Reply on RC3

on

Reply: There is indeed a variety of landscapes in the catchment, as you can see on Google Earth. The WEP-COR model generalizes this and divides the underlying surface into five classes: water body, soil-vegetation, irrigated farmland, non-irrigated farmland, and impervious area (Lines 655-Line 658). The water body class includes rivers, lakes and glaciers. The soil-vegetation class includes bare land, grassland and woodland. Impervious area consists of urban buildings and impervious surfaces. These five classes of underlying surfaces represent all landscapes of the catchment in the model. Simulation of soil water and heat transport is not applicable for water body and impervious area classes, while for other underlying surfaces, the improvement of the water and heat transport model was achieved in this study. Apologies for the lack of clarity regarding the surface classification in the previous version of the paper; we will describe this in the revised manuscript.

L76-77 "The geological features of the QTP are generally thin soil layers above the thick gravel layers with clear boundaries between them." -> Any study supporting this at a large scale?
Reply: Affected by geological structure and freeze-thaw cycles, the phenomenon of the gravel layer under the thin soil layer is prevalent in the Qinghai-Tibet Plateau (Sun, 1996;Yang et al., 2009;Chen et al., 2015). Moreover, we confirmed this phenomenon in the present study with field observations in the study area, as shown in the figure below.

I also think mentioning "gravel" might be misleading. I think the authors should find a name to describe this type of formation (I see Wang et al. 2013 uses "soil-gravel mixtures" which sounds much more informative to me) because I think most readers reading "gravel" won't think about this unsorted slope deposits but rather to well sorted alluvial gravel formation that are highly conductive and then it might sound very counter intuitive. Especially when the author says "since gravel can neither conduct nor store water" and that it "hinders the movement of water".
Reply: Thanks for the comment and suggestion. In the revised version, we will replace "gravel" with "soil-gravel mixture (SGM) layer".

Another big problem lies in the presentation of the models. Many points are unclear or don't make sense. I detail this below. A good example of this is the method used to calculate snow melt, which is first said to be based on a snow depth threshold between contour bands rather than on climatic variables (even though it is later the case, when the author mentions a PDD method). Not mentioning the so-called "snow sliding". I suspect the problem doesn't lie in the model itself since the output looks good, but rather in the description. I am happy to be shown I am wrong if it is the case, but from what I read and understand, there are major problems in the identifications of the processes and their representation as they are described (under the hypothesis that what is coded is different and correct).
Reply: Sorry for not being clear enough about this part in the previous version of the manuscript. There are some problems in our description, which we addressed in detail in answering questions 10 and 11.

Finally, I started correcting the English but I am not a native speaker and the task here is too important for a scientific reviewer. So I would recommend having the manuscript proofread by a native speaker because I think many formulations can be improved (see the examples I give below for the beginning).
Reply: Thanks for professional questions. This is a mistake, and we will correct it in the revised version.

Since the model uses different types of equations depending on a threshold for rain, what happens at the transition between normal and heavy rain ? What if during a rain event the threshold is crossed, how smooth is the model in this regard ? Are there some data processing methods to smooth a potentially sharp transition ?
Reply: Thanks for the insightful comment. The flow in the model was simulated by day. When the daily precipitation exceeded the threshold, the Green-Ampt model was used. A sharp transition in the flow process has not been detected in the current simulations, but your question is very insightful, and we will conduct a series of studies on it in our future research.

So how does all of this work together? The authors have been, since the beginning, using studies on soils containing rock fragments to support a certain type of behavior from their bottom layer and now they argue towards another behavior because they have been calling these fragments "gravels". And at this stage I am confused. Maybe I missed something here but if so, there is a lack of pedagogy/clarity in the way this model and stratigraphy works together.
Reply: Sorry for the unclearness on this part. There are macropores in the soil-gravel mixture (SGM) layer. In a saturated state, the macropores form a fast channel for transporting water. However, when the SGM layer is in an unsaturated state, the water mainly moves under the actions of the matrix potential and gravitational potential. Thus, in an unsaturated state, the macropores do not work, and the gravel will hinder the movement of water. In the revised version, we explain this in detail.

L255 "(the contour bands)"
The authors need to explain more clearly how the model works in the main text.

I had to read the appendix to get a clearer idea of how this works. It is an unusual approach so it needs to be commented on. What decides the shape and extension of a band ? And it also needs some statistics: How many bands ? Average size of a band ? Average elevation range within a band ? I would also like to see a map with all these bands to see how all this looks like. Otherwise, what is done here remains very abstract.
Reply: We thank you for your valuable suggestion. Each sub-basin was divided into 1~10 contour bands as the basic calculation unit according to the elevation. The basin was divided into 871 contour bands. The average area of the contour bands is 20 km 2 . The subbasins of the watershed and the division of the contour bands are shown in the figure below.

Why is melt based on thickness difference ? Melt should be based on the climate input. All this makes very little sense, but I suspect these problems lie in the model description and not in the model itself.
Reply: Sorry for the unclear description. We will correct it in the revised version as follows: "When the snow thickness difference between two adjacent contour bands exceeds this threshold, snow begins to slide between those contour bands. The snow in the higheraltitude contour band slides into the lower band until the two bands equalize in snow thickness."

What is snow sliding ? I never heard of that and found nothing relevant on the net. The 2 important redistribution mechanisms I can think of are wind drift and avalanche. Snow creep also exists but is marginal in comparison. So what is the process here ?
Reply: In our study, "snow sliding" means avalanche, i.e. snow collapse driven by elevation difference. In the model, we generalized the avalanche as the redistribution of snow between two adjacent contour bands. In the revised version, we will replace the terms with "avalanche".

L290-292 "G is the heat flux (MJ/m2/d) conducted into the snow or soil, which was determined by the temperature difference between the soil or snow and the atmosphere near the surface. The above equation was combined with the ground heat conduction and energy balance equations" I think this is wrong. G is the energy input in the ground that is used to drive heat conduction after the surface energy balance equation has been applied. So G is not derived from the atmosphere temperature near the surface, but H is. G is what you get when you sum the energy fluxes from the radiations and turbulent fluxes. Another problem is that the end of the sentence talks about heat conduction and energy balance equations. Conduction has not been introduced but energy balance is actually equation 7.
Reply: Sorry for the inaccurate description of this process. This part introduced the energy balance equation used in the model and the calculation method of each energy flux. G was calculated from the temperature difference between the underlying surface and the atmosphere near the surface and was the heat flux conducted into the underlying surface. The heat conduction in the underlying surface was only related to G. After it was determined, the heat flux and temperature of each layer were calculated via Equation 10. In the revised version, we will simplify this part, remove extraneous equations, and provide the calculation method for G.

Well that is not what the authors say before. Equation 9 is clearly a way to calculate H from temperature inputs. It is impossible to deduce H from the energy balance equation because you deduce G from this equation knowing all the other terms.
Reply: Sorry for not being clear in the previous version of the manuscript on this part pointed out by you. As we replied in question 12, we will supplement the calculation method of G as follows: where,C Vu is the volumetric heat capacity of the underlying surface (MJ/m 3 /°C); d u is the depth of the underlying surface affected by heat conduction (m); T a is the air temperature on the day of simulation (°C); and T u is the surface temperature of the underlying surface on the day before simulation (°C).

L297 "The temperature difference between the atmosphere and the surface is the source of heat conduction" Why say this after calculating the surface energy balance ? The surface energy balance enables to calculate the energy change of the top cell, to work with temperature, the authors can then do ΔT = ΔE/Cp. Saying what I quote here after detailing an SEB module is more than confusing. I don't have this level of problematic issues with the rest of the paper. Yet I think that in general, the text of the result and discussion section could be lighter and more concise.
Reply: Thank you for the insightful comments. For clarification of the energy calculation part, you can refer to our replies to questions 12 and 13 here. Sorry for being not clear enough in the previous version of the manuscript on the aspects pointed out by you, which will be clarified in the revised version.

Specific comments I don't know where to put it so I write it here: to be able to understand what the new stratigraphy brings we need to have access to the WEP-COR stratigraphy, on Figure 3 for example.
Reply: The WEP-COR stratigraphy was shown in the previous version. See Figure B1 in the appendix for details.

L16 "The Qinghai-Tibet Plateau has a thin soil layer on top of a thick gravel layer"
I have 2 problems with this abstract opening:

Problem 1: See my previous comments, this cannot be true at the scale of a region as large as the QTP where one can find mountain peaks, peatlands, moraines, alluvial fans, blocky terrain… I suggest writing something like "For hydrological purposes, simplifying the representation of the QTP subsurface conditions to a thin soil layer on top of a thick gravel layer…" but this needs to be either demonstrated in a previous paper or in the present paper.
Reply: Thank you for the insightful comments. For the proof of the geological structure in the Qinghai-Tibet Plateau and the generalization method of the different underlying surfaces in the model, you can refer to the answers to questions 1 and 2.

Problem 2: I guess this is just a matter of personal preference, but I would recommend to start the abstract with a bit of context on what big question this study works with. Hydrology in mountainous cold regions and climate change…
Reply: Thank you for your professional suggestion. In the revised version, we will revise the abstract according to your recommendation.

L41-42 "plays an important role in ensuring the security of water resources in China and Southeast Asia"
Needs to be supported by a reference.
Reply: We added more references as follows:

L43-44 "cannot be ignored" Needs also a reference. The sentence is also surprising. The authors could start the sentence by "The extensive glacier…" list the items and end the sentence with "have a major impact on the water cycle…"
Reply: Thanks for your professional suggestion. We will modify this sentence as you suggested: "The extensive glacier, snow cover, and permanent and seasonal frozen soil have a major impact on the water cycle." Additional references are as follows:

This is a convoluted way to say that both in permafrost and permafrost free areas, the ground undergoes seasonal freezing.
Reply: Thank you for pointing out this deficiency. We will modify it according to your suggestion.

My expertise on the topic is limited but this section on the links between tectonics, sedimentology and granulometry of the Quaternary sediments could be better phrased and states obvious things that don't show particular relevance for the study. I don't understand the message the authors want to convey that is important for the paper.
Reply: Thank you for the comments. This paragraph introduces geological factors affecting the formation of the special underlying surface structure of QTP. In the revised version, we will simplify the description of this part.

Add the average elevation associated to this mean temperature
Reply: The average elevation of the catchment is 4688.6 m and this will be added to the revised version.

Reference for this value ? Also I doubt one can reach such a precision in the significant numbers of the percentage.
Reply: This value was calculated from ground temperature in ArcGIS according to the definition of permafrost: ground that remains at or below 0 °C for at least two consecutive years (Biskaborn et al., 2019;Dobinski W, 2011). We will reduce the significant digits to the whole number, 24% of the area under permafrost.

It is the first time the authors mention this site, maybe introduce it first.
Reply: Thanks for the comments. In the revised version, we will incorporate this sentence into the introduction of experimental site in (Lines 114-118). Fig. 1 Reply: Figure 2 was taken near the experimental site shown in Figure 1, and we will make notes under this figure in the revised version.

Does this give access to volume changes along time ?
Reply: Yes, glacier volume changed over time. The area of glaciers was obtained from four remote sensing images from 1994, 2003, 2009 and 2015. We linearly interpolated the glacier volume calculated from the area, making its temporal change as the model input.

This is a conclusion, it should not be part of the methods.
Reply: Thank you for this comment. We will delete this part from the Methods section in the revised version.

L193-196
"In the non-freeze-thaw period, the calculation object of water movement was defined as the dualistic soil-gravel structure (Fig. 3a)

This is really hard to read/understand, rephrase, with examples and tangible elements.
Reply: We rewrote this sentence and illustrated it with Figure 3c: "In the non-freeze-thaw period, the calculation object of water movement was defined as the dualistic soil-gravel structure. The upper layer was soil, whose thickness was determined by the location of the calculation unit and which gradually decreased from the foot to the peak of the mountain (Fig. 3c). The lower layer was the gravel layer (mixed layer of soil and gravel)." Figure 3c: Snow-soil-gravel layered structure.

L233-234
"Until the water has the same potential energy in the soil and the gravel, the INF  breaks through the critical surface, and then the infiltration rate stabilizes (Fig.  4)."   I don't understand this part. First I am unsure that potential energy is the good  terminology (i.e. potential energy of the water in a dam)

, I assume it is the pressure head. And if it is so, the Green-Ampt model does not calculate the pressure head, it calculates the volume of infiltrated water or the depth of the infiltration front. So I don't understand this sentence. Maybe I did not understand the situation correctly but then please clarify this point.
Reply: Potential energy here refers to soil water potential, including solute potential (not considered in this study), matric, gravity, and pressure potentials. The Green-Ampt model was derived by combining the Darcy's law with the continuity principle (Green and Ampt, 1911). The volume of infiltrated water, depth of the infiltration front, and capillary suction pressure were used to calculate the potential gradient in Darcy's law. The specific derivation process can be found in the references.

Figure 4: Cumulative infiltration process of the WEP-QTP model I don't understand this figure. Please give an explanation in the caption. Reexplain the letters. Why are there 2 dashed lines, are they different scenarios ? I see now that this part of the figure is modified from Jia et al. (2001). I think that it should be cited as a source element of the figure. Also, now that I found this image from Jia, I understand that what is represented are the successive wetting fronts. Yet what I don't understand is why we see these dashed curves. In the Green-Ampt model, the wetting front is horizontal.
Reply: Sorry for the obscurity on this part. A dashed line represents the wetting front at a moment, and the dashed lines in the figure represent the wetting fronts at different times: t 1 , t m , and t itf . The dashed line is the actual wetting front, but the Green-Ampt model equates it to a straight line separating the saturated soil above from the soil below. In the revised version we will redraw this figure as you suggested.

Here again it is hard to understand what the authors are doing. Where does this equation come from? Classically, the infiltration rate tends towards K because F(t) is at the place of Fitf here. But Fitf is hard to understand as it is a finite quantity and not a variable (i.e. the cumulative infiltration when the front breaks through). Also why ksoil when working with the "gravel" layers ? And What are "error caused by the different soil moisture content of the soil above the interface". I think this paragraph needs more pedagogy to avoid giving the feeling that the authors are doing their own cooking with some well-established equations.
Reply: Equation 2 was modified from Equation B4, which calculates the stable infiltration rate after the infiltration front penetrates the soil-gravel interface. Because the saturated hydraulic conductivity of the upper soil was the upper limit of the infiltration rate of the lower soil-gravel mixtures layer, K soil was used. The detailed calculation process of the infiltration rate and the calculation method of each parameter can be found in Equations B4-B6 in the appendix. In the revised version, we will improve the presentation of this section to help readers better understand our work.

What is the approach ? Constant values ? Values derived from climate forcing data?
Reply: The thickness of the snow layer (cm) was calculated as follows: where S snow is cumulative snow water equivalent (mm); ρ l is the densities of water (10 3 kg/m 3 ); ρ s is the densities of snow (kg/m 3 ), which was derived from climate forcing data as shown in Equations 11.

Equation 7 "RN = LE + H + G" A more comprehensive way to write it is G=… because it shows how what the authors derive from the climate forcing data is used as an input from the model. Also what about this equation when there is snow ? Is it also applied ?
Reply: Thank you for your professional suggestion. In the reply to questions 12 and 13, we simplified the introduction of the energy calculation and provided the calculation method for G.

Equation 10
How is the ice content linked to the temperature ? I assume the authors also use a soil freezing curve.
Reply: The relationship between the water and heat transport of the frozen soil is mainly manifested in the dynamic balance of the moisture content of the unfrozen water and the negative temperature of the soil, which is shown in Equation B14 in the Appendix.

Where is this value used ? I feel some part of the model description is missing. From what I understood there is the soil and the gravel layers, now it seems there is a riverbed layer. I just looked for the word riverbed, it does not appear before part 3. I have the hardest time understanding how this model works and I am trying hard.
Reply: The model improvement in this study does not involve the water exchange process of the river channel, but the riverbed conductivity is a sensitive parameter of the model; hence, the parameter calibration results were presented here. at the mountaintop, mountainside, and foot of  the mountain was 0.4 m, 0.6 m, and 1.0 m, respectively."

Was there an attempt to characterize the stratigraphy based on the topography/morphology ? If so it is not mentioned in the method. And it is more than necessary for the message this study wants to convey. So if this effort has been made, please explain it in the methods. Also as I said before, the stratigraphic observations should be presented in detail somewhere. They really contribute to the added-value of the paper.
Reply: Yes, as mentioned in the reply to question 28, we have redrawn the generalized structure of the model, and in the revised version, we will supplement this part accordingly.

Figure 11
The legend is so small, with the resolution I got for the figure (which is low) I cannot read it.
Reply: Thank you for your suggestion. In the revised version, we will replace all the figures with high-quality figures and improve figure layout.

This discussion is really hard to follow. How speculative is the existence of this interlayer ? What is the physical process that makes it a relevant hypothesis ? I think if this suggestion is important it needs to be explained in more detail.
Reply: Thank you for the insightful comment. The soil-gravel mixtures (SGM) layer under the topsoil is not homogeneous, but in the model, we assume homogeneity of this layer and use a uniform set of parameters to describe its water and heat properties. In the SGM layers, the observed value of water content in the 160-cm layer was smaller than that in other layers and between the simulated values of the WEP-QTP and WEP-COR; hence, we speculate that there may be a soil interlayer. Reply: Sorry for the misunderstanding of the description here. The simulated and measured runoff values are compared in Figures 5 and 9. Due to the limitations of the experimental site, the hydrological cycle fluxes in Figure 10 have no measured values. Figure 10 presents the effect of model improvement on the runoff process. We will revise this in the revised version.