In: Stilla U, Rottensteiner F, Paparoditis N (Eds) CMRT09. IAPRS, Vol. XXXVIII, Part 3/W4 — Paris, France, 3-4 September, 2009
159
Eventually, the simulation process provides the following
output data for each reflection contribution detected in the 3D
object scene:
• coordinates in azimuth, slant range, and elevation
[units: meter]
• intensity data [dimensionless value between 0 and 1]
• bounce level information for every reflection
contribution [1 for single bounce, 2 for double
bounce, etc.]
• flags marking specular reflection effects [value 0 or
1]
Figure 2: left: Simulation using box model having a size of 20
m x 20 m x 20 m, line of sight indicated by arrow;
right: simulated reflectivity map simulated (slant-
range indicated by arrow)
Figure 3: simulation using step model (left), line of sight
indicated by arrow; simulated reflectivity map (right),
slant-range indicated by arrow
2.3 Reflectivity maps in azimuth and slant range
Firstly, all reflection contributions are mapped into the azimuth
- slant range plane. Afterwards, a regular grid is imposed onto
the plane and intensity contributions are summed up for each
image pixel. Figure 2 shows the resulting reflectivity map for a
cube (dimensions: 20 m x 20 m x 20 m) which has been
illuminated by the virtual SAR sensor using an incidence angle
of 45 degrees. The size of one resolution cell has been fixed to
cover 0.5 m x 0.5 m in azimuth and slant range. Surface
parameters are chosen in a way that box surfaces can be clearly
distinguished from ground parts, i.e. in the current example box
surfaces show stronger diffuse backscattering than the
surrounding ground. Following top-down in ground range
direction, diffuse single bounce contributions of the ground are
visible followed by a layover area of ground, wall of the box
and top of the box. At the end of the layover area, a strong
double bounce line is visible which is caused by the interaction
between the front wall and the ground in front of the box.
For this type of scene geometry, a 2D simulation and analysis is
usually sufficient. The next section will however illustrate
examples that underline the necessity of including the elevation
direction as third dimension into the simulation.
Figure 4: selection of pixel for elevation analysis (left);
definition of three slices (right) in slant-range (1),
azimuth (2), and elevation direction (3)
2.4 3D analysis of scattering effects
Figure 3 shows a reflectivity map simulated by illuminating a
step model (width: 10 m, length 20 m, height 20 m). For
providing the map, the same imaging geometry has been chosen
as for the box example, i.e. the step was oriented in direction to
the sensor and the incidence angle was fixed to 45 degrees in
order to obtain specific overlay effects for single and double
bounce contributions which are explained in the following.
Compared to the reflectivity map containing the box model
(Figure 2), the reflectivity map of the step shows similar
characteristics. Both the layover area of single bounce
contributions and the location of focused double bounce
contributions are identical. Only the size of the shadow zone
indicates a height difference between the illuminated objects. In
the case of the step model, separation of dihedrals - two right
angles at the steps - is impossible in the reflectivity map since
all double bounce effects are condensed in one single line.
Hence, separation of scattering effects in elevation direction
may be helpful since it enables to resolve layover effects for the
purpose of distinguishing several scatterers within one
resolution cell. To this end, an interactive click-tool has been
included into the simulator for defining two-dimensional slices
to be analyzed. In the case of the given reflectivity map for the
step model, one pixel is selected, e.g. located in the double
bounce area as shown in Figure 4. Based on the coordinates of
the pixel center, three slices are defined:
• slice no. 1 for displaying elevation data in slant-range
direction
• slice no. 2 for displaying elevation data in azimuth
direction
