Environmental Engineering Reference
In-Depth Information
FIGURE 5.5
Partition of the urban area of L'Aquila, Italy, overlaid on a COSMO/SkyMed spotlight image. GIS polygons are
outlined in red. The original radar data was generated by the Italian Space Agency and provided by the Italian Civil Protection
Department.
earthquake in May 2008. In that case more complex texture
measures were used, i.e. homogeneity on a 51
Evaluating the vulnerability of existing building stock has
a long history of method proposed along the years (Calvi
et al
., 2006), based either on empirical, analytical or even hybrid
approaches. The various methods proposed need a considerable
amount of information to be collected either on past events (for
empirical methods) or on the physical features and characteristics
of the considered buildings (for analytical approaches), or both
(for hybrid methods). This generally constrains the estimation
procedure to small areas because the
in-situ
collection of data is
too expensive and time-consuming to be practical beyond some
limits. It would be desirable to trade some accuracy for a wider
geographic scope; enabling such change is the goal of a research
activity (Polli, Dell'Acqua and Gamba, 2009) in progress in the
framework of GEO (GEO, 2005).
Recently, new algorithms have been developed for seismic
vulnerability assessment, requiring fewer data than their pre-
decessors. There is one named SP-BELA (Simplified Pushover-
Based Earthquake Loss Assessment) (Borzi, Crowley and Pinho
2008) capable of providing a sensible output for comparison
purposes with a smallest set of inputs including footprint of the
building and number of storeys, which can be extracted from
remotely sensed data.
In the scientific literature it is possible to find lots of building
height extraction methods, both for optical and SAR imagery.
Existingmethodologies are either based on shadow analysis (Hill,
Mote and Blacknell, 2008) or interferometric data (Bennet and
Blacknell, 2003). The interferogram calculation however fails if
all of the roof backscattering is sensed before the double bounce
area and therefore superimposes with the ground scattering in
the layover region, which is usually the case for high buildings.
To tackle the problem of signal mixture from different altitudes
methods founded on interferometric or polarimetric data or
stereoscopic SAR are proposed (Simonetto, Oirio and Garello,
2005; Cellier and Colin, 2006).
Generally speaking, as testified by the amount of relevant
literature, the problem of extracting a building 3D shape is
quite a complex one. For the purposes of seismic vulnerability
estimation, however, such problem can be split into two sub-
problems, namely footprint extraction and determination of the
number of storeys. This is where VHR SAR data may help with
its side-looking geometry. Some, still unpublished, experiments
have indeed shown that analyzing the rows of reflectors found
in the building fa¸ade with image processing methods one can
actually infer the number of storeys in the building without a
need for precise determination of its height, which is less relevant
when seismic vulnerability is concerned. The final goal of the
51 pixel window
and displacements d
x
=
21, d
y
=
21, yet this appears to be
a confirmation of the fact that post-event images alone may
provide statistical clues on the damage level. Suppressing the
elements with DAR
=
0 in the series, the correlation reaches
0.338. This means that if one had a criterion to select blocks
reporting actual damage, a more reliable estimate of the damage
level could be done.
The apparent correlation with variance may be tentatively
explainedwith a stronger speckle connected to the wider presence
of small reflectors due to the randomly shaped debris stacks. It is
reasonable to hypothesize that the stability of the texture statistics
with varying windows sizes comes from the large number of
sample at one's disposal, reckoning around 120 000 pixels on
average for every single block in the case at hand, to be compared
with less than one hundred for a corresponding ERS-like image.
The research is still inprogress; latest results, not published yet,
indicate that correlations as high as 0.7may be reached by suitable
selection of texture measures, and fusion with information from
optical data allows classifying damage level into three classes with
an overall accuracy of around 84%.
×
5.3.3
Vulnerabilitymapping with
VHR SAR data
As mentioned above, the concept of using remote sensing to
assess the effect of disasters is not new, as low-resolution systems
could be used e.g. to map flooding extent (Tholey, Clandillon
and De Fraipont, 1998) or to assess seismic damage (Trianni
and Gamba, 2008), forest fires (Kasischke, Bourgeau-Chavez and
French, 1994), and so on. The new generation of spaceborne SAR,
however, opens more possibilities to look also at the opposite side
of the disaster management cycle, and address preparedness and
vulnerability. It is known that low resolution SAR could be suc-
cessfully used to assess hazards, such as landslides (Metternicht,
Hurni and Gogu, 2005), because this kind of threat depends
on large scale features such as terrain slope and material; less
frequently it was used to assess actual vulnerability (Sanyal and
Lu, 2005), but again only where the considered threat depended
on large-scale features visible at ERS-like resolutions.
The ability of acquiring small details in the scene lets sci-
entists access new departments in vulnerability assessment. A
representative example is that of seismic vulnerability.
Search WWH ::
Custom Search