Prediction of low flows in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 171

Figure 8.6. Catchment groupings
in north-west Italy, based on four
grouping methods for low flows:
(a) weighted cluster analysis (WCA);
(b) residual pattern approach (RPA);
(c) classification and regression
trees (CRT); (d) seasonality indices
method (SI). From Vezza et al.
( 2010 ).

finding homogeneous regions. Laaha and Blöschl ( 2006b )
considered three seasonality indices, the variability and
mean day of occurrence of Q 95 , the seasonality ratio
between summer and winter low flows, and the seasonality
histogram. They found this method to work very well for
Austrian conditions, which is likely related to the strong
contrast in seasonality of low flows in their area. Vezza
et al.( 2010 ) examined similar seasonality indices and
classified their domain into two main units ( Figure 8.6d ).
Group 1 is the Apennine

all classification methods used by Vezza et al.( 2010 )
separate the south-eastern Apennine Mediterranean area

from the rest of the study domain (the Alps mountain
range), even though they used different attributes to carry
out this division (e.g., the percentage of forest, seasonality
of low flows, combination of several parameters) ( Figure
8.6 ). This suggests that

-

the obtained classification is

robust.
A number of studies compared the relative strengths of
the grouping methods (Nathan and McMahon, 1992 ,in
Australia; Laaha and Blöschl, 2007 , in Austria; Vezza
et al., 2010 , in north-west Italy; Engeland and Hisdal,

Mediterranean area where low
flows normally occur during summer and Group 2 is the
Alpine region, characterised by winter low flows. In fact,

-