Introduction - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 10

its role in regionalisation of models and parameters.
Hydrological similarity also appears explicitly in the com-
parative assessment of the performance of runoff predic-
tions in ungauged basins around the world in each of these
six chapters.
Synthesis across scales is reflected in the fact that statis-
tical methods and process-based methods are treated in a
consistent way throughout Chapters 5 to 10. Statistical
methods and process-based methods represent different
approaches to deal with scale issues that arise in predicting
runoff in ungauged basins. Statistical methods usually are
lumped or holistic representations of the entire catchment
system, or many catchments, based on the behaviour
observed at the catchment scale. They are therefore typic-
ally representative of the downward approach. Process-
based methods, in contrast, are mechanistic methods based
on a causal understanding of water flow at the process
scale, and are therefore more representative of the upward
approach. A common structure in these chapters has been
adopted for comparability of the statistical and process-
based methods, which may help in understanding the simi-
larities and differences in how they bridge the scales.
Chapter 2 presents and articulates the synthesis
framework used in the topic. As data are the doorway to
enhanced understanding and improved predictions,
Chapter 3 is devoted to the specific data needed for making
predictions in ungauged basins (and reflecting the PUB
theme of new data collection approaches). Chapter 4 deals
with the key issues of flow paths and storage in catch-
ments, and lays the foundations of the general process
insights used in the remaining chapters (and reflecting the
PUB theme of conceptualisation of process heterogeneity).
Much of what controls runoff is in the subsurface, so
understanding these

processes is particularly important for predictions in
ungauged basins. Chapters 5 to 10 are the main chapters
of the topic, each of them dealing with one runoff signa-
ture. The structure of each chapter is almost identical,
where first the practical needs of the particular signature
and its societal relevance are highlighted; in the next
section, the process interactions that underpin the signature
are reviewed, including how these can be used to define
hydrological similarity. The following two sections review
statistical and process-based methods for predictions in
ungauged basins (reflecting the PUB theme of new model-
ling approaches). Again, wherever possible and reason-
able, the types of methods are organised in a similar way
across all chapters. Chapters 5 to 10 all close with a
comparative assessment of the performance of methods
of runoff predictions in ungauged basins around the world,
based both on a literature review and on dedicated com-
parative analysis of numerous data sets that underpinned
many of these historical studies. The performance assess-
ment is carried out through a cross-validation of model
predictions in over 20 000 catchments, which represents
one measure of predictive uncertainty: it reflects the
PUB theme focused on uncertainty analysis and model
diagnostics. Chapter 11 contains several case studies from
around the world. The purpose here is to highlight the
societal relevance of predictions in ungauged basins in
different contexts, and to demonstrate that many of the
methods presented in the topic actually work for purposes
that are important to society. Finally, Chapter 12 synthe-
sises the findings of the previous chapters, and undertakes
a yet higher level synthesis to generate profound conclu-
sions and implications for hydrological science, and
recommendations for hydrological practice, as summarised
in Chapter 13 . flow paths and storage-related