Environmental Engineering Reference
In-Depth Information
2
0
1.8
2
Precipitation
1.6
4
Imperviousness 30%
1.4
6
Imperviousness 44%
1.2
8
1
10
0.8
12
0.6
14
0.4
16
0.2
18
0
20
0
4
8
12
16
20
24
28
32
36 40 44
Time [hours]
48
52
56
60
64
68
72
76
80
FIGURE 18.7
Hydrographs for scenario 1 with 30% imperviousness (default) and 44% imperviousness (IKONOS) for the
period from 3 May 2005 1.00 am till 6 May 2005 9.00 am.
was considered as covered by grass. The average degree of
imperviousness for different urban land-use classes obtained
from IKONOS and Landsat data are quite similar (Table 18.2).
For some classes, however, the values derived from the
proportion maps strongly differ from default values found in the
literature. This is especially the case for the low-density built-up
and city center classes, where levels of imperviousness are much
lower than assumed values. Also for roads and highways remote
sensing estimates are lower than the default value. For the high
density built-up class, for infrastructure and for industrial area
remote sensing estimates are higher than the assumed level of
imperviousness. The use of class-specific levels of imperviousness
produces maximum peak discharges that are higher than the
values obtained by defining one average level of imperviousness
for the whole built-up area (between 5% and 10% higher for
the two most important peaks in rainfall intensity, depending
on the method used for estimating class-specific levels of
imperviousness) (Fig. 18.8). The map of estimated runoff shows
a clear increase in the value of the runoff coefficient in the
urbanized area close to the river outlet in the northern part of the
catchment, and a decrease in runoff values in areas located further
away from the outlet, compared to scenario 1 (see Fig. 18.10).
The more pronounced spatial variation in runoff coefficient
values, however, seem to have only a minor effect on estimated
peak discharges for less intense rainfall events (Fig. 18.8).
impervious surfaces, bare soil and water, as obtained from the
IKONOS- and Landsat-derived proportion maps. The runoff
coefficient and depression storage for each cell are then estimated
as an area weighted average of the parameters for the four classes.
Analysing and comparing the results for the three scenarios, the
runoff calculated in scenario 3, and based on the IKONOS data,
produces the highest peak discharges. The peak discharges are
up to 15% higher than in the non-distributed simulation based
on IKONOS data (scenario 1) (Fig. 18.8). This demonstrates
that use of detailed information about the spatial distribution of
land-cover within urban areas may have a clear impact on the
modeling of peak discharges at the outlet of the catchment and,
therefore, on flood prediction. The spatial variation in runoff is
obvious from the map of runoff coefficient values obtained for
scenario 3 (see Fig. 18.10). High values for the runoff coefficient,
in the range of 0.8 - 1.0, prove to be linked to a closely connected
pattern of impervious areas. In other words, the spatial variation
in runoff and the connectivity between cells with high runoff
coefficients in particular, proves to be a major factor influencing
estimated discharge volume.
The hydrograph simulated in scenario 3, but based on Landsat
ETM
+
data, has lower peak values than the one based on
IKONOS data (Fig. 18.9). This is most likely due to a smoothing
effect in the distribution of impervious surface proportions
caused by the sub-pixel classification process (see also Fig. 18.5),
with a direct impact on the spatial variation of runoff coefficient
values (Van de Voorde, De Roeck and Canters, 2009). As can be
seen in Fig. 18.10, the spatial patterns of runoff for scenario 3,
using either IKONOS or Landsat ETM
18.5.3.3
Scenario 3: Fully distributed
approach
Finally, a fully distributed scenario was applied, where each
cell in the urban area is assigned its proportion of vegetation,
data are very similar,
yet for the Landsat scenario the pattern appears to be more
generalized. This suggests that sub-pixel estimation of land-
cover class proportions based on Landsat ETM
+
+
data is less
Search WWH ::
Custom Search