Modeling to Correct the Effect of Soil Moisture for Predicting Soil Total Nitrogen by Near-Infrared Spectroscopy

Round 1

Reviewer 1 Report

Dear

The authors are testing five correction methods to apply PLSR for predicting soil total nitrogen by diffuse reflectance spectroscopy in the near-infrared range, in samples under different moisture regimes. The idea is interesting and welcome. However, there are some concerns that require the authors' attention:

·       Lines 2-3: If the authors are testing five correction methods, why they presented only SST in the manuscript’s title? I would like to suggest: MODELLING TO CORRECT THE EFFECT OF SOIL MOISTURE FOR PREDICTING SOIL TOTAL NITROGEN BY NEAR-INFRARED SPECTROSCOPY.

·       Lines 9-12: The result of this study is only applied to the rubber forest brick-red soil? I think it is not. So, from my point of view, the goal of this study is to identify the best model to correct the effect of soil moisture for predicting soil total nitrogen by near-infrared spectroscopy.

·       Line 29: What do mean by “commonly used indicator in agriculture”? Indicator of what? It does not sound correct. Also, nitrogen is not an “essential characteristic”. It is a macronutrient, essential for any plant’s growth.

·       Line 33: Precision agriculture is much more than the determination of soil total nitrogen. I would suggest rewriting this paragraph.

·       Lines 37-39: What do you mean by “the last few years”? Also, you say “many researches”. However, you cited only one publication, from 2010, despite we have been in 2023. I believe this section should be revisited.

·       Lines 39-40: You refer to “several factors”, but you cite only water.

·       Lines 45-49: The idea is interesting. However, the main reason the samples should be dried before the reading is for equipment conservation. The presence of moisture in soil samples usually brings damage to the laboratory Vis-NIR-MIR spectrometers. So, you may not suggest reading spectra without drying the soil samples. However, you may propose finding a solution to the moisture problems for the cases where the spectra are taken on the field by portable equipment. Or as a strategy for correction when, for some reason, the samples were not enough dried and the spectra must be corrected.

·       Lines 50-55: If they still have a significant gap in model accuracy, why are you testing them?

·       Line 72: Which one? May you cite them?

·       I understand you are assuming that SST is the best model to correct the effect of soil moisture based on the literature. However, you are testing 5 models. Then I would suggest revisiting this section once it seems a little confusing.

·       Lines 91-92: Would you present a figure of the study area, showing the points where the samples were collected?

·       Line 96: Please provide details about it. Also, I would suggest to provide the soil parent material and the soil classification by WRB.

·       Lines 114-150: Are testing the different pretreatments too?

·       Please review the equations presented in the material and methods to check if their references were cited.

·       Lines 277-293: The test of pre-processing was not described as a goal of the study. It is not fitting in this work. If it is a goal, the objective must be included in the study aims. Also, the figure 2 is not well visible.

·       Table 2: RPD is observed in the calibration process. If it fits with a good prediction (combined with R2 and RMSE), then the model is applied to a validation set or prediction. So, the RPD should be added to the calibration.

·       Figure 3 is not well visible.

·       The results are mixed with discussion in some points. For example, in the from line 312 to 314. I would suggest the authors to replace them into the discussion topic.

·       Lines 321-323: I disagree. Usually, the absorption bands (valleys) in the spectra near to 1450 may be associated with 2:1 phyllosilicates and lattice OH (vide https://doi.org/10.1016/j.earscirev.2016.01.012; https://doi.org/10.36783/18069657rbcs20200115; https://doi.org/10.3390/land11122188)

·       Table 3: The RPD should be added in the calibration.

·       Lines 462-466: I would suggest removing this paragraph. It is redundant.

·       Lines 467-474: This was not foreseen in the goals.

·       Lines 475-477: What result do you have to prove that?

·       The discussion should be improved. There are too many assumptions, and it is redundant. There is only a few citations used to support the discussions.

·       The conclusion should be revisited to match with the aims of the study.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Authors have presented the effect of soil moisture on the predictive accuracy of total nitrogen content on the target of rubber forest brick-red soil collected from Hainan Province and further 11 studies the method of eliminating the impact. The study is important in terms of agricultural applications. Results seems to be good. However, some of the observations are as follows:

 

1.      P1L18, pls specify or write full form of the term “RMSEP”. What other parameters used for the accuracy.

2.     P2L76, “Therefore, this paper intends to take brick-red soil samples as the research object to explore the effects of SST on the removal of soil moisture effect in the soil spectra.” Recheck the term “object”.

3.     P2L78, any reason behind the selection of 107 brick-red soil samples.

4.     P3L91, if possible, add the geographical map with highlighting the soil sample collection points.

5.     Table 1, all the terms must be specified in the footnotes like TN.

6.     It will be better if you add the comparative table of different methods to remove the moisture effects. Some of the content can also be reduced from these methods.

7.      Figure 2 is not visible properly (resolution issue). Very difficult to analyses, it must be revised.

8.     Same comment for Figure 3.

9.     Make sure that sub figure/Table must be placed on same page. Check the errors for Fig. 5, 7 and Table 3.

 

10.  In Table 3, mention the importance of highlighted value in footnotes as there are two values have 0.74.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear

I congrats the authors to the work they have done. They attended the most of my suggestions and answered most of my questions as well. I appreciate that.

However, I still think that the WRB soil classification is important as well as the location map with the sampling points. 

So, if the authors do not mind, I would like to insist in those suggestions given the importance of this matter.

Regards

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have done all the relevant changes.

Author Response

Thanks.