Environmental Engineering Reference
In-Depth Information
Scenario I: no flooding
Scenario II: flooding
Rainfall/Evap
Rainfall/Evap
SW
SW
Recharge/ET
Recharge/ET
Fig. 22.11
Representation of the
simulation of saturated overland flow in
iSISMOD. Evap., evaporation; ET,
evapotranspiration; GW, groundwater;
SW, surface water.
GW
GW
(iSISMOD) in the local approach are better than
those generated using the uncoupled models in
regional approach. Figure 22.12 allows direct com-
parison of the FPMs obtained using the local and
regional approaches. Comparison reveals that ap-
plication of MODFLOW at a regional scale, with a
grid size of 5000 m (Fig. 22.12a), produces a poor
representation of flood probability. Specifically, it
under-predicts flooding in both extent and fre-
quency, which is to be expected given the coarse
representation of geomorphic landform features
provided by the 5000-m grid and the inability of
the model to reproduce the dynamics of water at
the surface. The finer resolution of the 500-m grid
(Fig. 22.12b) improves the identification of
areas subject to groundwater-induced flooding
markedly.
However, comparison with the FPM
generated using the fully coupled GW-SW model
(Fig. 22.12c) shows that this is also superior in
several other regards:
1 The predicted flooding has increased through-
out the test area in extent and frequency and better
matches observed patterns. The improvement is
particularly strong for the extent of areas flooded
during a 2-year return period event.
2 The coupled model accurately detects the
change in flood probability in low-lying, down-
stream areas, which is the logical consequence of
these areas receiving surface floodwaters from
upstream areas.
3 The appearance of flooding in low-lying, down-
stream areas (which was not depicted in the
regional modelling) for events of low probability
of occurrence is also an improvement emerging
from the fact that iSISMOD correctly simulates
what happens when surface water elevations ex-
ceed the threshold set by the crest elevations of
relict dune features in the micro-relief along the
troughs between the major dunes.
The improved FPMs produced through applica-
tion of iSISMOD suggest that coupled GW-SW
modelling provides the necessary basis for the
generation of realistic, physically based predic-
tions of flood probability, although limitations
inherent to FPM methodology mask some of
the advantages over using MODFLOW alone. Spe-
cifically, the fact that the maps are based on cal-
culation of the annual probability of the
occurrence of flooding means that the FPMs are
rather insensitive to the modelling approach
adopted. In fact, this was one of their strengths in
terms of the studies used to support the IMP for the
R´o Salado basin, as it made the results extremely
robust.
However, the annual series of flood probabil-
ities is not the best vehicle for demonstrating
the benefits of more physically based, coupled
models of floodingmechanisms in the generation
of flood probability maps. The benefits of im-
proved modelling would be better expressed for
flood mapping and management purposes if,
in the flood probability analysis, the annual
series were replaced by a peak-over-threshold
record, based on observations of each flood
occurrence.
