Prediction of annual runoff in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 77

water uptake (Harman et al., 2011a , b ). In Australia,
woody vegetation was shown to buffer annual transpiration
more effectively than non-woody vegetation, presumably
due to differences in root zone depth (Xu et al., 2012 ).
Vegetation cover is therefore both a response to the
partitioning of the water balance and a driver of the annual
water balance dynamics. Vegetation is also a significant
driver of weathering, of soil biogeochemistry, and a deter-
minant of soil hydraulic properties (Thompson et al., 2010 ;
Lucas, 2001 ). The role of vegetation in modifying its local
hydraulic environment can result in striking organisation
of the landscape. For instance, modification of soil
hydraulic properties by vegetation can result in the forma-
tion of spatial patterns in vegetation distribution, in which
bands of vegetation are interspersed with regions of bare
soil (Borgogno et al., 2009 ; Thompson et al., 2011a ). In
northern hemisphere rugged landscapes, the difference in
energy balance between north- and south-facing slopes
regularly leads to drought-adapted vegetation communities
on the south-facing slopes, and mesic vegetation on the
north-facing slopes. These vegetation differences are also
reflected in differences in soil depth, and carbon and nutri-
ent content (lower on the south-facing slopes) (Burnett
et al., 2008 ; Klemmedson and Weinhold, 1992 ; Franzme-
ier et al., 1969 ). These differences alter the storage cap-
acity and habitat quality of the slopes, providing a positive
feedback that exacerbates the differences between slopes
with different aspects, and ultimately driving both water
balance and catchment evolution (with vegetation, for
instance, suppressing erosion and runoff on north-facing
slopes, e.g., Cerdà, 1998 ; Istanbulluoglu et al., 2008 ).

The consequent increases in aridity in this strongly sea-
sonal Mediterranean climate led to more dramatic reduc-
tions in river flows to Perth

s dams. For example, the 16%
reduction in precipitation led to a 55% reduction in runoff.
Predicting runoff response to climatic changes is not
usually this straightforward (Montanari et al., 2010 ). For
example, reductions in precipitation could lead to
increased water stress on the vegetation, leading to pos-
sible forest thinning, changes in vegetation composition,
disease infestation and die-off, all of which can modify
annual runoff. The Budyko curve cannot capture the
transient changes in annual runoff associated with these
modifications, and may not be sensitive to vegetation or
soil changes even once the catchment reaches a new
equilibrium.
Increases in average temperature promote not only vege-
tation change, but also changes in snowfall, snow storage
and snowmelt regimes. These changes are likely to alter
seasonal runoff, and result in new patterns of annual runoff
as well. Several regions of the world have already seen
major dramatic changes as a result of increases in tempera-
ture, e.g., the Himalayas in India and Nepal, and California
in the western USA.
Since seasonality of climate, distribution of precipitation
throughout the year, and temporal patterns of precipitation
can be key determinants of inter-annual runoff variability
(Montanari et al., 2006 ), any changes in the seasonality of
these controls can also impact annual runoff. There are
historical examples where changes to the monsoon dynam-
ics and timing have led to huge changes in annual runoff
variability and collapse of entire civilisations, as in the
cases of the Indus Valley (Giosan et al., 2012 ) and the
Maya (Medina-Elizalde and Rohling, 2012 ).
Human-induced land use, water use and land cover
changes are the remaining catchment-scale factors altering
annual runoff. Examples include forest planting and har-
vesting, forest conversion to agricultural crops and urban
settlements, regulation of runoff by upstream storage, and
withdrawals for consumptive use (irrigation, municipal and
industrial use) (Peel et al., 2010 ; Vogel, 2011 ). Vegetation
change can be caused by human intervention, or may occur
because of adaptation to climate change. For example,
replacing a forest with crops or pasture typically reduces
evaporation, increasing annual runoff. This change can
manifest differently in different environments, depending
on the runoff generation processes. The effect of land use
and land cover change on runoff has been the subject of
many paired catchment studies around the world (e.g.,
Peck and Williamson, 1987 ; Brown et al., 2005 ; Bari and
Smettem, 2006 ), revealing, for instance, that transient
responses to land use change persist for longer during
afforestation than deforestation experiments, that land use
changes disproportionately affect low flows, and that there

'

Effects of global change
Given that the primary control of annual runoff variability
is climate, through the relative availability of water and
energy, changes in the magnitude or timing of precipitation
and temperature (or potential evaporation) could contribute
to major changes in annual runoff. A first-order indication
of the expected change can be approximated via the
Budyko curve. Changes in mean temperature (and hence
mean annual evaporation) and in mean annual precipitation
can be expressed as changes in the aridity index, E p /P.
Depending on the magnitude of this change, one could
'

and determine the new
value of E/P. For example, if the potential evaporation
remains constant and annual precipitation decreases, E p /P
will then increase (i.e., become more arid), and annual
runoff would be expected to decrease. A dramatic illustra-
tion of exactly this effect arises in south-west Western
Australia, as illustrated in Figure 5.8 . Observation records
in Jarrahdale (near Perth) over the past 100 years indicate
that annual precipitation went through a 16% step-change
reduction in 1975, and another small reduction in 1997.

move along the Budyko curve '