Geography Reference
In-Depth Information
threshold, and therefore may or may not be flooded, as
effectively they cannot be imaged by TerraSAR-X.
lack of adequate data distributed over space and time. As
a consequence it is in urban areas where observations of
flood dynamics are most needed. This is addressed in a
recent study by Schumann et al. (2011) where a unique
space-borne radar data set consisting of five images with
three additional aerial photographs over one single event
hydrograph of the summer 2007 flooding of the town of
Tewkesbury (England, UK) has been assembled.
Previous observations of urban flooding have used
single image and ground wrack mark data and have
therefore been unable to adequately chart the propaga-
tion and recession of flood waves through complex urban
topography. By using a combination of spaceborne radar
and aerial imagery Schumann et al. (2011) were able to
show that remotely sensed imagery, particularly from the
new TerraSAR-X radar, can reproduce dynamics ade-
quately and support flood modelling in urban areas. They
illustrated that image data from different remote sens-
ing platforms reveal sufficient information to distinguish
between models with varying spatial resolution, particu-
larly toward the end of the recession phase of the event
(Figure 6.11).
Findings also suggest that TerraSAR-X is able to com-
pete with aerial photography accuracies and can point to
structural model errors by revealing important hydraulic
process characteristics which might be missing in models;
however, as Schumann et al. (2011) note, this has only
been demonstrated for a particular urban setting and
different results should be anticipated for different types
of urban area.
Nevertheless, from this work it can be concluded that
SAR imagery from as far as several hundred kilometres
from the Earth's surface possesses the potential to deliver
important information about floodplain dynamics that
can be used to identify and help build suitable models,
even in urban environments.
A study by Pulvirenti et al. (2011b) using COSMO-
SkyMed images confirms the ability of metric resolution
SAR to map flooding with an unprecedented precision.
They successfully mapped flood dynamics over a period
of five days from a set of very fine resolution (1
6.4.3 Multi-temporal SARimages to inform
aboutfloodplaindynamics
Although most studies integrating remote sensing derived
flood parameters and flood inundation models have been
very successful in one way or the other, they are mostly
restricted to rural areas and single image data, and only
very few have attempted to observe and map flood dynam-
ics. The most notable efforts to date were undertaken by
Bates et al. (2006) on a predominantly rural reach of the
River Severn in west-central England, and more recently
by Schumann et al. (2011) in the town of Tewkesbury
and also by Pulvirenti et al. (2011b) for a river reach in
northern Italy.
On a 16 km reach of the River Severn, west-central
England, Bates et al. (2006) use airborne synthetic aper-
ture radar to map river flood inundation synoptically
at fine spatial resolution (1.2 m). Images were obtained
at four times through a large flood event between 8
and 17 November 2000 and processed using a statisti-
cal active contour algorithm (Horritt, 1999) to yield the
flood shoreline at each time. Intersection of these data
with a high vertical accuracy survey of floodplain topog-
raphy obtained from airborne laser altimetry permitted
the calculation of dynamic changes in inundated area,
total reach storage and rates of reach dewatering. In addi-
tion, comparison of the data to gauged flow rates, the
measured floodplain topography and map data giving the
location of embankments and drainage channels on the
floodplain yielded new insights into the factors control-
ling the development of inundation patterns at a variety
of scales.
Finally, the data were used to assess the performance
of a simple two-dimensional flood inundation model,
LISFLOOD-FP (Bates and de Roo, 2000), and allowed,
for the first time, to validate the dynamic performance
of the model (Figure 6.10). This process is shown to give
new information into structural weaknesses of the model
and suggests possible future developments, including
the incorporation of a better description of floodplain
hydrological processes in the hydraulic model to represent
more accurately the dewatering of the floodplain.
Like the case study reported above, remote sensing
studies of flooding dynamics to date have focused almost
exclusively on rural reaches. Urban areas, where most
assets are located and where flood risk is thus very high,
have been given relatively little attention, mainly due to
×
1 m) X-band SAR images for a flood event in northern
Italy in 2009. An unsupervised clustering algorithm was
developed to segment the images into homogeneous, non-
noisy regions. To associate the segments to different stages
of the flood evolution, some distinctive multi-temporal
backscattering trends, common to many regions, have
been singled out. The authors noted that changes in
vegetation and scattering mechanisms contribute most to
the radar signal as the water level changes and that these
Search WWH ::
Custom Search