Round 1

Reviewer 1 Report

The paper is interesting, well written and precisely focused on the presented topic. Two sections are very important: 2. Key phenomena for water retention curve models and its subsections, in which the authors' considerations on existing models emerge; 4.3 Validation, which provides how the authors controlled their proposed model.

The 5. Conclusions are clear, but I would suggest to expand the section with a reasoning around the potential applications of the developed model, being more accurate compared to the others (for istance, in the section 1. Introduction the transformation of coal mines areas to farmland is mentioned): how this model can provide benefits to whom and for what? What is missed in this model or in a supposed procedure that applies this model?

The Reference section is appropriate.

Author Response

Response to Reviewer 1 Comments

 

Point 1: The 5. Conclusions are clear, but I would suggest to expand the section with a reasoning around the potential applications of the developed model, being more accurate compared to the others (for instance, in the section 1. Introduction the transformation of coal mines areas to farmland is mentioned): how this model can provide benefits to whom and for what?

Response 1: We agree with these comments, thus the conclusions were expanded to more clearly show its benefits and how this model is linked to the transformation of post mining land to farmland.

Point 2: What is missed in this model or in a supposed procedure that applies this model?

Response 2: We’re sorry, we don’t understand your comment and are unsure how to reply.

Reviewer 2 Report

Refer to my comments in the attached file

Comments for author File: Comments.pdf

Author Response

A point by point response is given below, and the author's responses are also provided in the attached PDF document, which was the reviewer's original format.

Response to Reviewer 2 Comments

 

Point 1: Is the locally derived waste material organic or inorganic?

The biggest challenge with rehabilitated lands is poor soil structure and low soil organic matter. 


Response 1: Added further description of coal tailings stating they are comprised of mineral soil particles and coal particles.

Point 2: Do you mean "high water holding capacity"?

Response 2: No, the beneficiation processes are water intensive causing tailings to have an initially high water content.

Changed to "high initial water content" for clarity.

 Point 3: You should tell first what the acronym stands for before using it within your text.


Response 3: Fixed, acronym introduction used to be "SWRC", now it is "WRC"

 Point 4: Definitely. It is not taking clay mineralogy. That is why one size fits all does not work in soil physics.


Response 4: Yes this is a simplification from reality that was deemed necessary to have the proposed model use easily measurable soil properties, which was one of the goals outlined in the introduction.

Point 5: The methodology regarding the soil samples were packed is not clear. What was the bulk density of the soil during the packing?

How did you saturate the soil?

Did you apply suction in a vacuum in order to avoid any air entrapment during the wetting process? 


Response 5: Consolidation from a slurry is a fairly routine process for preparing reconstituted samples in geotechnical engineering and research. The soil is first made into a high water content slurry, then a consolidation pressure is placed upon the slurry. This pressure causes water to drain from the soil and increases the soil density. When consolidating soils, the density is usually not the controlled, instead the consolidation pressure is kept constant. Meaningful volumes of entrapped air within the soil does not occur within a sufficiently wet and well mixed slurry; the end result of consolidation is a saturated sample.

For clarity, a statement was added saying the consolidation process created a saturated sample.

After a sample was prepared by consolidation, most often it required some drying (with consequent shrinkage) before a suction could be detected reliably. That is, the density of the sample changes from immediately after consolidation to the very first data point of the water retention curves. Thus, the soil density after consolidation was deemed irrelevant for this research and is not provided, instead an initial void ratio (density) is provided, which corresponds to the very first suction measurement (as stated at the end of Section 4.3).

A sentence was added stating that some drying was required before a suction could be measured.

Point 6: May you please provide details on how the soil water content and the corresponding soil water potential were measured concurrently?

Regarding the WP4, which model did you use? Is it WP4, WP4T, or WP4C? Their resolution varies a lot, which will have implication on the accuracy of the data collected.

Response 6: Given that the initial water content of the material is known for the sample, it is possible to determine the water content of the sample after it dries somewhat, simply by measuring the sample's mass change (if the mass of the container is known). Immediately after measuring the samples mass, its suction is measured before any meaningful evaporation and mass change occurs.

The WP4C was used in this study, added this detail to manuscript. The error from the WP4C is estimated to be +- 50 kPa for suctions up to 5000 kPa; given that the WP4C was not used to measure suctions lower than about 2000 kPa, the error is negligible.

Point 7: Too low for a soil. That is a temperature used for plants and organic matter. Two days is too short at 60 degrees


Response 7: This 60 degree oven was used for partially drying a sample to increase the suction, it was not used to fully dry a sample for water content measurements. The two days is the time between partially drying and measuring suction again. This is done to allow the suction within the entire sample to equilibrate to a relatively uniform value

A sentence was extended to more clearly indicate the intent.

 Point 8: Earlier in your Methodology section, you stated that some of the models require calibration. How did you handle this issue with regard to your study? Have you taken this into consideration when comparing the models?


Response 8: One of the requirements for the proposed model is to avoid any calibration to water retention curve data. This is because calibration to water retention curve data is not feasible for the situation discussed in the introduction. Thus, if one of the comparison models was to be used for the same purpose as the proposed model, the comparison model must also avoid calibration. None of the models in this study are calibrated to water retention curve data.

A sentence was added in the methodology to make the purpose for a lack of calibration clearer

Point 9: Some of the models had low saturated volumetric water content as of the beginning. If saturated volumetric water content is one of the inputs parameters, then the problem could be with your parameterization. The same applies with other input parameters, which could be crucial for some of the models. Some models are sensitive to specific parameters.


Response 9: A material's saturated volumetric water content was the same value for each model; when the models are not calibrated to the data it is fitting (what is done in this paper) it is unlikely to provide an adequate prediction for every material.

Point 10: I rather advise a proper statistical analyses be contacted instead of using sum of squared error alone.

You should look at the coefficient of correlation and Wilmot's index of agreement that could provide a better assessment of the models.

Response 10: The coefficient of correlation and Wilmot's index are goodness of fit measures suitable for linear models. However, the proposed water retention curve model is non-linear, in this case the sum of squared errors is an appropriate tool to assess the goodness of fit between each of the models assessed. As the sum of squared errors is often used in non-linear regression where a numerical method minimises the sum of squared errors (or residuals) to optimise the fit of a non-linear model to some data.

Point 11: I have some reservations in terms of its predictive capability especially in relation to plant available water because it is not only the amount of clay that plays key role on the availability of water but also the clay mineralogy, which is not considered in the current model. Besides, previous studies have clearly shown that organic matter plays a crucial role in soil plant available water, which is not considered as one of the inputs in the current model. If the rehabilitated land is considered for agricultural production, organic matter is crucial to help the soil develop structure through microbial activity. Previous studies have already demonstrated that hard setting is the biggest challenge, which leads to high bulk density, low infiltration rate, low water holding capacity, stanted root growth, and poor vegetation coverage.

Response 11: Soil structure and organic matter are of course a major factor that should be considered in topsoil improvement. However, this research simply aims to present a new predictive water retention curve (WRC) model that was developed as part of our broader research. In addition, the ability to estimate a soil's plant available water from predictive WRC models was assessed when calibration to WRC data is not possible.

Our broader research is focusing upon developing a method to improve the available water in topsoils at minesites, as this is in our domain of research. Improving these topsoils in terms of organic matter and soil structure are outside the scope of our research and may be studied later or by others. The broader research aims to provide one piece of the solution, so to speak.

The introduction and conclusion was modified to more closely reflect the intent: that this study presents a WRC model developed as part of broader research on topsoil improvement.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Point 5: The methodology regarding the soil samples were packed is not clear. What was the bulk density of the soil during the packing?

How did you saturate the soil?

Did you apply suction in a vacuum in order to avoid any air entrapment during the wetting process? 


Response 5: Consolidation from a slurry is a fairly routine process for preparing reconstituted samples in geotechnical engineering and research. The soil is first made into a high water content slurry, then a consolidation pressure is placed upon the slurry. This pressure causes water to drain from the soil and increases the soil density. When consolidating soils, the density is usually not controlled, instead, the consolidation pressure is kept constant. Meaningful volumes of entrapped air within the soil do not occur within a sufficiently wet and well-mixed slurry; the end result of consolidation is a saturated sample.

For clarity, a statement was added saying the consolidation process created a saturated sample.

After a sample was prepared by consolidation, most often it required some drying (with consequent shrinkage) before a suction could be detected reliably. That is the density of the sample changes from immediately after consolidation to the very first data point of the water retention curves. Thus, the soil density after consolidation was deemed irrelevant for this research and is not provided, instead, an initial void ratio (density) is provided, which corresponds to the very first suction measurement (as stated at the end of Section 4.3).

A sentence was added stating that some drying was required before a suction could be measured.

 

Extension of Question 5

It is essential that you provide enough details in your methodology so that others should be able to read your Methodology section and repeat the work without the need to ask for the details. So please add more details.

Reconstituting from a slurry could lead to the destruction of soil aggregate stability, in your case the microaggregates. The implication is: it leads to abnormally higher bulk density during the drying process than the same soil which has been left out on the field to settle naturally. This will have implications on the porosity and pore size distribution. Since the study is dealing with the rehabilitation of lands to grow plants, I believe it is essential to note the issue with the methodology followed in the Methodology section.  

 

Conclusion and Recommendation

It is scientifically appropriate to bring forward the strengths and weaknesses as well as the need for further research in the conclusion. In this case, it would be appropriate to state that the model does not take into account soil organic matter as one of the factors and the need to include it in order to improve the model predictive capability in rehabilitated lands. 

Author Response

Point 5a: It is essential that you provide enough details in your methodology so that others should be able to read your Methodology section and repeat the work without the need to ask for the details. So please add more details.

Response 5a: A sentence was added in the methodology referring the reader to an appropriate textbook for experimental details.

Point 5b: Reconstituting from a slurry could lead to the destruction of soil aggregate stability, in your case the microaggregates. The implication is: it leads to abnormally higher bulk density during the drying process than the same soil which has been left out on the field to settle naturally. This will have implications on the porosity and pore size distribution. Since the study is dealing with the rehabilitation of lands to grow plants, I believe it is essential to note the issue with the methodology followed in the Methodology section.

Response 5b: A sentence was added reiterating that the model validation was performed on reconstituted samples, and validation does not apply to in-situ soils. The future research in the conclusion was extended to state the model may be validated for samples replicating in-situ soil structures.

Point 5c: It is scientifically appropriate to bring forward the strengths and weaknesses as well as the need for further research in the conclusion. In this case, it would be appropriate to state that the model does not take into account soil organic matter as one of the factors and the need to include it in order to improve the model predictive capability in rehabilitated lands. 

Response 5c: Two sentences were placed in the conclusion stating that adding a regression parameter for soil organic matter can potentially improve accuracy of the model.