Prediction of floods in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 209

Figure 9.15. Regional mean annual
flood (a) and 500-year flood
(b) scaled by mean annual flood as
a function of median annual
precipitation for regions around the
world. From Meigh et al.( 1997 ).

10000

a)

Korea 2

W Africa 1

1000 Korea 1

W Africa 4

Philippines 1
PNG
Java and Sumatra

Thailand 2

Brazil

W Africa 5

W Africa 3

Sri Lanka

Arid and
semi-arid

Iran

Kerala

Thailand 1

Saudi Arabia

W Africa 2 Phillippines 3

Phillippines 2

SA and Botswana

Swaziland

100

Malawi

Namibia

Zimbabwe

10

100 300 1000 3000 10000

Median annual rainfall (mm)

100

b)

30

Namibia 1

Arid and
semi-arid

SA and
Botswana 1

Saudi Arabia

Swaziland
Zimbabwe
Zimbabwe 2
Zimbabwe 3
Malawi

Sri Lanka 1

10

Namibia 2

Philippines 1

Iran 1

Philippines 2

Thailand 3
Thailand 5

Philippines 3
Philippines 4
Sri Lanka 3

Iran 2

SA and Botswana 2

Kerala

Thailand 4

Philippines 5

W Africa 5

Korea
W Africa 1
W Africa 2

3

Java and Sumatra 1

PNG

Thailand 1
Java and Sumatra 2

Brazil

W Africa 3

W Africa 4

Thailand 2

1

100 300 1000 3000 10000

Median annual rainfall (mm)

conjunction with ANNs to define a hydrological neigh-
bourhood and to define the relationship between catch-
ment/climate and flood quantiles (Cavadias, 1990 ;
Ribeiro-Corréa et al., 1995 ; Ouarda et al., 2008 ; Shu and
Ouarda, 2007 ). As an alternative, geostatistical methods can
be combined with regression methods as proposed by
Skøien et al.( 2006 ) . They correlated the flood moments
with catchment/climate characteristics such as mean annual
precipitation and the FARL (flood attenuation by reservoirs
and lakes) index and regionalised the residuals to the top-
kriging methods. The method allowed them to both account
for local catchment effects and exploit the spatial similarity

of floods along the stream network. Example applications
include Saxonia (Walther et al., 2011 ), Tirol (Rogger et al.,
2011 ) and Austria as a whole (Merz et al., 2008 ). Their
approach is illustrated in the case study of Section 11.10 in
the context of implementing the European Flood Frame-
work directive.

9.3.4 Estimation from short records
The index flood method can be used not only for estimat-
ing floods in ungauged basins but also in basins with short
records (Dalrymple, 1960 ). The approach is the same as for