiative transfer model like PROSPECT etc in Introduction and Discussion sections to support your findings.

Author Response

We expanded Introduction and Discussion by adding information and references related to radiative transfer models.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This manuscript presents a leave chlorophyll measurement work with AOTF-based hyperspectral imager. The advantage of the AOTF-base hyperspectral imager is that it collects spectral image data band-by-band (BSQ), and this strategy allows fast data acquisition, which indicates the AOTF-based hyperspectral imager may be a better choice to perform fast chlorophyll measurement.

1. The major concern is that the innovation of the manuscript requires specific and clearer delineation, especially over the [27].

2. The advantages of AOTF-based hyperspectral image are not comprehensively described here. How does it function? Why does this type of instrument can randomly access a specific wavelength, which results in a faster data acquisition? This is a remote sensing journal, and the readers might be unfamiliar with such instrument.

3. In section 2.2 on data acquisition, please check your reference [26], in which describing your hyperspectral imager's structure and essential details. Other details related to the imager need to be described, such as the polarization of the acquired light, aperture size used, and the AOTF hyperspectral imager have heavy stray light issue, did you remove the stray light background before processing the data.

4. In section 2.2 on data pre-processing, please specify the parameters of the median filter's length and your Gaussian smoothing.

5. The model building in section 3 could be improved. The CIs were not linearly correlated to the Chl concentration, therefore, linear regression with single CI factor may not be ideal. Please consider adding models building using multiple CIs and explore the use of nonlinear regression methods.

6. Minor issues such as type in line 132 “cErop”, and line 167 “Cis”.

Author Response

please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

My comments on this paper are as follows:

This paper presents a non-destructive method for estimating chlorophyll content using only 11 wavelengths and a linear regression model that may be easier to implement on mobile devices and use in the field.

Detailed explanation is needed of how sample data is collected and the conditions under which it is collected.

Listing the numerical results of each chlorophyll index in single-crop models and multi-crop models can provide more comprehensive information.

Among the single-crop models, only the wheat model showed a higher coefficient of determination. Is it possible that since the improved simple ratio was developed to examine winter wheat, which is similar to regular wheat, a higher coefficient could be produced.

Given that the data points for individual crops do not lie exactly on the linear model, how robust is the model to adapting to changes in chlorophyll concentration over a specific range?

In Figure 2, the data processing step mentions flat-field correction, but the text does not explain why flat-field correction is needed.

As can be seen from Figure 5.(a), an index containing R_750 and R_705 yields better results. Does the situation change when the linear model includes fewer than 6 study crops?

Whether it is a single-crop model or a multi-crop model, CI seems to be used to correspond to CC for model derivation. However, the article does not explain in detail why there is such a significant difference when using MSR705 in single-crop and multi-crop scenarios.

Foliage shot after shot may be worthy of CC analysis. This approach provides a clearer understanding of the relationship between model-predicted CC and actual detection data.

Author Response

please see the attachment.

Author Response File: Author Response.pdf