Prediction of floods in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 189

9 Prediction of floods in
ungauged basins

Contributors: D. Rosbjerg,* G. Blöschl, D. H. Burn,
A. Castellarin, B. Croke, G. Di Baldassarre, V. Iacobellis,
T. R. Kjeldsen, G. Kuczera, R. Merz, A. Montanari,
D. Morris, T. B. M. J. Ouarda, L. Ren, M. Rogger, J. L.
Salinas, E. Toth and A. Viglione

9.1 How high will the flood be?
Floods are one of the most burning societal issues catch-
ment hydrology has to face. Flood-related economic losses
have increased dramatically over the past decades in most
parts of the world, and the number of flood fatalities has
increased in some continents (Di Baldassarre et al., 2010 ).
Floods are shaping the patterns of human behaviour in
many ways. Flood risk is a major factor controlling settle-
ment patterns near rivers. Much of the infrastructure near
streams is susceptible to flood damage in some form or
another and floods may disrupt mobility and livelihood.
While floods are often only viewed from a perspective of
damage and destruction they may play an essential role in
ecosystems. In river wetlands, for example, regular
flooding sustains the dynamics of soil moisture and
groundwater necessary for ecosystem functioning.
Predictions of floods are therefore needed for a wide
variety of societal purposes. Integrated flood risk manage-
ment (e.g., EU, 2007 ) aims at coordinating the various
management goals related to floods. Part of the manage-
ment plan is the prudent design of the infrastructure involv-
ing design of dam spillways, bridges, road culverts and
levees. Residential area zoning, floodplain management
and urban design are other important aspects of integrated
flood management. For all of these purposes one needs to
know the level of flood water or flood runoff that may
occur with a given probability.
This chapter focuses on the prediction of floods in
ungauged catchments. In the context of this topic flood
prediction is defined as the estimation of flood runoff and
the associated exceedance probability at an unknown
future point in time, as opposed to real-time flood forecast-
ing where forecasts are made for the immediate future. The
focus is on river floods induced by heavy rain, sometimes
in association with snowmelt, but dam breach floods, ice

jam floods, and floods due to other processes are not dealt
with in this chapter.
One of the ways that humanity has learned to live with
floods is to understand the severity of flooding in terms
of the frequency or probability of flooding. Hydrology
then needs to understand the risk of rare or extreme
events, and factor that risk in cost-benefit analyses of
engineering decisions. Therefore, the quantity of interest
in flood estimation is the magnitude of the flood (nor-
mally flood runoff at a particular point at a river, or
the corresponding stage or water level) for a specified
return period (also called average recurrence interval).
An example is the so-called 100-year flood, the magni-
tude of the annual maximum flood that will be exceeded,
on average, once in 100 years; put another way, there is
a 1% chance in any year that the largest flood of the year
will exceed the 100-year magnitude. The choice of return
period in specific circumstances depends on the risk
society is prepared to accept, and is usually decided on
the basis of consensus or cost-benefit analysis. For
example, a lower return period might be accepted for
design of a stormwater drain than for design of a dam
spillway, since the potential damage is lower in the case
of the former. This chapter is particularly concerned with
the flood frequency curve.
From the perspective of runoff variability, the flood
frequency curve is an extreme value distribution that lies
at the long tail of the distribution of all flood peaks that
may occur in the catchment. It is constructed by plotting
the frequency of a flood (in terms of the return period)
against its runoff. Floods are a subset of the full spectrum
of runoff variability experienced in the catchment, as
reflected in the complete hydrograph. In spite of this
unique feature of representing extremes, the flood fre-
quency curve reflects the net result of the interactions
between rainfall variability, vegetation, soils and geology,
which in turn are related to climatic and landscape hydro-
logical processes. Floods are therefore closely related to