the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Future changes in flash flood frequency and magnitude over the European Alps
Abstract. Flash Floods are damaging natural hazards which often occur in the European Alps. Precipitation patterns and intensity may change in a future climate affecting their occurrence and magnitude. For impact studies, flash floods can be difficult to simulate due the complex orography and limited extent & duration of the heavy rainfall events which trigger them. The new generation convection-permitting regional climate models (CP-RCMs) improve the representation of the intensity and frequency of heavy precipitation. Therefore, this study combines such simulations with high-resolution distributed hydrological modelling to assess changes in flash flood frequency over the Alpine domain. We use output from a state-of-the-art CP-RCM to drive a high-resolution distributed hydrological wflow_sbm model covering most of the Alpine mountain range on an hourly resolution. First, the hydrological model was validated by comparing ERA5 driven simulation with streamflow observations from 130 stations (across Rhone, Rhine, Po, Adige and Danube basins). Second, a hourly wflow_sbm simulation driven by a CP-RCM downscaled ERAInterim simulation was compared to databases of past flood events to evaluate if the model can accurately simulate flash floods and to determine a suitable threshold definition for flash flooding. Finally, simulations of the future climate RCP 8.5 for the end-of-century (2096–2105) and current climate (1998–2007) are compared for which the CPRCM is driven by a Global Climate Model. The simulations are compared to assess if there are changes in flash flood frequency and magnitude using a threshold approach. Results show a similar flash flood frequency for autumn in the future, but a decrease in summer. However, the future climate simulations indicate an increase in the flash flood severity in both summer and autumn leading to more severe flash flood impacts.
- Preprint
(8309 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
CC1: 'Comment on hess-2022-207', Jie Chen, 04 Oct 2022
This manuscript combines simulations from new generation convection-permitting regional climate models with high-resolution distributed hydrological modelling to assess changes in flash flood frequency over the Alpine domain. Generally, the manuscript is well written. However, there are some issues need to be addressed before it can be accepted for publication.
- The Abstract spent lots of spaces to describe the methodology, while the methodology is not new. There were only two sentences of results and no conclusion. The authors should put more weights on results.
- The CP-RCM was driven by ERA-Interim simulation. Why not use the new generation of ERA5.
- The authors mentions that “the application of CP-RCMs in hydrological studies is novel.” It may be true that the CP-RCM was not used with hydrological models for hydrological studies. However, outputs of RCM has been used with hydrological models for many studies. What are the differences of this study compared to other studies?
- The study region has snow and glacial melt. How does the hydrological model simulate snowmelt and glacial melt? This should be clarified.
- Is there any calibration for the hydrological model? This should be clarified in the manuscript.
Citation: https://doi.org/10.5194/hess-2022-207-CC1 -
AC1: 'Reply on CC1', Marjanne Zander, 10 Oct 2022
Dear dr Jie Chen,
Thank you for your time and the feedback on the manuscript.
In reply to the comments:
- The abstract will be adjusted to put more emphasis on the results and conclusions.
- The choice of using ERA-Interim as boundary conditions was made by the climate modelling partner, as the ERA5 reanalysis product was not yet available at the start of the CP-RCM simulation experiment.
- In the revised version of the manuscript more attention will be paid to the differences between the current work with convection-permitting RCMs compared to previous work using high-resolution RCMs which are not convection-permitting e.g., the EURO-CORDEX RCMs (Di Sante et al., 2021):
- Especially in the summer and autumn seasons, when convection is a key process in rainfall generation, the 2.2 km resolution CP-RCM simulation employed in this study improves the peak hourly rainfall intensity, as well as the diurnal cycle of rainfall when compared to its courser 12km resolution RCM counterpart. The representation of intense rainfall events in the Mediterranean in autumn is thus improved (Berthou et al., 2018; Ban et al, 2021). These rainfall events are amongst the types of weather which can trigger flash floods, which alleviates caveats from previous studies that short duration intense rainfall events were not well represented in the modelling. Compared to previous hydrological modelling studies, which investigated riverine flooding, e.g., Di Sante et al., 2021, different research questions can thus be asked, delving into very local occurrence of flash flooding resulting from extreme rainfall events.
- Although this work focusses on the summer (JJA) and autumn (SON) seasons, in which snow and glacier melt dynamics play a lesser role than in (late) spring, we agree that the hydrological model description can be extended with a brief overview of the way in which we simulate snowmelt and glacier melt in wflow_sbm (using degree-day models), which is indeed not covered in the current version of the manuscript.
- In our study only one modelling parameter is studied for calibration. The parameter estimation in wflow_sbm is based on available spatial datasets providing information on soil properties, soil depth, rooting depth etc. Imhoff et al. (2021) developed a method for parameterization of the model using regionalization methods based on (pedo)transfer functions from literature and upscaling techniques to ensure fluxmatching. Only for the Horizontal Hydraulic Conductivity (“KsatHorFrac” multiplied with a priori estimated KSatVer is used as horizontal conductivity) there is no pedo-transfer function yet, while this is a rather sensitive parameter in the model. For a sensitivity analysis, the model was run changing the KsatHorFrac parameter with homogeneous values for the Rhine basin. The model was run using the ERA5 reanalysis as meteorological forcing and the performance was evaluated against two discharge observations, one at the catchment outlet (the Rhine at Basel), one for a subcatchment (the Thur river at Andelfingen). This is included as a supplement to the paper. The text will be clarified on this point.
On behalf of the authors,
Yours sincerely,
M.J. Zander
References:
Nikolina Ban, et al., “The first multi-model ensemble of regional climate simulations at kilometer-scale resolution, part I: evaluation of precipitation.” Climate Dynamics, pages 1–28, 2021.https://doi.org/10.1007/s00382-021-05708-w
Ségolène Berthou, Elizabeth J Kendon, Steven C Chan, Nikolina Ban, David Leutwyler, Christoph Schär, and Giorgia Fosser. “Pan-European climate at convection-permitting scale: a model intercomparison study”. Climate Dynamics, pages 1–25, 2018.https://doi.org/10.1007/s00382-018-4114-6
Ruben O. Imhoff, W.J. van Verseveld, B. van Osnabrugge, and A.H. Weerts. “Scaling point-scale (pedo)transfer functions to seamless large-domain parameter estimates for high-resolution distributed hydrologic modeling: An example for the Rhine river”. Water Resources Research, 56(4), April 2020. https://doi.org/10.1029/2019WR026807
Fabio Di Sante, Erika Coppola, and Filippo Giorgi. “Projections of river floods in Europe using EURO-CORDEX”, cmip5 and cmip6 simulations. International Journal of Climatology, 41(5):3203–3221, 2021. https://doi.org/10.1002/joc.7014
Citation: https://doi.org/10.5194/hess-2022-207-AC1
-
RC1: 'Review of hess-2022-207', Anonymous Referee #1, 17 Oct 2022
This very well written manuscript proposes a European scale analysis of flash flood frequency combining downscaled ERA5interim data with the distributed hydrological model wflow. The method and results are overall well described; there is however a considerable lack of reference to actual hydrological processes involved in flash flood generation and what this means in terms of requirements for the modelling chain. I cannot recommend publication of this manuscript. It has the merit to attempt a regional-scale analysis of an important phenomenon. But the modelling exercise is detached from flood generating mechanisms. I anticipate that revising the work would require a profound modification of the modeling chain.
As becomes clear below, I do not see evidence that the modelling framework is fit to model flash flood frequency. It rather seems like the model applies some filtering to the precipitation events but we do not know how realistically this generates flash floods. Accordingly, the added value of the hydrological model as compared to frequency analysis on the precipitation directly is not convincingly demonstrated or discussed. This problem is enhanced by the fact that according to the text (methods section), only 9 flash floods are contained in the used data base for the entire region and study period. For these floods, the model validation is summarized as: “For all recorded flash floods, the modelled peak daily discharge was heightened compared to the preceding and following days.” (Section 3.3). This seems a weak argument. The regional approach to flash flood validation (an observed event is validated if something happened within the three days in one of the 5 large subbasins, Fig. 1) is also not convincing: in as far does this demonstrate that the hydrological model has any added value compared to the precipitation input?
Furthermore I do not understand how the flash flood producing threshold is validated, i.e. I do not understand what is actually validated (lines 215-220). This threshold is however the key to judge if flash flood frequencies increases or decreases.
Finally: the study does not work with downscaling or bias correction (see discussion: “it was explicitly chosen not to apply a bias correction, downscaling, or a delta change approach to the climate model data as these techniques can disturb the change signal”). I have a hard time to understand how the output of the hydrological model is of any use in this case, in particular because it aggregates model outputs from catchments for which the bias of precip., temperature or their variability might be important but spatially different This problem is not reduced by comparing similar simulations for the reference and the future period since flash flood analysis corresponds to the analysis of extreme events that might be crucially influenced by biases, and differently under current climate compared to future climate.
Detailed comments
- Introduction: I would suggest to add references / arguments for “The intensity of flash floods and thereby their impacts may increase”. Is precipitation intensity the only driver? What is the role of soil moisture, infiltration capacity? Is looking at precipitation intensity sufficient or would you need to consider compound events of high precipitation intensity and low antecedent moisture? I do not mean to suggest that you need to address all this but it would be good to give the bigger picture and to not oversimplify the case of flash floods; we also would need to have information (in the methods part) if the chosen model can reliably reproduce Hortonian and saturation-excess floods and who this is validated for the selected catchments
- Introduction: the sentence “Although Kay et al. (2015) showed that finer resolution CP-RCMs (..) for large-scale river flooding,” is surprising, you just said before that using CP-RCMs for hydrological impact modelling is novel
- Line 56 following: unclear if you talk about your own study or about the study in Norway?
- Introduction: “However, they conclude to finding no clear added value of the CP-RCM simulations due to lacking realism in the temporal distribution of rainfall intensities at a sub-daily scale and/or total precipitation amount per rainfall event (Reszler et al., 2018).” Did you check in your work that the used precip series (without downscaling) have a realistic temporal distribution?
- Study area: a map with mean precip properties (annual totals, intensity of e.g. 1-day precip events) would complete the picture; with the presented information, we have no idea how spatially variable the meteorological drivers are
- Study area: “Most flash floods in the European Alps occur in late summer and autumn”; do you have a reference? later on, “in France xxxx” : these sentences are misleading since this analysis is only about a specific part of France, Italy etc. Do you have detailed information on flash floods in the catchments that you actually include in the study? What about flash floods in the Northern parts (Aare / Rhine catchments); are there many?
- Hydro model: could you perhaps mention the number of model parameters that are estimated / assigned following “same a priori parameter estimation methodology as Imhoff et al.”? which parameters potentially influence flash floods the most? Does the model have Hortonian and saturation-excess surface runoff?
- Can you elaborate on why the vertical hydraulic conductivity parameter is the most sensitive one? And is this result from previous work transferable to here? I guess the sensitivity of wflow parameters depends on what hydrological time scale is dominant for the considered catchment / scale: either processes leading to water mobilization at the hillslope scale or processes of lateral transfer (surface and subsurface) to the stream network or routing within the stream network; if you only assess hydraulic conductivity you implicitly assume that water mobilization is dominant? But what about lateral flow to the stream network?
- What is a “constant multiplication factor of 100”? how does it influence hydraulic conductivity?
- Why did you use ERA5 for the sensitivity study and not downscaled ERAinterim? Should you not test the model sensitivity with the type of input data that you use thereafter? What is the added value of the validation with ERA5? If the model performs better with ERA5, how is this useful for the final aim, generating flash floods under future climates?
- What was the time step of ERA5 combined with wflow? Hourly and daily, this is unclear? What is the value of daily model performance assessment for a model that is later on used to derive flash floods? I miss some convincing arguments that the analysis framework is actually able to reproduce extreme streamflow events of the flash flood type
- Is the quality of the modelling chain with CP-RCM driven by a Global Climate Model assessed in any way? E.g. in terms of the flash flood frequency for present day? Otherwise, how can you justify that this model chain gives valuable results?
- How are lakes treated in the model? Does flash flood analysis below large lakes make any sense?
- The paper would highly benefit from a sketch that summarizes the work scheme (what data went into what model and how the output was assessed / used)? Instead of Table 1, which does not mention to what the model output is compared to
- The focus of the paper is on the European mountain range where snowmelt processes will necessarily play a key role at elevations roughly above 1600 m asl., even in summer and especially in autumn; in summer, flash flood frequency will crucially depend on the saturation level of the catchment, which in turn is conditioned by the snowmelt season (duration, intensity); in autumn in exchange, potential early snowfall with or without subsequent rain-on-snow events can strongly modulate how intense precipitation events translate into flash floods or not
- 2.6: daily maximal specific discharge for flash flood identification: why would the maximum discharge of a daily time step be relevant for small catchments?
- Flash flood definition: I see that only rainfall induced flash floods are considered; are rain-on-snow events never assimilated to flash floods? Given that the analysis expands on the alpine area, this should be made very clear; besides: are you sure that flash floods cannot occur early in spring (the analysis is on summer (JJA) and autumn (SON) period)? Especially in the future?
- Results: the fact that the he annual cycle of discharge with low flows in winter and snowmelt in May to June does not validate the hydrological model; any model will do so as long as there is any form of freezing and snow model in it; given the high annual cycle, we have no idea what the reported KGE values mean (any model that has some annual cycle will have KGE values beyond 0.6 I would guess); how could you validate the flood generation frequency of the modelling chain otherwise?
- Line 211: incomplete sentence (and update all figure references)
- Discussion: I do not understand how you can judge if the frequency of events increases but at the same time admit that it is impossible to estimate return periods (“Just as in Rudd et al. (2020) we do not attempt to estimate return periods of extreme events as the simulation periods are too short to warrant such an approach.”). I guess that answer is in Rudd et al. but it would be good to understand this here also
Citation: https://doi.org/10.5194/hess-2022-207-RC1 - AC3: 'Reply on RC1', Marjanne Zander, 28 Nov 2022
-
RC2: 'Comment on hess-2022-207', Anonymous Referee #2, 31 Oct 2022
This study used CP-RCMs and hydrological model to study the flash flood frequency over the Alpine domain. The topic is interesting. I have some major comments shown below:
- Although this study focuses on the flash flood changes in the future climate, it is necessary to show the changes in annual precipitation and heavy precipitation between future climate and current climate conditions in the study area. These changes can be compared with the flood changes.
- What are the changes in snowmelt dominated flood in the future climate? In a warmer climate, it is expected to have significant changes in snow melt dominated flood.
- What are the differences between the changes in hourly and daily flood frequency and peaks?
- What are the differences in air temperature between the current climate and future climate conditions in the study area?
- The major parameters of the hydrological models should be shown in the manuscript. What are the differences of parameters used in different catchment? Are there any parameters related to snowmelt or glacier melt?
- Figure 2, it is better to add the boundary of catchment on the figure.
- Figure 3, it is better to add precipitation on the figure to compare with the simulated and observed flood.
- It is better to add more descriptions of RCM models, for example, the parameter scheme for convection and precipitation simulation.
Citation: https://doi.org/10.5194/hess-2022-207-RC2 - AC2: 'Reply on RC2', Marjanne Zander, 28 Nov 2022
Status: closed
-
CC1: 'Comment on hess-2022-207', Jie Chen, 04 Oct 2022
This manuscript combines simulations from new generation convection-permitting regional climate models with high-resolution distributed hydrological modelling to assess changes in flash flood frequency over the Alpine domain. Generally, the manuscript is well written. However, there are some issues need to be addressed before it can be accepted for publication.
- The Abstract spent lots of spaces to describe the methodology, while the methodology is not new. There were only two sentences of results and no conclusion. The authors should put more weights on results.
- The CP-RCM was driven by ERA-Interim simulation. Why not use the new generation of ERA5.
- The authors mentions that “the application of CP-RCMs in hydrological studies is novel.” It may be true that the CP-RCM was not used with hydrological models for hydrological studies. However, outputs of RCM has been used with hydrological models for many studies. What are the differences of this study compared to other studies?
- The study region has snow and glacial melt. How does the hydrological model simulate snowmelt and glacial melt? This should be clarified.
- Is there any calibration for the hydrological model? This should be clarified in the manuscript.
Citation: https://doi.org/10.5194/hess-2022-207-CC1 -
AC1: 'Reply on CC1', Marjanne Zander, 10 Oct 2022
Dear dr Jie Chen,
Thank you for your time and the feedback on the manuscript.
In reply to the comments:
- The abstract will be adjusted to put more emphasis on the results and conclusions.
- The choice of using ERA-Interim as boundary conditions was made by the climate modelling partner, as the ERA5 reanalysis product was not yet available at the start of the CP-RCM simulation experiment.
- In the revised version of the manuscript more attention will be paid to the differences between the current work with convection-permitting RCMs compared to previous work using high-resolution RCMs which are not convection-permitting e.g., the EURO-CORDEX RCMs (Di Sante et al., 2021):
- Especially in the summer and autumn seasons, when convection is a key process in rainfall generation, the 2.2 km resolution CP-RCM simulation employed in this study improves the peak hourly rainfall intensity, as well as the diurnal cycle of rainfall when compared to its courser 12km resolution RCM counterpart. The representation of intense rainfall events in the Mediterranean in autumn is thus improved (Berthou et al., 2018; Ban et al, 2021). These rainfall events are amongst the types of weather which can trigger flash floods, which alleviates caveats from previous studies that short duration intense rainfall events were not well represented in the modelling. Compared to previous hydrological modelling studies, which investigated riverine flooding, e.g., Di Sante et al., 2021, different research questions can thus be asked, delving into very local occurrence of flash flooding resulting from extreme rainfall events.
- Although this work focusses on the summer (JJA) and autumn (SON) seasons, in which snow and glacier melt dynamics play a lesser role than in (late) spring, we agree that the hydrological model description can be extended with a brief overview of the way in which we simulate snowmelt and glacier melt in wflow_sbm (using degree-day models), which is indeed not covered in the current version of the manuscript.
- In our study only one modelling parameter is studied for calibration. The parameter estimation in wflow_sbm is based on available spatial datasets providing information on soil properties, soil depth, rooting depth etc. Imhoff et al. (2021) developed a method for parameterization of the model using regionalization methods based on (pedo)transfer functions from literature and upscaling techniques to ensure fluxmatching. Only for the Horizontal Hydraulic Conductivity (“KsatHorFrac” multiplied with a priori estimated KSatVer is used as horizontal conductivity) there is no pedo-transfer function yet, while this is a rather sensitive parameter in the model. For a sensitivity analysis, the model was run changing the KsatHorFrac parameter with homogeneous values for the Rhine basin. The model was run using the ERA5 reanalysis as meteorological forcing and the performance was evaluated against two discharge observations, one at the catchment outlet (the Rhine at Basel), one for a subcatchment (the Thur river at Andelfingen). This is included as a supplement to the paper. The text will be clarified on this point.
On behalf of the authors,
Yours sincerely,
M.J. Zander
References:
Nikolina Ban, et al., “The first multi-model ensemble of regional climate simulations at kilometer-scale resolution, part I: evaluation of precipitation.” Climate Dynamics, pages 1–28, 2021.https://doi.org/10.1007/s00382-021-05708-w
Ségolène Berthou, Elizabeth J Kendon, Steven C Chan, Nikolina Ban, David Leutwyler, Christoph Schär, and Giorgia Fosser. “Pan-European climate at convection-permitting scale: a model intercomparison study”. Climate Dynamics, pages 1–25, 2018.https://doi.org/10.1007/s00382-018-4114-6
Ruben O. Imhoff, W.J. van Verseveld, B. van Osnabrugge, and A.H. Weerts. “Scaling point-scale (pedo)transfer functions to seamless large-domain parameter estimates for high-resolution distributed hydrologic modeling: An example for the Rhine river”. Water Resources Research, 56(4), April 2020. https://doi.org/10.1029/2019WR026807
Fabio Di Sante, Erika Coppola, and Filippo Giorgi. “Projections of river floods in Europe using EURO-CORDEX”, cmip5 and cmip6 simulations. International Journal of Climatology, 41(5):3203–3221, 2021. https://doi.org/10.1002/joc.7014
Citation: https://doi.org/10.5194/hess-2022-207-AC1
-
RC1: 'Review of hess-2022-207', Anonymous Referee #1, 17 Oct 2022
This very well written manuscript proposes a European scale analysis of flash flood frequency combining downscaled ERA5interim data with the distributed hydrological model wflow. The method and results are overall well described; there is however a considerable lack of reference to actual hydrological processes involved in flash flood generation and what this means in terms of requirements for the modelling chain. I cannot recommend publication of this manuscript. It has the merit to attempt a regional-scale analysis of an important phenomenon. But the modelling exercise is detached from flood generating mechanisms. I anticipate that revising the work would require a profound modification of the modeling chain.
As becomes clear below, I do not see evidence that the modelling framework is fit to model flash flood frequency. It rather seems like the model applies some filtering to the precipitation events but we do not know how realistically this generates flash floods. Accordingly, the added value of the hydrological model as compared to frequency analysis on the precipitation directly is not convincingly demonstrated or discussed. This problem is enhanced by the fact that according to the text (methods section), only 9 flash floods are contained in the used data base for the entire region and study period. For these floods, the model validation is summarized as: “For all recorded flash floods, the modelled peak daily discharge was heightened compared to the preceding and following days.” (Section 3.3). This seems a weak argument. The regional approach to flash flood validation (an observed event is validated if something happened within the three days in one of the 5 large subbasins, Fig. 1) is also not convincing: in as far does this demonstrate that the hydrological model has any added value compared to the precipitation input?
Furthermore I do not understand how the flash flood producing threshold is validated, i.e. I do not understand what is actually validated (lines 215-220). This threshold is however the key to judge if flash flood frequencies increases or decreases.
Finally: the study does not work with downscaling or bias correction (see discussion: “it was explicitly chosen not to apply a bias correction, downscaling, or a delta change approach to the climate model data as these techniques can disturb the change signal”). I have a hard time to understand how the output of the hydrological model is of any use in this case, in particular because it aggregates model outputs from catchments for which the bias of precip., temperature or their variability might be important but spatially different This problem is not reduced by comparing similar simulations for the reference and the future period since flash flood analysis corresponds to the analysis of extreme events that might be crucially influenced by biases, and differently under current climate compared to future climate.
Detailed comments
- Introduction: I would suggest to add references / arguments for “The intensity of flash floods and thereby their impacts may increase”. Is precipitation intensity the only driver? What is the role of soil moisture, infiltration capacity? Is looking at precipitation intensity sufficient or would you need to consider compound events of high precipitation intensity and low antecedent moisture? I do not mean to suggest that you need to address all this but it would be good to give the bigger picture and to not oversimplify the case of flash floods; we also would need to have information (in the methods part) if the chosen model can reliably reproduce Hortonian and saturation-excess floods and who this is validated for the selected catchments
- Introduction: the sentence “Although Kay et al. (2015) showed that finer resolution CP-RCMs (..) for large-scale river flooding,” is surprising, you just said before that using CP-RCMs for hydrological impact modelling is novel
- Line 56 following: unclear if you talk about your own study or about the study in Norway?
- Introduction: “However, they conclude to finding no clear added value of the CP-RCM simulations due to lacking realism in the temporal distribution of rainfall intensities at a sub-daily scale and/or total precipitation amount per rainfall event (Reszler et al., 2018).” Did you check in your work that the used precip series (without downscaling) have a realistic temporal distribution?
- Study area: a map with mean precip properties (annual totals, intensity of e.g. 1-day precip events) would complete the picture; with the presented information, we have no idea how spatially variable the meteorological drivers are
- Study area: “Most flash floods in the European Alps occur in late summer and autumn”; do you have a reference? later on, “in France xxxx” : these sentences are misleading since this analysis is only about a specific part of France, Italy etc. Do you have detailed information on flash floods in the catchments that you actually include in the study? What about flash floods in the Northern parts (Aare / Rhine catchments); are there many?
- Hydro model: could you perhaps mention the number of model parameters that are estimated / assigned following “same a priori parameter estimation methodology as Imhoff et al.”? which parameters potentially influence flash floods the most? Does the model have Hortonian and saturation-excess surface runoff?
- Can you elaborate on why the vertical hydraulic conductivity parameter is the most sensitive one? And is this result from previous work transferable to here? I guess the sensitivity of wflow parameters depends on what hydrological time scale is dominant for the considered catchment / scale: either processes leading to water mobilization at the hillslope scale or processes of lateral transfer (surface and subsurface) to the stream network or routing within the stream network; if you only assess hydraulic conductivity you implicitly assume that water mobilization is dominant? But what about lateral flow to the stream network?
- What is a “constant multiplication factor of 100”? how does it influence hydraulic conductivity?
- Why did you use ERA5 for the sensitivity study and not downscaled ERAinterim? Should you not test the model sensitivity with the type of input data that you use thereafter? What is the added value of the validation with ERA5? If the model performs better with ERA5, how is this useful for the final aim, generating flash floods under future climates?
- What was the time step of ERA5 combined with wflow? Hourly and daily, this is unclear? What is the value of daily model performance assessment for a model that is later on used to derive flash floods? I miss some convincing arguments that the analysis framework is actually able to reproduce extreme streamflow events of the flash flood type
- Is the quality of the modelling chain with CP-RCM driven by a Global Climate Model assessed in any way? E.g. in terms of the flash flood frequency for present day? Otherwise, how can you justify that this model chain gives valuable results?
- How are lakes treated in the model? Does flash flood analysis below large lakes make any sense?
- The paper would highly benefit from a sketch that summarizes the work scheme (what data went into what model and how the output was assessed / used)? Instead of Table 1, which does not mention to what the model output is compared to
- The focus of the paper is on the European mountain range where snowmelt processes will necessarily play a key role at elevations roughly above 1600 m asl., even in summer and especially in autumn; in summer, flash flood frequency will crucially depend on the saturation level of the catchment, which in turn is conditioned by the snowmelt season (duration, intensity); in autumn in exchange, potential early snowfall with or without subsequent rain-on-snow events can strongly modulate how intense precipitation events translate into flash floods or not
- 2.6: daily maximal specific discharge for flash flood identification: why would the maximum discharge of a daily time step be relevant for small catchments?
- Flash flood definition: I see that only rainfall induced flash floods are considered; are rain-on-snow events never assimilated to flash floods? Given that the analysis expands on the alpine area, this should be made very clear; besides: are you sure that flash floods cannot occur early in spring (the analysis is on summer (JJA) and autumn (SON) period)? Especially in the future?
- Results: the fact that the he annual cycle of discharge with low flows in winter and snowmelt in May to June does not validate the hydrological model; any model will do so as long as there is any form of freezing and snow model in it; given the high annual cycle, we have no idea what the reported KGE values mean (any model that has some annual cycle will have KGE values beyond 0.6 I would guess); how could you validate the flood generation frequency of the modelling chain otherwise?
- Line 211: incomplete sentence (and update all figure references)
- Discussion: I do not understand how you can judge if the frequency of events increases but at the same time admit that it is impossible to estimate return periods (“Just as in Rudd et al. (2020) we do not attempt to estimate return periods of extreme events as the simulation periods are too short to warrant such an approach.”). I guess that answer is in Rudd et al. but it would be good to understand this here also
Citation: https://doi.org/10.5194/hess-2022-207-RC1 - AC3: 'Reply on RC1', Marjanne Zander, 28 Nov 2022
-
RC2: 'Comment on hess-2022-207', Anonymous Referee #2, 31 Oct 2022
This study used CP-RCMs and hydrological model to study the flash flood frequency over the Alpine domain. The topic is interesting. I have some major comments shown below:
- Although this study focuses on the flash flood changes in the future climate, it is necessary to show the changes in annual precipitation and heavy precipitation between future climate and current climate conditions in the study area. These changes can be compared with the flood changes.
- What are the changes in snowmelt dominated flood in the future climate? In a warmer climate, it is expected to have significant changes in snow melt dominated flood.
- What are the differences between the changes in hourly and daily flood frequency and peaks?
- What are the differences in air temperature between the current climate and future climate conditions in the study area?
- The major parameters of the hydrological models should be shown in the manuscript. What are the differences of parameters used in different catchment? Are there any parameters related to snowmelt or glacier melt?
- Figure 2, it is better to add the boundary of catchment on the figure.
- Figure 3, it is better to add precipitation on the figure to compare with the simulated and observed flood.
- It is better to add more descriptions of RCM models, for example, the parameter scheme for convection and precipitation simulation.
Citation: https://doi.org/10.5194/hess-2022-207-RC2 - AC2: 'Reply on RC2', Marjanne Zander, 28 Nov 2022
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
816 | 336 | 47 | 1,199 | 36 | 35 |
- HTML: 816
- PDF: 336
- XML: 47
- Total: 1,199
- BibTeX: 36
- EndNote: 35
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1