Dam break – Outburst flood propagation and transient hydraulics: A geosciences perspective
Introduction
Dam break outburst floods are frequently generated by failures of natural dams (Carrivick and Rushmer, 2006). Particularly common sources of outburst floods include landslide-dammed lakes (e.g. Dai et al., 2005), glacial (e.g. Carrivick et al., 2009, Huggel et al., 2002, Tweed and Russell, 1999) and moraine lakes (e.g. Clague and Evans, 2000), artificial (e.g. Fread, 1996) and volcanic impoundments (e.g. Carrivick et al., 2009, Kataoka et al., 2008, Fenton et al., 2006, Macías et al., 2004, Waythomas, 2001), and the melting of ice by volcanic activity (e.g. Guðmundsson et al., 1997). They can be named ‘jökulhlaups’ if from a glacial source and as ‘lahars’ if incorporating volcaniclastic sediment. Generically such phenomena are referred to in the literature as ‘surging flow’, ‘flash/outburst flood’, or ‘dam break wave/flow/flood’. Across the different categorisations of these events by sources, trigger mechanisms and flow processes (e.g. Costa and Schuster, 1988, Tweed and Russell, 1999, Carrivick and Rushmer, 2006, Korup and Tweed, 2007) there remains what is essentially the propagation of a fluid wave over initially ‘dry’ terrain. Therefore the many types of dam break outburst floods will simply be referred to herein as ‘outburst floods’.
Outburst floods are the largest floods known to have occurred on Earth during the Quaternary (e.g. Baker, 2002, Korup and Tweed, 2007). Modern examples of these floods constitute a serious threat to life, property and infrastructure. Outburst floods can exert intense and widespread landscape change through erosion of both unconsolidated sediments (e.g. Gomez et al., 2002, Russell and Marren, 1999, Russell et al., 2006) and bedrock (e.g. Carrivick et al., 2004a, Tómasson, 1996, Baker, 1988). As climate change causes more ‘extreme’ weather outburst floods will likely increase in frequency, and perhaps also in magnitude (Bates et al., 2008).
Despite the pressing need to measure, model and to mitigate against outburst flood impacts, there are problems with each stage of this process. Such events are simply too sudden, powerful and short-lived to measure directly, and in situ field-deployed sensors and loggers are very likely to become destroyed. A few exceptional studies have succeeded in directly measured outburst flood properties; a most recent example being the March 18th lahar from Crater Lake on Mt. Ruapehu, New Zealand (Carrivick et al., 2009). In reaction to the problems of direct measurements, research aiming to better understand outburst floods has pursued two approaches. The first approach has sought to reconstruct characteristics of outburst floods from geological field evidence, using empirical relationships (e.g. Clague and Mathews, 1973, Walder and Costa, 1996), palaeocompetence (e.g. Carrivick, 2007) and various palaeohydraulic techniques including slope area methods. The same knowledge has also been incorporated into GIS-based hazard analyses (e.g. McKillop and Clague, 2007). The second approach has been adopted from the engineering community and is the application of fluid dynamics computational models, both with and without use of flume experiments. These studies are designed to develop simple physical-based equations that highlight the controls such as impounded water volume, reservoir geometry and dam height, for example, as well as to consider idealised dam breach evolution.
Nevertheless, individuals, authorities and organisations seeking to manage outburst flood hazards and to mitigate against outburst floods require knowledge of transient hydraulics. Specifically, the longitudinal evolution of a flood; i.e. flood propagation and time to inundation, time to peak discharge, persistence of peak discharge and rates of rising/falling stages, for example must be understood for effective outburst flood hazard analysis (e.g. McKillop and Clague, 2007). These properties, and the controls on them, have either not been directly measured or the literature has not reported them explicitly or coherently. Furthermore, although the fluid dynamics engineering provides some of the best approximations of the dam break processes, this literature is poorly known and understood by geomorphologists and sedimentologists who reconstruct outburst floods from field evidence.
Section snippets
Aim
The overall aims of this paper are to (1) To quantitatively analyse new experimental data on outburst flood propagation and transient hydraulics and (2) To summarise these analyses in a conceptual model of outburst flood controls and properties that will be of direct use for hazard management. These aims are achieved by explicitly testing a set of hypotheses derived from the literature concerning the key controls on longitudinal flow evolution. The framework for this work and the results are
Experimental insights to outburst floods
Outburst floods are unsteady non-uniform flow phenomena. In unsteady flows the key process parameter is friction velocity; which can be used to calculate bottom shear stress and must be measured within the (thin) viscous sublayer close to the channel bed. Friction velocity is frequently either not described or its dynamic behaviour is ignored (Schlütter, 1999). Additionally, turbulence, which is characterised by chaotic hydraulic changes; i.e. high momentum convection and rapid variation of
Methods
Flume experiments were conducted within the Sorby Environmental Fluid Dynamics Laboratory (SEFDL) at the University of Leeds. Experiments applied Froude scaling principles at 1:20 in order to best-match outburst flood conditions reported in the geological literature. Specifically, field observations and reconstructions of outburst floods commonly suggest Froude numbers of ∼1, high sediment concentrations approaching ∼25% by volume, and high channel roughness relative to flow depth (e.g.
Flow front character and propagation
The emphasis of the experiments in this study is not so much on absolute values; although these will be of use to modellers wishing to validate their results. Rather, the focus is on the controls on longitudinal flow propagation and hydraulics. For clarity and brevity, only results from a single experimental run of each combination of variables are graphically presented. However, the excellent repeatability of experiments can be demonstrated graphically as well as statistically. For example, an
Flow depth
Over fixed rough beds flow depths increase down channel (Fig. 6). Flow depths also increase with the height of the gate raise and with increasing bed roughness (Fig. 6). However, the duration of peak stages is rather more complicated. Whilst peak stage is long-lived with small gate raises, the stage hydrograph is more ‘flashy’ with large gate raises (Fig. 6). Increasing bed roughness does not appear to change the rate of rise to peak stage, nor the duration of peak stage and the recession from
Transient hydraulics and flow regime
So far the results described have emphasised effects on flow propagation and hydraulics of varying controls. However, varying flow conditions within a single outburst flood event are also important and best illustrated with recourse to high frequency velocity data. One example at 2 m channel distance from a 6 cm gate raise over a fine-grained fixed bed shows that such data is characteristically ‘noisy’ in its ‘raw’ form (Fig. 9A). Use of a running mean or standard deviation helps to compare these
Discussion
This section takes the results presented above and analyses them from a hazard management perspective. The embedded discussion emphasises external and internal controls on outburst flood propagation and transient hydraulics and seeks a process-based understanding of the observations. Both ‘at-a-point’ and ‘longitudinal’ variations are examined and the implications for managing outburst flood hazards are presented. Initially qualitative relationships are described, before expanding to consider
Conclusions
Previous experimental work has shown that outburst floods can be identified to possess three flow regimes. Firstly, there is a short acceleration due to the reservoir pressure level; i.e. the depth of impounded water. For the case of a narrow deep reservoir, such as an experimental lock box, this pressure rapidly falls and the channel flow quickly converges to an inertial regime (e.g. Stansby et al., 1998). The third phase is a viscous regime, dominated by channel bed friction and pronounced
Acknowledgements
This data presented in this paper is part of a project funded by a NERC New Investigator’s Research Grant NE/F000235/1 to J.L. Carrivick. Gareth Keevil is thanked for his tremendous technical assistance, and Jeff Peakall for his enthusiastic advice and support. Lee Brown and Lucy Rushmer gave advice on the quantitative analysis and flume modelling approaches, respectively. Insightful and helpful comments from two reviewers and an associate editor improved this manuscript.
References (92)
- et al.
Spatial and temporal variation of turbulence characteristics in combined sewer flow
Flow measurement and Instrumentation
(2008) - et al.
Palaeohydrology and sedimentary impacts of jökulhlaups from Kverkfjöll, Iceland
Sedimentary Geology
(2004) - et al.
Experimental study of the air–water shear flow in a hydraulic jump
International Journal of Multiphase Flow
(2000) Air–water and momentum exchanges in unsteady surging waters: an experimental study
Experimental Thermal and Fluid Science
(2005)- et al.
A review of catastrophic drainage of moraine-dammed lakes in British Columbia
Quaternary Science Reviews
(2000) - et al.
The 1786 earthquake-triggered landslide dam and subsequent dam-break flood on the Dadu River, southwestern China
Geomorphology
(2005) - et al.
Peak discharge of a Pleistocene lava-dam outburst flood in Grand Canyon, Arizona, USA
Quaternary Research
(2006) - et al.
Ice, moraine, and landslide dams in mountainous terrain
Quaternary Science Reviews
(2007) - et al.
Fluvial response to large rock-slope failures – examples from the Himalayas, the Tien Shan, and the Southern Alps in New Zealand
Geomorphology
(2006) - et al.
Contrasts between debris flows, hyperconcentrated flows and stream flows at a channel of Mount Semeru, East Java, Indonesia
Geomorphology
(2004)