Beyond precipitation: diversity of drivers of high river flows in European near-natural catchments

Lam'barki, Manal; Li, Wantong; O, Sungmin; Zhan, Chunhui; Orth, Rene

doi:https://doi.org/10.5194/hess-2022-404

Preprints

https://doi.org/10.5194/hess-2022-404

Preprints

20 Dec 2022

| 20 Dec 2022

Status: this discussion paper is a preprint. It has been under review for the journal Hydrology and Earth System Sciences (HESS). The manuscript was not accepted for further review after discussion.

Beyond precipitation: diversity of drivers of high river flows in European near-natural catchments

Manal Lam'barki, Wantong Li, Sungmin O, Chunhui Zhan, and Rene Orth

Abstract. High streamflow in rivers can lead to flooding, which may have severe impacts on economy, society and ecosystems. Therefore it is imperative to understand their underlying physical mechanisms. Previous research has illustrated the relevance of several hydrological drivers, such as precipitation, snowmelt and soil moisture. However, the relative importance of these drivers compared with each other is unclear. Moreover, the role of vegetation-related drivers is not well studied. In this study, we focus on high river flows and consider a comprehensive set of potential drivers and analyze their relative importance. This is done with streamflow observations from over 250 near-natural catchments located across Europe during 1984–2007, which are matched with driver data from various observation-based sources. Not surprisingly, we find that precipitation is the most relevant driver of high river flows in most catchments. In addition, and more interestingly, we show that next to precipitation a diversity of other drivers is relevant for high flows, including shallow soil moisture, deep soil moisture, snowmelt, evapotranspiration and leaf area index. These non-precipitation drivers tend to be even more relevant for more extreme high flows. The relative importance of most considered drivers is similar across daily, weekly and monthly time scales. The spatial patterns of the relevance of precipitation, snowmelt and soil moisture for supporting high river flows are controlled by vegetation types and terrain characteristics, while climate and basin area are less important. By analyzing a comprehensive selection of drivers of high river flow in a powerful framework which accounts for co-linearities between drivers, this study advances the understanding of flood generation processes and informs respective model development.

Received: 30 Nov 2022 – Discussion started: 20 Dec 2022

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 1517 KB)

Supplement (2251 KB)

Download & links

Manal Lam'barki, Wantong Li, Sungmin O, Chunhui Zhan, and Rene Orth

Status: closed

RC1:
'Comment on hess-2022-404', Anonymous Referee #1, 10 Jan 2023
This works presents an interesting investigation of floods across Europe. It aims to build on previous work by considering more factors as potential flood drivers (e.g. soil moisture at various depths, leaf area index, ET, and a metric precipitation variability.) The study concludes that:

This study provides a quantitative mapping of the importance of drivers of high river flow in near-natural European catchments.

[…] that antecedent precipitation anomalies are the most important driver of high flows in most catchments.

[ …] In some other catchments snowmelt and soil moisture are found to be the most relevant drivers.

[…] Moving beyond the state of the art we find a remarkable diversity of second-most important drivers across Europe. This includes vegetation-related drivers such as evapotranspiration.

[…] Overall, observed daily high flow dynamics can be explained similarly well using drivers from the daily, weekly and monthly time scales. This indicates that mechanisms acting at different time scales contribute similarly and jointly to high flow events.

[…] While the most important drivers are similar across time scales, we find interesting variations for the second-most relevant drivers where evapotranspiration and surface soil moisture become more relevant towards longer time scales while deep soil moisture gets less relevant. Furthermore, for more extreme high flows we find a greater diversity of most important drivers across the considered catchments.

[…] Therefore, while moderate high flows are strongly associated with antecedent precipitation, the most extreme events can only be fully understood when considering a comprehensive selection of drivers.

[…] The spatial variations in the relevance of considered high flow drivers can be attributed to vegetation and terrain characteristics of the catchments.

[…] Our findings thereby illustrate that it is beneficial for flood monitoring and prediction to jointly consider several time scales and a comprehensive set of drivers physically related to streamflow dynamics.

[…], Identifying the relative importance of high flow generating mechanisms can reveal regional patterns of causes of floods in Europe and inform future model development.

[…] Moving beyond the state of the art we find a remarkable diversity of second-most important drivers across Europe. This includes vegetation-related drivers such as evapotranspiration.

All aspects 1-11, listed above are potentially relevant for publication in HESS. However, all these aspects also require some substantial consideration before I can recommend them for publication in HESS. My main concerns is:

The chosen method that relies on removing the seasonal cycle sounds potentially useful, but it is unclear to me how this should assign a dominant driver. In places where particular processes are underlying the flood response (e.g. snowmelt in NE Europe, this process is not considered important anymore in the analysis), and processes that physically can have no meaningful effect on floods at the given timescale (e.g. ET at daily timescale) are sometimes identified as most important process. The paper should manage to explain the attribution method (and the logic of removing the seasonal cycle) better to ensure the reader can trust these findings. The whole paper hinges on these findings, so it would be good if the reader can be better convinced of the presented approach.

Some comments on the main conclusions

[1] Since the method seems to ignore relevant drivers and attribute irrelevant drivers, point 1 may not be not shown robustly

[2-3] this seems in line with earlier work. Is there anything that we learn here that we did not know from previous studies? This may be useful to better highlight.

[4] If ET is really important (at daily timescales) this needs to be physically argued. Otherwise it is hard to be convinced by this finding.

[5, 7] these are potentially very relevant and interesting findings, but they are only very briefly discussed and not very quantitatively shown in the paper. Can there be more explicit graphs/analyses that support these findings?

[8] Ok, but what do we learn from this attribution? Can this be stated?

[10] Can it be made a bit more explicit how models can benefit?

[11] This diversity of drivers hinges on my main concern of the paper listed above.

**Further comments**

Considering ET and LAI as drivers of soil moisture on daily timescales seems nonsensical. How would these processes physically affect floods as ET and LAI will be tiny components of the total water balance during flood conditions on such timescales. Are their effects not already captured in considering soil moisture (which integrates the effects of E(T), as also is acknowledged in section 3.1)

It is unclear to me why the model selection leads to a set of near natural catchments, instead of just a set of catchments with simple to model behaviour (independent of the degree of human interference). I would be careful in qualifying these as near-natural.

The choice of coarse spatial resolution of forcing data is understandable, but maybe problematic in the more mountains catchments. What are the potential consequences of this coarse data.

Why are seasonal cycles removed, as these seasonal cycles might be important underlying drivers of the extreme events (i.e these are the ~sum of a seasonal cycle + an individual event on top of that). In places where processes are dominantly driven by a particular seasonal cycle (e.g. snowmelt in NE Europe and large parts of Scandinavia, suddenly snow is not important anymore. How can you explain this to a reader?

Previous work across Europe also aggregates data across various time windows (e.g. Bloschl et al., 2017).

When daily values are used, should rainfall on the date of the flood be chosen, or on the day before, or does this depend on the catchment size?

Figure 2: The font color of soil moisture layer 1 is hard to read.

Figure 3-4: this color classification is hard to read. It would also be useful to guide the reader in what the conclusion is of the Figure (within the caption).

“Another interesting result is that the explained variance of high flows of the dominant drivers is similar across time scales. This indicates that studying drivers at different time scales is relevant to understand high flow dynamics, whereas daily, weekly and monthly time scales are similarly important. Multilayer soil moisture has a higher explained variance for events of the 99th percentile, suggesting the soil water storage is more relevant for the more extreme high flow generation.” This is an interesting statement, but I think it requires some more analysis to conclude this. Right now this result is based on hand wavy interpretations of the results, and no formal quantitative comparison.
Citation: https://doi.org/10.5194/hess-2022-404-RC1
- AC1: 'Reply on RC1', Wantong Li, 14 Mar 2023
  
  The comment was uploaded in the form of a supplement: https://hess.copernicus.org/preprints/hess-2022-404/hess-2022-404-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/hess-2022-404-AC1
RC2:
'Comment on hess-2022-404', Anonymous Referee #2, 01 Feb 2023

This paper presents a study about identifying different drivers of extreme flows over Europe by using observations from 250 near-natural catchments. It is well-written though; it basically uses some traditional and simple statistical approves say regression or correlation things to conduct some basic analysis between extreme flow and potential drivers or factors. Also, the paper did not transfer any new findings or new approach development. And no in-depth physical mechanisms have been introduced. Given that, it is not recommended for publication in this high-ranking journal.
Concerns:
Line 15, it goes “Therefore it is imperative to understand their underlying physical mechanisms”. After reading the paper, there is no information or analysis about the physical stuff.
Line 20. The paper goes like “And in these 11 articles the focus is mainly on regional and/or modelling studies, and they use some drivers for an explanation of the results rather than including them in the actual analysis.”, which induces a justification of the current like “This leaves a knowledge gap in the joint understanding of a variety of observation-based controls of high river flows across continental-scale areas.”. In fact, many existing studies have focused on identifying the possible contribution of extreme flow over the globe, including Europe as well. This paper should give a better explanation of why the current study should be done and why it is important.
The section of 3.2 attribution analysis is quite loose and hard to explain. tree over fraction is the most important in explaining spatial patterns of the relevance of precipitation. This result is not new. While about the elevation and slope, the paper has no in-depth explanation about the potential relationship of the streamflow. And the basin area is of important to affect the effect of elevation and slope to flow. This also needs more physical explanations.
This paper in fact did some basic statistical analysis, while all paper gives an impression about trying to link the physical mechanisms to the changes of flows. Yet, this is no physical analysis across the paper and no physical explanations but some statistical analysis.
Also see Fig. 6, the correlations of different time scales seem to be not consistent, even the direction (some positive or negative), this should be fully discussed.

Citation: https://doi.org/10.5194/hess-2022-404-RC2
- AC2: 'Reply on RC2', Wantong Li, 14 Mar 2023
  
  The comment was uploaded in the form of a supplement: https://hess.copernicus.org/preprints/hess-2022-404/hess-2022-404-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/hess-2022-404-AC2

Status: closed

RC1:
'Comment on hess-2022-404', Anonymous Referee #1, 10 Jan 2023
This works presents an interesting investigation of floods across Europe. It aims to build on previous work by considering more factors as potential flood drivers (e.g. soil moisture at various depths, leaf area index, ET, and a metric precipitation variability.) The study concludes that:

This study provides a quantitative mapping of the importance of drivers of high river flow in near-natural European catchments.

[…] that antecedent precipitation anomalies are the most important driver of high flows in most catchments.

[ …] In some other catchments snowmelt and soil moisture are found to be the most relevant drivers.

[…] Moving beyond the state of the art we find a remarkable diversity of second-most important drivers across Europe. This includes vegetation-related drivers such as evapotranspiration.

[…] Overall, observed daily high flow dynamics can be explained similarly well using drivers from the daily, weekly and monthly time scales. This indicates that mechanisms acting at different time scales contribute similarly and jointly to high flow events.

[…] While the most important drivers are similar across time scales, we find interesting variations for the second-most relevant drivers where evapotranspiration and surface soil moisture become more relevant towards longer time scales while deep soil moisture gets less relevant. Furthermore, for more extreme high flows we find a greater diversity of most important drivers across the considered catchments.

[…] Therefore, while moderate high flows are strongly associated with antecedent precipitation, the most extreme events can only be fully understood when considering a comprehensive selection of drivers.

[…] The spatial variations in the relevance of considered high flow drivers can be attributed to vegetation and terrain characteristics of the catchments.

[…] Our findings thereby illustrate that it is beneficial for flood monitoring and prediction to jointly consider several time scales and a comprehensive set of drivers physically related to streamflow dynamics.

[…], Identifying the relative importance of high flow generating mechanisms can reveal regional patterns of causes of floods in Europe and inform future model development.

[…] Moving beyond the state of the art we find a remarkable diversity of second-most important drivers across Europe. This includes vegetation-related drivers such as evapotranspiration.

All aspects 1-11, listed above are potentially relevant for publication in HESS. However, all these aspects also require some substantial consideration before I can recommend them for publication in HESS. My main concerns is:

The chosen method that relies on removing the seasonal cycle sounds potentially useful, but it is unclear to me how this should assign a dominant driver. In places where particular processes are underlying the flood response (e.g. snowmelt in NE Europe, this process is not considered important anymore in the analysis), and processes that physically can have no meaningful effect on floods at the given timescale (e.g. ET at daily timescale) are sometimes identified as most important process. The paper should manage to explain the attribution method (and the logic of removing the seasonal cycle) better to ensure the reader can trust these findings. The whole paper hinges on these findings, so it would be good if the reader can be better convinced of the presented approach.

Some comments on the main conclusions

[1] Since the method seems to ignore relevant drivers and attribute irrelevant drivers, point 1 may not be not shown robustly

[2-3] this seems in line with earlier work. Is there anything that we learn here that we did not know from previous studies? This may be useful to better highlight.

[4] If ET is really important (at daily timescales) this needs to be physically argued. Otherwise it is hard to be convinced by this finding.

[5, 7] these are potentially very relevant and interesting findings, but they are only very briefly discussed and not very quantitatively shown in the paper. Can there be more explicit graphs/analyses that support these findings?

[8] Ok, but what do we learn from this attribution? Can this be stated?

[10] Can it be made a bit more explicit how models can benefit?

[11] This diversity of drivers hinges on my main concern of the paper listed above.

**Further comments**

Considering ET and LAI as drivers of soil moisture on daily timescales seems nonsensical. How would these processes physically affect floods as ET and LAI will be tiny components of the total water balance during flood conditions on such timescales. Are their effects not already captured in considering soil moisture (which integrates the effects of E(T), as also is acknowledged in section 3.1)

It is unclear to me why the model selection leads to a set of near natural catchments, instead of just a set of catchments with simple to model behaviour (independent of the degree of human interference). I would be careful in qualifying these as near-natural.

The choice of coarse spatial resolution of forcing data is understandable, but maybe problematic in the more mountains catchments. What are the potential consequences of this coarse data.

Why are seasonal cycles removed, as these seasonal cycles might be important underlying drivers of the extreme events (i.e these are the ~sum of a seasonal cycle + an individual event on top of that). In places where processes are dominantly driven by a particular seasonal cycle (e.g. snowmelt in NE Europe and large parts of Scandinavia, suddenly snow is not important anymore. How can you explain this to a reader?

Previous work across Europe also aggregates data across various time windows (e.g. Bloschl et al., 2017).

When daily values are used, should rainfall on the date of the flood be chosen, or on the day before, or does this depend on the catchment size?

Figure 2: The font color of soil moisture layer 1 is hard to read.

Figure 3-4: this color classification is hard to read. It would also be useful to guide the reader in what the conclusion is of the Figure (within the caption).

“Another interesting result is that the explained variance of high flows of the dominant drivers is similar across time scales. This indicates that studying drivers at different time scales is relevant to understand high flow dynamics, whereas daily, weekly and monthly time scales are similarly important. Multilayer soil moisture has a higher explained variance for events of the 99th percentile, suggesting the soil water storage is more relevant for the more extreme high flow generation.” This is an interesting statement, but I think it requires some more analysis to conclude this. Right now this result is based on hand wavy interpretations of the results, and no formal quantitative comparison.
Citation: https://doi.org/10.5194/hess-2022-404-RC1
- AC1: 'Reply on RC1', Wantong Li, 14 Mar 2023
  
  The comment was uploaded in the form of a supplement: https://hess.copernicus.org/preprints/hess-2022-404/hess-2022-404-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/hess-2022-404-AC1
RC2:
'Comment on hess-2022-404', Anonymous Referee #2, 01 Feb 2023

This paper presents a study about identifying different drivers of extreme flows over Europe by using observations from 250 near-natural catchments. It is well-written though; it basically uses some traditional and simple statistical approves say regression or correlation things to conduct some basic analysis between extreme flow and potential drivers or factors. Also, the paper did not transfer any new findings or new approach development. And no in-depth physical mechanisms have been introduced. Given that, it is not recommended for publication in this high-ranking journal.
Concerns:
Line 15, it goes “Therefore it is imperative to understand their underlying physical mechanisms”. After reading the paper, there is no information or analysis about the physical stuff.
Line 20. The paper goes like “And in these 11 articles the focus is mainly on regional and/or modelling studies, and they use some drivers for an explanation of the results rather than including them in the actual analysis.”, which induces a justification of the current like “This leaves a knowledge gap in the joint understanding of a variety of observation-based controls of high river flows across continental-scale areas.”. In fact, many existing studies have focused on identifying the possible contribution of extreme flow over the globe, including Europe as well. This paper should give a better explanation of why the current study should be done and why it is important.
The section of 3.2 attribution analysis is quite loose and hard to explain. tree over fraction is the most important in explaining spatial patterns of the relevance of precipitation. This result is not new. While about the elevation and slope, the paper has no in-depth explanation about the potential relationship of the streamflow. And the basin area is of important to affect the effect of elevation and slope to flow. This also needs more physical explanations.
This paper in fact did some basic statistical analysis, while all paper gives an impression about trying to link the physical mechanisms to the changes of flows. Yet, this is no physical analysis across the paper and no physical explanations but some statistical analysis.
Also see Fig. 6, the correlations of different time scales seem to be not consistent, even the direction (some positive or negative), this should be fully discussed.

Citation: https://doi.org/10.5194/hess-2022-404-RC2
- AC2: 'Reply on RC2', Wantong Li, 14 Mar 2023
  
  The comment was uploaded in the form of a supplement: https://hess.copernicus.org/preprints/hess-2022-404/hess-2022-404-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/hess-2022-404-AC2

Manal Lam'barki, Wantong Li, Sungmin O, Chunhui Zhan, and Rene Orth

Supplement

https://doi.org/10.5194/hess-2022-404-supplement

Manal Lam'barki, Wantong Li, Sungmin O, Chunhui Zhan, and Rene Orth

Viewed

Total article views: 1,218 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
863	303	52	1,218	94	41	39

HTML: 863
PDF: 303
XML: 52
Total: 1,218
Supplement: 94
BibTeX: 41
EndNote: 39

Views and downloads (calculated since 20 Dec 2022)

Month	HTML	PDF	XML	Total
Dec 2022	239	54	5	298
Jan 2023	153	39	2	194
Feb 2023	75	28	2	105
Mar 2023	60	32	4	96
Apr 2023	39	19	0	58
May 2023	26	9	0	35
Jun 2023	21	7	1	29
Jul 2023	24	7	1	32
Aug 2023	13	16	1	30
Sep 2023	27	12	1	40
Oct 2023	15	9	1	25
Nov 2023	10	3	0	13
Dec 2023	10	11	2	23
Jan 2024	10	4	1	15
Feb 2024	7	9	1	17
Mar 2024	17	11	0	28
Apr 2024	14	4	6	24
May 2024	13	3	4	20
Jun 2024	31	2	3	36
Jul 2024	12	3	7	22
Aug 2024	8	4	4	16
Sep 2024	21	6	4	31
Oct 2024	8	4	1	13
Nov 2024	8	3	1	12
Dec 2024	2	4	0	6

Cumulative views and downloads (calculated since 20 Dec 2022)

Month	HTML	PDF	XML	Total
Dec 2022	239	54	5	298
Jan 2023	153	39	2	194
Feb 2023	75	28	2	105
Mar 2023	60	32	4	96
Apr 2023	39	19	0	58
May 2023	26	9	0	35
Jun 2023	21	7	1	29
Jul 2023	24	7	1	32
Aug 2023	13	16	1	30
Sep 2023	27	12	1	40
Oct 2023	15	9	1	25
Nov 2023	10	3	0	13
Dec 2023	10	11	2	23
Jan 2024	10	4	1	15
Feb 2024	7	9	1	17
Mar 2024	17	11	0	28
Apr 2024	14	4	6	24
May 2024	13	3	4	20
Jun 2024	31	2	3	36
Jul 2024	12	3	7	22
Aug 2024	8	4	4	16
Sep 2024	21	6	4	31
Oct 2024	8	4	1	13
Nov 2024	8	3	1	12
Dec 2024	2	4	0	6

Viewed (geographical distribution)

Total article views: 1,154 (including HTML, PDF, and XML) Thereof 1,154 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 13 Dec 2024

Short summary

We investigate the main drivers of high river flows in near-natural European catchments. While there are a lot of previous research in this area, the understanding of the relative relevance of high flow drivers other than precipitation is understudied. We find that the secondary drivers of high river flows are very diverse and become more relevant for more extreme events. This illustrates the necessity of flood management by considering a multitude of drivers in the context of climate change.


Total:	0
HTML:	0
PDF:	0
XML:	0