A Study on Climate-Driven Flash Flood Risks in the Boise River Watershed, Idaho
Round 1
Reviewer 1 Report
General recommendation
This manuscript explores the future projections of flash flood hazard. Authors use climate model projections in Boise River Watershed and force a hydrologic model. They consequently assess future streamflow. The research question is very interesting and is of interest of the hydrologic community and research in this area is much needed. However, more attention needs to be paid to the manuscript. My main concerns are that (1) Deterministic values of future streamflow are used for conclusion while addressing the changes or trends are of interest of projections. (2) The performance measures for validating the method is very weak. Since these two issues could change the results and conclusions, I do not recommend publishing this manuscript unless the authors address the following concerns.
Main concerns:
1. I do not understand why authors keep saying “flood risk”. Risk is defined as the probability of a bad outcome times its consequences. I do not consider this study as a risk analysis. This is a future exploration of flash flood hazard.
2. The results and conclusion of the paper could be improved significantly. More explanations/figures/tables are needed in the results section. For example, while the study is about future flash floods, I do not see even one figure showing the time series of future flash floods. Or the ensemble of projections. Please add those figures for all the stations showing all the ensemble members.
We cannot take just one value from one member of the ensemble from future projections of streamflow and use that as a deterministic prediction and base the conclusion of the paper on those numbers. Climate models are not designed for that. You need to analyze the changes to the spread of the ensemble. Or the trends of future flash floods. Picking one model and treating it as a deterministic prediction (as in table 5) does not make sense to me.
3. A simple correlation coefficient is not acceptable. You need to have two deterministic and two probabilistic measures. Please add this.
4. Line 93: I am not convinced why you considered flash floods independent. I do not think 3 days is enough to consider independence
5. Future changes in management, population, land cover and etc. are not considered in hydrologic modeling. Please discuss this in the discussions.
6. All of a sudden in line 207 (results section), authors talk about Monte Carlo simulations. This is not appropriate. If you conduct MCMC simulations, you need to explain it in the methods and show more results on it.
7. I always encourage open source research. Please indicate if/how you are planning to make data/code accessible to your readers?
Minor concerns:
The hydrologic model used is uncommon. Please provide more up-to-date citations if this model if it is currently being used.
Since there is flood control infrastructure (like dams) upstream of the USGS gages, are you considering naturalizing the streamflow observations? Please discuss it in the discussions.
Line 24: Add multiple citations
Line 28: add citation
Line 31: add citations
Line 44: Please cite scientific peer reviewed papers
Line 65-66: citation?
Figure 2: there are no x labels. Would be better to add the station number in the legend.
Line 89-90: cite papers who have used this method, if any.
Line 92: revise, please
Figure 3: While this plot looks good, I don’t understand why it is preferred to a simple bar chart. A bar chart takes less time to interpret for the reader and increases the readability of your paper. I recommend replacing this with a simple bar chart.
Line 142: revise
Line 142: provide citations for the model
Line 146,149: grammatical issue
Line 165: mention what criteria and give some explanation. It is an important part and should be explained briefly in the paper.
Line 168-170: I do not see a reason for underlining these lines.
Figures 4,5,6 do not have captions. They have low resolution. Increase the resolution. How about confidence intervals of other distributions than GEV? They are not included in the figures.
There is no table 3
Line 224: do you mean 4.5 and 8.5?
Thank you. I like the idea of the paper and it is well-written. I hope the authors can address these concerns and publish this paper as I believe more research on this is needed in the community.
Author Response
General recommendation
This manuscript explores the future projections of flash flood hazard. Authors use climate model projections in Boise River Watershed and force a hydrologic model. They consequently assess future streamflow. The research question is very interesting and is of interest of the hydrologic community and research in this area is much needed. However, more attention needs to be paid to the manuscript. My main concerns are that (1) Deterministic values of future streamflow are used for conclusion while addressing the changes or trends are of interest of projections. (2) The performance measures for validating the method is very weak. Since these two issues could change the results and conclusions, I do not recommend publishing this manuscript unless the authors address the following concerns.
Main concerns:
1. I do not understand why authors keep saying “flood risk”. Risk is defined as the probability of a bad outcome times its consequences. I do not consider this study as a risk analysis. This is a future exploration of flash flood hazard.
>>As a statistical term, yes, you are right. But, the term “flood risk” is acceptable broadly in engineering sense.
2. The results and conclusion of the paper could be improved significantly. More explanations/figures/tables are needed in the results section. For example, while the study is about future flash floods, I do not see even one figure showing the time series of future flash floods. Or the ensemble of projections. Please add those figures for all the stations showing all the ensemble members.
We cannot take just one value from one member of the ensemble from future projections of streamflow and use that as a deterministic prediction and base the conclusion of the paper on those numbers. Climate models are not designed for that. You need to analyze the changes to the spread of the ensemble. Or the trends of future flash floods. Picking one model and treating it as a deterministic prediction (as in table 5) does not make sense to me.
>>As per the reviewer’s suggestion, the manuscript is now revised significantly along with the added figures (Fig 7-9). Please see the revised manuscript in details.
2. A simple correlation coefficient is not acceptable. You need to have two deterministic and two probabilistic measures. Please add this.
>>We added the Kolmogorov-Smirnov statistic
3. Line 93: I am not convinced why you considered flash floods independent. I do not think 3 days is enough to consider independence
>>the 3-day peak flow at a given year is independent. Note that we used the annual 3-day running total flow (3D flow) for this analysis. So, we can say that 3D flow in 1982 and 3D flow in 1983 is independent.
4. Future changes in management, population, land cover and etc. are not considered in hydrologic modeling. Please discuss this in the discussions.
>>yes, we have not incorporated future land use change into this research. But, our previous publication addresses this issue. We added this in the revised manuscript. Thanks.
5. All of a sudden in line 207 (results section), authors talk about Monte Carlo simulations. This is not appropriate. If you conduct MCMC simulations, you need to explain it in the methods and show more results on it.
>>We added a table to show the result. Thank you.
6. I always encourage open source research. Please indicate if/how you are planning to make data/code accessible to your readers?
>>Table 1 would be useful for the reader to draw the same result. Regarding R code, we are happy to share with the reader as per request. Since we are not programmer, R code looks ugly. It is not fancy enough to put in the public domain. Please understand it, but you can request a code when you need after this article is published.
Minor concerns:
The hydrologic model used is uncommon. Please provide more up-to-date citations if this model if it is currently being used.
Since there is flood control infrastructure (like dams) upstream of the USGS gages, are you considering naturalizing the streamflow observations? Please discuss it in the discussions.
Line 24: Add multiple citations
>>Done. Thank you.
Line 28: add citation
>>Done. Thank you.
Line 31: add citations
>>Done. Thank you.
Line 44: Please cite scientific peer reviewed papers
>>Done. Thank you.
Line 65-66: citation?
>>Done. Thank you.
Figure 2: there are no x labels. Would be better to add the station number in the legend.
>>X labels are now added. Thank you.
Line 89-90: cite papers who have used this method, if any.
>>We thought that this approach is reasonable based on our engineering judgments.
Line 92: revise, please
>>The sentence and context is understandable. We would appreciate if the reviewer suggests any wording.
Figure 3: While this plot looks good, I don’t understand why it is preferred to a simple bar chart. A bar chart takes less time to interpret for the reader and increases the readability of your paper. I recommend replacing this with a simple bar chart.
>>we also thought about a simple bar chart, but all charts are too cluster together to analyze it. So, we decided to keep the polar chart as it is. Thanks for understanding.
Line 142: revise
Line 142: provide citations for the model
>>Done. Thank you.
Line 146,149: grammatical issue
>>Revised. Thank you.
Line 165: mention what criteria and give some explanation. It is an important part and should be explained briefly in the paper.
>>Thanks. Now, it is revised.
Line 168-170: I do not see a reason for underlining these lines.
>>Sorry. It is an editorial error.
Figures 4,5,6 do not have captions. They have low resolution. Increase the resolution. How about confidence intervals of other distributions than GEV? They are not included in the figures.
>>Now, it is fixed along with figures in higher resolution. Hopefully, the editorial office can make these image look better during in-press stage. Since confidence intervals of other distributions turn out a very similar, we didn’t include others. In general, all three distribution functions, including GEV, Pearson, and 3-param log normal perform very well.
There is no table 3
>>This is an editorial error by the publisher. Now, table 3 is imported from the original manuscript submitted to the publisher.
Line 224: do you mean 4.5 and 8.5?
>>yes, it is now fixed. Thank you.
Thank you. I like the idea of the paper and it is well-written. I hope the authors can address these concerns and publish this paper as I believe more research on this is needed in the community.
Reviewer 2 Report
I think the research results are important for practical applications. I can recommend several changes:
1- My main concern is that I could not identify what is the novelty of the proposed approach
2- There are numerous acronyms in the document. A single table with the acronyms listed would help with readability
3- The main quantitative results of the research should be added in Abstract and Conclusion Sections. Authors can give concrete results illustrating the novelty of research results
4- Lines 25-27: As global warming.....(BRW); please add a reference
5- It is recommended to change the table1 to a graph (x: date; y1:OBS1, y2:OBS2 and y3: OBS3)
6- The R2 of figure 5 is more than 0.98??? please add the equation and R2 to the figures
7- Line 230 – 232 & Table 5. This implies….variability: so what is your suggestion to reduce the uncertainties.
Author Response
Comments and Suggestions for Authors
I think the research results are important for practical applications. I can recommend several changes:
1- My main concern is that I could not identify what is the novelty of the proposed approach
>>Although there is a similar research project available somewhere, a flood frequency analysis at a different watershed (e.g., the Boise River Watershed) will build another case to benefit the hydro community in a changing climate.
2- There are numerous acronyms in the document. A single table with the acronyms listed would help with readability.
>>We are not sure that the editorial department allows us to add the acronym table. But, we spelled out when the acronyms appear at the first time.
3- The main quantitative results of the research should be added in Abstract and Conclusion Sections. Authors can give concrete results illustrating the novelty of research results
>>This is relevant to Question 1. Hope we answered the reviewer’s question properly.
4- Lines 25-27: As global warming.....(BRW); please add a reference
>>The reference is now added as per the reviewer’s request. Thank you.
5- It is recommended to change the table1 to a graph (x: date; y1:OBS1, y2:OBS2 and y3: OBS3)
>>The date in Table 1 is useful for the reader to apply a similar approach suggested by the authors. So, we decided to keep this table as it is. This question is also relevant to the question by the reviewer 1. Thanks for understanding.
6- The R2 of figure 5 is more than 0.98??? please add the equation and R2 to the figures
>>R2 is a typical statistic exercise. We added another matrix as the other reviewer suggests. We hope multiple measurement matrix would be good enough to convince the reader to understand our results.
7- Line 230 – 232 & Table 5. This implies….variability: so what is your suggestion to reduce the uncertainties.
>>We put some efforts to answer the reviewer’s question. Please see the revised manuscript.
Submission Date
15 March 2019
Reviewer 3 Report
This study investigates future changes in flood risk in Boise River Watershed, Idaho. My major concern is on the use of one hydrological model for future flood projections. It is well known that there are large uncertainties from hydrological models, especially in simulating future hydrological conditions. I suggest authors discuss this in detail in the revision. In addition, the non-stationarity of floods should be considered especially within the context of changing environment. Below are some references that should be included when addressing the above deficiencies of this study.
Zhao F., et al 2017. The critical role of the routing scheme in simulating peak river discharge in global hydrological models, Environmental Research Letters, 12 075003
Liu S., et al., 2019. Identification of the non-stationarity of floods: Changing patterns, causes, and implications. Water Resources Management 33(3), 939-953. DOI: 10.1007/s11269-018-2150-y.
Author Response
This study investigates future changes in flood risk in Boise River Watershed, Idaho. My major concern is on the use of one hydrological model for future flood projections. It is well known that there are large uncertainties from hydrological models, especially in simulating future hydrological conditions. I suggest authors discuss this in detail in the revision. In addition, the non-stationarity of floods should be considered especially within the context of changing environment. Below are some references that should be included when addressing the above deficiencies of this study.
Zhao F., et al 2017. The critical role of the routing scheme in simulating peak river discharge in global hydrological models, Environmental Research Letters, 12 075003
Liu S., et al., 2019. Identification of the non-stationarity of floods: Changing patterns, causes, and implications. Water Resources Management 33(3), 939-953. DOI: 10.1007/s11269-018-2150-y.
>>As per the reviewer’s request, we added the references above and revised the manuscript accordingly. Thanks for your time to review our manuscript.
