Prediction of annual runoff in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 84

Figure 5.12. Mean annual runoff
estimated from regressions vs.
observations. The dashed lines
correspond to the 90% prediction
intervals. The map shows the outlets
of 47 catchments in north-western
Italy. Adapted from Viglione et al.
( 2007a ).

1600

44

1400

43

44

43

39

1200

39

1000

800

35 34

600

600 800 1000 1200

Q A (mm/yr)

1400 1600

35

50 km

34

Figure 5.13. Coef cient of variation
of annual runoff versus mean annual
runoff. Dashed line relates to
Australia and Southern Africa
(ASA), solid line relates to the rest of
world (RoW). From McMahon et al.
( 2007b ), Koster and Suarez ( 1999 ).

10

ASA

3

1

Kalinin

RoW

0.3

0.1

1 10 100 1000 10000

Mean annual runoff (mm/yr)

indicated that arid (less rainy) catchments are characterised
by more variability. Figure 5.13 shows the results of
applying this approach to a global data set, stratified into
Australia and Southern Africa and the Rest of the World
(McMahon et al., 2007b ). The CV of annual runoff in
Australia and Southern Africa is significantly higher than
the rest of the world, for a given mean annual runoff.

Mean annual runoff
Budyko-type models Budyko-type models offer the
potential to estimate mean annual actual evaporation from
the aridity index and precipitation without calibration.
Mean annual runoff is then estimated as the residual of
precipitation and evaporation. Budkyo-type models
include: Schreiber (Schreiber, 1904 ), Ol

'

' dekop (Ol dekop,

-

1911 ), Turc Pike (Turc, 1954 ; Pike, 1964 ; Milly and

Dunne, 2002 ), Budyko (Budyko, 1974 ), Fu (Fu, 1981 ;
Zhang et al., 2004 ; Yang et al., 2007 ); Choudhury

Yang
(Choudhury, 1999 ; Yang et al., 2008 ), Zhang two-
parameter model (Zhang et al., 2001 ), and a linear model
by Potter and Zhang ( 2009 ). These models are driven by
the aridity index, and they do not use explicit conceptual-
isations of catchment processes. They typically include one
parameter, treated as fixed and not necessarily related to

-

5.3.2 Index methods
Index methods assume that the locally scaled signature
of interest, or some functional form of it, is the same for
all catchments in the group, which is called homoge-
neous if it fulfils this assumption. In the following, index
methods for mean and variability of the annual runoff
are discussed.