the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 May 2021
Benchmark tests for separating n time components of runoff with one stable isotope tracer
Abstract. A validation of the recently introduced iterative extension of the standard two-component hydrograph separation method is presented. The data for testing this method are retrieved from a random rainfall generator and a rainfall-runoff model composed of linear reservoirs. The results show that it is possible to reconstruct the simulated event water response of a given random model input by applying the iterative separation model and using a single stable isotope tracer. The benchmark model also covers the partially delayed response of event water so that a situation can be simulated in which pre-event water is rapidly mobilized. It is demonstrated how mathematical constraints, such as an ill-conditioned linear equation system, may influence the separation of the event water response. In addition, it is discussed how the volume weighted separated event water response can serve as an estimator for a time-varying backward travel time distribution.
- Preprint
(2769 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
AC1: 'Comment on hess-2021-213 - Additional graphics to better visualize the discussion in section 4.3', Simon Hoeg, 18 May 2021
Dear Editor, dear Reviewers and interested Readers,
actually I plan to improve section 4.3 with additional results that show how intrastorm variabilities regarding ce (e), ce-1 (e+1), ce-2 (e+2) and ce-3 (e+3) can be calculated on the basis of the simulated event water response, which relates to the right hand side of Niemi's theorem, J(tin) Θ(tin) h→(t-tin,t). Then by piece wise equating Niemi's left hand side
h←(φ,t)Qt = Qe(φ), φ∈ e
h←(φ,t)Qt = Qe-1(φ), φ∈ e+1
h←(φ,t)Qt = Qe-2(φ), φ∈ e+2
h←(φ,t)Qt = Qe-3(φ), φ∈ e+3
I can finally resolve and recalculate the tracer concentrations ce (e), ce-1 (e+1), ce-2 (e+2) and ce-3 (e+3). The mean deviation between the simulated event water response and separated event water response is then in a range of 1e-13 % also for a large delayed fractions of event water αmax, for instance 0.3, given that criterion 1 and 4 is fulfilled, which I think is a quite good result. Of course, in real field experiments a simulated event water response is not available. Instead, the hydrologist has to work with calibration procedures and/or Monte Carlo simulations.
Best regards,
Simon Hoeg
Citation: https://doi.org/10.5194/hess-2021-213-AC1 -
AC2: 'Reply on AC1', Simon Hoeg, 11 Jul 2021
Since the discussion phase has been extended to mid of August 2021, please find attached the corresponding additional Figure:
Simulation with the damped scenario and random 18O strongly delayed input applying a reverse adaptation of the tracer concentrations ce (e); ce-1 (e+1), ..., ce-3 (e+3) based on the separated event water response by piece wise equating Niemi’s left hand side - Mean deviation between the simulated event water response and separated event water response for each event in % - Delta e : Deviation related to Qe , Delta e-1 : Deviation related to Qe-1 , Delta e-2 : Deviation related to Qe-2 , Delta e-3 : Deviation related to Qe-3 , Delta: The mean of all deviations.
-
AC2: 'Reply on AC1', Simon Hoeg, 11 Jul 2021
-
RC1: 'Comment on hess-2021-213', Anonymous Referee #1, 28 Aug 2021
This manuscript presents, as the title makes clear, benchmark tests of the author's approach (originally proposed in WRR in 2019) for identifying the fractions of runoff from several previous rainfall inputs, using a single isotope tracer.
Although the title and abstract do not mention this, an important contribution is the matrix formulation Ax=b for the author's method (equation 20 and appendix B), which is much easier to understand, and much more readily extensible, than the equations in the 2019 paper. The application of the condition number, however, is not the conventional one (equation 22), but one that is much less widely known (equation 23). Unfortunately this is introduced with the ambiguous phrase "a similar estimation can be derived" (how? by whom?), leaving readers to wonder where this comes from – a reference is sorely needed. Likewise equation 24 is introduced by "it can be shown that" (how? by whom?) and readers are given no clue what "f" is, making the equation uninterpretable.
More generally, there would seem to be three important (and possibly disqualifying) limitations in the use of the condition number as a guide to the reliability of this analysis. First, equation 23 holds only for infinitesimal errors in the matrix A, but there is no guarantee that realistic errors in the tracer concentrations (which, along with 1's and -1's, make up the non-zero elements of A) are small enough that equation 23 is still realistic. Second, it is hard to see how equation 23 can give useful guidance in the case of equation 20, where some elements of the matrix A are dimensionless constants (1's and -1's), and others are dimensional quantities (the tracer concentrations), so their relative magnitudes (and thus which ones have greater influence on the matrix norm) are dependent on the arbitrary choice of units for the tracer concentrations. Since the 1's and -1's have no uncertainty (and thus their corresponding elements of deltaA will be zero), it would seem that we could make the ratio of norm(deltaA)/norm(A) as big or small as we want (and thus get any answer that we want from equation 23) just by changing the units of the tracer concentrations. Third, it is not clear how this works when the vector b (which will be multiplied by the inverse of A) is mostly composed of zeroes, as it is here. This part of the paper badly needs a numerical demonstration, with realistic input values (and realistic errors in those input values).
The benchmark tests presented here assume (as far as I can tell) that there are no errors in the event and pre-event tracer concentrations. This is fundamentally unrealistic, and makes the estimates of the errors in the event water fractions (Table A1, Figures A4, A5, A7, A8, A10, A11, A13, A14, A16, A17, A19, A20, A22) meaningless as guides to the real-world reliability of the method. By contrast, the author's previous paper (Hoeg, 2019) showed relative errors of up to 100% or more in event water fractions estimated from real-world data from an experimental catchment. A realistic analysis requires realistic simulated errors in the tracer concentrations. These errors go well beyond the analytical errors in the measurement, and should include the likely sampling errors (i.e., the rainfall that is sampled may not be the average of the rain that falls over the whole landscape). In any case, the errors are certainly not zero, and it is not helpful to assume that they are.
The benchmark tests presented here also assume that there is no isotopic fractionation of either the event water or the pre-event water. This again makes the results unrealistic as guides to what one might expect in the real world. In the real world, evapotranspiration (including interception losses) is often the dominant term in the water balance (rather than zero, as assumed here), and can significantly alter the isotopic composition of the water reaching the surface, relative to the sampled precipitation, and may also alter the composition of the pre-event water over time. Any change in the isotopic composition of either the event water or the pre-event water would seem to pose serious challenges for the approach presented here.
The benchmark simulations are unrealistic in other ways as well. The precipitation events are large and regularly spaced, with very long rainless intervals in between, and the event water fractions are large compared to those that are typically observed in many real-world studies (including the author's 2019 study). And the behavior of the benchmark model itself is unrealistically simple; since it consists of two linear reservoirs with a constant partitioning coefficient, the forward transit time distributions of all precipitation events are identical. Readers would be more confident in the results if they were based on a benchmark model that is nonlinear and nonstationary, as real-world hydrologic systems are.
It seems that in some ways the benchmark tests have been designed to conform to the assumptions of the method. But to the extent that this is the case, the benchmark tests only show that the method would work in a world that conformed to the method's assumptions. Readers will be far more interested in whether the method is reliable in the real world, which requires benchmark tests under more realistic conditions.
The method is presented as being "based on an iterative balance of catchment input and output mass flows along the time axis" (line 84). This is not the case. As with conventional hydrograph separation, there is no mass balance of inputs and outputs in the sense of equations 1 and 2, but just conservative mixing of the event water and pre-event water.
It is odd to see all the figures put into an appendix, but on the other hand they are too numerous and repetitive to all go in the main text. Presumably this could be straightened out in an eventual revision.
This review has not considered the additional material that has been supplied as author comments, because the HESS review process is – as far as I know – based on the assumption that submitted manuscripts are complete and final, not drafts that are still undergoing revision.
Citation: https://doi.org/10.5194/hess-2021-213-RC1 -
AC3: 'Reply on RC1', Simon Hoeg, 08 Sep 2021
Dear Referee #1,
thanks for your constructive and detailed feedback on the manuscript hess-2021-213. By considering your points I believe that the quality of the manuscript will be further enhanced.
Please find my reply to your comments below.
> Although the title and abstract do not mention this, an important contribution is the matrix formulation Ax=b for the author's method (equation 20 and appendix B), which is much easier to understand, and much more readily extensible, than the equations in the 2019 paper. The application of the condition number, however, is not the conventional one (equation 22), but one that is much less widely known (equation 23). Unfortunately this is introduced with the ambiguous phrase "a similar estimation can be derived" (how? by whom?.), leaving readers to wonder where this comes from – a reference is sorely needed. Likewise equation 24 is introduced by "it can be shown that" (how? by whom?) and readers are given no clue what "f" is, making the equation uninterpretable.
The reference for equation 23, which describes the relative error in vector x in case of small disturbances in the coefficients of matrix A, would be Stoer (1994). It is true that equation 23 is not that often mentioned in the literature, maybe I can replace it by a more popular form that would be:
norm(delta x)/norm(x) <= cond(A)/(1-cond(A)*norm(delta A)/norm(A))*norm(delta A)/norm(A)
Equation 24 can be found at Deuflhard and Hohmann (2019) and holds for disturbances in matrix A, in which f'(A) is the derivation of matrix A, also known as the Jacobian matrix. With equation 24, I wanted to emphasize that the norm of the first derivative of the matrix A is also limited by its condition. This is interesting in terms of the Gaussian error propagation, in which the entries of f'(A) are applied to the uncertainties of the known variables. This way, I wanted to point out analytically that a basic Gaussian error analysis of the system A*x = b would not produce different results than an analysis based on the condition number. Certainly, I could better describe this in the corresponding section. However, in section 4.2, I have written that an "ill-conditioned system will lead to larger gradients in a Jacobian matrix and to potentially higher errors in a Gaussian error propagation".
> More generally, there would seem to be three important (and possibly disqualifying) limitations in the use of the condition number as a guide to the reliability of this analysis. First, equation 23 holds only for infinitesimal errors in the matrix A, but there is no guarantee that realistic errors in the tracer concentrations (which, along with 1's and -1's, make up the non-zero elements of A) are small enough that equation 23 is still realistic. Second, it is hard to see how equation 23 can give useful guidance in the case of equation 20, where some elements of the matrix A are dimensionless constants (1's and -1's), and others are dimensional quantities (the tracer concentrations), so their relative magnitudes (and thus which ones have greater influence on the matrix norm) are dependent on the arbitrary choice of units for the tracer concentrations. Since the 1's and -1's have no uncertainty (and thus their corresponding elements of deltaA will be zero), it would seem that we could make the ratio of norm(deltaA)/norm(A) as big or small as we want (and thus get any answer that we want from equation 23) just by changing the units of the tracer concentrations. Third, it is not clear how this works when the vector b (which will be multiplied by the inverse of A) is mostly composed of zeroes, as it is here. This part of the paper badly needs a numerical demonstration, with realistic input values (and realistic errors in those input values).As indicated already in the Introduction section, the condition number is used only as a measure that describes how strongly an input error can affect the calculated output in the worst case. The condition number is not used to calculate absolute or relative errors, maybe I should even better emphasize this in the text. Instead, I exactly calculate absolute and relative errors on the basis of a comparison between the separated event water response with the simulated event water response for different test scenarios in the results section 3. The absolute errors Δ [mm/h] and relative errors Δ [%] are visualized and described for every test scenario. My intention for using the condition number is to evaluate the mathematical constraints that arise directly from the applied linear equation system 20 (A*x=b) by detecting ill-conditioned and well-conditioned situations. It is a very elementary approach, but necessary for every kind of mathematical model, since there is a point at which even exact knowledge of physical input parameters might not lead to reliable results any more. Imagine a catchment in which the bulk mass flow and isotope concentration of water would be exactly known at each point in time. Then, in case of an ill-conditioned linear equation system A*x=b, the proposed iterative separation method might not produce reliable event and pre-event water fractions any more, regardless of any other factor. I have mentioned this in section 2.3 "Condition number and error estimation". Later in the results section 3, I have shown this by simply violating Criterion 1.
> The benchmark tests presented here assume (as far as I can tell) that there are no errors in the event and pre-event tracer concentrations. This is fundamentally unrealistic, and makes the estimates of the errors in the event water fractions (Table A1, Figures A4, A5, A7, A8, A10, A11, A13, A14, A16, A17, A19, A20, A22) meaningless as guides to the real-world reliability of the method. By contrast, the author's previous paper (Hoeg, 2019) showed relative errors of up to 100% or more in event water fractions estimated from real-world data from an experimental catchment. A realistic analysis requires realistic simulated errors in the tracer concentrations. These errors go well beyond the analytical errors in the measurement, and should include the likely sampling errors (i.e., the rainfall that is sampled may not be the average of the rain that falls over the whole landscape). In any case, the errors are certainly not zero, and it is not helpful to assume that they are.> The benchmark tests presented here also assume that there is no isotopic fractionation of either the event water or the pre-event water. This again makes the results unrealistic as guides to what one might expect in the real world. In the real world, evapotranspiration (including interception losses) is often the dominant term in the water balance (rather than zero, as assumed here), and can significantly alter the isotopic composition of the water reaching the surface, relative to the sampled precipitation, and may also alter the composition of the pre-event water over time. Any change in the isotopic composition of either the event water or the pre-event water would seem to pose serious challenges for the approach presented here.
I agree that the paper (Hoeg, 2019) appears far more descriptive and familiar to experimental researchers, because it is based on data from a field study. Even I personally prefer analyzing experimental lab or field data. However, in the current study I present the results of an elementary benchmark test for the iterative separation model introduced in the year 2019, which from my point of view is an absolutely necessary exercise. In the discussion section "Design and applicability of the rainfall-runoff model", I have described why the applied synthetically generated data do not reflect the complexity of a natural hydrological system.
> The benchmark simulations are unrealistic in other ways as well. The precipitation events are large and regularly spaced, with very long rainless intervals in between, and the event water fractions are large compared to those that are typically observed in many real-world studies (including the author's 2019 study). And the behavior of the benchmark model itself is unrealistically simple; since it consists of two linear reservoirs with a constant partitioning coefficient, the forward transit time distributions of all precipitation events are identical. Readers would be more confident in the results if they were based on a benchmark model that is nonlinear and nonstationary, as real-world hydrologic systems are.
The study validates the iterative separation method on the basis of numerical simulations. The applied rainfall-runoff model has one important advantage, it enables an exact determination of the event water response across an arbitrary number of subsequent rainfall runoff events. Therefore, the focus of the current investigation can be on the capabilities of the iterative separation method, and not on the exactness of the benchmark model.
I am aware that hydrologists prefer realistic rainfall-runoff models, but I am sure that they also prefer reliable separation methods. Here, I had to find a suitable compromise and have chosen a rather elementary setup that will be replicable for everyone. Actually, the events are not regularly spaced in section 3.2 "Random rainfall and 18O input", but they are indeed regularly spaced for technical reasons in section 3.3 "Random rainfall and 18O input and delayed response of event water". If desired, I can put more randomnes in section 3.2 regarding the event intervals and the partition coefficient, it will not touch the outcome in the results section. However, in section 3.3. random event intervals would definitely lead to different results, since for technical reasons the rapid mobilization of pre-event water would not be in sync with the rainfall impulses any more, which from my point view would appear physically unrealistic.
> It seems that in some ways the benchmark tests have been designed to conform to the assumptions of the method. But to the extent that this is the case, the benchmark tests only show that the method would work in a world that conformed to the method's assumptions. Readers will be far more interested in whether the method is reliable in the real world, which requires benchmark tests under more realistic conditions.
I have done thousands of simulations with very different parameter sets. For the manuscript, I have picked some examples to demonstrate where the iterative separation model shows good results, but also where it should not be applied from a mathematical and numerical point of view. Maybe I should make this more clear in the conclusions. One the one hand, it is a violation of the model criteria, on the other hand, it may be large fractions of rapidly mobilized pre-event water. The latter was also a surprising result for me. Nevertheless, on the basis of Niemis theorem it is then possible to find the exact event water response by applying the corresponding age functions, respectively recalculated tracer concentrations. However, in the field practice, where the event water response, respectively travel time distribution, is normally unknown, this might require the use of Monte Carlo simulations.
> The method is presented as being "based on an iterative balance of catchment input and output mass flows along the time axis" (line 84). This is not the case. As with conventional hydrograph separation, there is no mass balance of inputs and outputs in the sense of equations 1 and 2, but just conservative mixing of the event water and pre-event water.
I think as long as all the criteria (1-4) are fulfilled, the above sentence should hold. For instance, I could reword the sentence to "based on an iterative balance of catchment event and pre-event mass flows along the time axis"?
> It is odd to see all the figures put into an appendix, but on the other hand they are too numerous and repetitive to all go in the main text. Presumably this could be straightened out in an eventual revision.
If some of the figures might appear too redundant, then I could certainly thin out a bit.
> This review has not considered the additional material that has been supplied as author comments, because the HESS review process is – as far as I know – based on the assumption that submitted manuscripts are complete and final, not drafts that are still undergoing revision.
This is fine for me. Thank you for your feedback.
With kind regards,
Simon HoegCitation: https://doi.org/10.5194/hess-2021-213-AC3
-
AC3: 'Reply on RC1', Simon Hoeg, 08 Sep 2021
-
RC2: 'Comment on hess-2021-213', Anonymous Referee #2, 18 Oct 2021
Author validated the iterative extension of standard two-component hydrograph separation method using a single isotopic signature (published in WRR in 2019) through a random rainfall generator and a rainfall-runoff model. I am happy to see that author has addressed a critical question raised during the WRR review why the new modeling approach did not overall improve the model’s uncertainty but instead increased uncertainties for some events. This current study confirmed the speculation at the time via “condition number” that it is the difference between new and old water that caused the increase in uncertainty. In addition, author demonstrated in the current study how the volume weighted separated event water response can serve as an estimator for a time-varying backward travel time distribution, which I consider is a novel contribution and will be of great interests to watershed hydrologists.
One defect in the design, however, is that ET (evaporation + transpiration) is entirely ignored in the rainfall-runoff model (Figure A2, though considered in equations 1 and 2), which is not realistic in the real-world scenario. It is unclear how consideration of ET will affect the modeling outcomes. Not to mention the impact of evaporation on isotopic fractionation (e.g., via bare soil evaporation and canopy interception), the influence of ET on streamflow quantity alone should affect the magnitude and timing of hydrograph and thus in my opinion possibly the isotopic signature of pre-event water, which is taken from streamflow right before an event. This defect will limit the application of this method to concept proofing and prevent it from broad application to real world-problem solving. But I am aware of the fact that adding ET impact in current study would make it way too complicated and extremely hard to follow. I think author should talk about any possible ET effect in a very general way in the current study and let readers be aware of its potential impact and explore whether or not this could be one of future studies.
Additionally, I am curious if both isotopic tracers (2H and 18O) are used, whether or not isotopic fractionation due to evaporation and phase change can be incorporated into the hydrograph separation model. If so, this would significantly boost its application and extend to snowmelt-dominated catchments. I do not think the current model is applicable to streamflow generated primarily from snowmelt.
I am also curious about the potential impact of prolonged droughts on the model results. What if there are two events that occur several months apart?
Though I am not suggesting significant revisions to address the above issues in the current study, any potential impact and future extension should be discussed in the discussion section and the limitation should be cautioned in the conclusion section.
To promote this study and extend the use of the method, I suggest author to add more details to some of the mathematical equations. I have some questions on these equations (see below), which basically require clarification or more information.
P3/L77: The bracket not closed at the end?
P3/Equation 3: The time factor is muted for all quantities in the equations. That is fine, but readers must be reminded that within an event isotopic signature in all components is assumed to be constant. This implicit assumption was discussed later in the discussion section, but I think this assumption should be explicitly stated here and then discussed later in the discussion section.
P5/L122: Isn’t “sequential event water response” a better term than “separated event water response”?
P5/Equation 12: Why does the integral start from negative infinite, not tin (in as subscript)?
Why not dt but dtin instead? Need to explain. I have the same question for equation 15 in page 6.P7/L188: As “condition number” appears for the first time in the text and is not a well-known term in hydrology, it needs either a reference or a further explanation or at least an indication it is being explained below.
P8/Equation 23: It is unclear how the term after the plus sign is derived or defined.
P8/L208: What is gamma here? Explain.
P8/Equation 24: “f” should be better explained.
P11/L313: ET was not considered, but why were precipitation and runoff not equal?
P15/L411-421: Need to mention that the current model setup does not work for snowmelt-dominated system.
P16/L471: It is not constant due to the variable nature of Qt? If so, explicitly say so.
P18/Conclusions: The limitations of current model need to be briefly summarized as well.
Figure A1: This figure gave me hard time at the beginning, as it is depicted by forward concept not backward notation, while the latter is dominated the text. Need more information in the caption to explain this so that readers follow it easily.
Figure A2: Though this model was explained in the text, enough information should be given in the caption to make it stand by itself. At least parameters (alpha and eta) need to be explained in the caption for better readability.
Figure A3: Missing the second y-axis labels.
Figure A4: For one or two curves, why not smoothed near the end? I do not remember if this has been explained in the text.
Figure A5: Extremely hard to distinguish the curves, which occurs in other figures as well.
Figure A9: Need to say what the two arrows are for in the caption.
Figure A15: Where is precipitation? Also, missing mentioning ce (e as subscript) in the caption. The same issue exists for Figure A18.
Citation: https://doi.org/10.5194/hess-2021-213-RC2 -
AC4: 'Reply on RC2', Simon Hoeg, 31 Oct 2021
Dear Referee #2,
I appreciate your very positive and motivating feedback on my manuscript hess-2021-213.
> Author validated the iterative extension of standard two-component hydrograph separation method using a single isotopic signature (published in WRR in 2019) through a random rainfall generator and a rainfall-runoff model. I am happy to see that author has addressed a critical question raised during the WRR review why the new modeling approach did not overall improve the model’s uncertainty but instead increased uncertainties for some events. This current study confirmed the speculation at the time via “condition number” that it is the difference between new and old water that caused the increase in uncertainty. In addition, author demonstrated in the current study how the volume weighted separated event water response can serve as an estimator for a time-varying backward travel time distribution, which I consider is a novel contribution and will be of great interests to watershed hydrologists.
Thank you for summarizing the major outcomes.
> One defect in the design, however, is that ET (evaporation + transpiration) is entirely ignored in the rainfall-runoff model (Figure A2, though considered in equations 1 and 2), which is not realistic in the real-world scenario. It is unclear how consideration of ET will affect the modeling outcomes. Not to mention the impact of evaporation on isotopic fractionation (e.g., via bare soil evaporation and canopy interception), the influence of ET on streamflow quantity alone should affect the magnitude and timing of hydrograph and thus in my opinion possibly the isotopic signature of pre-event water, which is taken from streamflow right before an event. This defect will limit the application of this method to concept proofing and prevent it from broad application to real world-problem solving. But I am aware of the fact that adding ET impact in current study would make it way too complicated and extremely hard to follow. I think author should talk about any possible ET effect in a very general way in the current study and let readers be aware of its potential impact and explore whether or not this could be one of future studies. Additionally, I am curious if both isotopic tracers (2H and 18O) are used, whether or not isotopic fractionation due to evaporation and phase change can be incorporated into the hydrograph separation model. If so, this would significantly boost its application and extend to snowmelt-dominated catchments.
Yes, ET and the impact of evaporation on isotopic fractionation is not explicitly considered in the current iterative separation equations. Instead, the effects of phase change on tracer concentrations would be subject to an Gaussian error analysis. Additionally, in the introduction I wrote that in the context of a chronological sequence of rainfall-runoff events, the event water concentration c_e can be related to the tracer concentration in the precipitation c_J. This relation (e.g. f: c_J -> c_e) is always subject to certain assumptions, which are discussed in almost every hydrograph separation study. Based on function f, I would consider processes such as base soil evaporation and canopy interception when using the current model setup. In this numerical study, I have simply defined that c_J = c_e. However, mid and long term goal would be an extension of the current iterative mixing of the end members Q_e, Q_e-1, ... Q_e-n and Q_p, Q_p-1, ... Q_p-n towards an iterative splitting (see e.g. Kirchner and Allen (2020)) that would also explicitly include evapotranspiration processes.
> I do not think the current model is applicable to streamflow generated primarily from snowmelt.
If the quantity and isotope composition of snowmelt has been sufficiently monitored, then it can be also applied to the current model. In this case the above mentioned relation f could be extended to f(c_J, c_m) -> c_e.
> I am also curious about the potential impact of prolonged droughts on the model results. What if there are two events that occur several months apart?
This is not a problem, as long as the age function remain stable (see e.g. line 155), that as long as piece-wise steady state conditions apply. If the age function has changed, which of course will happen in a prolonged drought, then it should be considered in the calculations. See also my comment at https://doi.org/10.5194/hess-2021-213-AC1
> Though I am not suggesting significant revisions to address the above issues in the current study, any potential impact and future extension should be discussed in the discussion section and the limitation should be cautioned in the conclusion section.
Yes, the discussion and conclusion section might additionally address the above mentioned topics.
> To promote this study and extend the use of the method, I suggest author to add more details to some of the mathematical equations. I have some questions on these equations (see below), which basically require clarification or more information.
P3/L77: The bracket not closed at the end?
Yes, it is not closed, since [t_n, t_(n+1)[ would be the next semantic interval.
> P3/Equation 3: The time factor is muted for all quantities in the equations. That is fine, but readers must be reminded that within an event isotopic signature in all components is assumed to be constant. This implicit assumption was discussed later in the discussion section, but I think this assumption should be explicitly stated here and then discussed later in the discussion section.
It is true that the intra-storm time variability is muted for the end member concentrations c_e, c_e-1, ... c_e-n and c_p, c_p-1, ... c_p-n in most of the simulations. However, on page 4, Criteria 2 and 3 say that the "event/pre-event component maintains a constant isotopic signature in space and time, or any variations can be accounted for". From line 458 on page 16, I discuss for the strong delayed fractions scenario that with an increasing proportion of rapidly mobilized pre-event water, such as Q_(e-1) , the pre-event water concentration c_p (assumed to be constant during the event) may decreasingly represent the bulk pre-event concentration of the considered control volume. Therefore, in https://doi.org/10.5194/hess-2021-213-AC1 and https://doi.org/10.5194/hess-2021-213-AC2 I demonstrate a reverse and time variable adaptation of the tracer concentrations ce (e); ce-1 (e+1), ..., ce-3 (e+3) by piece wise equating Niemi’s left hand side. The mean deviation between the simulated event water response and separated event water response is then in a range of 1e-13% also for a large delayed fractions of event water alpha_max.
> P5/L122: Isn’t “sequential event water response” a better term than “separated event water response”?
Basically yes, but I wanted to emphasize that this variable is derived from a hydrograph separation. Therefore, I would like to keep the term.
> P5/Equation 12: Why does the integral start from negative infinite, not tin (in as subscript)?
Here, I simply adopted the notation of Niemi (1977). I have to say that it is the more exact reference, since Maloszweski and Zuber (1982) applied similar equations, but used the expression "exit age" instead of "injection time". Also Botter et al. (2011) used the notation of Niemi (1977). Later van der Velde (2012) or Harmann (2015) defined backward and forward approaches by using the actual time (t) instead of the injection time (t_in), in the context of an age ranked distribution of the water and associated components being based on cumulative probabilities. To answer your question: The integral itself may not start from t_in, as t_in is the integration variable, but in practice only the values c_j(t_in) * h(t - t_in) > 0 will play a role for the result.
> Why not dt but dtin instead? Need to explain. I have the same question for equation 15 in page 6.
The variable t refers to the actual time, however in the backward perspective we look back to the injection time t_in.
> P7/L188: As “condition number” appears for the first time in the text and is not a well-known term in hydrology, it needs either a reference or a further explanation or at least an indication it is being explained below.
I agree, here I could add the references of Stoer (1994) and Higham (1995). The latter is currently available free of charge from Elsevier.
> P8/Equation 23: It is unclear how the term after the plus sign is derived or defined.
The Landau notation is used to describe the asymptotic behavior, when approaching a finite or infinite limit value. The big O is used to indicate a maximum order of magnitude. Actually, equation 23 is not that often mentioned in the literature, therefore I will replace it by a more popular form that would be: norm(Δx)/norm(x) <= cond(A)/(1-cond(A)*norm(ΔA)/norm(A))*norm(ΔA)/norm(A)
> P8/L208: What is gamma here? Explain.
Gamma is a scalar, a real number. The property cond(gamma*A) = cond(A) says that the size of the condition number for matrix A cannot determined by any scaling factor.
> P8/Equation 24: “f” should be better explained.
See also my response at https://hess.copernicus.org/preprints/hess-2021-213/#AC3: Equation 24 can be found at Deuflhard and Hohmann (2019) and holds for disturbances in matrix A, in which f'(A) is the derivation of matrix A, also known as the Jacobian matrix. With equation 24, I wanted to emphasize that the norm of the first derivative of the matrix A is also limited by its condition. This is interesting in terms of the Gaussian error propagation, in which the entries of f'(A) are applied to the uncertainties of the known variables. This way I wanted to point out analytically that a basic Gaussian error analysis of the system A*x = b would not produce different results than an analysis based on the condition number. Certainly, I could better describe this in the corresponding section. However, in section 4.2, I have written that an "ill-conditioned system will lead to larger gradients in a Jacobian matrix and to potentially higher errors in a Gaussian error propagation".
> P11/L313: ET was not considered, but why were precipitation and runoff not equal?
The simulation is stopped exactly after four years (4*365*24 hours). During the last time step the outflow Q is still 0.08 mm, and the missing 2 mm belong to the storages S_U and S_L.
> P15/L411-421: Need to mention that the current model setup does not work for snowmelt-dominated system.
I could mention that the current model setup has not been tested yet in snowmelt-dominated systems. Provided that the precipitation, snow layer and meltwater are adequately observed, I see no reason why it should not work.
> P16/L471: It is not constant due to the variable nature of Qt? If so, explicitly say so.
The time variability of Qt is not the only reason. In the simulations from Sections 3.1 and 3.2, the effect (intra-storm time-varying concentrations c_e(e), c_e-1(e+1), c_e-2(e+2)) does not occur or only to a small extent.
> P18/Conclusions: The limitations of current model need to be briefly summarized as well.
Yes, I agree.
> Figure A1: This figure gave me hard time at the beginning, as it is depicted by forward concept not backward notation, while the latter is dominated the text. Need more information in the caption to explain this so that readers follow it easily.
I agree, the presentation is not that accurate. I should emphasize the backward perspective. It is not necessary to focus so much on the precipitation J.
> Figure A2: Though this model was explained in the text, enough information should be given in the caption to make it stand by itself. At least parameters (alpha and eta) need to be explained in the caption for better readability.
Yes, I agree.
> Figure A3: Missing the second y-axis labels.
There is only y-axis in this figure, but I will improve or remove the tickmarks on the right hand side.
> Figure A4: For one or two curves, why not smoothed near the end? I do not remember if this has been explained in the text.
Near the end the condition numbers are very high (up to 2x10^6). In line 289, I have mentioned that "there are increasing (numerical) instabilities in the last events".
> Figure A5: Extremely hard to distinguish the curves, which occurs in other figures as well.
Yes, sometimes not that easy to distinguish, you have to zoom in. I will try with wider lines.
> Figure A9: Need to say what the two arrows are for in the caption.
Yes, I agree.
> Figure A15: Where is precipitation? Also, missing mentioning ce (e as subscript) in the caption. The same issue exists for Figure A18.
I agree, and will improve the figures A15 and A18.Thank you for your help and feedback.
With kind regards,
Simon HoegAdditional References:
Kirchner, J. W. and Allen, S. T.: Seasonal partitioning of precipitation between streamflow and evapotranspiration, inferred from end-member splitting analysis, Hydrol. Earth Syst. Sci., 24, 17–39, https://doi.org/10.5194/hess-24-17-2020, 2020van der Velde, Y., Torfs, P. J. J. F., van der Zee, S. E. A. T. M., and Uijlenhoet, R. (2012), Quantifying catchment-scale mixing and its effect on time-varying travel time distributions, Water Resour. Res., 48, W06536, doi:10.1029/2011WR011310.
Citation: https://doi.org/10.5194/hess-2021-213-AC4
-
AC4: 'Reply on RC2', Simon Hoeg, 31 Oct 2021
Status: closed
-
AC1: 'Comment on hess-2021-213 - Additional graphics to better visualize the discussion in section 4.3', Simon Hoeg, 18 May 2021
Dear Editor, dear Reviewers and interested Readers,
actually I plan to improve section 4.3 with additional results that show how intrastorm variabilities regarding ce (e), ce-1 (e+1), ce-2 (e+2) and ce-3 (e+3) can be calculated on the basis of the simulated event water response, which relates to the right hand side of Niemi's theorem, J(tin) Θ(tin) h→(t-tin,t). Then by piece wise equating Niemi's left hand side
h←(φ,t)Qt = Qe(φ), φ∈ e
h←(φ,t)Qt = Qe-1(φ), φ∈ e+1
h←(φ,t)Qt = Qe-2(φ), φ∈ e+2
h←(φ,t)Qt = Qe-3(φ), φ∈ e+3
I can finally resolve and recalculate the tracer concentrations ce (e), ce-1 (e+1), ce-2 (e+2) and ce-3 (e+3). The mean deviation between the simulated event water response and separated event water response is then in a range of 1e-13 % also for a large delayed fractions of event water αmax, for instance 0.3, given that criterion 1 and 4 is fulfilled, which I think is a quite good result. Of course, in real field experiments a simulated event water response is not available. Instead, the hydrologist has to work with calibration procedures and/or Monte Carlo simulations.
Best regards,
Simon Hoeg
Citation: https://doi.org/10.5194/hess-2021-213-AC1 -
AC2: 'Reply on AC1', Simon Hoeg, 11 Jul 2021
Since the discussion phase has been extended to mid of August 2021, please find attached the corresponding additional Figure:
Simulation with the damped scenario and random 18O strongly delayed input applying a reverse adaptation of the tracer concentrations ce (e); ce-1 (e+1), ..., ce-3 (e+3) based on the separated event water response by piece wise equating Niemi’s left hand side - Mean deviation between the simulated event water response and separated event water response for each event in % - Delta e : Deviation related to Qe , Delta e-1 : Deviation related to Qe-1 , Delta e-2 : Deviation related to Qe-2 , Delta e-3 : Deviation related to Qe-3 , Delta: The mean of all deviations.
-
AC2: 'Reply on AC1', Simon Hoeg, 11 Jul 2021
-
RC1: 'Comment on hess-2021-213', Anonymous Referee #1, 28 Aug 2021
This manuscript presents, as the title makes clear, benchmark tests of the author's approach (originally proposed in WRR in 2019) for identifying the fractions of runoff from several previous rainfall inputs, using a single isotope tracer.
Although the title and abstract do not mention this, an important contribution is the matrix formulation Ax=b for the author's method (equation 20 and appendix B), which is much easier to understand, and much more readily extensible, than the equations in the 2019 paper. The application of the condition number, however, is not the conventional one (equation 22), but one that is much less widely known (equation 23). Unfortunately this is introduced with the ambiguous phrase "a similar estimation can be derived" (how? by whom?), leaving readers to wonder where this comes from – a reference is sorely needed. Likewise equation 24 is introduced by "it can be shown that" (how? by whom?) and readers are given no clue what "f" is, making the equation uninterpretable.
More generally, there would seem to be three important (and possibly disqualifying) limitations in the use of the condition number as a guide to the reliability of this analysis. First, equation 23 holds only for infinitesimal errors in the matrix A, but there is no guarantee that realistic errors in the tracer concentrations (which, along with 1's and -1's, make up the non-zero elements of A) are small enough that equation 23 is still realistic. Second, it is hard to see how equation 23 can give useful guidance in the case of equation 20, where some elements of the matrix A are dimensionless constants (1's and -1's), and others are dimensional quantities (the tracer concentrations), so their relative magnitudes (and thus which ones have greater influence on the matrix norm) are dependent on the arbitrary choice of units for the tracer concentrations. Since the 1's and -1's have no uncertainty (and thus their corresponding elements of deltaA will be zero), it would seem that we could make the ratio of norm(deltaA)/norm(A) as big or small as we want (and thus get any answer that we want from equation 23) just by changing the units of the tracer concentrations. Third, it is not clear how this works when the vector b (which will be multiplied by the inverse of A) is mostly composed of zeroes, as it is here. This part of the paper badly needs a numerical demonstration, with realistic input values (and realistic errors in those input values).
The benchmark tests presented here assume (as far as I can tell) that there are no errors in the event and pre-event tracer concentrations. This is fundamentally unrealistic, and makes the estimates of the errors in the event water fractions (Table A1, Figures A4, A5, A7, A8, A10, A11, A13, A14, A16, A17, A19, A20, A22) meaningless as guides to the real-world reliability of the method. By contrast, the author's previous paper (Hoeg, 2019) showed relative errors of up to 100% or more in event water fractions estimated from real-world data from an experimental catchment. A realistic analysis requires realistic simulated errors in the tracer concentrations. These errors go well beyond the analytical errors in the measurement, and should include the likely sampling errors (i.e., the rainfall that is sampled may not be the average of the rain that falls over the whole landscape). In any case, the errors are certainly not zero, and it is not helpful to assume that they are.
The benchmark tests presented here also assume that there is no isotopic fractionation of either the event water or the pre-event water. This again makes the results unrealistic as guides to what one might expect in the real world. In the real world, evapotranspiration (including interception losses) is often the dominant term in the water balance (rather than zero, as assumed here), and can significantly alter the isotopic composition of the water reaching the surface, relative to the sampled precipitation, and may also alter the composition of the pre-event water over time. Any change in the isotopic composition of either the event water or the pre-event water would seem to pose serious challenges for the approach presented here.
The benchmark simulations are unrealistic in other ways as well. The precipitation events are large and regularly spaced, with very long rainless intervals in between, and the event water fractions are large compared to those that are typically observed in many real-world studies (including the author's 2019 study). And the behavior of the benchmark model itself is unrealistically simple; since it consists of two linear reservoirs with a constant partitioning coefficient, the forward transit time distributions of all precipitation events are identical. Readers would be more confident in the results if they were based on a benchmark model that is nonlinear and nonstationary, as real-world hydrologic systems are.
It seems that in some ways the benchmark tests have been designed to conform to the assumptions of the method. But to the extent that this is the case, the benchmark tests only show that the method would work in a world that conformed to the method's assumptions. Readers will be far more interested in whether the method is reliable in the real world, which requires benchmark tests under more realistic conditions.
The method is presented as being "based on an iterative balance of catchment input and output mass flows along the time axis" (line 84). This is not the case. As with conventional hydrograph separation, there is no mass balance of inputs and outputs in the sense of equations 1 and 2, but just conservative mixing of the event water and pre-event water.
It is odd to see all the figures put into an appendix, but on the other hand they are too numerous and repetitive to all go in the main text. Presumably this could be straightened out in an eventual revision.
This review has not considered the additional material that has been supplied as author comments, because the HESS review process is – as far as I know – based on the assumption that submitted manuscripts are complete and final, not drafts that are still undergoing revision.
Citation: https://doi.org/10.5194/hess-2021-213-RC1 -
AC3: 'Reply on RC1', Simon Hoeg, 08 Sep 2021
Dear Referee #1,
thanks for your constructive and detailed feedback on the manuscript hess-2021-213. By considering your points I believe that the quality of the manuscript will be further enhanced.
Please find my reply to your comments below.
> Although the title and abstract do not mention this, an important contribution is the matrix formulation Ax=b for the author's method (equation 20 and appendix B), which is much easier to understand, and much more readily extensible, than the equations in the 2019 paper. The application of the condition number, however, is not the conventional one (equation 22), but one that is much less widely known (equation 23). Unfortunately this is introduced with the ambiguous phrase "a similar estimation can be derived" (how? by whom?.), leaving readers to wonder where this comes from – a reference is sorely needed. Likewise equation 24 is introduced by "it can be shown that" (how? by whom?) and readers are given no clue what "f" is, making the equation uninterpretable.
The reference for equation 23, which describes the relative error in vector x in case of small disturbances in the coefficients of matrix A, would be Stoer (1994). It is true that equation 23 is not that often mentioned in the literature, maybe I can replace it by a more popular form that would be:
norm(delta x)/norm(x) <= cond(A)/(1-cond(A)*norm(delta A)/norm(A))*norm(delta A)/norm(A)
Equation 24 can be found at Deuflhard and Hohmann (2019) and holds for disturbances in matrix A, in which f'(A) is the derivation of matrix A, also known as the Jacobian matrix. With equation 24, I wanted to emphasize that the norm of the first derivative of the matrix A is also limited by its condition. This is interesting in terms of the Gaussian error propagation, in which the entries of f'(A) are applied to the uncertainties of the known variables. This way, I wanted to point out analytically that a basic Gaussian error analysis of the system A*x = b would not produce different results than an analysis based on the condition number. Certainly, I could better describe this in the corresponding section. However, in section 4.2, I have written that an "ill-conditioned system will lead to larger gradients in a Jacobian matrix and to potentially higher errors in a Gaussian error propagation".
> More generally, there would seem to be three important (and possibly disqualifying) limitations in the use of the condition number as a guide to the reliability of this analysis. First, equation 23 holds only for infinitesimal errors in the matrix A, but there is no guarantee that realistic errors in the tracer concentrations (which, along with 1's and -1's, make up the non-zero elements of A) are small enough that equation 23 is still realistic. Second, it is hard to see how equation 23 can give useful guidance in the case of equation 20, where some elements of the matrix A are dimensionless constants (1's and -1's), and others are dimensional quantities (the tracer concentrations), so their relative magnitudes (and thus which ones have greater influence on the matrix norm) are dependent on the arbitrary choice of units for the tracer concentrations. Since the 1's and -1's have no uncertainty (and thus their corresponding elements of deltaA will be zero), it would seem that we could make the ratio of norm(deltaA)/norm(A) as big or small as we want (and thus get any answer that we want from equation 23) just by changing the units of the tracer concentrations. Third, it is not clear how this works when the vector b (which will be multiplied by the inverse of A) is mostly composed of zeroes, as it is here. This part of the paper badly needs a numerical demonstration, with realistic input values (and realistic errors in those input values).As indicated already in the Introduction section, the condition number is used only as a measure that describes how strongly an input error can affect the calculated output in the worst case. The condition number is not used to calculate absolute or relative errors, maybe I should even better emphasize this in the text. Instead, I exactly calculate absolute and relative errors on the basis of a comparison between the separated event water response with the simulated event water response for different test scenarios in the results section 3. The absolute errors Δ [mm/h] and relative errors Δ [%] are visualized and described for every test scenario. My intention for using the condition number is to evaluate the mathematical constraints that arise directly from the applied linear equation system 20 (A*x=b) by detecting ill-conditioned and well-conditioned situations. It is a very elementary approach, but necessary for every kind of mathematical model, since there is a point at which even exact knowledge of physical input parameters might not lead to reliable results any more. Imagine a catchment in which the bulk mass flow and isotope concentration of water would be exactly known at each point in time. Then, in case of an ill-conditioned linear equation system A*x=b, the proposed iterative separation method might not produce reliable event and pre-event water fractions any more, regardless of any other factor. I have mentioned this in section 2.3 "Condition number and error estimation". Later in the results section 3, I have shown this by simply violating Criterion 1.
> The benchmark tests presented here assume (as far as I can tell) that there are no errors in the event and pre-event tracer concentrations. This is fundamentally unrealistic, and makes the estimates of the errors in the event water fractions (Table A1, Figures A4, A5, A7, A8, A10, A11, A13, A14, A16, A17, A19, A20, A22) meaningless as guides to the real-world reliability of the method. By contrast, the author's previous paper (Hoeg, 2019) showed relative errors of up to 100% or more in event water fractions estimated from real-world data from an experimental catchment. A realistic analysis requires realistic simulated errors in the tracer concentrations. These errors go well beyond the analytical errors in the measurement, and should include the likely sampling errors (i.e., the rainfall that is sampled may not be the average of the rain that falls over the whole landscape). In any case, the errors are certainly not zero, and it is not helpful to assume that they are.> The benchmark tests presented here also assume that there is no isotopic fractionation of either the event water or the pre-event water. This again makes the results unrealistic as guides to what one might expect in the real world. In the real world, evapotranspiration (including interception losses) is often the dominant term in the water balance (rather than zero, as assumed here), and can significantly alter the isotopic composition of the water reaching the surface, relative to the sampled precipitation, and may also alter the composition of the pre-event water over time. Any change in the isotopic composition of either the event water or the pre-event water would seem to pose serious challenges for the approach presented here.
I agree that the paper (Hoeg, 2019) appears far more descriptive and familiar to experimental researchers, because it is based on data from a field study. Even I personally prefer analyzing experimental lab or field data. However, in the current study I present the results of an elementary benchmark test for the iterative separation model introduced in the year 2019, which from my point of view is an absolutely necessary exercise. In the discussion section "Design and applicability of the rainfall-runoff model", I have described why the applied synthetically generated data do not reflect the complexity of a natural hydrological system.
> The benchmark simulations are unrealistic in other ways as well. The precipitation events are large and regularly spaced, with very long rainless intervals in between, and the event water fractions are large compared to those that are typically observed in many real-world studies (including the author's 2019 study). And the behavior of the benchmark model itself is unrealistically simple; since it consists of two linear reservoirs with a constant partitioning coefficient, the forward transit time distributions of all precipitation events are identical. Readers would be more confident in the results if they were based on a benchmark model that is nonlinear and nonstationary, as real-world hydrologic systems are.
The study validates the iterative separation method on the basis of numerical simulations. The applied rainfall-runoff model has one important advantage, it enables an exact determination of the event water response across an arbitrary number of subsequent rainfall runoff events. Therefore, the focus of the current investigation can be on the capabilities of the iterative separation method, and not on the exactness of the benchmark model.
I am aware that hydrologists prefer realistic rainfall-runoff models, but I am sure that they also prefer reliable separation methods. Here, I had to find a suitable compromise and have chosen a rather elementary setup that will be replicable for everyone. Actually, the events are not regularly spaced in section 3.2 "Random rainfall and 18O input", but they are indeed regularly spaced for technical reasons in section 3.3 "Random rainfall and 18O input and delayed response of event water". If desired, I can put more randomnes in section 3.2 regarding the event intervals and the partition coefficient, it will not touch the outcome in the results section. However, in section 3.3. random event intervals would definitely lead to different results, since for technical reasons the rapid mobilization of pre-event water would not be in sync with the rainfall impulses any more, which from my point view would appear physically unrealistic.
> It seems that in some ways the benchmark tests have been designed to conform to the assumptions of the method. But to the extent that this is the case, the benchmark tests only show that the method would work in a world that conformed to the method's assumptions. Readers will be far more interested in whether the method is reliable in the real world, which requires benchmark tests under more realistic conditions.
I have done thousands of simulations with very different parameter sets. For the manuscript, I have picked some examples to demonstrate where the iterative separation model shows good results, but also where it should not be applied from a mathematical and numerical point of view. Maybe I should make this more clear in the conclusions. One the one hand, it is a violation of the model criteria, on the other hand, it may be large fractions of rapidly mobilized pre-event water. The latter was also a surprising result for me. Nevertheless, on the basis of Niemis theorem it is then possible to find the exact event water response by applying the corresponding age functions, respectively recalculated tracer concentrations. However, in the field practice, where the event water response, respectively travel time distribution, is normally unknown, this might require the use of Monte Carlo simulations.
> The method is presented as being "based on an iterative balance of catchment input and output mass flows along the time axis" (line 84). This is not the case. As with conventional hydrograph separation, there is no mass balance of inputs and outputs in the sense of equations 1 and 2, but just conservative mixing of the event water and pre-event water.
I think as long as all the criteria (1-4) are fulfilled, the above sentence should hold. For instance, I could reword the sentence to "based on an iterative balance of catchment event and pre-event mass flows along the time axis"?
> It is odd to see all the figures put into an appendix, but on the other hand they are too numerous and repetitive to all go in the main text. Presumably this could be straightened out in an eventual revision.
If some of the figures might appear too redundant, then I could certainly thin out a bit.
> This review has not considered the additional material that has been supplied as author comments, because the HESS review process is – as far as I know – based on the assumption that submitted manuscripts are complete and final, not drafts that are still undergoing revision.
This is fine for me. Thank you for your feedback.
With kind regards,
Simon HoegCitation: https://doi.org/10.5194/hess-2021-213-AC3
-
AC3: 'Reply on RC1', Simon Hoeg, 08 Sep 2021
-
RC2: 'Comment on hess-2021-213', Anonymous Referee #2, 18 Oct 2021
Author validated the iterative extension of standard two-component hydrograph separation method using a single isotopic signature (published in WRR in 2019) through a random rainfall generator and a rainfall-runoff model. I am happy to see that author has addressed a critical question raised during the WRR review why the new modeling approach did not overall improve the model’s uncertainty but instead increased uncertainties for some events. This current study confirmed the speculation at the time via “condition number” that it is the difference between new and old water that caused the increase in uncertainty. In addition, author demonstrated in the current study how the volume weighted separated event water response can serve as an estimator for a time-varying backward travel time distribution, which I consider is a novel contribution and will be of great interests to watershed hydrologists.
One defect in the design, however, is that ET (evaporation + transpiration) is entirely ignored in the rainfall-runoff model (Figure A2, though considered in equations 1 and 2), which is not realistic in the real-world scenario. It is unclear how consideration of ET will affect the modeling outcomes. Not to mention the impact of evaporation on isotopic fractionation (e.g., via bare soil evaporation and canopy interception), the influence of ET on streamflow quantity alone should affect the magnitude and timing of hydrograph and thus in my opinion possibly the isotopic signature of pre-event water, which is taken from streamflow right before an event. This defect will limit the application of this method to concept proofing and prevent it from broad application to real world-problem solving. But I am aware of the fact that adding ET impact in current study would make it way too complicated and extremely hard to follow. I think author should talk about any possible ET effect in a very general way in the current study and let readers be aware of its potential impact and explore whether or not this could be one of future studies.
Additionally, I am curious if both isotopic tracers (2H and 18O) are used, whether or not isotopic fractionation due to evaporation and phase change can be incorporated into the hydrograph separation model. If so, this would significantly boost its application and extend to snowmelt-dominated catchments. I do not think the current model is applicable to streamflow generated primarily from snowmelt.
I am also curious about the potential impact of prolonged droughts on the model results. What if there are two events that occur several months apart?
Though I am not suggesting significant revisions to address the above issues in the current study, any potential impact and future extension should be discussed in the discussion section and the limitation should be cautioned in the conclusion section.
To promote this study and extend the use of the method, I suggest author to add more details to some of the mathematical equations. I have some questions on these equations (see below), which basically require clarification or more information.
P3/L77: The bracket not closed at the end?
P3/Equation 3: The time factor is muted for all quantities in the equations. That is fine, but readers must be reminded that within an event isotopic signature in all components is assumed to be constant. This implicit assumption was discussed later in the discussion section, but I think this assumption should be explicitly stated here and then discussed later in the discussion section.
P5/L122: Isn’t “sequential event water response” a better term than “separated event water response”?
P5/Equation 12: Why does the integral start from negative infinite, not tin (in as subscript)?
Why not dt but dtin instead? Need to explain. I have the same question for equation 15 in page 6.P7/L188: As “condition number” appears for the first time in the text and is not a well-known term in hydrology, it needs either a reference or a further explanation or at least an indication it is being explained below.
P8/Equation 23: It is unclear how the term after the plus sign is derived or defined.
P8/L208: What is gamma here? Explain.
P8/Equation 24: “f” should be better explained.
P11/L313: ET was not considered, but why were precipitation and runoff not equal?
P15/L411-421: Need to mention that the current model setup does not work for snowmelt-dominated system.
P16/L471: It is not constant due to the variable nature of Qt? If so, explicitly say so.
P18/Conclusions: The limitations of current model need to be briefly summarized as well.
Figure A1: This figure gave me hard time at the beginning, as it is depicted by forward concept not backward notation, while the latter is dominated the text. Need more information in the caption to explain this so that readers follow it easily.
Figure A2: Though this model was explained in the text, enough information should be given in the caption to make it stand by itself. At least parameters (alpha and eta) need to be explained in the caption for better readability.
Figure A3: Missing the second y-axis labels.
Figure A4: For one or two curves, why not smoothed near the end? I do not remember if this has been explained in the text.
Figure A5: Extremely hard to distinguish the curves, which occurs in other figures as well.
Figure A9: Need to say what the two arrows are for in the caption.
Figure A15: Where is precipitation? Also, missing mentioning ce (e as subscript) in the caption. The same issue exists for Figure A18.
Citation: https://doi.org/10.5194/hess-2021-213-RC2 -
AC4: 'Reply on RC2', Simon Hoeg, 31 Oct 2021
Dear Referee #2,
I appreciate your very positive and motivating feedback on my manuscript hess-2021-213.
> Author validated the iterative extension of standard two-component hydrograph separation method using a single isotopic signature (published in WRR in 2019) through a random rainfall generator and a rainfall-runoff model. I am happy to see that author has addressed a critical question raised during the WRR review why the new modeling approach did not overall improve the model’s uncertainty but instead increased uncertainties for some events. This current study confirmed the speculation at the time via “condition number” that it is the difference between new and old water that caused the increase in uncertainty. In addition, author demonstrated in the current study how the volume weighted separated event water response can serve as an estimator for a time-varying backward travel time distribution, which I consider is a novel contribution and will be of great interests to watershed hydrologists.
Thank you for summarizing the major outcomes.
> One defect in the design, however, is that ET (evaporation + transpiration) is entirely ignored in the rainfall-runoff model (Figure A2, though considered in equations 1 and 2), which is not realistic in the real-world scenario. It is unclear how consideration of ET will affect the modeling outcomes. Not to mention the impact of evaporation on isotopic fractionation (e.g., via bare soil evaporation and canopy interception), the influence of ET on streamflow quantity alone should affect the magnitude and timing of hydrograph and thus in my opinion possibly the isotopic signature of pre-event water, which is taken from streamflow right before an event. This defect will limit the application of this method to concept proofing and prevent it from broad application to real world-problem solving. But I am aware of the fact that adding ET impact in current study would make it way too complicated and extremely hard to follow. I think author should talk about any possible ET effect in a very general way in the current study and let readers be aware of its potential impact and explore whether or not this could be one of future studies. Additionally, I am curious if both isotopic tracers (2H and 18O) are used, whether or not isotopic fractionation due to evaporation and phase change can be incorporated into the hydrograph separation model. If so, this would significantly boost its application and extend to snowmelt-dominated catchments.
Yes, ET and the impact of evaporation on isotopic fractionation is not explicitly considered in the current iterative separation equations. Instead, the effects of phase change on tracer concentrations would be subject to an Gaussian error analysis. Additionally, in the introduction I wrote that in the context of a chronological sequence of rainfall-runoff events, the event water concentration c_e can be related to the tracer concentration in the precipitation c_J. This relation (e.g. f: c_J -> c_e) is always subject to certain assumptions, which are discussed in almost every hydrograph separation study. Based on function f, I would consider processes such as base soil evaporation and canopy interception when using the current model setup. In this numerical study, I have simply defined that c_J = c_e. However, mid and long term goal would be an extension of the current iterative mixing of the end members Q_e, Q_e-1, ... Q_e-n and Q_p, Q_p-1, ... Q_p-n towards an iterative splitting (see e.g. Kirchner and Allen (2020)) that would also explicitly include evapotranspiration processes.
> I do not think the current model is applicable to streamflow generated primarily from snowmelt.
If the quantity and isotope composition of snowmelt has been sufficiently monitored, then it can be also applied to the current model. In this case the above mentioned relation f could be extended to f(c_J, c_m) -> c_e.
> I am also curious about the potential impact of prolonged droughts on the model results. What if there are two events that occur several months apart?
This is not a problem, as long as the age function remain stable (see e.g. line 155), that as long as piece-wise steady state conditions apply. If the age function has changed, which of course will happen in a prolonged drought, then it should be considered in the calculations. See also my comment at https://doi.org/10.5194/hess-2021-213-AC1
> Though I am not suggesting significant revisions to address the above issues in the current study, any potential impact and future extension should be discussed in the discussion section and the limitation should be cautioned in the conclusion section.
Yes, the discussion and conclusion section might additionally address the above mentioned topics.
> To promote this study and extend the use of the method, I suggest author to add more details to some of the mathematical equations. I have some questions on these equations (see below), which basically require clarification or more information.
P3/L77: The bracket not closed at the end?
Yes, it is not closed, since [t_n, t_(n+1)[ would be the next semantic interval.
> P3/Equation 3: The time factor is muted for all quantities in the equations. That is fine, but readers must be reminded that within an event isotopic signature in all components is assumed to be constant. This implicit assumption was discussed later in the discussion section, but I think this assumption should be explicitly stated here and then discussed later in the discussion section.
It is true that the intra-storm time variability is muted for the end member concentrations c_e, c_e-1, ... c_e-n and c_p, c_p-1, ... c_p-n in most of the simulations. However, on page 4, Criteria 2 and 3 say that the "event/pre-event component maintains a constant isotopic signature in space and time, or any variations can be accounted for". From line 458 on page 16, I discuss for the strong delayed fractions scenario that with an increasing proportion of rapidly mobilized pre-event water, such as Q_(e-1) , the pre-event water concentration c_p (assumed to be constant during the event) may decreasingly represent the bulk pre-event concentration of the considered control volume. Therefore, in https://doi.org/10.5194/hess-2021-213-AC1 and https://doi.org/10.5194/hess-2021-213-AC2 I demonstrate a reverse and time variable adaptation of the tracer concentrations ce (e); ce-1 (e+1), ..., ce-3 (e+3) by piece wise equating Niemi’s left hand side. The mean deviation between the simulated event water response and separated event water response is then in a range of 1e-13% also for a large delayed fractions of event water alpha_max.
> P5/L122: Isn’t “sequential event water response” a better term than “separated event water response”?
Basically yes, but I wanted to emphasize that this variable is derived from a hydrograph separation. Therefore, I would like to keep the term.
> P5/Equation 12: Why does the integral start from negative infinite, not tin (in as subscript)?
Here, I simply adopted the notation of Niemi (1977). I have to say that it is the more exact reference, since Maloszweski and Zuber (1982) applied similar equations, but used the expression "exit age" instead of "injection time". Also Botter et al. (2011) used the notation of Niemi (1977). Later van der Velde (2012) or Harmann (2015) defined backward and forward approaches by using the actual time (t) instead of the injection time (t_in), in the context of an age ranked distribution of the water and associated components being based on cumulative probabilities. To answer your question: The integral itself may not start from t_in, as t_in is the integration variable, but in practice only the values c_j(t_in) * h(t - t_in) > 0 will play a role for the result.
> Why not dt but dtin instead? Need to explain. I have the same question for equation 15 in page 6.
The variable t refers to the actual time, however in the backward perspective we look back to the injection time t_in.
> P7/L188: As “condition number” appears for the first time in the text and is not a well-known term in hydrology, it needs either a reference or a further explanation or at least an indication it is being explained below.
I agree, here I could add the references of Stoer (1994) and Higham (1995). The latter is currently available free of charge from Elsevier.
> P8/Equation 23: It is unclear how the term after the plus sign is derived or defined.
The Landau notation is used to describe the asymptotic behavior, when approaching a finite or infinite limit value. The big O is used to indicate a maximum order of magnitude. Actually, equation 23 is not that often mentioned in the literature, therefore I will replace it by a more popular form that would be: norm(Δx)/norm(x) <= cond(A)/(1-cond(A)*norm(ΔA)/norm(A))*norm(ΔA)/norm(A)
> P8/L208: What is gamma here? Explain.
Gamma is a scalar, a real number. The property cond(gamma*A) = cond(A) says that the size of the condition number for matrix A cannot determined by any scaling factor.
> P8/Equation 24: “f” should be better explained.
See also my response at https://hess.copernicus.org/preprints/hess-2021-213/#AC3: Equation 24 can be found at Deuflhard and Hohmann (2019) and holds for disturbances in matrix A, in which f'(A) is the derivation of matrix A, also known as the Jacobian matrix. With equation 24, I wanted to emphasize that the norm of the first derivative of the matrix A is also limited by its condition. This is interesting in terms of the Gaussian error propagation, in which the entries of f'(A) are applied to the uncertainties of the known variables. This way I wanted to point out analytically that a basic Gaussian error analysis of the system A*x = b would not produce different results than an analysis based on the condition number. Certainly, I could better describe this in the corresponding section. However, in section 4.2, I have written that an "ill-conditioned system will lead to larger gradients in a Jacobian matrix and to potentially higher errors in a Gaussian error propagation".
> P11/L313: ET was not considered, but why were precipitation and runoff not equal?
The simulation is stopped exactly after four years (4*365*24 hours). During the last time step the outflow Q is still 0.08 mm, and the missing 2 mm belong to the storages S_U and S_L.
> P15/L411-421: Need to mention that the current model setup does not work for snowmelt-dominated system.
I could mention that the current model setup has not been tested yet in snowmelt-dominated systems. Provided that the precipitation, snow layer and meltwater are adequately observed, I see no reason why it should not work.
> P16/L471: It is not constant due to the variable nature of Qt? If so, explicitly say so.
The time variability of Qt is not the only reason. In the simulations from Sections 3.1 and 3.2, the effect (intra-storm time-varying concentrations c_e(e), c_e-1(e+1), c_e-2(e+2)) does not occur or only to a small extent.
> P18/Conclusions: The limitations of current model need to be briefly summarized as well.
Yes, I agree.
> Figure A1: This figure gave me hard time at the beginning, as it is depicted by forward concept not backward notation, while the latter is dominated the text. Need more information in the caption to explain this so that readers follow it easily.
I agree, the presentation is not that accurate. I should emphasize the backward perspective. It is not necessary to focus so much on the precipitation J.
> Figure A2: Though this model was explained in the text, enough information should be given in the caption to make it stand by itself. At least parameters (alpha and eta) need to be explained in the caption for better readability.
Yes, I agree.
> Figure A3: Missing the second y-axis labels.
There is only y-axis in this figure, but I will improve or remove the tickmarks on the right hand side.
> Figure A4: For one or two curves, why not smoothed near the end? I do not remember if this has been explained in the text.
Near the end the condition numbers are very high (up to 2x10^6). In line 289, I have mentioned that "there are increasing (numerical) instabilities in the last events".
> Figure A5: Extremely hard to distinguish the curves, which occurs in other figures as well.
Yes, sometimes not that easy to distinguish, you have to zoom in. I will try with wider lines.
> Figure A9: Need to say what the two arrows are for in the caption.
Yes, I agree.
> Figure A15: Where is precipitation? Also, missing mentioning ce (e as subscript) in the caption. The same issue exists for Figure A18.
I agree, and will improve the figures A15 and A18.Thank you for your help and feedback.
With kind regards,
Simon HoegAdditional References:
Kirchner, J. W. and Allen, S. T.: Seasonal partitioning of precipitation between streamflow and evapotranspiration, inferred from end-member splitting analysis, Hydrol. Earth Syst. Sci., 24, 17–39, https://doi.org/10.5194/hess-24-17-2020, 2020van der Velde, Y., Torfs, P. J. J. F., van der Zee, S. E. A. T. M., and Uijlenhoet, R. (2012), Quantifying catchment-scale mixing and its effect on time-varying travel time distributions, Water Resour. Res., 48, W06536, doi:10.1029/2011WR011310.
Citation: https://doi.org/10.5194/hess-2021-213-AC4
-
AC4: 'Reply on RC2', Simon Hoeg, 31 Oct 2021
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 808 | 355 | 41 | 1,204 | 24 | 25 |
- HTML: 808
- PDF: 355
- XML: 41
- Total: 1,204
- BibTeX: 24
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1