Quantifying the Impact of Cascade Reservoirs on Streamflow, Drought, and Flood in the Jinsha River Basin
Round 1
Reviewer 1 Report
Dear Authors,
Please find my comments below. Please address all of them, as mentioned in the document.
Best regards
Comments for author File: 
Comments.pdf
Author Response
Dear Reviewer：
Thank you for the evaluation of our manuscript. We have taken all your comments into consideration and tried to address each point in the revised manuscript.
- Comments from the Reviewer #1:
General comments:
(1) “The manuscript is relatively interesting and addresses modern techniques used in the quantitative analysis of hydrological extremes events. It uses artificial neural networks, which have already become a normal trend for such types of studies, but combines different kinds of data, to properly address this type of analysis.”
Response: Thank you for the positive evaluation of our manuscript. The runoff of Jinsha River is mainly affected by precipitation and snowmelt, etc. In this paper, precipitation, temperature, snowmelt, soil moisture and evaporation are used as driving factors to simulate the runoff. Input data were obtained from the open databases, precipitation data were from Climate Hazards Group Infrared Precipitation with Stations(CHIRPS 2.0), temperature, snow melt and evaporation data were from the land component of the fifth generation of European Reanalysis (ERA5-Land), soil moisture data were obtained from the NASA Global Land Data Assimilation System (GLDAS). They were unified into a daily scale of 0.5°×0.5° spatial resolution. The input and output data were then standardized and normalized before modeling.
(2) “It has the main parts of a complete study (hypothesis, methodology, results, discussion, impact evaluation, and validation), and addresses each chapter responsibly.”
Response: Thanks for your reminder. The main parts of my study have been modified and refined (hypothesis, methodology, results, discussion, impact evaluation, and validation).
(3) “English language needs minor improving - from spelling, to phrase topic, and overall linguistic corrections.”
Response: Thank you for your suggestion. The paper has been polished by professionals.
(4) “Please verify all the editorial requirements again, because there are some formatting sections missing, according to the final journal template, before resubmitting the manuscript, to insure not to skip any formatting aspects.”
Response: Thanks for this comment. We have checked and supplemented all formatting issues.
(5) “The novelty aspects of the analysis should be described in more technical detail, so that it could possibly be referenced in future studies, in this field.”
Response: Thank you for your guidance. The introduction has been revised to highlight the innovation of this paper. As the upper reaches of China’ largest river(the Yangtze River), the Jinsha River are an important supplier and determinant of water resources in the downstream urban agglomeration in the middle and lower reaches of the Yangtze River, has implications for the economic stability and regional security of southern China. The Jinsha River is also the main water source of the West Route of China's South-to-North Water Diversion Project, and the rational allocation of water re-sources from the Jinsha River is important for the successful implementation of this project. We propose a hydrological model to construct different scenarios in the basin, obtain the impact of power plants on runoff excluding the influence of climate change (by controlling input variables), and analyze the impact of power plants construction on different periods. This method can be used as a reference for other hydrological models of hydropower stations.
Line75~77:
The objectives of this study were to: i) build a hydrological model suitable for the JRB based on neural networks; ii) quantify the influence of cascade reservoirs on streamflow in the JRB; and iii) analyze the effect of reservoirs on streamflow, flood, and drought events in the JRB. The innovation point is that this paper analyzes the influence of power stations at different times by constructing two scenarios: power station and power station-free. The results of our study could provide technical support for the construction of hydrological models in regions with cascade reservoirs, reference for the construction and operation of hydropower stations in the JRB, data for the West Route of the South-to-North Water Diversion project, and research directions for water security in the Yangtze River basin.
(6) “The conclusions are not ground-breaking, they just reveal a quantitative comparison of before-after situations, in relation to the dams along the main river.”
Response: Thank you for your comments. In this paper, the influence of power station on runoff is quantitatively analyzed by comparing the runoff in the scenario with and without power stations at the same time period. Compared with other studies, the innovation of our study is that it evaluates the influence of cascade reservoirs on runoff in different periods, which is worthy to discussion. We have added to the conclusion and thank you for your suggestions.
Line 456~459:
- The construction of cascade reservoirs on the JRB reduced the average daily streamflow by 5.69% from 4937.97 m3/s to 4657.10 m3/s. During the rainy and dry seasons, average daily streamflow decreased by 10.93% and 10.95%, respectively. The cumulative storage capacity and quantity of the reservoirs affect the change degree of runoff. On the whole, with the increase of the cumulative storage capacity, the effect of the reservoirs on runoff increases, with a non-linear relationship.
- The cascade reservoirs on the JRB reduced the flood POT from an average of 2.6 to 2.4 per year, reducing the frequency of floods along the Jinsha River by 7.69%. In terms of flood magnitude, the construction of reservoirs reduced the average annual runoff exceeding the threshold from 5.25×105 m3 to 3.26×105 m3, reduced by 37.86%. In other words, cascade reservoirs in the JRB reduce the frequency and magnitude of flood events.
- The effect of reservoir construction on drought duration is mainly reflected in the decreasing effect on the duration of extreme drought and increasing effect on the duration of moderate and severe drought on all time scales. The effect of reservoirs on drought severity is mainly reflected in the mitigation of the severity of extreme drought and aggravation of the severity of moderate and severe drought. The reservoirs have a mitigating effect on the severity of mild drought on the SRI-1 and SRI-6 timescales and an intensifying effect on the SRI-3 and SRI-6 timescales.
- It is suggested that in the JRB, when the Central Yunnan Water diversion Project and the South to North Water Transfer West Line Project are com-pleted, the water transfer volume and time should be fully considered to coordinate with the storage and operation of the cascade reservoirs.
(7) “Overall, the study focuses on mathematically sound approaches on the methodological side of things, but partially ignores the hydrological and geographical reality of the flow processes, and regime.”
Response: We use the LSTM method in the data-driven model. The principle of this method is to find the mapping relationship between input and output through machine learning, so it contains more mathematical content. The process-driven model is modeled by the physical process of precipitation-convergence in the basin. This is the difference between data-driven model and physical model. Because of the mathematical characteristics of the data-driven model, it has the advantages of using open source data, simple rate setting and high simulation accuracy. In addition, hydrological methods such as POT have also been adopted.
Specific comments:
(1) “L65 – different font sizes? Please verify”
Response: Thank you for your correction. We have verified the font size.
(2) “L81-82 – the phrasing is ambiguous. The subject is the river, but you continue by talking about the drainage basin. But is it referring to the location, or the morphometric aspects? Please rephrase.”
Response: Thank you very much for your correction. We have revised the wording.
Line 85~86:
The Jinsha River constitutes the upper reaches of the Yangtze River above the Pingshan hydrological station with a length of 3481 km. The JRB is located at 90–105°E, 24–36°N with a drainage area of 47×104 km2 (Figure.1). In the JRB, precipitation and temperature present a heterogeneous distribution [20], with precipitation ranging from 300 mm to more 1,300 mm and the annual minimum temperature ranging from -5.6°C to 21.9°C [21]. The average annual runoff of the JRB is 145 billion m3, mainly governed by precipitation and melting snow (ice).
(3) “L89 – Another ambiguous phrase – that altitude is the height range? Or maximum? Please clarify.”
Response: Sorry to cause you trouble. Here refers to the height difference between upstream and downstream, we have modified this place. (Line91)
Line 93:
The Jinsha River originates from the Tibetan Plateau, and spans the Yunnan-Guizhou Plateau and the western edge of the Sichuan Basin with a complex and variable topography. The altitude difference between upstream and downstream can reach 6329 m. The JRB has abundant water resources, and it is the largest water and electricity energy base in China, with a potential capacity of approximately 100 million kW. A series of cascade reservoirs have been constructed in the basin. In this study, 15 reservoirs, with a total storage capacity of 572 billion m3, were considered, and their major parameters are shown in Table 2.
(4) “Figure 1 – the colors of the height range are not representative from a cartographic perspective. The do not associate with the earth tones required in such a vast altitude range (from dark green, all the way to white – in the alpine domain). This should be corrected.”
Response: Thank you for your advice. We've made revisions.
(5) “You state several aspects regarding climate change and hydrological flow/regime (“With the use of historical time 49 series analysis, changes in hydrological indices before and after dam construction are 50 compared, but this method is not ideal because it cannot ignore the influence of other 51 factors such as climate change on runoff [9, 10]. The impact of climate change and cascade 52 power station on runoff can be determined using model simulation by controlling input 53 variables [11, 12].”). Furthermore, in Figure 2, you describe that the assessment period is between 1998-2020. How can you validate this is relevant, in the context of climate change, which you have mentioned multiple times in the manuscript？”
Response: Sorry to cause you trouble. During the assessment period, the same meteorological data entered in both dammed and dam-free scenarios. And then we compare dammed streamflow with natural streamflow. According to the control variable method, the impact of climate change could be considered to exclude.
(6) “L193 - the frequency and magnitude of floods – have you tried incorporating classic statistical methods, such as Pearson III recurrence, or Weibull, to see if there are any different results?”
Response: Thank you for your constructive comments. Pearson III recurrence and Weibull are both distribution function and probability density methods to calculate flood frequency and intensity. A flood lasting for several days can be easily identified as multiple floods, while POT's method sets a 15-day time window, that is, only one peak value is considered within 15 days, which avoids counting the same flood multiple times. Therefore, we chose the POT method instead of Pearson III recurrence and Weibull in this paper.
(7) “Figure 4 – what time unit is depicted on the horizontal axis? I understand you talk about “Different timescales” but this should be more demystified.”
Response: I'm sorry for causing you trouble. The time unit on the horizontal axis of Figure 4 is month, and we have modified it here.
(8) “L309-310 – you state “In other words, cascade reservoirs in the JRB 309 reduce the frequency and magnitude of flood events.” This is a very simplistic conclusion, that is generally true, and not necessarily dependent on such a complex analysis. Would you say that the identification of the reduction of the flood magnitude by 37.86% is an important scientific achievement, derived from this paper? Because all man-made reservoirs, managed correctly, would have a significant reduction of flood magnitudes.”
Response: Thank you for your comments. We quantitatively evaluated the change of flood magnitude, which is reduced by 37.86%. This is part of our quantitative results. Compared with other studies, the innovation of our research is to evaluate the impact of cascade reservoirs on runoff in different periods, which is worth discussing. In addition, this paper constructs two scenarios with and without power stations, which provides a reference for the study of river basins with power stations. Therefore, this study is valuable.
(9) “In the Discussions section, you mentioned a few shortcomings of the study, but they are quite relevant, such as: “water diversion activities in the JRB may also reduce runoff.”, which is obvious, alongside with the phrase “However, the risk of drought 400 in the dry season is higher than that in the rainy season” which is a general truth. You do not have to undergo such a study, just to affirm such statements.”
Response: We agree with your statement that "it is obvious that water diversion activities reduce runoff". The reason why we emphasize water diversion activities is that there are a series of large-scale water diversion projects under construction in the Jinsha River basin, such as the Central Yunnan Water diversion Project. However, the exact extent of the impact of these drinking water projects remains to be studied. In the future, these projects may have an impact on runoff that cannot be ignored, and the following research needs attention.
(10) “Conclusions are too general and only reflect numbers that back up a general truth related to dams built along river courses.”
Response: Thanks to your suggestion, we have revised the conclusion.
Line 456~459:
(1) The construction of cascade reservoirs on the JRB reduced the average daily streamflow by 5.69% from 4937.97 m3/s to 4657.10 m3/s. During the rainy and dry seasons, average daily streamflow decreased by 10.93% and 10.95%, respectively. The cumulative storage capacity and quantity of the reservoirs affect the change degree of runoff. On the whole, with the increase of the cumulative storage capacity, the effect of the reservoirs on runoff increases, with a non-linear relationship.
(2) The cascade reservoirs on the JRB reduced the flood POT from an average of 2.6 to 2.4 per year, reducing the frequency of floods along the Jinsha River by 7.69%. In terms of flood magnitude, the construction of reservoirs reduced the average annual runoff exceeding the threshold from 5.25×105 m3 to 3.26×105 m3, reduced by 37.86%. In other words, cascade reservoirs in the JRB reduce the frequency and magnitude of flood events.
(3) The effect of reservoir construction on drought duration is mainly reflected in the decreasing effect on the duration of extreme drought and increasing effect on the duration of moderate and severe drought on all time scales. The effect of reservoirs on drought severity is mainly reflected in the mitigation of the severity of extreme drought and aggravation of the severity of moderate and severe drought. The reservoirs have a mitigating effect on the severity of mild drought on the SRI-1 and SRI-6 timescales and an intensifying effect on the SRI-3 and SRI-6 timescales.
(4) It is suggested that in the JRB, when the Central Yunnan Water Diversion Project and the South to North Water Transfer West Line Project are completed, the water transfer volume and time should be fully considered to coordinate with the storage and operation of the cascade reservoirs.
Author Response File: Author Response.pdf
Reviewer 2 Report
Thank you for the opportunity of reviewing the interesting paper. It covers the important issue of qualifying the impact of cascade reservoirs in Jinsha river basin.
I have the following comments:
1. The title is fine, explaining the contents of the paper. But the title could explain the key message clearer. For example, it can be modifying by adding “reducing droughts and floods”
2. The structure of the paper is well organized and clear. Methods are well explained.
3. Abstract is also fine.
4. Considerations and conclusions are supported by findings.
5. Please stress academic novelty. The paper used the methodology of LSTM model already developed. Please explain what new contribution to scholarship is.
6. 2.4 flood characteristics index “The criteria for threshold selection is that there are 2-3 floods every year in the baseline period”. Please explain more about this volume. Is this the volume causing flood damage on towns or agriculture fields along the river?
7. 2.5 drought characteristic index “mild, moderate, severe and extreme drought correspond”. Similarly, please explain more about these meanings. What damage on people’s lives and agriculture and economic activities are caused by different levels of droughts?
8. Figure 5c: It shows some big differences of observed flood and simulated floods. Several observed data of 15-20 million m3 show some 10 million m3 for simulated. These are big gaps that are related to reliability and applicability of the model. Please discuss the factors of causing these gaps and justify using this model.
9. Discussion: What did dam development improved people’s lives and economic activities by reducing floods and droughts? Whether are these effects planned or unexpected?
10. Conclusions should include policy implications.
Author Response
Dear Reviewer，
Thank you for the evaluation of our manuscript. We have taken all your comments into consideration and tried to address each point in the revised manuscript.
Response to Reviewer：
- Comments from the Reviewer #2:
(1) The title is fine, explaining the contents of the paper. But the title could explain the key message clearer. For example, it can be modifying by adding “reducing droughts and floods”
Response: Sincerely thank you for your suggestion. We have seriously considered this problem, because the conclusion in the drought part of this paper is that moderate drought and severe drought increase and extreme drought decrease. It may not be appropriate to change it to "reduce drought and flood". Thank you very much for your valuable comments.
(2) “The structure of the paper is well organized and clear. Methods are well explained.”
Response: Thank you for the positive evaluation of our manuscript. We have taken all your comments into consideration and tried to address each point in the revised manuscript.
(3) “Abstract is also fine".”
Response: Thank you very much for your comments. Your suggestions are all valuable and helpful for improving our article.
(4) “Considerations and conclusions are supported by findings.”
Response: Thank you for the positive evaluation.
(5) “Please stress academic novelty. The paper used the methodology of LSTM model already developed. Please explain what new contribution to scholarship is.”
Response: Thank you for your guidance. The introduction has been revised to highlight the innovation of this paper. As the upper reaches of China’ largest river(the Yangtze River), the Jinsha River are an important supplier and determinant of water resources in the downstream urban agglomeration in the middle and lower reaches of the Yangtze River, has implications for the economic stability and regional security of southern China. The Jinsha River is also the main water source of the West Route of China's South-to-North Water Diversion Project, so the rational allocation of water resource from the Jinsha River is important for the successful implementation of this project. In this basin, we first used the scenario simulation method to analyze the impact of cascade reservoirs on runoff in different periods. This method can be used as a reference for other basins with reservoirs.
(6) “2.4 flood characteristics index “The criteria for threshold selection is that there are 2-3 floods every year in the baseline period”. Please explain more about this volume. Is this the volume causing flood damage on towns or agriculture fields along the river?”
Response: I'm very sorry to cause you doubt. We have revised the expression. This index doesn’t refer to the volume causing flood damage on towns or agriculture fields along the river. We mean that the threshold must be set under a conditions that flood events extracted by POT are 2–3 times per. If the threshold is too large, there will be less than two flood events per year on average, and the flood peak will be missed in years of low discharge. If the threshold is too small, there will be more than 3 floods per year on average, which will lead to excessive extraction of flood events in multi-flood peak years. Therefore, in order to accurately reflect the real situation, the threshold we select should ensure that 2-3 floods can be extracted on average every year in the base period. This index we refer to articles by Xiaobo Yun et al and Xu Yuan et al.
Line 201~202:
To quantify the impact of reservoirs on floods, the frequency and magnitude of floods should be taken into account. The Peaks Over Threshold (POT) [29, 30, 31] is a commonly used index for quantifying the frequency of flood events. POT is the number of flood events that exceed a specified threshold. According to this threshold, there are 2-3 flood events exceeding it every year on average during the base period. Then the threshold is used during the assessment period to detect flood events. In particular, in the 15-day window, only one peak was considered to avoid double counting of the same flood [31]. By comparing the POT of natural flow and dammed flow, the influence of reservoirs in the JRB on flood frequency can be determined. The magnitude of the flood was mainly evaluated by calculating the total runoff exceeding the flood threshold. By comparing this value between natural flow and dammed flow, the impact of reservoirs on flood magnitude can be obtained.
(7) “2.5 drought characteristic index “mild, moderate, severe and extreme drought correspond”. Similarly, please explain more about these meanings. What damage on people’s lives and agriculture and economic activities are caused by different levels of droughts?”
Response: I'm sorry I didn't express it clearly. Drought characteristic index “mild, moderate, severe and extreme drought correspond”. It is only an indicator calculated according to SRI, which is obtained according to the experience of the basin and the literature of others(Sun et al. 2018). It can only reflect the degree of drought to a certain extent, and cannot completely correspond to people's life, agriculture and economic activities. (Sun, Z.; Zhu, X.; Pan, Y.; Zhang, J.; Liu, X. Drought evaluation using the GRACE terrestrial water storage deficit over the Yangtze River Basin, China. Science of the Total Environment (2018), 634, 727-738.)
(8) “Figure 5c: It shows some big differences of observed flood and simulated floods. Several observed data of 15-20 million m3 show some 10 million m3 for simulated. These are big gaps that are related to reliability and applicability of the model. Please discuss the factors of causing these gaps and justify using this model.”
Response: Thank you for your comments. We understand your doubts. High value simulation is a common difficulty in hydrological simulation. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation. In the discussion, we added the comparison of the accuracy with other models in the Jinsha River basin, as well as the analysis of model uncertainty. Sincerely thank you for your question.
Line 373~376, 387~394:
In this study, the performance of LSTM model is compared with that of other physical process models in the daily scale flow simulation of the Jinsha River basin. As shown in Table 3, the performance of the LSTM model is higher than that of physical process models [6, 38, 39, 40, 41]. The performance improvement can be attributed to the capability of the LSTM model to control input data flexibility based on watershed characteristics [4]. For example, the JRB originates from the Qinghai Tibet Plateau, where snowmelt affects runoff generation. Accordingly, snowmelt data can be incorporated into the input matrix. As the main principle of the LSTM model is to establish an optimal mapping of driving forces and runoff instead of causation, more diverse data can be introduced, and the construction and calibration of the model are relatively simple [42]. Meanwhile, physical process models generalize the process of precipitation-runoff through mathematical and physical methods [43]. Consequently, physical process models have a fixed form, with high data requirements [44]. Moreover, the construction and calibration of physical process models are complex, and the flexibility of these models requires extensive verification. However, in the current hydrological model research, the poor simulation effect of high value is a common problem. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation.
Table 3 Performances of the LSTM and physical process models reported in the literature on daily streamflow forecasting.
|
Model |
Testing period |
NSE |
Source |
|
LSTM |
1994–1998 |
0.92 |
This study |
|
SWAT (Semi-distributed Model) |
2000–2016 |
0.90 |
Chen et al. (2020) |
|
SWAT (Semi-distributed Model) |
2012–2012 |
0.71 |
Wu et al. (2020) |
|
Xinanjiang model (Distributed model) |
1986–2000 |
0.84 |
Feng et al. (2018) |
|
VIC model (Distributed model) |
2004–2006 |
0.72 |
Maza et al. (2020) |
|
MIKE 11 NAM model (Distributed model) |
2009–2015 |
0.83 |
Aredo et al. (2021) |
(9) “Discussion: What did dam development improved people’s lives and economic activities by reducing floods and droughts? Whether are these effects planned or unexpected?”
Response: When the flood comes, the cascade reservoirs has coordinated and coordinated the residents by means of peak shifting, peak cutting and flood blocking, which has greatly reduced the flooded area and protected people's lives and property safety. The cascade reservoirs in the Jinsha River provides a large amount of clean energy, which is an important base of “power transmission from west to east” and an important guarantee of national energy strategy. The construction of reservoirs extended the “golden waterway” of the mainstream of the upper reaches of the Yangtze River, forms a wide lake, greatly improves the natural river channel condition of the rapids, and can be used for navigation of thousand-tonnage ships all year round. This waterway promoted the integration of the East and the West of the Yangtze River, built a new economic support belt of China based on the Yangtze River. These effects are generally nationally planned.
(10) “Conclusions should include policy implications.”
Response: Thank you for your advice and the conclusions have been revised with policy implications.
Line 456~459:
(1) The construction of cascade reservoirs on the JRB reduced the average daily streamflow by 5.69% from 4937.97 m3/s to 4657.10 m3/s. During the rainy and dry seasons, average daily streamflow decreased by 10.93% and 10.95%, respectively. The cumulative storage capacity and quantity of the reservoirs affect the change degree of runoff. On the whole, with the increase of the cumulative storage capacity, the effect of the reservoirs on runoff increases, with a non-linear relationship.
(2) The cascade reservoirs on the JRB reduced the flood POT from an average of 2.6 to 2.4 per year, reducing the frequency of floods along the Jinsha River by 7.69%. In terms of flood magnitude, the construction of reservoirs reduced the average annual runoff exceeding the threshold from 5.25×105 m3 to 3.26×105 m3, reduced by 37.86%. In other words, cascade reservoirs in the JRB reduce the frequency and magnitude of flood events.
(3) The effect of reservoir construction on drought duration is mainly reflected in the decreasing effect on the duration of extreme drought and increasing effect on the duration of moderate and severe drought on all time scales. The effect of reservoirs on drought severity is mainly reflected in the mitigation of the severity of extreme drought and aggravation of the severity of moderate and severe drought. The reservoirs have a mitigating effect on the severity of mild drought on the SRI-1 and SRI-6 timescales and an intensifying effect on the SRI-3 and SRI-6 timescales.
(4) It is suggested that in the JRB, when the Central Yunnan Water Diversion Project and the South to North Water Transfer West Line Project are com-pleted, the water transfer volume and time should be fully considered to coordinate with the storage and operation of the cascade reservoirs.
Author Response File: Author Response.pdf
Reviewer 3 Report
The Jinsha River Basin (JRB) is the largest hydropower base in China and the main water source for the South-to-North Water Diversion West Project. This article aims to elucidate the effects of reservoirs in the Jinsha River basin on runoff, floods and droughts. It has certain guiding significance for regional development, and the selected topic has certain theoretical value and practical significance. However, the following improvements are needed：
1. Lines 136-137，In the established model, meteorological and subsurface data are entered into the model. Whether the assessment of human activities on the runoff of the basin is taken into account.
2.Figure 5 shows that the simulation results in low value period are good, but the simulation results in high value period, especially in peak period, are poor. Whether such results will have an impact on the study of flood conditions.
3.Whether the construction of the reservoir will have a stable impact on the long-term series of runoff
4.Judging from the results of the article, whether the model is good for flood simulation
5.The position of the images in the article should be consistent, all on the left side of the page, or all in the middle of the page.
6. So far, there are many researches in this field, and this research needs to highlight the innovation point.
7. The language of the paper needs to be polished by native English speakers.
8. The introduction part of the paper needs to be adjusted logically, and the different paragraphs need to be logically connected.
9. Figure 1, the elevation color logo is wrong, suggest to fix (the more red color means the higher elevation)
10. In order to increase the reliability of the paper, it is necessary to add a discussion on the uncertainty of the model simulation results.
Author Response
Dear Reviewer #3，
Thank you for the evaluation of our manuscript. We have taken all your comments into consideration and tried to address each point in the revised manuscript.
Response to Reviewer #3:
- Comments from the Reviewer #3:
(1) “Lines 136-137，In the established model, meteorological and subsurface data are entered into the model. Whether the assessment of human activities on the runoff of the basin is taken into account.”
Response: Thank you for your question. There are few large cities along the Jinsha River and large-scale water diversion projects such as the Central Yunnan Water Diversion Project and the Western Route of the South-to-North Water Diversion Project are under planning and construction, there is no large-scale water diversion project yet. Moreover, it is difficult for us to find relevant data about human activities at present, so human activities have not been taken into account. We have brought this point into the discussion.
(2) “Figure 5 shows that the simulation results in low value period are good, but the simulation results in high value period, especially in peak period, are poor. Whether such results will have an impact on the study of flood conditions.”
Response: Thank you for your comments. We understand your doubts. High value simulation is a common difficulty in hydrological simulation. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. If longer time series runoff data can be obtained in the future, the performance of the model will be better. Compared with other models, the LSTM model is improved in high value simulation. In my opinion, this result can be used for flood analysis. In the discussion, we added the comparison of the accuracy with other models in the Jinsha River basin, as well as the analysis of model uncertainty. Sincerely thank you for your question.
Line 373~376, 387~394:
In this study, the performance of LSTM model is compared with that of other physical process models in the daily scale flow simulation of the Jinsha River basin. As shown in Table 3, the performance of the LSTM model is higher than that of physical process models [6, 38, 39, 40, 41]. The performance improvement can be attributed to the capability of the LSTM model to control input data flexibility based on watershed characteristics [4]. For example, the JRB originates from the Qinghai Tibet Plateau, where snowmelt affects runoff generation. Accordingly, snowmelt data can be incorporated into the input matrix. As the main principle of the LSTM model is to establish an optimal mapping of driving forces and runoff instead of causation, more diverse data can be introduced, and the construction and calibration of the model are relatively simple [42]. Meanwhile, physical process models generalize the process of precipitation-runoff through mathematical and physical methods [43]. Consequently, physical process models have a fixed form, with high data requirements [44]. Moreover, the construction and calibration of physical process models are complex, and the flexibility of these models requires extensive verification. However, in the current hydrological model research, the poor simulation effect of high value is a common problem. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation.
Table 3 Performances of the LSTM and physical process models reported in the literature on daily streamflow forecasting.
|
Model |
Testing period |
NSE |
Source |
|
LSTM |
1994–1998 |
0.92 |
This study |
|
SWAT (Semi-distributed Model) |
2000–2016 |
0.90 |
Chen et al. (2020) |
|
SWAT (Semi-distributed Model) |
2012–2012 |
0.71 |
Wu et al. (2020) |
|
Xinanjiang model (Distributed model) |
1986–2000 |
0.84 |
Feng et al. (2018) |
|
VIC model (Distributed model) |
2004–2006 |
0.72 |
Maza et al. (2020) |
|
MIKE 11 NAM model (Distributed model) |
2009–2015 |
0.83 |
Aredo et al. (2021) |
(3) “Whether the construction of the reservoir will have a stable impact on the long-term series of runoff”
Response: In the long run, the operation of reservoirs will have a stable influence on the long-term runoff series, but the impact of special incoming water and regulation changes will not be excluded.
(4) “Judging from the results of the article, whether the model is good for flood simulation”
Response: Thank you for your comments. We understand your doubts. High value simulation is a common difficulty in hydrological simulation. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation. In my opinion, this result can be used for flood analysis. In the discussion, we added the comparison of the accuracy with other models in the Jinsha River basin, as well as the analysis of model uncertainty. Sincerely thank you for your question.
Line 373~376, 387~394:
In this study, the performance of LSTM model is compared with that of other physical process models in the daily scale flow simulation of the Jinsha River basin. As shown in Table 3, the performance of the LSTM model is higher than that of physical process models [6, 38, 39, 40, 41]. The performance improvement can be attributed to the capability of the LSTM model to control input data flexibility based on watershed characteristics [4]. For example, the JRB originates from the Qinghai Tibet Plateau, where snowmelt affects runoff generation. Accordingly, snowmelt data can be incorporated into the input matrix. As the main principle of the LSTM model is to establish an optimal mapping of driving forces and runoff instead of causation, more diverse data can be introduced, and the construction and calibration of the model are relatively simple [42]. Meanwhile, physical process models generalize the process of precipitation-runoff through mathematical and physical methods [43]. Consequently, physical process models have a fixed form, with high data requirements [44]. Moreover, the construction and calibration of physical process models are complex, and the flexibility of these models requires extensive verification. However, in the current hydrological model research, the poor simulation effect of high value is a common problem. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation.
Table 3 Performances of the LSTM and physical process models reported in the literature on daily streamflow forecasting.
|
Model |
Testing period |
NSE |
Source |
|
LSTM |
1994–1998 |
0.92 |
This study |
|
SWAT (Semi-distributed Model) |
2000–2016 |
0.90 |
Chen et al. (2020) |
|
SWAT (Semi-distributed Model) |
2012–2012 |
0.71 |
Wu et al. (2020) |
|
Xinanjiang model (Distributed model) |
1986–2000 |
0.84 |
Feng et al. (2018) |
|
VIC model (Distributed model) |
2004–2006 |
0.72 |
Maza et al. (2020) |
|
MIKE 11 NAM model (Distributed model) |
2009–2015 |
0.83 |
Aredo et al. (2021) |
(5) “The position of the images in the article should be consistent, all on the left side of the page, or all in the middle of the page.”
Response: Thank you very much for your suggestion. We have unified all the picture positions in the article into the middle of the page.
(6) “So far, there are many researches in this field, and this research needs to highlight the innovation point.”
Thank you for your guidance. The introduction has been revised to highlight the innovation of this paper. As the upper reaches of China’ largest river(the Yangtze River), the Jinsha River are an important supplier and determinant of water resources in the downstream urban agglomeration in the middle and lower reaches of the Yangtze River, has implications for the economic stability and regional security of southern China. The Jinsha River is also the main water source of the West Route of China's South-to-North Water Diversion Project, and the rational allocation of water re-sources from the Jinsha River is important for the successful implementation of this project. We propose a hydrological model to construct different scenarios in the basin, obtain the impact of power plants on runoff excluding the influence of climate change (by controlling input variables), and analyze the impact of power plants construction on different periods. This method can be used as a reference for other hydrological models of hydropower stations.
Line75~77:
The objectives of this study were to: i) build a hydrological model suitable for the JRB based on neural networks; ii) quantify the influence of cascade reservoirs on streamflow in the JRB; and iii) analyze the effect of reservoirs on streamflow, flood, and drought events in the JRB. The innovation point is that this paper analyzes the influence of power stations at different times by constructing two scenarios: power station and power station-free. The results of our study could provide technical support for the construction of hydrological models in regions with cascade reservoirs, reference for the construction and operation of hydropower stations in the JRB, data for the West Route of the South-to-North Water Diversion project, and research directions for water security in the Yangtze River basin.
(7) “The language of the paper needs to be polished by native English speakers.”
Response: Thank you for your suggestions. The paper has been polished by professional institutions.
(8) “The introduction part of the paper needs to be adjusted logically, and the different paragraphs need to be logically connected.”
Response: Thank you for your suggestions. he introduction has been adjusted and connected. (Line 47-48)
Line 48:
Therefore, changes in the hydrological regime of the JRB caused by the construction of hydropower stations have received extensive attention, a method is urgently needed to quantify these changes.
(9) “Figure 1, the elevation color logo is wrong, suggest to fix (the more red color means the higher elevation)”
Response: Thank you for pointing this out. We have modified Figure 1.
Line 99:
Figure 1. Hydropower plants and Pingshan hydrological station in the Jinsha River Basin.
(10) “In order to increase the reliability of the paper, it is necessary to add a discussion on the uncertainty of the model simulation results.”
Response: Your recommendations are very good and we have included a model uncertainty analysis in the discussion. The uncertainty of LSTM model in this study is mainly reflected in the uncertainty of input data. The learning performance of LSTM model is closely related to the quality of data and the size of data samples. Increasing the number of effective training samples can effectively improve the simulation accuracy of LSTM model, especially increasing the number of flood events in training samples can improve the ability of the model to find flood peaks and significantly improve the accuracy of peak simulation.
Line 373~376, 387~394:
In this study, the performance of LSTM model is compared with that of other physical process models in the daily scale flow simulation of the Jinsha River basin. As shown in Table 3, the performance of the LSTM model is higher than that of physical process models [6, 38, 39, 40, 41]. The performance improvement can be attributed to the capability of the LSTM model to control input data flexibility based on watershed characteristics [4]. For example, the JRB originates from the Qinghai Tibet Plateau, where snowmelt affects runoff generation. Accordingly, snowmelt data can be incorporated into the input matrix. As the main principle of the LSTM model is to establish an optimal mapping of driving forces and runoff instead of causation, more diverse data can be introduced, and the construction and calibration of the model are relatively simple [42]. Meanwhile, physical process models generalize the process of precipitation-runoff through mathematical and physical methods [43]. Consequently, physical process models have a fixed form, with high data requirements [44]. Moreover, the construction and calibration of physical process models are complex, and the flexibility of these models requires extensive verification. However, in the current hydrological model research, the poor simulation effect of high value is a common problem. On one hand, due to the small number of high values in training samples, it is difficult for neural network to find accurate high value mapping relationship; On the other hand, in the process of simulation, we choose the best fitting effect between over-fitting and under-fitting. Compared with other models, the LSTM model is improved in high value simulation.
Table 3 Performances of the LSTM and physical process models reported in the literature on daily streamflow forecasting.
|
Model |
Testing period |
NSE |
Source |
|
LSTM |
1994–1998 |
0.92 |
This study |
|
SWAT (Semi-distributed Model) |
2000–2016 |
0.90 |
Chen et al. (2020) |
|
SWAT (Semi-distributed Model) |
2012–2012 |
0.71 |
Wu et al. (2020) |
|
Xinanjiang model (Distributed model) |
1986–2000 |
0.84 |
Feng et al. (2018) |
|
VIC model (Distributed model) |
2004–2006 |
0.72 |
Maza et al. (2020) |
|
MIKE 11 NAM model (Distributed model) |
2009–2015 |
0.83 |
Aredo et al. (2021) |
Author Response File: Author Response.pdf
Reviewer 4 Report
The authors studied the impact of cascade reservoirs on streamflow, drought, and flood in the Jinsha River Basin, China, using a framework coupling long short-term memory (LSTM) and flood drought assessment techniques. The authors also built a hydrological model suitable for the JRB based on a neural network. The overall study is very interesting and helpful for water resouce manager and policy maker. I have only a few minor comments:
In Table 1, please re-check the spatial resolution of CHIRPS satellite products, If I am not wrong its spatial resolution is 0.05 degrees (about 5 km)
For equations 7-10, please use the same font size as you used in the text instead of large size font.
In Figure 5ab graphs, in legend, you represented dammed flow by a dashed black line. it should be simulated flow as dams were built after 1998.
Author Response
Dear Reviewer #4，
Thank you for the evaluation of our manuscript. We have taken all your comments into consideration and tried to address each point in the revised manuscript.
Response to Reviewer #4:
- Comments from the Reviewer #4:
(1) “In Table 1, please re-check the spatial resolution of CHIRPS satellite products, If I am not wrong its spatial resolution is 0.05 degrees (about 5 km)”
Response: Thank you for pointing this out. We have checked the spatial resolution of CHIRPS satellite data, Chirps 2.0 has two resolution data for download, one is 0.05 degrees, the other is 0.25 degrees, the latter is used in this paper.
(2) “For equations 7-10, please use the same font size as you used in the text instead of large size font.”
Response: Thank you very much for your reminder. We have modified Formula 7-10.
(3) “In Figure 5ab graphs, in legend, you represented dammed flow by a dashed black line. it should be simulated flow as dams were built after 1998.”
Response: Thank you for pointing this out. We have corrected it.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Dear Authors,
Thank you for your corrections. I have read all the comments you have addressed, and they are reasonable and valid.
Best regards
Reviewer 3 Report
Moderate English changes required
