THE DESIGN FLOOD 193
with the worst flood producing catchment conditions that can be realisti-
cally expected in the prevailing meteorological conditions. A discussion of
the rainfall–runoff method and unit hydrograph procedure is beyond the
scope of this text but for further particulars see, e.g., Linsley, Kohler and
Paulhus (1988), Wilson (1990), Shaw (1994), Institute of Hydrology
(1999), Gosschalk (2002) and US Army Corps of Engineers (2002).
The PMF may also approximately be determined as a multiple of a
flood of a certain return period (e.g. 150 years, Q 150 (m^3 s^1 )) which, in turn,
is given by flood frequency analysis or by an equation correlating the flood
with the catchment area, the index of catchment permeability, the stream
frequency (number of junctions on a 1:25 000 scale map divided by catch-
ment area), the net one-day rainfall of a certain return period (e.g. 5 years),
and incorporating a regional constant (NERC, 1993). In UK conditions the
PMF is then about 5 times the Q 150 value; similarly the PMF corresponds to
about 3 times the 1000 year flood and twice the 10 000 year flood.
The definition of PMF implies that it is not a fixed value, as its deter-
mination (apart from location) depends on the reliability of information,
the advance of technical knowledge, and the accuracy of analysis; thus
PMF can – and should – be periodically reviewed (see also Reed and Field
(1992)). Its probability cannot be determined, as it represents events that
are so rare that no observed data are available to establish it, and thus it
can only be treated deterministically. (For estimating probabilities of
extreme floods, see also National Research Council (1988).)
Table 4.1 is an example of the current recommendations of the Insti-
tution of Civil Engineers (UK) for reservoir flood standards (ICE, 1996).
In the case of category b and c dams (Table 4.1) an alternative eco-
nomic optimization study could be used (see also ICOLD, 1997) in which
the chosen flood minimizes (on a probability basis) the sum of the spillway
and damage costs; however, this reservoir inflow flood may not drop below
the specified minimum.
In some cases of dams in category d (Table 4.1) a cost–benefit analy-
sis and economic evaluation may also be used as a complementary
measure.
For small embankments designed for a limited life of L(years) (e.g.
cofferdams for the diversion of rivers during dam construction) the risk r
of exceeding the design flood of a return period T(years) can be com-
puted as
r 1 (1 1/T)L. (4.1)Fahlbusch (1999) points out that the safety levels adopted for major
dam construction cofferdams are often far too low (usually 30 years for
embankments and 20 years for concrete dams), whereas to be consistent
with the safety levels for the finished dams, diversion works for construc-
tion periods 2–4 years should be designed for floods 200–800 years return