(SP3 of Research Unit “Space-time Dynamics of Extreme Floods SPATE”)
The subproject 3 (SP3) of the research unit SPATE (Space-time Dynamics of Extreme Floods) has been started in September 2017 at the GeoForschungsZentrum Potsdam of the Helmholtz Association. The first phase of the project is approved for three years. The overarching goal of SP3 is to achieve a better understanding of causes for the occurrence of extreme floods in comparison to small and medium floods. A comprehensive analysis on the main factors and whose interactions controlling the probability of extreme floods is conducted. Scientific exchange with other researchers of SPATE as well as international scientists is achieved by working in research clusters, focusing on event scale and spatial and temporal variability in this subproject.
The research work at GFZ Potsdam is structured into three working packages:
WP1 is focused on tail behavior of extreme value distributions of precipitation and discharge. Upper tails of distributions are termed ‘heavy-tailed’, if extreme values are more likely to occur than would be predicted by distributions with exponential asymptotic behavior. Several studies show that heavy-tail behavior plays an important role in extreme value distributions of hydro-meteorological variables (Katz et al., 2002) which is of great importance since potential impacts of such extreme events can be considerably high. The upper tail behavior of distributions is found to be a complex interplay of climate and catchment characteristics. Bernadara et al.
(2008) detect regional pattern of heavy-tail behavior in a data set of 173 catchments, proposing that catchments with snow melt as dominant driver tend to have lighter tails. Further, Villarini and Smith (2010) state that heavy tails of flood distributions of Eastern US are strongly influenced by atmospheric situation and that landfalling cyclones are an important factor on observed tail behavior. Thus to consider the spatio-temporal variability of flood-driving processes, further research combining multi-variate data analysis and process understanding is important to achieve a better understanding of heavy tails in distributions of hydro-meteorological variables. For working package WP1, the aim is to identify major factors and processes for the occurrence of heavy tails in flood and heavy precipitation distributions. Therefore, a large data set of German and Austrian catchments representing a wide range of catchment and climate characteristics is derived. Additionally, a number of catchment and event signatures are derived and a comprehensive analysis of signatures and tail behavior is taken out by means of multivariate data analysis.
The main tasks of WP1 are:
- Derivation of a data base of streamflow and precipitation time series for a large number of catchments,
- Using different methods to estimate the upper tail of distributions
- Derivation of catchment and event signatures from various sources, including for instance antecedent soil moisture, base flow
index and large-scale atmospheric situation
- Comparative analyses of upper tail behavior and signatures by means of data-mining approaches including Random Forest and
Bayesian networks and
- Comparison of upper tail behavior of flood and heavy precipitation distributions
In WP2, the superposition of flood waves at the confluence of rivers is investigated as a potential driver for the occurrence of extreme floods which can influence the tail behavior of flood distributions. In general, river network processes like channel-floodplain interactions and superposition of waves can have substantial impact on flood characteristics including the volume, peak and shape of the hydrograph as well as the extent of inundation. Important factors controlling superposition are spatial and temporal rainfall patterns, flood generation processes, topographic source areas and timing and peak heights of tributary hydrographs. Large catchments with braided river network and large storage capacity are expected to have higher potential to be affected by flood wave superposition in comparison to smaller catchments. Several studies show that especially river regulation by measures like deepening, straitening and reduction of floodplain storage has significant influence on flood characteristics and can lead to enhanced peaks of large floods as was shown by Lammersen et al. (2002) and Vorogushyn and Merz (2013) for the river Rhine. Vorogushyn and Merz (2013) could further show that this increase was partly caused by superposition of flood waves from river Rhine and Neckar. River regulation was found to exacerbate this effect due to increased wave celerity and reduced time lags between peaks (Kalweit et al., 1993). Flood wave transformation is influenced by complex interactions of several factors and is not yet sufficiently understood. Sincerely, flood wave superposition is a potential risk factor in flood generation downstream and thus, gaining further knowledge on the main processes controlling flood wave superposition is important.
In WP2, the aim is to deepen the knowledge of important factors and interactions controlling flood wave transformation. Superposition patterns will be detected in large river systems in Germany and Austria and atmospheric conditions and catchment state will be analyzed to identify dominant factors. Further, the scaling of flood wave superposition with flood magnitude will be analyzed in order to find differences between the impact of superposition on extreme and small or medium floods. Therefore, specific conditions in the past are identified in which a superposition of flood waves of main river and tributary lead to an increase in the flood volume and peak and superposition patterns are contrasted with flood generation processes in order to find main controls. The main tasks of WP2 are:
- Characterization of flood waves in time series of Germany and Austria
- Detection of superposition patterns in historical events of large catchments
- Analyses of potentially responsible flood generation processes
- Contrasting superposition patterns with flood generation processes
Fig. 1: Scheme of flood wave superposition in river Rhine (from Vorogushyn und Merz, 2013)
In WP3, the knowledge derived from WP1 and WP2 is combined to study non-linearity behavior of extreme floods in comparison to small and medium floods. Heavy-tail behavior of flood distributions is the result a complex interplay of different processes. It is further expected, that non-linear superposition of processes can lead to extreme floods whereas the processes individually do not cause this extremeness in floods (Blöschl and Zehe, 2005). Several studies point out the complex interplay between atmospheric processes and catchment conditions in the case of extreme flood events. Especially the interaction between antecedent soil moisture has been found to be causing large-scale floods as for instance the floods in August 2002 und June 2013 in Germany (Schröter et al., 2015; Stohl and James, 2004).
Important causes of non-linear behavior in extreme flood generation are threshold processes that can lead to a switch in the hydrological system such as from subsurface runoff to surface runoff, if precipitation rates exceed infiltration, causing infiltration excess overflow. In flood frequency curves, non-linear behavior could be detected as breaks or steps in the slope of curve (Rogger et al. 2012). This has for instance been detected by Kusumastuti et al. (2007), who suggest that observed step changes may be caused by a switch of runoff generation processes. Threshold processes can also rise from the river network such as dike breaching. Merz and Blöschl (2003, 2009) suggest that different flood types also influence the shape of the flood frequency curve and that upper tail behavior could be dominated by a particular flood type such as flash floods.
Non-linearities can occur along the whole flood generation cascade. Further research in this field is required since non-linearity behavior can have a considerable impact on the generation of extreme floods. In WP3, the aim is to analyze source and degree of non-linearity behavior along the complete flood generation cascade to detect differences in the generation processes of extreme floods compared to small and medium floods. Again, heavy-tail behavior is analyzed but with a focus on process understanding. Therefore, catchment conditions (e.g. soil moisture conditions) and atmospheric situations (e.g. precipitation volume and weather conditions) are jointly considered.
The main tasks of WP3 are:
- Selection of extreme floods and contrasting them to the remaining smaller floods
- Derivation of space-time fields of processes along the flood process cascade from observations and simulations, characteristics
include: Atmospheric and initial catchment conditions, Event catchment meteorology and runoff generation in the catchment
- Comparison of small and extreme floods by investigating threshold processes, process interactions, patterns and anomalies.
Fig. 2: Different possible extrapolations of flood extremes. Extrapolation 3 refers to a step change in the flood frequency curve due
to threshold exceedance of water storage (from Rogger et al., 2012, WRR)
Dr. Heidi Kreibich (Lead of subproject)
Prof. Dr. Bruno Merz
Dr. Sergjy Vorogushyn
Dr. Björn Guse (funded as Post-Doc in SPATE)
Luzie Wietzke (funded as PhD in SPATE)
Bernardara, P., Schertzer, D., Sauquet, E., Tchiguirinskaia, I. and Lang, M.: The flood probability distribution tail: how heavy is it? Stoch Environ Res Risk Assess, 22, 107–122, 2008.
Blöschl, G. and Zehe, E.: On hydrological predictability, Hydrol. Process., 19, 3923–3929, 2005.
Kalweit, H., Buck, W., Felkel, K., Gerhard, H., van Malde, J., Nippes, K.-R., Ploeger, B., and Schmitz, W: Der Rhein unter der Einwirkung des Menschen – Ausbau, Schifffahrt, Wasserwirtschaft, Report Nr. I-11, Internationale Kommission für die Hydrologie des Rheingebietes, Lelystad, The Netherlands, 1993.
Kusumastuti, D. I., Struthers, I., Sivapalan, M., and Reynolds, D. A.: Threshold effects in catchment storm response and the occurrence and magnitude of flood events: implications for flood frequency, Hydrol. Earth Syst. Sci., 11, 1515-1528, 2007.
Lammersen, R., Engel, H., van de Langemheen, W., and Buiteveld, H.: Impact of river training and retention measures in flood peaks along the Rhine, J. Hydrol., 267, 115–124, 2002.
Merz, R. and Blöschl, G.: A process typology of regional floods, Water Resources Research 39 (12), 1340, 2003
Merz, R. and Blöschl, G.: Process controls on the statistical flood moments – a data based analysis,Hydrol. Process, 23, 675-696, 2008
Rogger, M., H. Pirkl, A. Viglione, J. Komma, B. Kohl, R. Kirnbauer, R. Merz, and G. Blöschl, Step changes in the flood frequency curve: Process controls, Water Resour. Res., 48, W05544, 2012
Schröter, K., Kunz, M., Elmer, F., Mühr, B., and Merz, B.: What made the June 2013 flood in Germany an exceptional event? A hydro-meteorological evaluation. Hydrol. Earth Syst. Sci., 19, 309-327, 2015
Stohl, A. and James, P., A Lagrangian analysis of the atmospheric branch of the global water cycle.
Part I: Method, description, validation, and demonstration for the August 2002 flooding in Central Europe, Journal of Hydrometeorology, 5, 656-678, 2004
Villarini, G. and Smith, J. A.: Flood peak distributions for the eastern United States, Water Resources Research, Vol. 46, W06504, 2010
Vorogushyn, S. and Merz, B.: Flood trends along the Rhine: the role of river training, Hydrol. Earth Syst. Sci. 17, 3871-3884, 2013.