Outcomes of synthesis - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 377

identifying the key predictors, and exploring their relative
performances, we gain insights that help build generalised
understanding into the behaviour of catchments, in a most
Darwinian sense.
The power of the Newtonian approach derives ultim-
ately from the fact that it is based on Newtonian mechan-
ics, universal balance laws of mass, momentum and energy
for simple systems with a clear causality. Specifically in
hydrology, it benefits from the understanding gained of
individual processes. While the Newtonian approach can
be carried out both locally and spatially (e.g., by compar-
ing predictions of distributed models with observed spatial
patterns), for logistical reasons any new insights are gained
from one or more local process studies. It will continue to
gain predictive power through advances in observing and
understanding processes, including through detailed obser-
vations at all scales, and improved theories to handle
landscape heterogeneity and the complexity of flow path-
ways and residence times.
In contrast, the Darwinian approach recognises that catch-
ments are complex systems that have co-evolved from the
interplay of climate, landform, vegetation, soils and geology.
It is much harder to learn about Darwinian systems from a
single location; there is too much complexity. A viable
approach to shedding light on complex systems is a compara-
tive hydrology approach that contrasts different catchments
in different regions around the world to learn about the
processes of co-evolution. Thus, the Darwinian approach is
synoptic, amenable to generating generalised understanding
about the co-evolution of catchments, through comparative
studies along climatic, geological and other gradients. Inter-
estingly, Chapter 6 presented examples of two schools of
thought in respect of seasonal runoff: the geographic, which
focuses on classifying seasonal runoff into regime types and
mapping them across the landscape, and the engineering,
which focuses on quantitative estimation and is generally
place-based. They are both valid approaches, and reflect
competing paradigms, the Newtonian and Darwinian.
Of course, each paradigm has strengths and weaknesses.
The Darwinian learns from contrasting different regions and
interprets them in terms of the processes across many time
and space scales. It allows for the results of feedback pro-
cesses in the landscape, for co-dependencies and adaptive
processes. However, because of the complexity, causality
may not always be as clear as in the Newtonian approach.
On the other hand, the power of the Newtonian approach is
in explicitly including cause

Figure 12.10. Synthesis of the Newtonian and Darwinian approaches
through understanding, prediction and assessment.

the benefits of both, while overcoming their weaknesses
(Harte, 2002 ). The synthesis may gain potency through a
combination of simple causality and complex co-
dependencies. For example, Gaál et al. (2012) illustrated
how the comparison of catchments of contrasting character-
istics can help to recognise the combined effect and inter-
play of flood processes on the landscape. They showed that,
in one region, landform had adapted to the flashiness of
floods, producing efficient drainage networks, which in turn
enhanced the flashiness of the flood response. In other
regions, tortuous drainage networks have evolved, which
in turn retarded the flood response and impeded the evolu-
tion of an efficient drainage network. This is a result not
likely to have been discovered by the Newtonian approach.
The notion of a Newtonian Darwinian synthesis has

been a recurring theme right through the PUB decade
(Sivapalan, 2003a , 2005 ; Sivapalan et al., 2003; McDon-
nell et al., 2007 ) as the possible framework to overcome
the major challenges facing hydrological predictions and
as the possible foundation for a new theory of hydrology at
the catchment scale. This received a fillip through the
Hydrological Synthesis Project of the US National Science
Foundation (Sivapalan et al., 2011b ), and these ideas are
further reinforced through the outcomes of the synthesis
carried out for this topic. Throughout the topic, the ana-
lyses of prediction performances have indicated strong
elements of both the Newtonian and Darwinian approaches
to predictions in ungauged basins.
As we study differences in behaviour between catch-
ments as legacies of co-evolution, the goal is to discover
patterns and connections. The Budyko curve is the best
example of an empirical co-evolutionary pattern, which
remains to be deciphered at a fundamental level. When
we combine it with related drainage density patterns
(Wang and Wu, 2012 ), vegetation patterns (Xu et al.,
2012 ), and soil and vegetation catena effects (Hwang
et al., 2012 ), one is tempted to ask if there are deeper
organising principles in action. Hydrologists have been

-

effect relationships, but it
quickly loses effectiveness in making general predictions
by not being able to account for process interactions, feed-
backs and parameter co-dependency that one invariably
finds in natural co-evolved catchments. Therefore, a synthe-
sis of both paradigms (see Figure 12.10 ) would help to
advance hydrological prediction in a way that brings out

-