Prediction of seasonal runoff in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 109

Figure 6.7. Lag in evaporation (E)
relative to leaf area index (LAI)
increase and potential evaporation
(E p ) increase at the deciduous
Morgan Monroe State Forest,
Indiana, USA, during the first 6
months of the year. Results are
averages of E p computations (using
the Penman Monteith equation) and

measurements over a 6-year period.
Photos: D. Dragoni.

-

changes in seasonal flow regime could be used to track
vegetation phenology and the onset of spring in eastern
USA. Three runoff metrics were examined: (i) the differ-
ence between precipitation and runoff volume (P

to allow seasonal changes and bud-burst dynamics to be
inferred directly from runoff observations. The implication
of these observations for seasonal runoff predictions is that
the spring increase in evaporation associated with trees
leafing out significantly impacts runoff volumes and
dynamics, and cannot be ignored under these circum-
stances. On the other hand, phenology does not appear to
have a strong effect in areas with uniformly high potential
evaporation and leaf area, such as the Amazon rainforest
(Czikowsky et al. , unpublished data). In areas with signifi-
cant deciduous forests (including dry or drought deciduous
forests in the tropics and deciduous Mediterranean environ-
ments), however, the rapid increase or decrease in transpir-
ation associated with leaf dynamics can impose significant
changes on seasonal runoff (Cayan et al., 2001 ).

R), cal-
culated with a 30-day moving average; (ii) a measure of the
mean recession constant, estimated as the time period
required for a flood hydrograph to decay to 1/e of its peak;
and (iii) the amplitude of diurnal runoff oscillations; all of
which are impacted by increased transpiration activity.
These runoff metrics were computed for 73 locations where
comparison could be made between hydrological

−

' spring

onset and independent estimates of spring increase. The

results are schematised in Figure 6.8 and demonstrate
coherent relationships between the phenological signature
identified in the runoff record (i.e., panels a, b and c, relating
to the above runoff metrics) and that inferred from terrestrial
measurements. The delay of 20

'

-

25 days in the hydrological
spring compared to the atmospheric conditions mirrors the
delay between potential and actual evaporation observed at
the Morgan Monroe site ( Figure 6.7 ) , and likely reflects the
time scale needed for increased evaporation to change
catchment storage at the end of winter, and may be influ-
enced by the proportion of deciduous species in the catch-
ment. The runoff-based estimates of spring date can also be
used to track the onset of spring conditions across eastern
USA (Czikowsky and Fitzjarrald, 2004 ).
The most important message from Figure 6.8 is that the
signature of phenology in runoff data is thus strong enough

Inter-annual variability in the flow regime
There are often significant variations in the seasonal
patterns of runoff between years (see the grey bands in
Figure 6.3 ). Inter-annual variability of seasonal runoff can
be quantified by calculating the stability (or regularity) of
the seasonal flow regimes (Krasovskaia, 1995 ; Pfaundler
et al., 2006 ). Changes in the timing of the seasonal flow
regime can be related to the seasonal distribution and
quantity of precipitation, while changes in the magnitude
of the flow regime curve have been linked to large-scale
atmospheric circulation, strength and timing of monsoonal
systems, and the effectiveness of catchment storage in