Outcomes of synthesis - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 375

Table 12.4. Examples of Newtonian and Darwinian (co-evolutionary) similarity measures/predictors for the runoff
signatures

Similarity
measures/
predictors

Ch. 5
Annual

Ch. 6
Seasonal

Ch. 7
FDC

Ch. 8
Low flows

Ch. 9
Floods

Ch. 10
Hydrographs

Newtonian annual

precipitation

air temperature,
subsurface
storage

air temperature,
geology

precipitation, air
temperature,
geology

event character-
istics, soil
characteristics

soil character-
istics, wetness
index

Darwinian (co-
evolutionary)

area, aridity,
drainage
density,
vegetation

vegetation
cover,
phenology

aridity,
hydraulic
channel
geometry

riparian
vegetation,
wetlands

mean annual
precipitation,
drainage density

aridity,
hydrological
landscape units

the Newtonian approach has remained the dominant para-
digm since Freeze and Harlan

co-evolved objects is not rare in the literature. While
Chapter 7 uses Newtonian similarity measures such as
geology, an interesting Darwinian measure would be the
hydraulic channel geometry, at-a-site and downstream, as
it is indeed the result of the co-evolution of runoff along
with the development of the stream network itself. Predict-
ors of low flows ( Chapter 8 ) depend on whether we are
dealing with winter or summer low flows, so a first-order
similarity measure is one that is able to discriminate
between the two (such as air temperature, or elevation as
a proxy). Storage, while hard to quantify, is a Newtonian
index. Interestingly, the observation that low flow reces-
sion is often log-linear may, in fact, be a result of co-
evolution, and can be used to predict low flows in the case
of persistent droughts. Floods ( Chapter 9 ) have particularly
interesting Darwinian similarity measures because of the
strong evolutionary interactions between floods, the land-
scape and soils. Mean annual precipitation is an excellent
predictor of flood frequency, almost always much better
than event-scale extreme precipitation. From a Newtonian
perspective this is counterintuitive, but from a Darwinian
perspective it is not. In many climatic regions (in particular
humid regions) mean annual precipitation may be a good
predictor because of a close correlation with extreme
event-scale precipitation, antecedent (seasonal-scale) soil
wetness, and (decadal-scale) evolutionary interactions with
the landscape and soils. Other important Darwinian type
similarity measures are catchment area and the stream
network density. Out of all the runoff signatures, the case
of floods provides the clearest sign that co-evolution is
indeed operative and that the resulting emergent patterns
can be used to make predictions. For runoff hydrographs
( Chapter 10 ), Newtonian predictors such as soil character-
istics estimated through pedo-transfer functions are used,
but their areal representation in terms of hydrological
response units is a Darwinian way of breaking up the
landscapes into parts where similar hydrological processes
operate as a result of co-evolution.

s( 1969 ) blueprint of a
physically based model. Several pieces of evidence taken
from this topic show the value of the Darwinian approach
in catchment hydrology and PUB. These are presented
next.

'

Newtonian vs. Darwinian (co-evolutionary) similarity
measures/predictors from the topic
The most important evidence is probably the similarity
measures discussed throughout the topic. These similarity
measures are also predictors of runoff in ungauged basins.
They fall into two categories. The first type is based on the
Newtonian paradigm of invoking direct causality; an
example is the topographic wetness index of Beven and
Kirkby ( 1979 ) that is based on the local competition of
recharge and drainage mediated by Darcy

s law ( Table
12.4 ). The second type does not start from causality based
on local- or small-scale equations, but arises from the
treatment of the catchment as a complex system in which
Darwinian (co-evolutionary) similarity measures are found
or chosen that account for the diverse feedbacks between
hydrology, climate, geomorphology, soils and vegetation
( Table 12.4 ).
In the case of annual runoff ( Chapter 5 ), the aridity
index appears to be the clearest similarity measure, and
therefore predictor, through the Budyko relationship. Arid-
ity is thus a similarity measure of the Darwinian kind. As
annual runoff is the foundation of all runoff signatures,
aridity was also found to be useful in analysing all other
signatures in terms of comparative performance. In dis-
cussing seasonal runoff ( Chapter 6 ), the main similarity
measures used (such as air temperature, or elevation as a
proxy) are Newtonian, but vegetation cover and phenology
may provide the basis for Darwinian type similarity meas-
ures. The use of Chernoff faces as a way to classify
seasonal flow regimes (as seen in Chapter 6 ) is another
indication that

'

the holistic view of catchments as