Multi-Parameter Health Assessment of Jujube Trees Based on Unmanned Aerial Vehicle Hyperspectral Remote Sensing
Round 1
Reviewer 1 Report
Some of the details of what the authors want to do, and the procedures explained in the abstract and introduction are not very clear in my opinion. For example, line 45, or the sentence in lines 59-62.
Lines 229-232: this part of the procedure is not explained clearly, and what is shown in the image below seems to be slightly different from what is written in the text. For example, the feature map in the image is an output of RPN, while in the text it is an input of RPN.
Lines 330-331: “DCVI” is used two times, but I think it is an error and authors meant “DCNI”.
Lines 431-444: from the title of this paragraph, I would expect the 4 variables (SPAD, LAI, tree height, and canopy area) to be used together as input of the models. From the explanation however it seems that they are used separately. From what is said in the conclusion, authors probably used all the variables together and it was not explained enough in this section.
The idea of the study is quite relevant and interesting, but like many other papers that have been published recently, it seems to be excessively complicated, with many steps, each of which is a non-trivial problem by itself. Understanding if each step works, and if the united algorithm really works, is very difficult. Finally, when working with Machine Learning algorithms, there will always generate a percentage of errors. Therefore, concatenating many of them often gives worst results than having just one model that goes directly from your data to the desired output in a single step.
Just minor revisions are needed.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This manuscript (agriculture-2536490) introduces a health assessment model for jujube trees based on multiple physical and chemical parameters. This aims to address the limitations of single-parameter models, which are often imprecise. By using UAV hyperspectral images, various attributes of jujube trees, such as leaf chlorophyll content, Leaf Area Index (LAI), tree height, and canopy area, have been evaluated to assess tree health. Several spectral indices and models, including Mask R-CNN and the Canopy Height Model (CHM), were employed. Four health assessment models were assessed: Partial Least Squares Regression Analysis (PLSR), Random Forest (RF), Support Vector Machines (SVM), and Decision Tree Regression (DT). The PLSR model exhibited the highest R^2 value at 0.853 with an RMSE of 0.3, surpassing the other models. This method offers a rapid and accurate health evaluation of jujube trees using UAV hyperspectral imagery, indicating the PLSR model's capability in accurately gauging the trees' health.
The introduction, materials and methods, and results sections are acceptable. However, the discussion needs revising, as do the conclusions.
Could you provide a clearer explanation regarding the correlation coefficients and the variables in Figures 5 and 7? Why not use a matrix for this representation? Could you clarify further?
In Figure 6, what does "Inversion values" mean?
Please check the standardisation of the images (axes) and within the manuscript text.
What is the relationship between SPAD and quantified chlorophyll? Do the values differ when compared with the tested vegetation indices?
In Table 6, what do the multiplication values indicate for SPAD, LAI, Height, and Canopy area? Which value should be added to apply this equation?
Please explain the abbreviations in Table 7 to assist readers.
The current discussion is insufficient. There are no references; in fact, no papers have been cited at all. There are numerous studies on these modelling techniques, yet the authors have not compared any with existing literature. Moreover, how has your study progressed concerning quantification and modelling aspects? How can the chemometrics explored be directly implemented in the field? You outline future perspectives, but there is a lack of discussion about potential challenges that may arise with the application of this method.
The conclusions are also lacking. It's essential to describe and summarise the content, rather than presenting it as bullet points or topics.
Review the references, as some are outdated. Many recent papers, especially from the past five years, have dealt with this type of modelling.
The English needs corrections, such as grammar, verbosity, and spelling.
Author Response
请参阅附件。
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
I believe the authors made relevant modifications. I think it is acceptable.
Minor changes in grammar.
