Geoscience Reference
In-Depth Information
to hydrological and fluvial processes (Frostick and Reid,
1987; Cooke, Warren and Goudie, 1993; Tooth, 2000; Bull
and Kirkby, 2002; Parsons and Abrahams, 2009; and also
in this volume). This is important not only because flu-
vial processes are a cause of so many problems in desert
areas, but also because the peculiarities of dryland envi-
ronments cause sufficient differences in river behaviour
that the lessons learnt in humid areas are not reliably
translated (Pilgrim, Chapman and Doran, 1988).
the dense networks of gauges that would be required to
measure such spotty rainfalls are extremely rare in dry-
lands. Nevertheless, there is good evidence to suggest that
these cells have a diameter that is generally less than 10-
14 km (Diskin and Lane, 1972; Renard and Keppel, 1966)
(Figure 13.1(a)). This means that rainfall measured at one
gauge cannot be used to predict rainfall even a few kilome-
tres away, in contrast with temperate regions affected by
ubiquitous frontal storms (Wheater
et al
., 1991a) (Figures
13.1(b) and 13.2). Because of the discrete nature of each
convective cell, an individual storm may be unlikely to
affect the entire drainage net, while successive storms are
more than likely to wet different parts of a river catchment
(Schick and Lekach, 1987) (Figure 13.1(a)).
In catchments of modest dimensions (10
2
to 10
3
km
2
),
this has implications for the flood hydrograph, because
a different part of the drainage basin will contribute wa-
ter during each event. Indeed, some tributaries have been
noted first to run and then to remain dry during a period
of successive floods in a trunk stream (Frostick, Reid and
Layman, 1983). As a result of this and other variables, the
shape of the flood hydrograph must change considerably.
13.2
Rainfall and river discharge
13.2.1
Storm characteristics
One way in which desert streams differ from their peren-
nial counterparts is the generation and propagation of the
flood wave. However, the peculiarities of flash floods are
not entirely a function of processes on the ground. Of
considerable significance for runoff is the fact that rain
is more often than not associated with discrete convec-
tive cells (Sharon, 1972, 1974). Data are few because
.6
.6
.4
.2
.2
.4
(a)
.8
(b)
0
0
1.0
.2
0
.4
.4
0
.2
5km
0.5
0
0
.6
.4
.2
.4
.6
.8
.8
0
0
10
20
.2
Distance between rainguages, km
.2
.2
Illinois
Sukumaland
Negev
1.6
.4
.8
1.4
.6
1.0
1.2
1.0
.4
Rainguage
Catchment boundary
Isohyet
.6
.8
1.0
Figure 13.1
(a) Raingauge net, catchment boundary and the isohyets (in inches) of two storms over Walnut Gulch, Arizona
(modified and redrawn
after Renard and Keppel, 1966
). (b) Correlogram of rainfall caught by point gauges spaced at distances
up to 20 km for winter frontal rainfall (Illinois) and cellular convective storms (Negev Desert and Sukumaland, western Tanzania)
