Practical Implications of the Availability of Multiple Measurements to Classify Agricultural Soil Compaction: A Case-Study in The Netherlands

Round 1

Reviewer 1 Report

The manuscript is interesting and contibutes in the search for critical values of soil resistance to penetration for plants. The link between bulk density and resistance to penetration is interesting, and including the misclassification costs value even more the manuscript.

The results and observations of this manuscript open the possibility for future researches.

Some appointments were realized along the manuscript, and it has conditions to be accepted for publication.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The work presented for review raises a very interesting issue related to the assessment of excessive soil compaction made on the basis of the parameters of penetration resistance and bulk density. Bulk density was taken as a reference and penetration resistance as a result of a simple and easy to apply measuring method.

The introduction does not raise any objections, it is synthetic and justifies the purpose of the work.

The methodology of the research in a clearly written form.

Figures 5 and 6 can be combined in one figure, figure 7 should be deleted, it is not needed.

It is very interesting to try to estimate the costs of an incorrect diagnosis of soil compaction.

The results chapter is written correctly.

I think that the publication is generally too long, it could be divided into two thematically linked publications.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

 

This paper proposes a methodical framework for comparison between the soil compaction that allows to assess the level of agreement between classification of the state of soil compaction based on two substantially different methods. Overall, the paper is well written and has the potential to be highly cited and also has many practical implications for practitioners. Therefore, my first impression is very positive. I tried to be as constructive as possible, because I think that the paper could be improved, and that, if improved, it could have a high impact.

The methodology is sound, and results are discussed in an appropriate and balanced way. However, some criticisms emerge. A potential weakness of the method, which has been mentioned by the authors, is that the approach of authors ignored/reduced the soil variability on the side of clay-content correction of bulk density threshold. As authors stated: “Texture-specific BDd thresholds … are generally accepted in both the scientific literature … and (inter)national quality standards…” But authors assigned the BD single-threshold based on the borrowed clay content from single measurement from the Dutch national soil information database, from the same database they stated as outdated for bulk-density (Line 77). I wonder how robust all the measurements of bulk density in 5 depths with elaborated methodology were (incl. the clean site with no-walk side of soil pit) and how simple the threshold assignment to BD measurement using single clay value was. It seems to me unbalanced methodology to omit short-distance and vertical variation of clay. The natural variation of soil is inherently involved in the soil strength (PR) measurement using penetrologger and highly reduced using the single-value of clay for corrected threshold. There was really no measurement of clay content during the very intensive soil pit sampling? Did authors used some proxy field method to estimate the soil texture (i.e hand, finger estimation of soil texture in the field) during the soil pit description? I checked the model-based prediction of clay on the website at the site: https://maps.opendatascience.eu/?base=BingMaps%20(Aerial)&layer=Soil%20clay%20content&zoom=17.5&eye=5000000&center=52.7025,5.7295&opacity=83&time=2020

Being aware of the uncertainty propagation from covariates and model, there can still be expected a degree of clay variation in both horizontal and vertical sense. Concerning the fact that authors gave some economic consideration of the misclassification, authors should broadly discuss the potential effect of bias of using single clay threshold. Why authors did not also the threshold range similarly to PR thresholds? If I make the opposite extreme case than authors I will attribute all the BDs variability (roughly estimated from Figure 11 to range 1.32 –1.68 mg/m3) to natural conditions of soil I can approximated the soil clay range between 7 – 25 %.  Authors did a lot of effort to “tune” optimal threshold for PR value but settled with single-corrected BDs threshold, with assumption of low clay variability. It would be beneficial to prove that such assumption was legitimate and give evidence on clay content stability in both horizontal and vertical orientation - authors have some useful covariates used for cLHS, can you utilise them to assess the in-field variability of soil condition? To be fair, authors are aware of the mentioned limitation of study and briefly discussed it (Lines 702–711) but especially to connection with horizontal short-range variation of clay. The vertical variation should be also discussed (form Figure 4 one can guesstimate that some vertical variation of clay occur)…While all the mentioned limitations are indisputable, the experimental set-up and the validation/confusion study are very carefully designed. The results are well discussed. All illustrations are relevant. From a formal point, the manuscript is written in good English (at least as I can state) with minimum typological errors. I believe that this paper is a nice methodological guideline but the economically consideration of misclassification costs should be interpreted with care considering the simplifying assumption of constant clay content.

I have some minor comments:

Figure 1 – authors manipulated the width/height ratio of the field map (upper part of Fig- 1), and hence the graphical scale is not valid – the field is not in reality 100 x 400 m …

Line 131 – There is no need to introduce the study objective with redundant tittle Objectives.

Lines 195 – The mentioned data-loss is the same as the data exclusion based on the soil moisture issue (Lines 272–274)?

Lines 199 – What resolution of covariates inputs was used for the cLHS?

Line 201 – I am confused which sampling scheme was used - k-means clustering to coordinate-wise geographical partition of the area or covariate-wise conditioned Latin hypercube sample (cLHS)?

Lines 244–246 – “The arithmetic mean of the triplets of soil samples was determined for each depth, providing the measured BDd at 0.001 Mg m-3 resolution. As previously stated, only samples from the loamy topsoil were used for further analysis in this research.” It is confusing – you used the depth profiles for further analysis but only for samples where the topsoil was loamy? How did authors estimated the soil texture since they did no particle size distribution analysis…?

Line 280 – On which base the averages were weighted (measurement depth)?

Line 283 – Maybe I missed something, but I did not fully understand the selection of optimal bandwidth for depth-aware kernel smoothing… Why it should be connected to the Kopecky soil core diameter?

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx