International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
scatterers are involved, the signal contributions of different
objects can hardly be separated analysing a single SAR image
alone. The InSAR DEM is unreliable in layover areas. Layover
is a major hindrance for the analysis of SAR data in urban
areas, e.g. for building reconstruction.
Fig. 3: a) ground range magnitude image and building layer
(red); b) DEM and vector map (roads: green).
The choice of a large viewing angle minimizes layover effects.
But, large viewing angles lead to extended occlusion areas,
caused by the cast shadow from elevated objects on the ground
behind. From shadow regions, no signal is returned to the
sensor. Therefore, these areas in the SAR data contain no
information but noise only. The choice of the viewing angle in a
dense urban area is a trade-off between layover and shadow
effects [Soergel et al, 2003a].
Since the viewing angle is known, the building height can be
determined from the size of layover or shadow areas. However,
these features are often disturbed by the signal of other objects.
Parallel to the sensor track orientated building walls together
with the ground in front of them build a so-called dihedral
corner reflector: the signal is reflected back to the sensor by
double-bounce scattering. The path length is identical for all
double-bounce signals, causing bright lines in azimuth direction
in the SAR image located at the building footprints (e.g. at
buildings F and G). At building / a dihedral corner is build from
the tall main building and the flat roof of the entrance hall.
Another kind of dominant scattering frequently observed in
urban areas is specular reflection, e.g. at the roof structures on
buildings B and J. These single-bounce signals can be
502
discriminated from double-bounce scattering by their
polarimetric properties [Guillaso et al., 2003].
In case of a small radar signal wavelength like X-band, there is
a significant influence of trees on the visibility of buildings. For
example, building K is occluded from a large tree, which also
covers some small buildings of group / in front by layover.
4. CHOICE OF OPTIMAL SAR ACQUISITION
PARAMETERS
It is essential to determine a-priori the optimal SAR acquisition
parameters in order to minimize the influence of layover and
shadow for a selected area of interest or an object class. For this
purpose, GIS data are required. Here, we focus on best effort
mapping of buildings and roads by SAR for the test area shown
in Fig. 3b.
Based on the DEM, layover and shadow are simulated [Meier et
al., 1993]. From the sensor position, the DEM grid is sampled
in range direction. Layover and shadow regions are detected
analysing the distance and the viewing angle. By intersection of
these results with the map data, the affected areas of buildings
and roads are identified.
In order to find the best SAR parameters the simulations were
systematically repeated with varying aspect angles a and
viewing angles 9 (Fig. 4). About 650 simulations have been
carried out for the entire test site [Soergel et al, 2003b].
Different viewing angles 0
o
Different aspect angles a
Flight &
trajectory
(0,9) *
Fig. 4: Simulation of SAR phenomena layover and shadow
With the given sensor parameters, only 43% of the building
roof area and about 20% of the road area can be sensed without
distortions by layover and shadow. These numbers increase to
86% for roofs and 62% for roads in case of a combination of
four chosen SAR acquisitions (Tab. 1).
SAR measurements ] 2 3 4
Undistorted roof area .|51.6 172.8 181.5.186.5
Undistorted road area | 39 47 55.6 | 62
Table !. Portion of roof and road areas visible without
distortions for several SAR acquisitions in per cent.
Two simulation results for the same area as shown in Fig. 3b
are depicted in Fig. Sa,b. The first simulation was carried out
with the parameters of the acquired InSAR data and the other
one for an aspect angle from the right with same carrier altitude
and viewing angle. The dependence of SAR data acquisition
from the aspect angle can clearly be observed comparing e.g.
the shadow and layover areas caused by the buildings D and E.
Intern
The
withc
An ir
of €
How:
areas
the a
data |
Ui
-
In th:
a lay
sensc
into |
shift
due t
visib
of th
road
road: