Prediction of annual runoff in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 94

Figure 5.22 and Tables A5.1 and A5.2 indicate that
overall a good global coverage was achieved. For annual
runoff, the assessments in humid and cold climates domin-
ate, while the inter-annual variability is mostly assessed
globally over several climate zones.
The analyses from the literature were stratified using the
climate classes based on the updated Köppen

2007 ). In this assessment, the locations with a Mediterra-
nean climate (Csa and Csb) have been considered ,

even though they would belong to the C climate in Peel
et al.( 2007 ). Finally, locations where the temperature of
the coldest month does not exceed 0 C are defined as
'

' arid '

(D and E climates in Peel et al., 2007 ). In studies
where catchments from different climate classes were ana-
lysed together, the dominant climate class was used in the
assessment. In the cases where catchments from all around
the world were analysed in the same study, the climate
class has been defined as

cold '

Geiger cli-
mate classification of Peel et al.( 2007 ). The following
classification criteria were applied. All locations with mean
annual precipitation lower than a threshold value are clas-
sified as belonging to the B climate in Peel et al.( 2007 )
and as

-

' global ' .

in this assessment. If 70% of the mean annual
precipitation occurs in winter, the threshold (in mm) is 20
times the mean annual temperature (in degrees Celsius). If
70% of the mean annual precipitation occurs in summer,
the threshold is 20 times the mean annual temperature +
28. Otherwise the threshold is 20 times the mean annual
temperature + 14. For areas where mean annual precipita-
tion exceeds the threshold, other climate types are possible.
In this assessment, the locations where the temperature of
the coldest month is not

' arid '

How good are the predictions in different climates?
Figure 5.23 shows that the highest performances are
obtained for the cold and humid catchments, while they
tend to be lower in arid regions. This indicates that the
prediction of annual runoff in regions with a

' surplus of

water over energy is easier than in drier climates. The

main reason behind this is that in arid regions a higher
variability of dominant processes and the possibility of
feedbacks with physiographic characteristics (i.e., evapor-
ation, geology and vegetation) complicate the estimation of
annual runoff.

'

lower than 18 C are called

' (A climate in Peel et al., 2007 ). The locations

where the temperature of the hottest month is above 10 C
and the temperature of the coldest month is between 0 and
18 C are defined as

tropical '

' humid ' (C climate in Peel et al.,

Which method performs best?
Figure 5.24 compares the performance of different
methods for estimating annual runoff and inter-annual run-
off variability. The regionalisation methods used here are
regression, index methods, spatial proximity and process-
based methods for the performance of spatial predictions
and methods using proxy data (i.e., tree rings) for the
temporal accuracy of runoff estimation. The analysis
includes ten results for annual runoff and ten results for
inter-annual runoff variability that used different regres-
sion approaches to estimate annual runoff characteristics.
The index method group includes eight results for annual
runoff and nine results for inter-annual runoff variability
that applied different Budyko-type models. The spatial
proximity group consists of ten results for annual runoff.

Figure 5.22. Map indicating the countries included in the Level 1
assessment.

Figure 5.23. Squared correlation
coefficient (r 2 ) of predicting annual
runoff (left) and inter-annual runoff
(right) in ungauged basins stratified by
climate. Each symbol refers to a result
from the studies shown in Table A5.1
(annual runoff) and Table A5.2 (inter-
annual runoff). Triangles indicate the
temporal variability assessment based
on tree rings. Lines indicate studies
where the same method was applied
across different climate regions.
Boxes show 25% - 75% quantiles.

Annual runoff Interannual variability

1.0

0.8

0.6

0.4

0.2

Arid Cold Humid Global Arid Tropical Cold Humid Global

Climate region Climate region