Geoscience Reference
In-Depth Information
catchments, it is much more difficult if parameter values have to be estimated
a priori
, with no guarantee
of equally good performance in prediction (of course if a catchment is really ungauged then there will be
little reason to question the predictions, but researchers often play the game of pretending a catchment
is ungauged to test different prediction methods as if that catchment was ungauged).
The ungauged catchment problem has a long history. Early methods were mostly based on regressions
of model parameter values or runoff coefficients determined for gauged catchments against variables
representing the characteristics of those catchments. Once the regression equations had been developed
then they could be used to estimate the parameter values for the ungauged catchments of interest. This is
a purely empirical approach to the regionalisation problem, but it would be nice to think that we might
also be able to bring some hydrological science to bear on the problem. After all, a science that has to
rely on calibration for every site that its theory is applied to cannot be said to be mature (even if there are
very good reasons for these difficulties). This was recognised in the late 1990s when the International
Association of Scientific Hydrologists (IAHS) started to develop plans for the predictions in ungauged
basins (PUB) project. The PUB project has stimulated a lot of research activity, including attempts to
develop methods based on the classification of catchment hydrological responses and the investigation
of whether the problem could be solved by a minimum amount of gauging of the ungauged catchment.
These approaches are discussed in this chapter.
10.2 The PUB Initiative
The prediction of ungauged basins (PUB) project is an initiative of the International Association of
Hydrological Sciences (IAHS). Plans for the decade-long project were approved in 2001 and it started
formally in 2003 (Sivapalan, 2003).The science plan for PUB (see www.iahs-pub.org) recognises that
predicting the response of ungauged basins involves inherent uncertainties and that improving the pre-
dictions is primarily an exercise in constraining uncertainty, which may involve the collection of new
types of data as well as improving the scientific basis for hydrology.
The aims of PUB are summarised as:
•
to advance the ability of hydrologists worldwide to predict fluxes of water and associated constituents
from ungauged basins, qualified by estimates of predictive uncertainty;
•
to advance knowledge and understanding of climatic and landscape controls on hydrological processes
at all scales, in order to constrain predictive uncertainty;
•
to demonstrate the value of data for hydrological predictions and provide a rational basis for future data
acquisitions, including alternative data sources, by quantifying the links between data and predictive
uncertainty;
•
to advance the scientific foundations of hydrology and provide a scientific basis for sustainable river
basin management;
•
to promote and encourage capacity-building activities in the development of appropriate scientific
knowledge and technology to the areas and communities where it is needed most.
Two primary scientific targets were identified:
•
Target 1: Examine and improve existing models in terms of their ability to predict in ungauged basins
through appropriate measures of predictive uncertainty.
•
Target 2: Develop new, innovative models to capture space-time variability of hydrological processes
for making predictions in ungauged basins, with a concomitant reduction in predictive uncertainty.
