the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Catchment water storage dynamics and its role in modulating streamflow generation in spectral perspective: a case study in the headwater of Baiyang Lake, China
Zhuping Sheng
Yanmin Yang
Shumin Han
Qingzhou Zhang
Huilong Li
Yonghui Yang
Abstract. Although it is important in hydrological cycles, catchment water storage dynamics is still not fully understood because it is affected by multiple drivers simultaneously and is difficult to be estimated using field hydrometric observations and hydrological models. Taking the headwater of Baiyang Lake, China as an example, this study employed a spectral approach to illustrate how catchment water storage was influenced by rainfall and vegetation, and how water storage modulated streamflow for the period of 1982–2015. The competence of the spectral approach in characterizing causality was verified and a more holistic understanding of hydrological cycles was gained. Results showed that under different climatic phases (wet/dry), catchment water storage dynamics were controlled by different factors and dominant streamflow generation mechanisms were not invariant. In the wet phase, catchment water storage dynamics was determined by rainfall. And groundwater flow was the most important part of streamflow, followed by subsurface flow and surface flow. Nevertheless, in the dry phase, catchment water storage dynamics was modulated by evapotranspiration. And the surface flow was the most important part of streamflow, followed by subsurface flow and groundwater flow. The land use change induced by human activities could alter the streamflow sensitivity to rainfall, but could not cause fundamental changes to hydrological cycles. We concluded that the spectral approach can be an effective supplement to the experimental methods and their integration can provide systematic insights into hydrological cycles in the study area and other watershed systems.
- Preprint
(4844 KB) - Metadata XML
- BibTeX
- EndNote
Xinyao Zhou et al.
Status: closed
-
RC1: 'Comment on hess-2022-357', Anonymous Referee #1, 31 Dec 2022
Taking several catchments from North China, this paper used a spectral analysis approach, trying to illustrate how catchment water storage was influenced by rainfall and vegetation, and how water storage modulated streamflow for the period of 1982 - 2015. The topic is important and interesting. However, there are several major concerns to be further clarified before it can be considered for publication. The comments and suggestions are listed as following, hopefully they will be helpful to improve the manuscript.
Major comments:
- The focus and novelty of this paper is not clear. Are you trying to show that the spectral analysis is an effective method in demonstrating water storage dynamics? Or some new findings about catchment storage dynamics and runoff generation mechanisms?
- The manuscript is overall not well organized, and a clearer logic is needed. For example, the land use change or vegetation succession within four different catchments should be in the part of “Methods, Study Area, and Data”. Meanwhile, before further discussion, it is important to show what components are included in the catchment water storage in the study area, such as lakes, groundwater, soil water and weathered bedrock water. What is the most important one among them for catchment water storage dynamics? And why it is reasonable to attribute the water storage dynamics differences to vegetation root water uptake patterns between wet and dry phases?
- Data quality determines the reliability of final results. I don’t think it is appropriate to use precipitation data from only one station to represent the average level for the whole catchment, especially with large altitude gradient as shown in Figure 1.
- The language needs to be improved. For example, in line 24-25, what does “fundamental changes to hydrological cycles” mean?
Specific Comments
Line 103-105: It is not clear how spectral analysis method can be used to detect causality. Please explain more.
In section 2.3, it should be more clearly demonstrated how you processed different spatiotemporal resolution data. For example, why data from 1960-2015 were used for rainfall periodicity detection while others were based on data from 1982-2015? Does it have impact on final results? Similarly, different spatial-resolution ET products were used among the time series, the impact should be evaluated to make the results more convincing.
Line 196: How could ET lags behind water storage by 4 years? Can you interpretate this result with some physical phenomenon or evidence? This is also the base of using such spectral analysis methods to illustrate hydrological processes.
Line 212: Why there is a trough in April in the wet phase?
Line 236-237: why it is not “water storage determines ET”?
Line 242: Do you mean a higher contribution of subsurface flow to stream flow in the wet phase when compared with that in a dry phase?
Line 255-256 References are needed.
Line 267: How will the catchment characteristics impact storage dynamics and streamflow responses among 4 catchments?
Line 318: Is there any irrigation or groundwater/stream water pumping activity in the study area, which may also influence water storage dynamics.
Line 330: see major comment (2).
Line 322-327: the impact of land use change can be long-term lasting, so I doubt this argument.
Line 338: What do you mean “these plants decide that rainfall will increase…”?
Section 6: Conclusions need to be re-organized. One sentence for one paragraph is not a good way at least in my opinion.
Citation: https://doi.org/10.5194/hess-2022-357-RC1 -
AC1: 'Reply on RC1', Xinyao Zhou, 18 Apr 2023
We highly appreciate the two anonymous reviewers for their invaluable comments, which enlightened us on lots of points. In this letter, we will briefly summarize how their insightful suggestions help us improve our manuscript, and then answer specific questions.
Firstly, in response to the concerns about scarce rainfall data raised by both reviewers, we have included satellite products such as CHIRPS to cross-validate rainfall data in the revision. Furthermore, we have used GRACE data to validate catchment water storage result as suggested.
Secondly, we have reorganized the context of the revised manuscript to make it more logical, for example, introducing vegetation succession in Study Area section instead of Results section.
Thirdly, we have highlighted novelty as suggested by reviewer 1, by demonstrating the effectiveness of systematic approaches when they are applied in hydrology due to the commonality between the two disciplines. And a full discussion is included about the new understanding of hydrologic cycle inspired by the multidisciplinary integration.
Fourthly, in response to the issues related to methods raised by reviewer 2, we have emphasized the importance of system approach and added causal discovery method to infer system structure.
We realized that "spectral perspective" may not be enough to tell what we are trying to unearth and that "systematic perspective" would provide better insights into hydrological processes. From the systematic perspective, everything can be considered as a system which typically contains two major parts, a structure composed of multiple interlocking elements, and signals passing through the structure. Hydrological process is no exception. Thus we consider the hydrological process as a system with four-elements structure (rainfall, evapotranspiration, catchment water storage and streamflow) and signals carried by water. We recognized the importance of the system structure because it determines how signals propagate and how the system acts. Therefore, in the revision we have added system structure analysis, which uses causal discovery methods to infer system structure. With such enhancement we can show how system structure evolves over time and affects water transfer within system or among subsystems.
The interlocking elements, flows, and feedback loops make a system more than the sum of its parts and may exhibit adaptive, dynamic, goal-seeking, self-preserving, and evolutionary behavior with covarying structure and signals. This can allow us jumping out the pattern of traditional rigid hydrologic framework and understanding the growth, stagnation, decline, oscillation, randomness, or evolution of hydrological system at long-term behavior-level rather than short-term event-level.
Finally, we have asked native English hydrologists to proofread the revised manuscript to improve the manuscript.
More answers to specific comments are listed below.
Responses to reviewer 1
Taking several catchments from North China, this paper used a spectral analysis approach, trying to illustrate how catchment water storage was influenced by rainfall and vegetation, and how water storage modulated streamflow for the period of 1982 - 2015. The topic is important and interesting. However, there are several major concerns to be further clarified before it can be considered for publication. The comments and suggestions are listed as following, hopefully they will be helpful to improve the manuscript.
Major comments:
(1)The focus and novelty of this paper is not clear. Are you trying to show that the spectral analysis is an effective method in demonstrating water storage dynamics? Or some new findings about catchment storage dynamics and runoff generation mechanisms?
Response: Thank you for the comment. As mentioned above, the revised manuscript have focused on the effectiveness of systematic approaches when they are applied in hydrology. And the new understanding of hydrological processes brought by the integration of multidisciplinary methods have been fully discussed.
(2)The manuscript is overall not well organized, and a clearer logic is needed. For example, the land use change or vegetation succession within four different catchments should be in the part of “Methods, Study Area, and Data”. Meanwhile, before further discussion, it is important to show what components are included in the catchment water storage in the study area, such as lakes, groundwater, soil water and weathered bedrock water. What is the most important one among them for catchment water storage dynamics? And why it is reasonable to attribute the water storage dynamics differences to vegetation root water uptake patterns between wet and dry phases?
Response: Thank you for the comment. According to the comments, we have reorganized the context in the revised manuscript, including more related detailed information such as vegetation and geology in the part of "Methods, Study Area, and Data".
(3)Data quality determines the reliability of final results. I don’t think it is appropriate to use precipitation data from only one station to represent the average level for the whole catchment, especially with large altitude gradient as shown in Figure 1.
Response: Thank you for the comment. We have added other datasets such as CHIRPS to verify the rainfall data.
(4)The language needs to be improved. For example, in line 24-25, what does “fundamental changes to hydrological cycles” mean?
Response: Thank you for the comment. We have addressed this in the revision.
Specific Comments
(5)Line 103-105: It is not clear how spectral analysis method can be used to detect causality. Please explain more.
Response: Thank you for the comment. We have used causal discovery methods such as Granger Causality Test to detect the causality.
(6)In section 2.3, it should be more clearly demonstrated how you processed different spatiotemporal resolution data. For example, why data from 1960-2015 were used for rainfall periodicity detection while others were based on data from 1982-2015? Does it have impact on final results? Similarly, different spatial-resolution ET products were used among the time series, the impact should be evaluated to make the results more convincing.
Response: Thank you for the comment. We have made revision to keep the data span consistent within the period of 1982-2015 and spatial resolution of 1 km.
(7)Line 196: How could ET lags behind water storage by 4 years? Can you interpretate this result with some physical phenomenon or evidence? This is also the base of using such spectral analysis methods to illustrate hydrological processes.
Response: Thank you for the comment. We have used causal discovery methods to interpret the result.
(8)Line 212: Why there is a trough in April in the wet phase?
Response: Thank you for the comment. We have added references to explain the trough in April in wet phase.
(9)Line 236-237: why it is not “water storage determines ET”?
Response: Thank you for the comment. We have used causal discovery methods to determine the causal direction between water storage and ET.
(10)Line 242: Do you mean a higher contribution of subsurface flow to stream flow in the wet phase when compared with that in a dry phase?
Response: Thank you for the comment. The delays show the existence of subsurface flow. We have made revision to make it clear.
(11)Line 255-256 References are needed.
Response: Thank you for the comment. References were added.
(12)Line 267: How will the catchment characteristics impact storage dynamics and streamflow responses among 4 catchments?
Response: Thank you for the comment. We have checked other references for the characteristics of 4 catchments and added them in the revision.
(13)Line 318: Is there any irrigation or groundwater/stream water pumping activity in the study area, which may also influence water storage dynamics.
Response: Thank you for the comment. We have checked other references for these human activities and added them in the revision.
Line 330: see major comment (2).
(14)Line 322-327: the impact of land use change can be long-term lasting, so I doubt this argument.
Response: Thank you for the comment. We have considered the long-term effects in the revision.
(15)Line 338: What do you mean “these plants decide that rainfall will increase…”?
Response: Thank you for the comment. The sentence refers to the reinforcing feedback of vegetation. We have revised it accordingly.
(16)Section 6: Conclusions need to be re-organized. One sentence for one paragraph is not a good way at least in my opinion.
Response: Thank you for the comment. We have revised the conclusion section.
Citation: https://doi.org/10.5194/hess-2022-357-AC1
-
RC2: 'Comment on hess-2022-357', Anonymous Referee #2, 28 Mar 2023
The manuscript “Catchment water storage dynamics and its role in modulating streamflow generation in spectral perspective: a case study in the headwater of Baiyang Lake, China” aims to quantify and describe how catchment water storage changes over time and to identify the relative importance of concurrent changes in system drivers.
The overall objective of the study is highly relevant and the authors also analyse the research question from a conceptually/methodologically very interesting perspective. While the majority of studies that have previously addressed the topic analysed absolute changes in the magnitudes of system drivers, such as precipitation or atmospheric/canopy water demand, the authors place a much stronger focus on the role of changes in timing and the resulting changes in phase shifts between the various variables.
In spite of these interesting aspects, the analysis unfortunately does not live up to its potential and remains quite superficial with quite some assumptions that remain speculative or poorly substantiated by data. In the end, this leaves the reader a bit disappointed as a much more thorough would have been possible, including a more detailed description of the experiment and more in-depth analysis of the results which could have generated much more interesting and robust insights.
As such, I do have several major concerns that need to be addressed in detail to develop the analysis to the point:
(1) The current experiment is based on precipitation data from 5 precipitation gauges. In contrast, the study region has an area of ~ 10.000km2. While I do understand the challenge of scarce data, it remains opaque to me, why the authors did not choose to complement their in-situ observations with globally available fused precipitation data products, such as CHIRPS (1981-present) or comparable? This would at least allow them to demonstrate that their point-scale data are at least sort of representative for their study basins. Similarly, although only available at larger spatial scales, the authors did not even mention nor use the GRACE product, providing global estimates of total water storage variations, as complementary data source to at least plausibility check their results.
(2) Little effort is made to consider and report uncertainties throughout the analysis. This should, in the year 2023, be a standard in scientific publishing. For example, in Figure 5, merely one line is shown for each variable. The reader is left to assume that these lines represent the (multi-annual?) average (mean? median?) regime curves. Why are the distributions (or ranges) around these average not shown? This would be so much more helpful for the reader to assess if the reported changes between wet and dry regimes are indeed significant. Similarly, no quantification of the significance of the wet/dry differences is provided. What are the p-values for the monthly differences between wet and dry?
(3) The authors use wavelets to analyse temporal shifts in the power spectral properties of the various variables. While they are not the first ones to use wavelet analysis in hydrology, it nevertheless remains a rather under-exploited tool. As such, many readers will not be familiar with the idea behind it and how to interpret its results. Thus a much more detailed description of the method and in particular of how the results have to be interpreted is necessary. Otherwise most readers will not be able to assess the results as well as the reasoning behind the interpretation.
(4) The authors invested quite some effort in discussing and interpreting their results in a very intriguing storyline. However, many assumptions are made in this interpretation, which leaves much of it very speculative. This is exacerbated by not substantiating the arguments made with specific numbers that emerge from the underlying analysis. In particular, the role of land cover changes remains addressed in only a surprisingly simplistic and underdeveloped discussion that does neither refer to existing literature in an insightful way nor offer any specific numbers.
(5) The structure of the manuscript needs a major overhaul as it is difficult to find a red line and difficult to follow the flow of the argument. For example, why are the results of the individual catchments shown only after the analysis of one of them? Why is change of land cover only discussed in section 4.2, while it is first mentioned in section 3.4? This causes confusion for the reader.
(5) The level of English does not yet meet the standard required for scientific publications and requires a detailed round of proofreading by a native English speaker with hydrology background.
(6) Figures could be developed with much more care and could offer quite some more detail.
Specific points:
(7) L.18: in hydrology close to nothing can actually be “verified”. Given the uncertainties in our data, the best we can hope for is that any results are more or less consistent with the available data.
(8) L.21: a clear definition of “subsurface flow” is needed, as of course, groundwater is also a subsurface flow (and not an above-surface flow…).
(9) L.57-62: this is not correct. There are several studies that have previously shown the opposite (e.g. Hulsman et al., 2021).
(10) L.74: the authors could at least have mentioned the GRACE satellite product that provides estimates of total water storage variations (e.g. Landerer and Swenson, 2012). Ideally it could also be used as complementary data source to at least plausibility check the results.
(11) L.129: what does that mean? “The unit of streamflow data was 108m3”??? This is not being very kind to the reader.
(12) L.130: Five stations is not a lot and I am not sure that it is sufficient to plausibly describe precipitation across this large area. Why was this data set not complemented by gridded data? See Comment (1) above
(13) L.136: Seriously? Can the authors please try to use units that are standard in scientific hydrology, such as mm d-1?!
(14) L.147: what is meant by “old water” and “young water”? I do not believe the authors have made any attempt to actually date the water here in the sense of hydrograph separation (e.g. Klaus and McDonnell, 2013) or transit time distributions (e.g. Benettin et al., 2022). The terms “old water” and “young water” have very specific, narrowly defined meaning in hydrology. Please do not use them in a different context.
(15) L.148: sure, but which data product was this exactly?
(16) L.150: “outperformance among” - what is this supposed to mean?
(17) L.150: “the units […] were in 108m3 […]” ????
(18) L.155: how was “representative” objectively determined?
(19) L.158: the trends are not at all distinguishable in Figure 2a
(20) L.158: “water storage […] fluctuates around zero […]”. Does this make sense?? How can there be a negative water storage?? Please be a bit more careful with how to describe things: it is not the water storage that fluctuates around zero but the water storage anomaly.
(21) L.162: what is a “high wavelet power spectrum”??? Is this good or bad? What do the coefficients mean? Are the equivalent to coefficients of a Fourier transform which are directly related to the spectral power of specific frequencies? How can their fluctuations be interpreted? This is a relevant part of the analysis and needs to be described in much more detail to allow the reader to follow and fully appreciate the interpretation.
(22) L.169: on basis of which criteria were the wet and dry phases defined?
(23) L.177: I do not think that a 30-yr periodicity can be meaningfully and robustly inferred from a data record of merely 33 years.
(24) L.198: what is meant by “hysteresis effect” here?
(25) L.208: Ok, all coefficients are negative. So what? what does this mean?
(26) L.211, Figure 5: are these average values? Which average? Mean? Median? Long-term? Why are only these lines shown and not the ranges, envelopes or distributions? How significant are the differences between wet and dry?
(27) L.214: water storage cannot be negative…see Comment (20) above.
(28) L.227-228: How was this done?
(29) L.237-238: which vegetation cover? How can the reader know that if it is only shown in detail much later in the manuscript??
(30) L.244-245: How would the authors know if this is indeed “surface flow”?? Although occurring locally, this is a very uncommon phenomenon at the catchment scale, as the authors also point out later in the manuscript. Could just as well be any other fast responding flow component in the system.
(31) L.300, Figure 10: Panel (a) is surprisingly uninformative. Why not find a way to show the differences between the decades in these maps? Panel (b) has incorrect units on the y-axis – this should probably read as “Fraction”.
(32) L.306-307: “dramatic decrease in annual ET in ~2004-2005”??? I cannot see that anywhere. If anything, the major decrease happens somewhere between 1998 and 2002, but even that is difficult to see in Figure 2a.
(33) L.300ff, Sections 4.2-4.3: nice storyline but mostly based on assumptions and speculation and not sufficiently substantiated by data.
(34) L.344-345: what is “surface stream flow”??
(35) L.345ff: how can results of such plot-scale studies be reasonably assumed to give meaningful insights at the river basin scale??
References:
Benettin, P., Rodriguez, N. B., Sprenger, M., Kim, M., Klaus, J., Harman, C. J., ... & McDonnell, J. J. (2022). Transit time estimation in catchments: Recent developments and future directions. Water Resources Research, 58(11), e2022WR033096.
Hulsman, P., Hrachowitz, M., & Savenije, H. H. (2021). Improving the representation of long‐term storage variations with conceptual hydrological models in data‐scarce regions. Water Resources Research, 57(4), e2020WR028837.
Klaus, J., & McDonnell, J. J. (2013). Hydrograph separation using stable isotopes: Review and evaluation. Journal of hydrology, 505, 47-64.
Landerer, F. W., & Swenson, S. C. (2012). Accuracy of scaled GRACE terrestrial water storage estimates. Water resources research, 48(4).
Citation: https://doi.org/10.5194/hess-2022-357-RC2 -
AC2: 'Reply on RC2', Xinyao Zhou, 18 Apr 2023
We highly appreciate the two anonymous reviewers for their invaluable comments, which enlightened us on lots of points. In this letter, we will briefly summarize how their insightful suggestions help us improve our manuscript, and then answer specific questions.
Firstly, in response to the concerns about scarce rainfall data raised by both reviewers, we have included satellite products such as CHIRPS to cross-validate rainfall data in the revision. Furthermore, we have used GRACE data to validate catchment water storage result as suggested.
Secondly, we have reorganized the context of the revised manuscript to make it more logical, for example, introducing vegetation succession in Study Area section instead of Results section.
Thirdly, we have highlighted novelty as suggested by reviewer 1, by demonstrating the effectiveness of systematic approaches when they are applied in hydrology due to the commonality between the two disciplines. And a full discussion is included about the new understanding of hydrologic cycle inspired by the multidisciplinary integration.
Fourthly, in response to the issues related to methods raised by reviewer 2, we have emphasized the importance of system approach and added causal discovery method to infer system structure.
We realized that "spectral perspective" may not be enough to tell what we are trying to unearth and that "systematic perspective" would provide better insights into hydrological processes. From the systematic perspective, everything can be considered as a system which typically contains two major parts, a structure composed of multiple interlocking elements, and signals passing through the structure. Hydrological process is no exception. Thus we consider the hydrological process as a system with four-elements structure (rainfall, evapotranspiration, catchment water storage and streamflow) and signals carried by water. We recognized the importance of the system structure because it determines how signals propagate and how the system acts. Therefore, in the revision we have added system structure analysis, which uses causal discovery methods to infer system structure. With such enhancement we can show how system structure evolves over time and affects water transfer within system or among subsystems.
The interlocking elements, flows, and feedback loops make a system more than the sum of its parts and may exhibit adaptive, dynamic, goal-seeking, self-preserving, and evolutionary behavior with covarying structure and signals. This can allow us jumping out the pattern of traditional rigid hydrologic framework and understanding the growth, stagnation, decline, oscillation, randomness, or evolution of hydrological system at long-term behavior-level rather than short-term event-level.
Finally, we have asked native English hydrologists to proofread the revised manuscript to improve the manuscript.
More answers to specific comments are listed below.
Responses to reviewer 2
The manuscript “Catchment water storage dynamics and its role in modulating streamflow generation in spectral perspective: a case study in the headwater of Baiyang Lake, China” aims to quantify and describe how catchment water storage changes over time and to identify the relative importance of concurrent changes in system drivers.
The overall objective of the study is highly relevant and the authors also analyse the research question from a conceptually/methodologically very interesting perspective. While the majority of studies that have previously addressed the topic analysed absolute changes in the magnitudes of system drivers, such as precipitation or atmospheric/canopy water demand, the authors place a much stronger focus on the role of changes in timing and the resulting changes in phase shifts between the various variables.
In spite of these interesting aspects, the analysis unfortunately does not live up to its potential and remains quite superficial with quite some assumptions that remain speculative or poorly substantiated by data. In the end, this leaves the reader a bit disappointed as a much more thorough would have been possible, including a more detailed description of the experiment and more in-depth analysis of the results which could have generated much more interesting and robust insights.
As such, I do have several major concerns that need to be addressed in detail to develop the analysis to the point:
(1) The current experiment is based on precipitation data from 5 precipitation gauges. In contrast, the study region has an area of ~ 10.000km2. While I do understand the challenge of scarce data, it remains opaque to me, why the authors did not choose to complement their in-situ observations with globally available fused precipitation data products, such as CHIRPS (1981-present) or comparable? This would at least allow them to demonstrate that their point-scale data are at least sort of representative for their study basins. Similarly, although only available at larger spatial scales, the authors did not even mention nor use the GRACE product, providing global estimates of total water storage variations, as complementary data source to at least plausibility check their results.
Response: Thank you for the comment. We have added other rainfall products to check the reliability of point-scale rainfall and used GRACE to verify the catchment water storage changes.
(2) Little effort is made to consider and report uncertainties throughout the analysis. This should, in the year 2023, be a standard in scientific publishing. For example, in Figure 5, merely one line is shown for each variable. The reader is left to assume that these lines represent the (multi-annual?) average (mean? median?) regime curves. Why are the distributions (or ranges) around these average not shown? This would be so much more helpful for the reader to assess if the reported changes between wet and dry regimes are indeed significant. Similarly, no quantification of the significance of the wet/dry differences is provided. What are the p-values for the monthly differences between wet and dry?
Response: Thank you for the comment. As suggested, we have discussed the uncertainties and highlighted the significance of the wet/dry differences in the revision.
(3) The authors use wavelets to analyse temporal shifts in the power spectral properties of the various variables. While they are not the first ones to use wavelet analysis in hydrology, it nevertheless remains a rather under-exploited tool. As such, many readers will not be familiar with the idea behind it and how to interpret its results. Thus a much more detailed description of the method and in particular of how the results have to be interpreted is necessary. Otherwise most readers will not be able to assess the results as well as the reasoning behind the interpretation.
Response: Thank you for the comment. As mentioned above, we have adopted systematic perspective to integrate the signal analysis and system structure. On this basis, more introduction of signal analysis were added.
(4) The authors invested quite some effort in discussing and interpreting their results in a very intriguing storyline. However, many assumptions are made in this interpretation, which leaves much of it very speculative. This is exacerbated by not substantiating the arguments made with specific numbers that emerge from the underlying analysis. In particular, the role of land cover changes remains addressed in only a surprisingly simplistic and underdeveloped discussion that does neither refer to existing literature in an insightful way nor offer any specific numbers.
Response: Thank you for the comment. On the basis of new results from causal discovery methods, we have included additional discussion of the results in the revision and more references were added.
(5) The structure of the manuscript needs a major overhaul as it is difficult to find a red line and difficult to follow the flow of the argument. For example, why are the results of the individual catchments shown only after the analysis of one of them? Why is change of land cover only discussed in section 4.2, while it is first mentioned in section 3.4? This causes confusion for the reader.
Response: Thank you for the comment. We have reorganized the context in our revised manuscript.
(5) The level of English does not yet meet the standard required for scientific publications and requires a detailed round of proofreading by a native English speaker with hydrology background.
Response: Thank you for the comment. We have a hydrologist with native English to proofread the revised manuscript.
(6) Figures could be developed with much more care and could offer quite some more detail.
Response: Thank you for the comment. We have revised the figures to provide more information.
Specific points:
(7) L.18: in hydrology close to nothing can actually be “verified”. Given the uncertainties in our data, the best we can hope for is that any results are more or less consistent with the available data.
Response: Thank you for the comment. We have revised the sentence accordingly.
(8) L.21: a clear definition of “subsurface flow” is needed, as of course, groundwater is also a subsurface flow (and not an above-surface flow…).
Response: Thank you for the comment. We have added a definition for the "subsurface flow".
(9) L.57-62: this is not correct. There are several studies that have previously shown the opposite (e.g. Hulsman et al., 2021).
Response: Thank you for the comment. We have included these references to show the debates.
(10) L.74: the authors could at least have mentioned the GRACE satellite product that provides estimates of total water storage variations (e.g. Landerer and Swenson, 2012). Ideally it could also be used as complementary data source to at least plausibility check the results.
Response: Thank you for the comment. We have inclued GRACE product to verify the catchment water storage.
(11) L.129: what does that mean? “The unit of streamflow data was 108m3”??? This is not being very kind to the reader.
Response: Thank you for the comment. We have revised it accordingly as well as the others similarly.
(12) L.130: Five stations is not a lot and I am not sure that it is sufficient to plausibly describe precipitation across this large area. Why was this data set not complemented by gridded data? See Comment (1) above
Response: Thank you for the comment. We have added other gridded datasets to complement the rainfall data.
(13) L.136: Seriously? Can the authors please try to use units that are standard in scientific hydrology, such as mm d-1?!
Response: Thank you for the comment. We have changed to the standard unit of mm/year.
(14) L.147: what is meant by “old water” and “young water”? I do not believe the authors have made any attempt to actually date the water here in the sense of hydrograph separation (e.g. Klaus and McDonnell, 2013) or transit time distributions (e.g. Benettin et al., 2022). The terms “old water” and “young water” have very specific, narrowly defined meaning in hydrology. Please do not use them in a different context.
Response: Thank you for the comment. We have revised it accordingly.
(15) L.148: sure, but which data product was this exactly?
Response: Thank you for the comment. More explanations about the dataset were provided.
(16) L.150: “outperformance among” - what is this supposed to mean?
Response: Thank you for the comment. We have revised it with better expression.
(17) L.150: “the units […] were in 108m3 […]” ????
Response: Thank you for the comment. We have changed it to the standard unit of mm/year.
(18) L.155: how was “representative” objectively determined?
Response: Thank you for the comment. We have reorganized the manuscript to show all four catchments here.
(19) L.158: the trends are not at all distinguishable in Figure 2a
Response: Thank you for the comment. We have revised the figure to make it clear.
(20) L.158: “water storage […] fluctuates around zero […]”. Does this make sense?? How can there be a negative water storage?? Please be a bit more careful with how to describe things: it is not the water storage that fluctuates around zero but the water storage anomaly.
Response: Thank you for the comment. We have rewiritten the sentence.
(21) L.162: what is a “high wavelet power spectrum”??? Is this good or bad? What do the coefficients mean? Are the equivalent to coefficients of a Fourier transform which are directly related to the spectral power of specific frequencies? How can their fluctuations be interpreted? This is a relevant part of the analysis and needs to be described in much more detail to allow the reader to follow and fully appreciate the interpretation.
Response: Thank you for the comment. We have provided more details about the wavelet results.
(22) L.169: on basis of which criteria were the wet and dry phases defined?
Response: Thank you for the comment. The wet and dry phases were separated according to the change of rainfall and the separation was verified by other studies.
(23) L.177: I do not think that a 30-yr periodicity can be meaningfully and robustly inferred from a data record of merely 33 years.
Response: Thank you for the comment. We have focused on the difference between wet and dry phases and weakened the periodicity in the later revision.
(24) L.198: what is meant by “hysteresis effect” here?
Response: Thank you for the comment. Additional explanations were added in the revision.
(25) L.208: Ok, all coefficients are negative. So what? what does this mean?
Response: Thank you for the comment. More explanation have been added here.
(26) L.211, Figure 5: are these average values? Which average? Mean? Median? Long-term? Why are only these lines shown and not the ranges, envelopes or distributions? How significant are the differences between wet and dry?
Response: Thank you for the comment. Uncertainty analysis was added here and figure 5 was revised.
(27) L.214: water storage cannot be negative…see Comment (20) above.
Response: Thank you for the comment. We have made revisions accordingly.
(28) L.227-228: How was this done?
Response: Thank you for the comment. Causal discovery method was used in the revision.
(29) L.237-238: which vegetation cover? How can the reader know that if it is only shown in detail much later in the manuscript??
Response: Thank you for the comment. The structure of paper was adjusted in the revision.
(30) L.244-245: How would the authors know if this is indeed “surface flow”?? Although occurring locally, this is a very uncommon phenomenon at the catchment scale, as the authors also point out later in the manuscript. Could just as well be any other fast responding flow component in the system.
Response: Thank you for the comment. We have separated streamflow into rapid-response component and slow-response component in the revision.
(31) L.300, Figure 10: Panel (a) is surprisingly uninformative. Why not find a way to show the differences between the decades in these maps? Panel (b) has incorrect units on the y-axis – this should probably read as “Fraction”.
Response: Thank you for the comment. We have changed the figure according to the suggestions.
(32) L.306-307: “dramatic decrease in annual ET in ~2004-2005”??? I cannot see that anywhere. If anything, the major decrease happens somewhere between 1998 and 2002, but even that is difficult to see in Figure 2a.
Response: Thank you for the comment. We have changed the figure to include more information.
(33) L.300ff, Sections 4.2-4.3: nice storyline but mostly based on assumptions and speculation and not sufficiently substantiated by data.
Response: Thank you for the comment. More discussions were included and more references were added.
(34) L.344-345: what is “surface stream flow”??
Response: Thank you for the comment. It refers to rapid-response component of streamflow here. It has been revised here.
(35) L.345ff: how can results of such plot-scale studies be reasonably assumed to give meaningful insights at the river basin scale??
Response: Thank you for the comment. We have included some catchment-scale references in discussion section.
Citation: https://doi.org/10.5194/hess-2022-357-AC2
-
AC2: 'Reply on RC2', Xinyao Zhou, 18 Apr 2023
Status: closed
-
RC1: 'Comment on hess-2022-357', Anonymous Referee #1, 31 Dec 2022
Taking several catchments from North China, this paper used a spectral analysis approach, trying to illustrate how catchment water storage was influenced by rainfall and vegetation, and how water storage modulated streamflow for the period of 1982 - 2015. The topic is important and interesting. However, there are several major concerns to be further clarified before it can be considered for publication. The comments and suggestions are listed as following, hopefully they will be helpful to improve the manuscript.
Major comments:
- The focus and novelty of this paper is not clear. Are you trying to show that the spectral analysis is an effective method in demonstrating water storage dynamics? Or some new findings about catchment storage dynamics and runoff generation mechanisms?
- The manuscript is overall not well organized, and a clearer logic is needed. For example, the land use change or vegetation succession within four different catchments should be in the part of “Methods, Study Area, and Data”. Meanwhile, before further discussion, it is important to show what components are included in the catchment water storage in the study area, such as lakes, groundwater, soil water and weathered bedrock water. What is the most important one among them for catchment water storage dynamics? And why it is reasonable to attribute the water storage dynamics differences to vegetation root water uptake patterns between wet and dry phases?
- Data quality determines the reliability of final results. I don’t think it is appropriate to use precipitation data from only one station to represent the average level for the whole catchment, especially with large altitude gradient as shown in Figure 1.
- The language needs to be improved. For example, in line 24-25, what does “fundamental changes to hydrological cycles” mean?
Specific Comments
Line 103-105: It is not clear how spectral analysis method can be used to detect causality. Please explain more.
In section 2.3, it should be more clearly demonstrated how you processed different spatiotemporal resolution data. For example, why data from 1960-2015 were used for rainfall periodicity detection while others were based on data from 1982-2015? Does it have impact on final results? Similarly, different spatial-resolution ET products were used among the time series, the impact should be evaluated to make the results more convincing.
Line 196: How could ET lags behind water storage by 4 years? Can you interpretate this result with some physical phenomenon or evidence? This is also the base of using such spectral analysis methods to illustrate hydrological processes.
Line 212: Why there is a trough in April in the wet phase?
Line 236-237: why it is not “water storage determines ET”?
Line 242: Do you mean a higher contribution of subsurface flow to stream flow in the wet phase when compared with that in a dry phase?
Line 255-256 References are needed.
Line 267: How will the catchment characteristics impact storage dynamics and streamflow responses among 4 catchments?
Line 318: Is there any irrigation or groundwater/stream water pumping activity in the study area, which may also influence water storage dynamics.
Line 330: see major comment (2).
Line 322-327: the impact of land use change can be long-term lasting, so I doubt this argument.
Line 338: What do you mean “these plants decide that rainfall will increase…”?
Section 6: Conclusions need to be re-organized. One sentence for one paragraph is not a good way at least in my opinion.
Citation: https://doi.org/10.5194/hess-2022-357-RC1 -
AC1: 'Reply on RC1', Xinyao Zhou, 18 Apr 2023
We highly appreciate the two anonymous reviewers for their invaluable comments, which enlightened us on lots of points. In this letter, we will briefly summarize how their insightful suggestions help us improve our manuscript, and then answer specific questions.
Firstly, in response to the concerns about scarce rainfall data raised by both reviewers, we have included satellite products such as CHIRPS to cross-validate rainfall data in the revision. Furthermore, we have used GRACE data to validate catchment water storage result as suggested.
Secondly, we have reorganized the context of the revised manuscript to make it more logical, for example, introducing vegetation succession in Study Area section instead of Results section.
Thirdly, we have highlighted novelty as suggested by reviewer 1, by demonstrating the effectiveness of systematic approaches when they are applied in hydrology due to the commonality between the two disciplines. And a full discussion is included about the new understanding of hydrologic cycle inspired by the multidisciplinary integration.
Fourthly, in response to the issues related to methods raised by reviewer 2, we have emphasized the importance of system approach and added causal discovery method to infer system structure.
We realized that "spectral perspective" may not be enough to tell what we are trying to unearth and that "systematic perspective" would provide better insights into hydrological processes. From the systematic perspective, everything can be considered as a system which typically contains two major parts, a structure composed of multiple interlocking elements, and signals passing through the structure. Hydrological process is no exception. Thus we consider the hydrological process as a system with four-elements structure (rainfall, evapotranspiration, catchment water storage and streamflow) and signals carried by water. We recognized the importance of the system structure because it determines how signals propagate and how the system acts. Therefore, in the revision we have added system structure analysis, which uses causal discovery methods to infer system structure. With such enhancement we can show how system structure evolves over time and affects water transfer within system or among subsystems.
The interlocking elements, flows, and feedback loops make a system more than the sum of its parts and may exhibit adaptive, dynamic, goal-seeking, self-preserving, and evolutionary behavior with covarying structure and signals. This can allow us jumping out the pattern of traditional rigid hydrologic framework and understanding the growth, stagnation, decline, oscillation, randomness, or evolution of hydrological system at long-term behavior-level rather than short-term event-level.
Finally, we have asked native English hydrologists to proofread the revised manuscript to improve the manuscript.
More answers to specific comments are listed below.
Responses to reviewer 1
Taking several catchments from North China, this paper used a spectral analysis approach, trying to illustrate how catchment water storage was influenced by rainfall and vegetation, and how water storage modulated streamflow for the period of 1982 - 2015. The topic is important and interesting. However, there are several major concerns to be further clarified before it can be considered for publication. The comments and suggestions are listed as following, hopefully they will be helpful to improve the manuscript.
Major comments:
(1)The focus and novelty of this paper is not clear. Are you trying to show that the spectral analysis is an effective method in demonstrating water storage dynamics? Or some new findings about catchment storage dynamics and runoff generation mechanisms?
Response: Thank you for the comment. As mentioned above, the revised manuscript have focused on the effectiveness of systematic approaches when they are applied in hydrology. And the new understanding of hydrological processes brought by the integration of multidisciplinary methods have been fully discussed.
(2)The manuscript is overall not well organized, and a clearer logic is needed. For example, the land use change or vegetation succession within four different catchments should be in the part of “Methods, Study Area, and Data”. Meanwhile, before further discussion, it is important to show what components are included in the catchment water storage in the study area, such as lakes, groundwater, soil water and weathered bedrock water. What is the most important one among them for catchment water storage dynamics? And why it is reasonable to attribute the water storage dynamics differences to vegetation root water uptake patterns between wet and dry phases?
Response: Thank you for the comment. According to the comments, we have reorganized the context in the revised manuscript, including more related detailed information such as vegetation and geology in the part of "Methods, Study Area, and Data".
(3)Data quality determines the reliability of final results. I don’t think it is appropriate to use precipitation data from only one station to represent the average level for the whole catchment, especially with large altitude gradient as shown in Figure 1.
Response: Thank you for the comment. We have added other datasets such as CHIRPS to verify the rainfall data.
(4)The language needs to be improved. For example, in line 24-25, what does “fundamental changes to hydrological cycles” mean?
Response: Thank you for the comment. We have addressed this in the revision.
Specific Comments
(5)Line 103-105: It is not clear how spectral analysis method can be used to detect causality. Please explain more.
Response: Thank you for the comment. We have used causal discovery methods such as Granger Causality Test to detect the causality.
(6)In section 2.3, it should be more clearly demonstrated how you processed different spatiotemporal resolution data. For example, why data from 1960-2015 were used for rainfall periodicity detection while others were based on data from 1982-2015? Does it have impact on final results? Similarly, different spatial-resolution ET products were used among the time series, the impact should be evaluated to make the results more convincing.
Response: Thank you for the comment. We have made revision to keep the data span consistent within the period of 1982-2015 and spatial resolution of 1 km.
(7)Line 196: How could ET lags behind water storage by 4 years? Can you interpretate this result with some physical phenomenon or evidence? This is also the base of using such spectral analysis methods to illustrate hydrological processes.
Response: Thank you for the comment. We have used causal discovery methods to interpret the result.
(8)Line 212: Why there is a trough in April in the wet phase?
Response: Thank you for the comment. We have added references to explain the trough in April in wet phase.
(9)Line 236-237: why it is not “water storage determines ET”?
Response: Thank you for the comment. We have used causal discovery methods to determine the causal direction between water storage and ET.
(10)Line 242: Do you mean a higher contribution of subsurface flow to stream flow in the wet phase when compared with that in a dry phase?
Response: Thank you for the comment. The delays show the existence of subsurface flow. We have made revision to make it clear.
(11)Line 255-256 References are needed.
Response: Thank you for the comment. References were added.
(12)Line 267: How will the catchment characteristics impact storage dynamics and streamflow responses among 4 catchments?
Response: Thank you for the comment. We have checked other references for the characteristics of 4 catchments and added them in the revision.
(13)Line 318: Is there any irrigation or groundwater/stream water pumping activity in the study area, which may also influence water storage dynamics.
Response: Thank you for the comment. We have checked other references for these human activities and added them in the revision.
Line 330: see major comment (2).
(14)Line 322-327: the impact of land use change can be long-term lasting, so I doubt this argument.
Response: Thank you for the comment. We have considered the long-term effects in the revision.
(15)Line 338: What do you mean “these plants decide that rainfall will increase…”?
Response: Thank you for the comment. The sentence refers to the reinforcing feedback of vegetation. We have revised it accordingly.
(16)Section 6: Conclusions need to be re-organized. One sentence for one paragraph is not a good way at least in my opinion.
Response: Thank you for the comment. We have revised the conclusion section.
Citation: https://doi.org/10.5194/hess-2022-357-AC1
-
RC2: 'Comment on hess-2022-357', Anonymous Referee #2, 28 Mar 2023
The manuscript “Catchment water storage dynamics and its role in modulating streamflow generation in spectral perspective: a case study in the headwater of Baiyang Lake, China” aims to quantify and describe how catchment water storage changes over time and to identify the relative importance of concurrent changes in system drivers.
The overall objective of the study is highly relevant and the authors also analyse the research question from a conceptually/methodologically very interesting perspective. While the majority of studies that have previously addressed the topic analysed absolute changes in the magnitudes of system drivers, such as precipitation or atmospheric/canopy water demand, the authors place a much stronger focus on the role of changes in timing and the resulting changes in phase shifts between the various variables.
In spite of these interesting aspects, the analysis unfortunately does not live up to its potential and remains quite superficial with quite some assumptions that remain speculative or poorly substantiated by data. In the end, this leaves the reader a bit disappointed as a much more thorough would have been possible, including a more detailed description of the experiment and more in-depth analysis of the results which could have generated much more interesting and robust insights.
As such, I do have several major concerns that need to be addressed in detail to develop the analysis to the point:
(1) The current experiment is based on precipitation data from 5 precipitation gauges. In contrast, the study region has an area of ~ 10.000km2. While I do understand the challenge of scarce data, it remains opaque to me, why the authors did not choose to complement their in-situ observations with globally available fused precipitation data products, such as CHIRPS (1981-present) or comparable? This would at least allow them to demonstrate that their point-scale data are at least sort of representative for their study basins. Similarly, although only available at larger spatial scales, the authors did not even mention nor use the GRACE product, providing global estimates of total water storage variations, as complementary data source to at least plausibility check their results.
(2) Little effort is made to consider and report uncertainties throughout the analysis. This should, in the year 2023, be a standard in scientific publishing. For example, in Figure 5, merely one line is shown for each variable. The reader is left to assume that these lines represent the (multi-annual?) average (mean? median?) regime curves. Why are the distributions (or ranges) around these average not shown? This would be so much more helpful for the reader to assess if the reported changes between wet and dry regimes are indeed significant. Similarly, no quantification of the significance of the wet/dry differences is provided. What are the p-values for the monthly differences between wet and dry?
(3) The authors use wavelets to analyse temporal shifts in the power spectral properties of the various variables. While they are not the first ones to use wavelet analysis in hydrology, it nevertheless remains a rather under-exploited tool. As such, many readers will not be familiar with the idea behind it and how to interpret its results. Thus a much more detailed description of the method and in particular of how the results have to be interpreted is necessary. Otherwise most readers will not be able to assess the results as well as the reasoning behind the interpretation.
(4) The authors invested quite some effort in discussing and interpreting their results in a very intriguing storyline. However, many assumptions are made in this interpretation, which leaves much of it very speculative. This is exacerbated by not substantiating the arguments made with specific numbers that emerge from the underlying analysis. In particular, the role of land cover changes remains addressed in only a surprisingly simplistic and underdeveloped discussion that does neither refer to existing literature in an insightful way nor offer any specific numbers.
(5) The structure of the manuscript needs a major overhaul as it is difficult to find a red line and difficult to follow the flow of the argument. For example, why are the results of the individual catchments shown only after the analysis of one of them? Why is change of land cover only discussed in section 4.2, while it is first mentioned in section 3.4? This causes confusion for the reader.
(5) The level of English does not yet meet the standard required for scientific publications and requires a detailed round of proofreading by a native English speaker with hydrology background.
(6) Figures could be developed with much more care and could offer quite some more detail.
Specific points:
(7) L.18: in hydrology close to nothing can actually be “verified”. Given the uncertainties in our data, the best we can hope for is that any results are more or less consistent with the available data.
(8) L.21: a clear definition of “subsurface flow” is needed, as of course, groundwater is also a subsurface flow (and not an above-surface flow…).
(9) L.57-62: this is not correct. There are several studies that have previously shown the opposite (e.g. Hulsman et al., 2021).
(10) L.74: the authors could at least have mentioned the GRACE satellite product that provides estimates of total water storage variations (e.g. Landerer and Swenson, 2012). Ideally it could also be used as complementary data source to at least plausibility check the results.
(11) L.129: what does that mean? “The unit of streamflow data was 108m3”??? This is not being very kind to the reader.
(12) L.130: Five stations is not a lot and I am not sure that it is sufficient to plausibly describe precipitation across this large area. Why was this data set not complemented by gridded data? See Comment (1) above
(13) L.136: Seriously? Can the authors please try to use units that are standard in scientific hydrology, such as mm d-1?!
(14) L.147: what is meant by “old water” and “young water”? I do not believe the authors have made any attempt to actually date the water here in the sense of hydrograph separation (e.g. Klaus and McDonnell, 2013) or transit time distributions (e.g. Benettin et al., 2022). The terms “old water” and “young water” have very specific, narrowly defined meaning in hydrology. Please do not use them in a different context.
(15) L.148: sure, but which data product was this exactly?
(16) L.150: “outperformance among” - what is this supposed to mean?
(17) L.150: “the units […] were in 108m3 […]” ????
(18) L.155: how was “representative” objectively determined?
(19) L.158: the trends are not at all distinguishable in Figure 2a
(20) L.158: “water storage […] fluctuates around zero […]”. Does this make sense?? How can there be a negative water storage?? Please be a bit more careful with how to describe things: it is not the water storage that fluctuates around zero but the water storage anomaly.
(21) L.162: what is a “high wavelet power spectrum”??? Is this good or bad? What do the coefficients mean? Are the equivalent to coefficients of a Fourier transform which are directly related to the spectral power of specific frequencies? How can their fluctuations be interpreted? This is a relevant part of the analysis and needs to be described in much more detail to allow the reader to follow and fully appreciate the interpretation.
(22) L.169: on basis of which criteria were the wet and dry phases defined?
(23) L.177: I do not think that a 30-yr periodicity can be meaningfully and robustly inferred from a data record of merely 33 years.
(24) L.198: what is meant by “hysteresis effect” here?
(25) L.208: Ok, all coefficients are negative. So what? what does this mean?
(26) L.211, Figure 5: are these average values? Which average? Mean? Median? Long-term? Why are only these lines shown and not the ranges, envelopes or distributions? How significant are the differences between wet and dry?
(27) L.214: water storage cannot be negative…see Comment (20) above.
(28) L.227-228: How was this done?
(29) L.237-238: which vegetation cover? How can the reader know that if it is only shown in detail much later in the manuscript??
(30) L.244-245: How would the authors know if this is indeed “surface flow”?? Although occurring locally, this is a very uncommon phenomenon at the catchment scale, as the authors also point out later in the manuscript. Could just as well be any other fast responding flow component in the system.
(31) L.300, Figure 10: Panel (a) is surprisingly uninformative. Why not find a way to show the differences between the decades in these maps? Panel (b) has incorrect units on the y-axis – this should probably read as “Fraction”.
(32) L.306-307: “dramatic decrease in annual ET in ~2004-2005”??? I cannot see that anywhere. If anything, the major decrease happens somewhere between 1998 and 2002, but even that is difficult to see in Figure 2a.
(33) L.300ff, Sections 4.2-4.3: nice storyline but mostly based on assumptions and speculation and not sufficiently substantiated by data.
(34) L.344-345: what is “surface stream flow”??
(35) L.345ff: how can results of such plot-scale studies be reasonably assumed to give meaningful insights at the river basin scale??
References:
Benettin, P., Rodriguez, N. B., Sprenger, M., Kim, M., Klaus, J., Harman, C. J., ... & McDonnell, J. J. (2022). Transit time estimation in catchments: Recent developments and future directions. Water Resources Research, 58(11), e2022WR033096.
Hulsman, P., Hrachowitz, M., & Savenije, H. H. (2021). Improving the representation of long‐term storage variations with conceptual hydrological models in data‐scarce regions. Water Resources Research, 57(4), e2020WR028837.
Klaus, J., & McDonnell, J. J. (2013). Hydrograph separation using stable isotopes: Review and evaluation. Journal of hydrology, 505, 47-64.
Landerer, F. W., & Swenson, S. C. (2012). Accuracy of scaled GRACE terrestrial water storage estimates. Water resources research, 48(4).
Citation: https://doi.org/10.5194/hess-2022-357-RC2 -
AC2: 'Reply on RC2', Xinyao Zhou, 18 Apr 2023
We highly appreciate the two anonymous reviewers for their invaluable comments, which enlightened us on lots of points. In this letter, we will briefly summarize how their insightful suggestions help us improve our manuscript, and then answer specific questions.
Firstly, in response to the concerns about scarce rainfall data raised by both reviewers, we have included satellite products such as CHIRPS to cross-validate rainfall data in the revision. Furthermore, we have used GRACE data to validate catchment water storage result as suggested.
Secondly, we have reorganized the context of the revised manuscript to make it more logical, for example, introducing vegetation succession in Study Area section instead of Results section.
Thirdly, we have highlighted novelty as suggested by reviewer 1, by demonstrating the effectiveness of systematic approaches when they are applied in hydrology due to the commonality between the two disciplines. And a full discussion is included about the new understanding of hydrologic cycle inspired by the multidisciplinary integration.
Fourthly, in response to the issues related to methods raised by reviewer 2, we have emphasized the importance of system approach and added causal discovery method to infer system structure.
We realized that "spectral perspective" may not be enough to tell what we are trying to unearth and that "systematic perspective" would provide better insights into hydrological processes. From the systematic perspective, everything can be considered as a system which typically contains two major parts, a structure composed of multiple interlocking elements, and signals passing through the structure. Hydrological process is no exception. Thus we consider the hydrological process as a system with four-elements structure (rainfall, evapotranspiration, catchment water storage and streamflow) and signals carried by water. We recognized the importance of the system structure because it determines how signals propagate and how the system acts. Therefore, in the revision we have added system structure analysis, which uses causal discovery methods to infer system structure. With such enhancement we can show how system structure evolves over time and affects water transfer within system or among subsystems.
The interlocking elements, flows, and feedback loops make a system more than the sum of its parts and may exhibit adaptive, dynamic, goal-seeking, self-preserving, and evolutionary behavior with covarying structure and signals. This can allow us jumping out the pattern of traditional rigid hydrologic framework and understanding the growth, stagnation, decline, oscillation, randomness, or evolution of hydrological system at long-term behavior-level rather than short-term event-level.
Finally, we have asked native English hydrologists to proofread the revised manuscript to improve the manuscript.
More answers to specific comments are listed below.
Responses to reviewer 2
The manuscript “Catchment water storage dynamics and its role in modulating streamflow generation in spectral perspective: a case study in the headwater of Baiyang Lake, China” aims to quantify and describe how catchment water storage changes over time and to identify the relative importance of concurrent changes in system drivers.
The overall objective of the study is highly relevant and the authors also analyse the research question from a conceptually/methodologically very interesting perspective. While the majority of studies that have previously addressed the topic analysed absolute changes in the magnitudes of system drivers, such as precipitation or atmospheric/canopy water demand, the authors place a much stronger focus on the role of changes in timing and the resulting changes in phase shifts between the various variables.
In spite of these interesting aspects, the analysis unfortunately does not live up to its potential and remains quite superficial with quite some assumptions that remain speculative or poorly substantiated by data. In the end, this leaves the reader a bit disappointed as a much more thorough would have been possible, including a more detailed description of the experiment and more in-depth analysis of the results which could have generated much more interesting and robust insights.
As such, I do have several major concerns that need to be addressed in detail to develop the analysis to the point:
(1) The current experiment is based on precipitation data from 5 precipitation gauges. In contrast, the study region has an area of ~ 10.000km2. While I do understand the challenge of scarce data, it remains opaque to me, why the authors did not choose to complement their in-situ observations with globally available fused precipitation data products, such as CHIRPS (1981-present) or comparable? This would at least allow them to demonstrate that their point-scale data are at least sort of representative for their study basins. Similarly, although only available at larger spatial scales, the authors did not even mention nor use the GRACE product, providing global estimates of total water storage variations, as complementary data source to at least plausibility check their results.
Response: Thank you for the comment. We have added other rainfall products to check the reliability of point-scale rainfall and used GRACE to verify the catchment water storage changes.
(2) Little effort is made to consider and report uncertainties throughout the analysis. This should, in the year 2023, be a standard in scientific publishing. For example, in Figure 5, merely one line is shown for each variable. The reader is left to assume that these lines represent the (multi-annual?) average (mean? median?) regime curves. Why are the distributions (or ranges) around these average not shown? This would be so much more helpful for the reader to assess if the reported changes between wet and dry regimes are indeed significant. Similarly, no quantification of the significance of the wet/dry differences is provided. What are the p-values for the monthly differences between wet and dry?
Response: Thank you for the comment. As suggested, we have discussed the uncertainties and highlighted the significance of the wet/dry differences in the revision.
(3) The authors use wavelets to analyse temporal shifts in the power spectral properties of the various variables. While they are not the first ones to use wavelet analysis in hydrology, it nevertheless remains a rather under-exploited tool. As such, many readers will not be familiar with the idea behind it and how to interpret its results. Thus a much more detailed description of the method and in particular of how the results have to be interpreted is necessary. Otherwise most readers will not be able to assess the results as well as the reasoning behind the interpretation.
Response: Thank you for the comment. As mentioned above, we have adopted systematic perspective to integrate the signal analysis and system structure. On this basis, more introduction of signal analysis were added.
(4) The authors invested quite some effort in discussing and interpreting their results in a very intriguing storyline. However, many assumptions are made in this interpretation, which leaves much of it very speculative. This is exacerbated by not substantiating the arguments made with specific numbers that emerge from the underlying analysis. In particular, the role of land cover changes remains addressed in only a surprisingly simplistic and underdeveloped discussion that does neither refer to existing literature in an insightful way nor offer any specific numbers.
Response: Thank you for the comment. On the basis of new results from causal discovery methods, we have included additional discussion of the results in the revision and more references were added.
(5) The structure of the manuscript needs a major overhaul as it is difficult to find a red line and difficult to follow the flow of the argument. For example, why are the results of the individual catchments shown only after the analysis of one of them? Why is change of land cover only discussed in section 4.2, while it is first mentioned in section 3.4? This causes confusion for the reader.
Response: Thank you for the comment. We have reorganized the context in our revised manuscript.
(5) The level of English does not yet meet the standard required for scientific publications and requires a detailed round of proofreading by a native English speaker with hydrology background.
Response: Thank you for the comment. We have a hydrologist with native English to proofread the revised manuscript.
(6) Figures could be developed with much more care and could offer quite some more detail.
Response: Thank you for the comment. We have revised the figures to provide more information.
Specific points:
(7) L.18: in hydrology close to nothing can actually be “verified”. Given the uncertainties in our data, the best we can hope for is that any results are more or less consistent with the available data.
Response: Thank you for the comment. We have revised the sentence accordingly.
(8) L.21: a clear definition of “subsurface flow” is needed, as of course, groundwater is also a subsurface flow (and not an above-surface flow…).
Response: Thank you for the comment. We have added a definition for the "subsurface flow".
(9) L.57-62: this is not correct. There are several studies that have previously shown the opposite (e.g. Hulsman et al., 2021).
Response: Thank you for the comment. We have included these references to show the debates.
(10) L.74: the authors could at least have mentioned the GRACE satellite product that provides estimates of total water storage variations (e.g. Landerer and Swenson, 2012). Ideally it could also be used as complementary data source to at least plausibility check the results.
Response: Thank you for the comment. We have inclued GRACE product to verify the catchment water storage.
(11) L.129: what does that mean? “The unit of streamflow data was 108m3”??? This is not being very kind to the reader.
Response: Thank you for the comment. We have revised it accordingly as well as the others similarly.
(12) L.130: Five stations is not a lot and I am not sure that it is sufficient to plausibly describe precipitation across this large area. Why was this data set not complemented by gridded data? See Comment (1) above
Response: Thank you for the comment. We have added other gridded datasets to complement the rainfall data.
(13) L.136: Seriously? Can the authors please try to use units that are standard in scientific hydrology, such as mm d-1?!
Response: Thank you for the comment. We have changed to the standard unit of mm/year.
(14) L.147: what is meant by “old water” and “young water”? I do not believe the authors have made any attempt to actually date the water here in the sense of hydrograph separation (e.g. Klaus and McDonnell, 2013) or transit time distributions (e.g. Benettin et al., 2022). The terms “old water” and “young water” have very specific, narrowly defined meaning in hydrology. Please do not use them in a different context.
Response: Thank you for the comment. We have revised it accordingly.
(15) L.148: sure, but which data product was this exactly?
Response: Thank you for the comment. More explanations about the dataset were provided.
(16) L.150: “outperformance among” - what is this supposed to mean?
Response: Thank you for the comment. We have revised it with better expression.
(17) L.150: “the units […] were in 108m3 […]” ????
Response: Thank you for the comment. We have changed it to the standard unit of mm/year.
(18) L.155: how was “representative” objectively determined?
Response: Thank you for the comment. We have reorganized the manuscript to show all four catchments here.
(19) L.158: the trends are not at all distinguishable in Figure 2a
Response: Thank you for the comment. We have revised the figure to make it clear.
(20) L.158: “water storage […] fluctuates around zero […]”. Does this make sense?? How can there be a negative water storage?? Please be a bit more careful with how to describe things: it is not the water storage that fluctuates around zero but the water storage anomaly.
Response: Thank you for the comment. We have rewiritten the sentence.
(21) L.162: what is a “high wavelet power spectrum”??? Is this good or bad? What do the coefficients mean? Are the equivalent to coefficients of a Fourier transform which are directly related to the spectral power of specific frequencies? How can their fluctuations be interpreted? This is a relevant part of the analysis and needs to be described in much more detail to allow the reader to follow and fully appreciate the interpretation.
Response: Thank you for the comment. We have provided more details about the wavelet results.
(22) L.169: on basis of which criteria were the wet and dry phases defined?
Response: Thank you for the comment. The wet and dry phases were separated according to the change of rainfall and the separation was verified by other studies.
(23) L.177: I do not think that a 30-yr periodicity can be meaningfully and robustly inferred from a data record of merely 33 years.
Response: Thank you for the comment. We have focused on the difference between wet and dry phases and weakened the periodicity in the later revision.
(24) L.198: what is meant by “hysteresis effect” here?
Response: Thank you for the comment. Additional explanations were added in the revision.
(25) L.208: Ok, all coefficients are negative. So what? what does this mean?
Response: Thank you for the comment. More explanation have been added here.
(26) L.211, Figure 5: are these average values? Which average? Mean? Median? Long-term? Why are only these lines shown and not the ranges, envelopes or distributions? How significant are the differences between wet and dry?
Response: Thank you for the comment. Uncertainty analysis was added here and figure 5 was revised.
(27) L.214: water storage cannot be negative…see Comment (20) above.
Response: Thank you for the comment. We have made revisions accordingly.
(28) L.227-228: How was this done?
Response: Thank you for the comment. Causal discovery method was used in the revision.
(29) L.237-238: which vegetation cover? How can the reader know that if it is only shown in detail much later in the manuscript??
Response: Thank you for the comment. The structure of paper was adjusted in the revision.
(30) L.244-245: How would the authors know if this is indeed “surface flow”?? Although occurring locally, this is a very uncommon phenomenon at the catchment scale, as the authors also point out later in the manuscript. Could just as well be any other fast responding flow component in the system.
Response: Thank you for the comment. We have separated streamflow into rapid-response component and slow-response component in the revision.
(31) L.300, Figure 10: Panel (a) is surprisingly uninformative. Why not find a way to show the differences between the decades in these maps? Panel (b) has incorrect units on the y-axis – this should probably read as “Fraction”.
Response: Thank you for the comment. We have changed the figure according to the suggestions.
(32) L.306-307: “dramatic decrease in annual ET in ~2004-2005”??? I cannot see that anywhere. If anything, the major decrease happens somewhere between 1998 and 2002, but even that is difficult to see in Figure 2a.
Response: Thank you for the comment. We have changed the figure to include more information.
(33) L.300ff, Sections 4.2-4.3: nice storyline but mostly based on assumptions and speculation and not sufficiently substantiated by data.
Response: Thank you for the comment. More discussions were included and more references were added.
(34) L.344-345: what is “surface stream flow”??
Response: Thank you for the comment. It refers to rapid-response component of streamflow here. It has been revised here.
(35) L.345ff: how can results of such plot-scale studies be reasonably assumed to give meaningful insights at the river basin scale??
Response: Thank you for the comment. We have included some catchment-scale references in discussion section.
Citation: https://doi.org/10.5194/hess-2022-357-AC2
-
AC2: 'Reply on RC2', Xinyao Zhou, 18 Apr 2023
Xinyao Zhou et al.
Xinyao Zhou et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 402 | 130 | 19 | 551 | 9 | 10 |
- HTML: 402
- PDF: 130
- XML: 19
- Total: 551
- BibTeX: 9
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1