Geoscience Reference
In-Depth Information
Figure 4.5
Observed and predicted discharges using the IHACRES model for (a) Coweeta Watershed 36 and
(b) Coweeta Watershed 34: Top panel: observed and predicted flows; middle panel: model residual series;
lower panel: predicted total flow and model identified slow flow component (after Jakeman and Hornberger,
1993, with kind permission of the American Geophysical Union).
linear models of Box 4.1 with the total discharge as the output. The parameters of the complete model
are calibrated by fitting transfer functions to different values of
c, τ
w
and
f
until the best results are
achieved. Standard errors and covariances for the transfer function model parameters can be estimated,
but uncertainty in the
c, τ
w
and
f
parameters has not generally been considered.
The IHACRES model has now been applied to a wide variety of catchments (Jakeman
et al.
, 1990,
1993a; Jakeman and Hornberger, 1993; Post and Jakeman, 1996, 1999; Post
et al.
, 1998; Sefton and
Howarth, 1998; Littlewood, 2002; Kokkonen
et al.
, 2003; and see, for example, Figure 4.5), including
catchments subject to significant snowmelt inputs (Schreider
et al.
, 1997; Steel
et al.
, 1999) and in pre-
dictions of the impacts of climate change on catchment hydrology (Jakeman
et al.
, 1993b; Schreider
et al.
, 1996). The model has also been linked to erosion and water quality components (Jakeman
et al.
1999). The results generally show that the parallel transfer function of Figure 4.4 is a suitable structure
for rainfall-runoff simulation at the catchment scale, with one fast flow pathway and one slower flow
pathway (see Box 4.1). The fast flow pathway provides the major part of the predicted storm hydrograph;
the slower pathway, the major part of the recession discharge between storm periods. Note again that this
does not imply anything about whether surface or subsurface flow processes, old or new water, direct
flow or displacement are involved in the fast and slow flow responses. The fast flow pathway should not
be interpreted as a surface flow pathway; it could equally be a response controlled by the time scale of
displacement of old water from subsurface storage. This type of parallel pathway linear routing is also
used in many other models (including the Xinanjiang, Arno, VIC model described in Box 2.2; HyMOD,
which has been widely used in studies on the calibration of hydrological models (e.g. Boyle
et al.
, 2000;
