ET Partitioning Assessment Using the TSEB Model and sUAS Information across California Central Valley Vineyards

Round 1

Reviewer 1 Report

The paper presents substantial new results in evapotranspiration partitioning assessment. The overall quality of the paper is very good; besides a few details, the paper is clear and well-supported. I see no reason not to accept the manuscript for publication in Remote Sensing, subject to a few minor corrections.

Specific Comments:

1. The research questions are clear enough, but three fairly complex objectives seem excessive for a single research paper. Having read through the manuscript, I know all the objectives are addressed to some degree, but they are an unfocused introduction to the aim and scope of the presented research.

2. Section 2.1: About three study areas: Add more information about these areas as table 1: Latitude, Longitude, elevation above sea level, plant density, or other valuable information. 

Author Response

Thank you so much for your comments and suggestions. Study site description, such as latitude, longitude, and elevation above sea level, is added in the appendix.

Reviewer 2 Report

Accurate evapotranspiration estimation is very important for agriculture and water resources management, but it is particularly difficult to accurately estimate the components of evapotranspiration. In this study, evapotranspiration and transpiration were calculated based on the TSEB model, combined with sUAS estimates of soil and vegetation temperatures. More accurate estimates of transpiration were obtained based on the improved separation method of soil temperature and vegetation temperature. The analysis is clean and interesting. There are only a few places requiring some further clarification. please see bellow:

1. Since accurate separation of components of evapotranspiration is very important for evaluation, it is necessary to introduce these partition methods of EC observations. Eddy does not directly observe ET and Es. Are the components obtained by these methods reliable?

2. The evapotranspiration components at the EC scale is estimated based on TSEB model and sUAS observation. Although references on how to extract sUAS observation information over EC footprint are mentioned in the paper, it is necessary to present the footprint information of the study sites.

3. TSEB-PT is one of the main methods evaluated in this paper and should be described in more detail.

4. There is a section "Time-based performance of the TSEB-2TQ NK model", I wonder why these results from TSEB-2TQ are time-based? 

5. Although the method (TSEB-2TQ) proposed by the author improved simulation of transpiration, mainly reflected in the obvious reduction of BIAS, the correlation coefficient between simulated ET or transpiration and observed EC values is only 0.5-0.6, which means that the simulation can only explain about 30% of observed ET/T changes. Can such simulation results effectively meet application requirements such as irrigation demand?

Author Response

Comments and suggestions for authors:

Accurate evapotranspiration estimation is very important for agriculture and water resources management, but it is particularly difficult to accurately estimate the components of evapotranspiration. In this study, evapotranspiration and transpiration were calculated based on the TSEB model, combined with sUAS estimates of soil and vegetation temperatures. More accurate estimates of transpiration were obtained based on the improved separation method of soil temperature and vegetation temperature. The analysis is clean and interesting. There are only a few places requiring some further clarification. please see bellow:

1. Since accurate separation of components of evapotranspiration is very important for evaluation, it is necessary to introduce these partition methods of EC observations. Eddy does not directly observe ET and Es. Are the components obtained by these methods reliable?

Yes, eddy-covariance data does not provide ET partitioning, but the method, such as CEC, can separate the ET into E and T. One of our co-authors, Einara Zahn, has published one work about using eddy-covariance data for ET partitioning in 2022 (DOI: 10.1016/J.AGRFORMET.2021.108790). In that research, three methods (FVS, MREA, and CEC) which are also mentioned in this research are discussed deeply. Specifically, the result shows that the CEC and MREA framework can be used as a qualitative measure of the importance of ET partitioning. Her research results are fundamental to our research. For this research, we are more focusing on the remote-sensing-based result since there are still several challenges (i.e., temperature separation).

2. The evapotranspiration components at the EC scale is estimated based on TSEB model and sUAS observation. Although references on how to extract sUAS observation information over EC footprint are mentioned in the paper, it is necessary to present the footprint information of the study sites.

It is modified accordingly in the manuscript to show the footprint-area calculation work.

3. TSEB-PT is one of the main methods evaluated in this paper and should be described in more detail.

This is possible, but it will dramatically increase the number of words for this article. TSEB-PT and TSEB-2T are considered in this research, and three type of resistance models are also considered. Plus, the upgraded algorithm is used for the TSEB-2T, and the upgraded algorithm is a key point in this research. In this case, there are 9 different experiments, which is described by Figure 5 in the article.

The difference between TSEB-PT, TSEB-2T, and TSEB-2T_Q is discussed in section 2.3.2. Additionally, results, such as Figure 5, has described that the results from TSEB-2T_Q is better than that from TSEB-PT. Therefore, more attention is focused on TSEB-2T_Q and the following issue, ET partitioning.

4. There is a section "Time-based performance of the TSEB-2TQ NK model", I wonder why these results from TSEB-2TQ are time-based? 

All sUAS flights are divided into 3 groups: LS, SN, and AF. The reason for splitting flights into 3 groups and the meaning of each of them are explained in the article. The term, “Time-based performance,” is seeing the TSEB-2T_Q performance at different periods (LS, SN, and AF). From this aspect, we can discuss whether the parameters (e.g., G ratio) in the TSEB model are appropriate.

5. Although the method (TSEB-2TQ) proposed by the author improved simulation of transpiration, mainly reflected in the obvious reduction of BIAS, the correlation coefficient between simulated ET or transpiration and observed EC values is only 0.5-0.6, which means that the simulation can only explain about 30% of observed ET/T changes. Can such simulation results effectively meet application requirements such as irrigation demand?

This is hard to answer from this single paper. Regarding the model evaluation metrics, it does not bring very exciting information for us. Because there are still limitations and challenges regarding the field measurements (i.e., EC tower measurements for vine vegetation, interrow vegetation, and ET partitioning based on EC tower data), and they are discussed in this article. But the meaningful point is this article is one of the first for evaluating ET partitioning estimation from sUAS imagery based on EC-based partitioning methods, and it provides some thoughts for future research. Therefore, future relating results should be better than current results.

Speaking of ET, the upgraded temperature separation can improve the TSEB-2T performance within the footprint area. In other words, the ET simulation at the vineyard scale is better than before. Speaking of transpiration, the same as mentioned above, addressing the limitations and challenges can sufficiently support accurate ET partitioning based on the sUAS imagery, and this needs further efforts.

Reviewer 3 Report

Quantifying evaporation and transpiration is a challenging endeavor and a very important component of evapotranspiration estimation studies. In this study, a new coupling estimation method, TSEB model, was developed based on the TSEB model based on the EC data. It is somewhat innovative. However, it is also needed to be aware of the following issues:

1. In Figures 3 and 4, please plot the exact size of the shadows.

 2. Figure 4, please re-number each icon and arrange it neatly, can you remove the arrows and dashed lines?

 3. In the Abstract section, "improved sensible heat flux estimation", please give the specific percentage of increase.

 4. As can be seen from Figure A 5, TSEB-2TQ improves the accuracy of Bias, but does not increase the correlation coefficient. So, in the summary section, please describe more accurately and clearly "the TSEB-2TQ shows better agreement ....... via the CEC method”.

5. In the Discussion section, please add the main reasons why the new method improves the accuracy of estimation compared to other methods.

Author Response

Comments and suggestions for authors:

Quantifying evaporation and transpiration is a challenging endeavor and a very important component of evapotranspiration estimation studies. In this study, a new coupling estimation method, TSEB model, was developed based on the TSEB model based on the EC data. It is somewhat innovative. However, it is also needed to be aware of the following issues:

1. In Figures 3 and 4, please plot the exact size of the shadows.

Figure 3 is just an example to show the ideal case for the temperature separation: the vegetation and soil temperatures will be calculated based on the “black” points within the pure vegetation and pure soil zone, respectively. In other words, previous relating research are using this methodology.

Figure 4 is re-modeled, and the shadow location at the 0.15 m scale is displayed on the map via small red pixels. (P.S., the shadow pixel is recognized at the 0.6 m scale at the end)

2. Figure 4, please re-number each icon and arrange it neatly, can you remove the arrows and dashed lines?

Figure 4 is re-modeled. The arrows are removed, but the dashed lines remain because they are supposed to easily show readers information.

3. In the Abstract section, "improved sensible heat flux estimation", please give the specific percentage of increase.

It is modified in the article.

4. As can be seen from Figure A 5, TSEB-2TQ improves the accuracy of Bias, but does not increase the correlation coefficient. So, in the summary section, please describe more accurately and clearly "the TSEB-2TQ shows better agreement ....... via the CEC method”.

It is modified in the article.

5. In the Discussion section, please add the main reasons why the new method improves the accuracy of estimation compared to other methods.

The main reasons why the new method improves the accuracy of estimation compared to other methods are explained in section 2.3.1 (when explaining the temperature separation algorithm). So, the author was thinking to point out other points in the Discussion section.