Geography Reference
In-Depth Information
12 Outcomes of synthesis
Contributors: H. V. Gupta,* G. Blöschl, J. J. McDonnell,
H. H. G. Savenije, M. Sivapalan, A. Viglione and
T. Wagener
12.1 Learning from synthesis
This topic has addressed the problem of predictions of
runoff in ungauged basins, i.e., predictions of runoff at
those locations where no runoff data are available, and is a
contribution to the IAHS Predictions in Ungauged Basins
(PUB) initiative (Sivapalan et al., 2003b ) . It has strived to
bring together the outcomes of extant research on the
problem of making predictions of runoff in ungauged
basins that had previously remained disparate and frag-
mented, like scattered pieces of a jigsaw puzzle yet to
reveal any coherent picture.
An original feature of the topic is that it provides exten-
sive and comparative assessments (in the form of blind
tests) of the relative performance of runoff prediction
methods in a large number of catchments from around
the world. Rather than applying the prediction methods
ourselves, we analysed the outcomes of many previous
prediction efforts, and the collective experiences of their
authors, who thus became partners in a unique community
effort. The outcomes of this topic therefore demonstrate
the potential for new knowledge and insights to be gained
through the practice of hydrology when a community of
researchers organises around a pressing topic.
To respond to the fragmentation of knowledge implicit
in multiple disparate studies, which became particularly
obvious in the results of the comparative assessment, we
have taken recourse to
snapshots of the full spectrum of catchment processes that
contribute to runoff variability at all scales, and have
common origins in catchment hydrological processes
(Jothityangkoon et al., 2001 ; Wagener et al., 2007 ). The
dominant theme of hydrological similarity arises as a
framework to perform synthesis across places. Hydro-
logical similarity acts as a vehicle to advance understand-
ing and predictions, and serves as the foundation for
regionalisation, which remains at the heart of predictions
in ungauged basins. Finally, a consistent treatment of statis-
tical and process-based predictive techniques reflects the
need to address a synthesis across scales. Statistical
methods and process-based methods raise different chal-
lenges in terms of how to deal with scale issues, and require
different approaches when used for predictions of runoff in
ungauged basins. Intermediate approaches such as index
methods are also adopted. By separating time scales of
hydrological signatures, defining and relying upon hydro-
logical similarity, and by suitably scaling the statistical and
process-based predictive techniques used, the three core
research issues or programmes outlined in the PUB Science
Plan (SSG, 2003 , p. 10) can be addressed:
(1) Evaluating the performance of existing prediction
methods that could be used in ungauged basins, through
comparative analysis of a full range of methods and
detailed investigations of the processes governing the
quantity of interest
;
(2) demonstrating the value of data, knowledge and process
understanding for improved hydrological predictions ;
(3) understanding the hydrological functioning, i.e., dominant
processes, of basins at different time/space scales, and
how these vary across scales in the different hydroclimatic
regions of the world
Common amongst all three programmes will be the
sharp focus on realistic estimation and eventual reduction
of the level of predictive uncertainty. Robust measures of
predictive uncertainty will be adopted as the criteria
of success of PUB.
As outlined in the PUB Science Plan, predictive uncer-
tainty, measured in this topic by the cross-validation
, which had the aim of
unifying existing, diverse pieces of information to discover
previously unrecognised connections, and to develop sci-
entific understanding that is valid across multiple pro-
cesses, places and scales (Blöschl, 2006 ; Sivapalan et al.,
2011b ).
Synthesis across processes is reflected in the independ-
ent exploration of runoff signatures at different temporal
scales, ranging from annual runoff, to seasonal runoff, to
the flow duration curve, to low flows, to floods and to the
detailed runoff hydrograph. These signatures provide
'
synthesis
'
 
Search WWH ::




Custom Search