Data Grouping Method for the Purpose of Forecasting the Mechanical Strength of Plastic Soils

Round 1

Reviewer 1 Report

A brief summary

This new version is much less long, more intuitive and mostly comprehensible than that previous. The use of new figures and tables has lightened the task of reading it. 

Broad comments

I have done a quick (not detailed) review of the text. From what I have seen, most of the proposed changes have been introduced. I add below some very specific improvement proposals. 

Specific Comments

Lines 14-15: A field measurements campaign was conducted in subsoils of the Szczecin Lowland, north-west Poland and soil samples were taken from subsoil on 43 profiles.

[It is important to locate the study area in the summary]

Line 20:  …humus next, 10 soil parameters have been calculated

[Add coma after 'next']

Line 28: Keywords: soil, strength, penetration resistance, predicting, data grouping, measurement

[Delate the last keyword, it is not necessary]

Line 71: change e.t.c. by etc.

Line 83 (Figure 2): There are multiple spelling errors. Review it completely.

Line 129: Be careful with the font size in the figure. It is too small and many readers will not be able to read it clearly if they print the article. Enlarge it.

Lines 167-169: Review this sentence. I propose cut it into two different phrases. Hence, other parameters (indicators) were used to arrange the cases, which better diversified the soils, were highly correlated with content of fine particles and calculated on the basis of largest possible number different granulometric fractions.

Line 198: …only one parameter was used at the a time.

[I think the ‘a’ is not necessary in this sentence. It sounds strange]

Author Contributions Statement. The journal’s instructions stablish that authors are encouraged to include a short paragraph before the References section with specific author contributions. The authors indicate in their response that they have created this section, but in the PDF sent to the reviewer the contributions of each author do not appear. Maybe a simple editing error?

Author Response

Dear Reviewer, thank you for your valuable comments and suggesstions. We've analyzed all of them and our response is included in the attached file.

We wish you happy Easter and all the best,

authors

Author Response File: Author Response.docx

Reviewer 2 Report

I reviewed the MS entitled “Data grouping method for the purpose of forecasting the mechanical strength of plastic soils” by Błażejczaket al. The MS deals with the development of data grouping method enabling the selection of best regression equation for forecasting soil penetration resistance based on a small set of soil properties (e.g., soil moisture content, bulk density and particle size distribution). The Authors concluded that the dry bulk density is less useful for predicting the penetration resistance of plastic soils than soil moisture.

I think the MS is interesting and might give new insight to scientific community but I found it severely impaired by i) methods used for soil physical analysis and, ii) the lack of actual data description. Regarding the first point, a pair of examples might be the sample dimension issue and the presence of humic substances, please see below for further details. I think you need to include at least a new paragraph in which you warn the reader about the methodological limitation that might have affected your results

Abstract

The abstract is well written but exceed the Journal limit of 200 words. Please summarize to fit the Journal standard

Keywords

Please use the alphabetical order

Introduction

The introduction is generally well written and structured

Materials and Methods

Material and methods section needs to to be modified. I suggest to make the text more fluid removing the sentences like “The methodology of the study is shown in Figure 1” or again “Table 1 gives number of profiles and pits per object, the location of arable fields of individual objects, soil types and depth of soil cultivation.” This type of sentence makes the text heavy to read, please re-phrase and describe your data/methodology also if information is reported in Table/Figure

L88 Please give geographical coordinates

L88 What is “object”? It is not clear if you refer to distinct pedological distinct object or if you mean that Soil was taken from 13 plots/positions. Please specify and give more information about the sampling scheme, does it include replicates?

LL97-98 You used 50*50 mm cylinder to collect undisturbed soil samples. Dimension of samples matter more than ever in soil physics. I think the Author must discuss this aspect and include a paragraph on the discussion including the methodological limitations (e.g., sample dimension) that might have affected their results. What if you sampled with 10*10 or 20*20 cm? The measurement of bulk density in plastic soil might implies a representative elementary volume exceeding 5*5 cm. I think you need also to include recent relevant literature:

Piccoli I., Schjønning P., Lamandé M., Zanini F., Morari F., 2019. Coupling gas transport measurements and X-ray tomography scans for multiscale analysis in silty soils. Geoderma, DOI 10.1016/j.geoderma.2018.09.029

Lucas M., Vetterlein D., Vogel H.-J., Schlüter S., 2020. Revealing pore connectivity across scales and resolutions with X-ray CT. European Journal of Soil Science, DOI: 10.1111/ejss.12961

L105 Company details missing (city and or country)

LL109-110 The application of not-commonly used method for particle size distribution determination (Bouyoucos Casagrande method modified by Prószyński) makes this paper usable just for local application. This is another important aspect that should be included in the discussion on the methodological limitations that might have affected your results

L110 Missing details on Pycnometer, e.g., company, models, …

LL111 Humic substances has been reviewed during the last 20 years. Spectroscopy and scanning transmission microscopy demonstrated that soil organic matter is composed of smaller and simpler biomolecules (Sutton and Sposito 2005; Lehmann et al. 2008; Kleber and Johnson 2010), rather than large and complex molecules (humic substances). Over the recent years scientific community agreed in viewing organic matter persistence in soil no longer as intrinsic property of SOM, but rather as a property of the ecosystem (Schmidt et al. 2011). This is another methodological impairment of the paper this must be included in the discussion

Cited literature:

Sutton R., Sposito G., 2005. Molecular structure in soil humic substances: The new view. Environ Sci Technol, 39:9009–9015. doi: 10.1021/es050778q

Lehmann J., Solomon D., Kinyangi J., et al., 2008. Spatial complexity of soil organic matter forms at nanometre scales. Nat Geosci 1:238–242. doi: 10.1038/ngeo155

Kleber M., Johnson M.G., 2010. Advances in understanding the molecular structure of soil organic matter: implications for interactions in the environment. Adv Agron 106:77–142. doi: DOI: 10.1016/S0065-2113(10)06003-7

Schmidt M.W.I., Torn M.S., Abiven S., et al., 2011. Persistence of soil organic matter as an ecosystem property. Nature 478:49–56

L119,120, 125… Please specify the equation used for calculation also if you reported reference paper

Results and discussion

You need to reformulate these section by i) including a chapter informing about the methodological details that might have affected your results and, ii) adding more actual data also if they are present in some Tables and/or Figures, you need to better explicit you results

LL221-227 Actual data are missing. Before discussing about grouping test you need to describe how measured soil properties varied across depths and locations

LL221-222 “The results of measurements of independent variables (w_w, ρ_d) and dependent variable (PR) of the regression model are given in Table 2” please avoid do write sentence like that, check all the MS

LL249-257 Again the is lack of actual data

LL272-273 Describe the selected parameters

Conclusion

Please modify your conclusions by not stating results again but expanding on the discussion and stating implications of your study and ending in 1-2 sentences of future research recommendations

Author Response

Dear Reviewer,

thank you very much for your review.

All your comments and suggesstions have been regarded and our reply in included in the attached file. Of course, all necessary corrections, changes and new content have been included in the manuscript.

Yours sincerely,

authors.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Dear Authors,

congratulations, you have done a good job in reviewing your MS. I think your MS can be accepted after minor revision. Indeed, you should polish a little bit your conclusion. Please modify your conclusions by not stating results again but expanding on the discussion and stating implications of your study and ending in 1-2 sentences of future research recommendations. The Conclusions in this journal are not a summary of the paper.

Best regards

Author Response

Dear reviewer,

thank you for your comments.

We've modified (re-written) the Conclusions - please see the final version of the manuscript.

Warmest regards,

authors

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

The aim of this paper is to develop a forecasting method of the strength of plastic soils, which allows avoiding conducting tedious lab work or using a Penetrologger to obtain this data at field. This soil property, strength, was described by the penetrometric resistance, and the two proposed independent variables to forecast it were moisture content and bulk density. For this purpose, a wide sampling was conducted in subsoils from the region of Szczecin Lowland (Poland). Ten soil properties were directly recorded, and other 10 secondary soil parameters were after calculated. The experimental data collected were divided into a basic set (275 cases) and a validation set (77 cases). Then, data were subjected to a grouping procedure trying to maximize the coefficient of determination value for dependence of penetration resistance on gravimetric moisture content and dry bulk density, obtaining than the best option was using four sets with 7 subsets. Six grouping parameters were selected, three for each one of the two stages of ordering applied. Different combinations of these grouping parameters were tested, allowing the selection of a variant for which the obtained multiple regression equation was characterized by a high matching to the experimental data. Using the selected combination (#9), ranges for selected soil parameters values (soil particle fraction content and soil humus content), independent variables (moisture content and bulk density), and the dependent variable (penetrometric resistance) were defined for the particular 7 data subsets. Finally, the regression equations to predict the soil penetration resistance were obtained for each subset after excluding the outlieres (±2SD). The R² was in the range of 0.48 to 0.84. The validation set was used to evaluate these equations, obtaining that the mean relative errors of the prognosis of soil penetration resistance were less than 20%. In view of this result, authors consider the procedure as satisfactory. This is a well-written article that nevertheless presents some important substantive problems that must be reviewed by the authors before publication.

BROAD COMMENTS:

This paper try to reduce the number of analyzes to be performed to obtain a valid forecast of the penetration resistance. However, the proposed final equations (Table 8) include the soil dry bulk density when its impact on the dependent variable is not significant (Lines 305-311). Wouldn't it be better then to eliminate it from the equations and check how good this new forecast is using only one independent variable? It makes no sense to try to simplify and finally leave an independent variable in the equations just because it has traditionally been considered important. These equations should be obtained again by calculating the relative error of this new simplified forecasts to see if those adjustment differs much or little from those previously obtained.

These independent variables (moisture content and bulk density) were already identified as important in previous researches, as is commented in the introduction. What is the novelty of this article regarding existing scientific knowledge? The only progress is how the data subsets are chosen to obtain the equations, but little else. Although this approach may be interesting, its application presents certain possible errors that are discussed below. In the methodology, the number of soil samples taken is not clear. In theory, there are 275 cases in the main set and 77 cases in the validation set, but that does not match with the number of soils specified in Table 6 (237 in total). It is also not clear how many cases were sampled in each one of the 13 objects included in Table 1. This is important to know the sampling effort by soil type/location. This prevents assessing if the sampling design is well balanced. In fact, looking at the textures in Table 6 seems that the vast majority of sampled soils were loam (129 of 237, 54%). This can be an important sampling bias.

What was the reason to choose the division into four sets as the best option at the stacking of the observations in respect of the particle content <0.02 mm? Viewing the values of the coefficients of determination showed in Table 3, the highest arithmetic mean value for the R² values corresponds to the use of two sets (B1 and B2), not to the use of four sets. Unless the authors have a good explanation for this, the rest of the article's calculations would no longer be valid. Even, the explanation given in Lines 305-311 support the importance of combination #5 when it is said that sampled soils have a high content of clay particles and this reduce the importance of bulk density for forecasting soil strength.

The selected combination for grouping (i.e., combination #9) is not that with the highest matching to the experimental data, as the authors propose. The combination with the highest sum of coefficients of determination is the number 5 (see Table 5). The combination #5 uses the inverse of soil average grain diameter as the grouping parameter for the stage I, and the content of readily-dispersible clay for the stage II. Meanwhile, the combination #9 uses the field water capacity calculated without the participation of the organic matter mixed with the specific surface in the stage I, and the field water capacity calculated with the participation of the organic matter for the stage II. Therefore, the used subsets could not be the better to identify the regression equations to forecast the penetration resistance. This is a second mistake that can (again) seriously affect to the overall results of the article.

SPECIFIC COMMENTS:

The summary is very large and focused on identifying the soil parameters included in the data set, but does not indicate some aspects of the followed methodology (number of samples, for example). And, above all, does not include some of the main conclusions of the article, specifically that the dry bulk density is much less useful for predicting the penetration resistance than soil moisture.

Lines 20-21, the same abbreviation is used for the dry bulk density and the density of solid particles. The density of solid particles appears twice in the summary, into lines 21 and 23. If the second one is a secondary (calculated) parameter, it should have another name to avoid misunderstandings.

Lines 86-87: the cases are divided between the main set and the validation set using the period of sampling (2003-2006 or 2007-2012). Statistically, it would be better applying a random assignment mixing all data.

Section 3, Data grouping method, should be included in the material and methods, although some of its content could really be included in the introduction (Lines 177 to 182, for example).

Be careful with the use of commas instead of points for decimals. For example, in Tables 1 and 6 for the content of fraction, or in Table 8 for the equations. There are other places where the same thing happens. Review this throughout the whole document.

How were the parameters of Table 4 selected? I understand that the parameters less used in the scientific literature and that are more correlated with others were discarded, but we have no information on these correlations. The considered parameters are showed in Lines 246-247. How did authors decide to assign them to the first or second stage? This is not explained in the text.

Lines 273-274. Are authors sure that the subsets Zp and Zi have the largest range of values? What happens with the parameter <0.02?

Line 285: Change “I” for “and”.

Lines 324-327: The Variance Inflation Factor is low since it doesn’t exceed the value of 10. Therefore, there isn’t multicollinearity. But this low VIF is not enough to assert that the used method is correct (Lines 327-329).

Line 342 and Table 8: the relative prediction errors are calculated as a mean, so it would be advisable to show a measure of the variance of this result.

In theory, the choice of the equation to be used among those proposals in Table 8 is made using the criteria given in Table 9. However, it is not said how to act if the parameters of a soil point to different subsets according to the considered parameter. In addition, the parameters shown in this table are only a partial selection of all analyzed, hindering this task to those researchers trying to apply the equations to real cases.

Author Contributions are not described.

References are generally old. 26 of 31 (84%) are before 2009. This does not speak well of the novelty of research.

Reviewer 2 Report

Review to manuscript agronomy-648494

Title: Data grouping method for the purpose of forecasting the strength of plastic soils

By Dariusz Błażejczak et al.

I appreciate the editor to give me a chance to review an interesting paper.

The authors have taken a very high number of soil parameters from soil samples taken from 13 different sites in the northwest of Poland. The aim of the study is to establish a relationship between mechanical soil strength (more precisely penetrometric resistance = PR) on the one hand and soil parameters such as bulk density, grain size distribution and many more on the other hand.

The approx. 11 different soil parameters were divided into groups (Soil data grouping method (DGM) and it was looked with which parameter groups a prediction to the real PR can be made best. The aim was to be able to optimally estimate PR with only a few parameter combinations in the future.

---------------------------------------------------------------------------------

The manuscript is weakly written and contains many incomprehensible passages. However, the data obtained is impressively good and overwhelming in its accuracy and variety. Unfortunately, the methods are not sufficiently comprehensible.

What is immediately noticeable is that the manuscript is sometimes much too long and contains virtually no diagrams and illustrations, which makes it less intuitive and less comprehensible. Especially the summary is unusually long. The summary does not follow a classical sequence of motivation, material, methods and results. The summary lacks descriptions of the number of samples, sample size and sample analysis methods. There are hardly any results available that refer to the announcement in the title. What is the relationship between soil parameters and soil strength? The many lists and introductions of soil parameters and their symbols are far too long and far too detailed, and some parameters are explained twice.

There are superfluous repetitions in the manuscript, e.g. the penetrometric resistance is abbreviated with PR. The parameter is introduced several times in a row (Line 20, Line 37, Line 51-52, Line 157). Please change, it is sufficient to introduce an abbreviation once.

Line 41: Wheel compaction as motivation is reported here, there is no mention of it in the title of the manuscript or in the abstract. Unfortunately this does not fit together so well, at least in the abstract the problem should be mentioned as motivation.

After reading the manuscript I noticed a lot of inaccuracies and not very precise sentences and a detailed commentary is too complex.

Here are a few examples:

Table 1: the ranges for the measured values are given, but it remains unclear what this range is, an error range with standard errors? It remains unclear and should be explained in more detail in the method or at least in the table.

Line 98: it is not clear at what time the moisture content was measured. Is it the field freshness or the moisture content measured in the laboratory, how were the samples stored in the meantime? Please give more details here.

Table 1 and Line 99: the parameter "wpF" is used for water content. Previously "wp" was called plastic limit, which is confusing and ambiguous. Also in line 99, w is not written in italics later, that is unnecessary, please use it uniformly.

Line 102: cm2: unfortunately the two is not superscripted here and shows that the manuscript was not carefully reviewed before it was submitted. Please avoid such small mistakes.

In the further course of the manuscript too many columns of numbers are presented in the form of tables. These many values belong rather in an appendix and are not so well suited to represent the results in the center of an paper, it would be useful here to convert the many number series into diagrams, in order to make a visual intuitive comparability between the locations and the parameters clearly easier.

As mentioned before, the data quality and the basic idea of the study are good and interesting, but the presentation form is unfavorable and confusing.

However, I think that there are some merits in the study and it has a potential to publish in Agronomy after an accurate major revision.

I recommend to revise the manuscript as a whole:

Shortening the manuscript and the abstract, translating numbers into diagrams, explaining methods in a comprehensible way, avoiding repetitions, correcting formal errors.

-----------------------------------------------------------------------------------------------------