In their revised version of the manuscript, the authors satisfactorily responded to the comments of the reviewers by including extra measurements to confirm their findings. I think this paper is of importance for the soil science community since it addresses an important topic, namely the hydraulic conductivity of stony soils. The authors present interesting data that are in contradiction with currently available models that are used to relate the hydraulic properties of stony soils to the properties of the fine earth in these soils.
However, there are still a few main points that need to be clarified.
First, the author speak of replicate samples when they discuss the effect of stoniness on the saturated hydraulic conductivity. But, it is not clear what these replications in fact represent. Neither is it clear what the common factor is between these so-called replication.
Second, there are still some problems with the numerical simulations of the evaporation experiments. In fact, the authors should be able to evaluate the spatial variation of locally measured hydraulic heads due to the presence of stones and the effect that this may have on the estimation of effective or ‘average’ unsaturated hydraulic properties. But, the authors are referring to numerical instabilities. In case these numerical instabilities are important, the quality and relevance of the numerical simulations should be questioned.
Third, concerning the experimental procedure, the authors write that all samples were packed to the same bulk density. But, doesn’t this lead to lower bulk densities of the fine earth in the samples with high stone content and couldn’t this be the reason for the increase in saturated conductivity in samples with higher stone content?
Fourth, the presentation of the experiments and simulations that were carried out is not yet very well streamlined and should be better structured.
I am confident that the author can address these remaining issues relatively easily.
Detailed comments:
P1 Ln 22: Change to: ‘We pointed out several factors that determine saturated hydraulic conductivity of stony soil but that are not considered by these models.’ To my understanding, a driver is something that will lead to a dynamic change.
P2 ln 24: …other factors: include here which factors.
P2 ln 26- 27: ‘… the reduced volume available for flow might compensated by other factors.’ If you think of increased tortuousity, the effect of reduced volume is in fact amplified and not compensated. I think you must explicitly mention here the different factors you have in mind.
P 2 ln 27: ‘contradictory effect’ Change to ‘compensation factor’
P3 ln 2: add saturated and unsaturated hydraulic conductivity
P3 ln 3: aforementioned models: Change to: predictive models that have been proposed in the literature.
P4 ln 14-15: change non-porous to non-conductive since heat flow also takes place in non-porous media.
P5 Eq 7. The second equation for h>0 is not needed since K is given as a function of Se and Se cannot be larger than 1.
P5 Ln 17: add: ‘ for soil with rock fragments’ I suppose you do not mean the water content and water potential in the rock fragments.
P5 ln 21: ‘coarse inclusions’ add > ?? mm. You could also think of inclusions of pockets of sand in the clay.
P5 ln 27: Was the bulk density of the overall sample 1.51 g/cm³ or was the bulk density of the fine earth in the sample 1.51 g/cm³? In case there are stones in the sample (which have a larger density than the fine earth), packing the stony soil to the same density of the fine earth will lead to a lower density of the fine earth in the stony soil. Hence the porosity of the fine earth will be larger, which of course influences the hydraulic properties.
P 6 ln 12: add unsaturated hydraulic conductivity.
P 7 ln 17: Since you use K already for the hydraulic conductivity, I propose to change K to H here (to be consistent with Eq. 12).
P8 ln 18: change this sentence to: HYDRUS 2D solves the two dimensional Richards equation using the Galerkin finite element method.
P8 ln 21: ‘…rock fragments were supposed to be circular.’ This is confusing since you also did simulations for non-circular rock fragments.
P8 ln 24: ‘fitting on’ --> fitting of.
P9 paragraph 2.3.1. This paragraph should be rewritten.
P9: ln 12: what do you mean with setup extension?
P9 ln 14: What do you mean by numerical instability? Numerical instability refers to a numerical error that is made when the flow equation is solved numerically. But, in the stony sample, you may as well have spatial variations in the hydraulic head at one depth because of the influence that the stones have on the flow field. It is not clear from your line of argumentation whether you are talking about spatial variations of hydraulic heads or numerical errors. In case of the latter, you cannot argue that you use simulated pressure heads only at the top and bottom of the sample since the pressure heads inside the sample are prone to numerical errors. If there are important numerical errors at some locations, then you cannot use the results at other locations either because these will be impacted by the errors that you make. If the variations in pressure heads inside the soil columns are due to the impact of the stones on the pressure heads, then they represent a true effect that can be expected in such soils and that will have an impact on the measured pressure heads. In that case, you can use these simulations to evaluate the effect of these variations on the estimated hydraulic properties. You write that you use only the simulated pressure heads at one observation node. I think this is reasonable if you want to evaluate the effect of the tensiometer location on the estimated hydraulic functions. But then I think you should use data from different observation nodes and compare the obtained results if you consider these different data.
P10 ln 20: write: ‘and four volumetric stone fractions’
P 10 ln 19-27: This paragraph is not clear. First, it is not clear from the start that this is about true experiments. Second, it is not clear which stone content was considered in the samples with glass beads. Third, it is not clear how many replicates were actually considered. I am not saying that after reading, it could not be figured out: 5 replicates for the samples with rock fragments and 1 for the glass beads and the volumetric stone content of the glass bead samples is 20%. But, the way it is formulated is very indirect and not straightforward.
P10 ln 23-25. This I cannot follow. What do you mean by: The four replications were processed all together? What do you mean by: between replications, the soil was oven dried for 24 hours and passed through a 2 mm sieve? Do you mean that for one stone content you packed 4 times a sample using the same soil material?
P11 ln 5: change ‘developed’ into ‘presented’.
P11 ln 18: What does the 95% refers to? This always requires some statistical inference. But, you only have a few replicates (and for the predictions by the different models, you can hardly assume that the replicates are truly random repetitions). Furthermore, for the experiments, can you assume that the Kse values are normally distributed? Mostly, Kse values are lognormally distributed. To avoid all these problems and discussions, I propose to plot just the range between the minimal and maximal value and the median.
Ln 23 and figure 3: I do not understand why you can treat different samples with different stone contents as replicates of each other. In other words, what is the relation between the samples denoted by ‘gravels 1’? Why are these replicates? Replicate may be the wrong wording here. It seems as if you assume that there are two factors: one is the stone content and one is another factor but I could not figure out what it should represent. What is common among measurements that belong to the set or factor level ‘gravels 1’? First, you need to explain what these other factor levels represent. However, at this moment, I cannot find a reason why there should be other factor levels.
Ln 24-28: Since I do not understand what the replications represent and whether a common factor can be assumed between these samples, it does not seem to make sense to discuss how different individual samples compare with each other.
P12 ln 11: I would propose to replace ‘sampling’ by ‘packing’. Sampling rather refers to taking an soil sample whereas here, you ‘make’ a sample.
P12 ln 25: change ‘depending on’ to ‘with’.
P12 ln 17: again, I don’t understand the mean of ‘replicate’.
P13 ln 3: correct ‘permeability’
P13 ln 15: Change ‘drivers’. Table 3 illustrates the effect of different factors on the simulated Kse.
P15 ln 7: Which effect? Do you mean that with higher stone contents, the unsaturated conductivity will increase with increasing stone content as the Kse did?
Table 2: Use footnotes to explain Rv and n and the meaning of the circles, triangles and bars. Add also the number of replicates used in the experiments.
Figure 2: Check whether the graph becomes clearer when the Ks axis is logarithmically scaled. The 95% interval in the caption is not well defined. I would propose to change the 95% interval simply by the range between the maximal and minimal Kse that is measured or predicted by the different models.
Figure 5: explain in the figure caption what the dotted line refers to and explain that the triangles are saturated conductivities (closed is measured with black for the stony and grey for the fine earth, and open is predicted by the model (indicate which model exactly). Make also a distinction between the data points that are obtained from the two replicates. |