ISPRS Commission III, Vol.34, Part 3A ,,Photogrammetric Computer Vision", Graz, 2002
of polystyrene. The central area is a car park made of asphalt,
and the square enclosing this park is a bare soil.
S.1 Spectral band impact
The four following images (figure 6) are simulated the 12% of
February, at 2 p.m. after a sunny day, and after shadows have
scanned half of the scene. All images are viewed with a zenith
angle of 30 ?, and an azimuth angle of 60 ? East. Figure 6a
represents the surface temperature, ranging from 0 ?C in the
shadows of the facades to 24 ?C for sunlit exterior walls. Figure
6b is an image simulated between 3 and 5 um. Radiance range
from approximately 0.6 to 2.8 W.m*sr'. Solar radiation
dominates in the signal in this band; contrast between shadowed
and sunlit areas is important.
shadowed. This is due to thermal inertia. In the illustration,
remanence effect is emphasised on the bare soil, due to the high
capacity of this material to store energy.
8-12 um 12:00 || 8-12um © 15:00
(c) (d)
Figure 6: illustration of spectral band impact; (a) image
representing surface temperatures, (b), (c) and (d),
images in bands 3-5 um, 8-10 um and 10-12 um.
Figures 6c and 6d are images simulated in thermal bands, from
8 to 10 um and 10 to 12 um respectively. Influences of the
different physical phenomena are approximately the same in
these two bands. Radiance range from 12 to 18 W.m*.sr!. The
difference of contrast between the park made of asphalt and the
bare soil is due to the difference of the ratio of emissivity values
for those materials in these two bands. It leads to observe a
higher contrast from 8 to 10 between the park and the bare soil,
and a lower contrast from 10 to 12, compared to temperature
image; this is due to self-emission process.
5.2 Temporal shadow effects
The following sequence of images illustrates the shadow
variations during the day (figure 7). Images are simulated
between 8 and 12 pm, the 21% of June. Not only spatial
influence of shadows varies during the day, but also the contrast
between shaded and sunlit areas. Another typical effect in the
infrared range is the remanence of shadows. Simulations exhibit
continuous variations of radiometry in the areas previously
A- 244
(c) (d)
Figure 7: illustration of the shadow variations during the day;
images are simulated between 8 and 12 um, at 6 and
9 a.m. and at 12 and 15 p.m.
6. CONCLUSION
The study demonstrates that realistic representation of the
landscape is only possible with a 3-D representation, as the
sampling rate increases in both time and space. In this respect,
objects that form the landscape interact each other. A simulator
of realistic images should reproduce faithfully these physical
interactions between objects. In the infrared range, the main
physical phenomena affecting the signal emitted by the scene
are shadow effect, wind disturbance linked to landscape relief,
multiple reflections or heat conduction.
The above conclusions were used as a starting point in the
specification, the design and the development of a simulator of
landscapes in the thermal infrared range with a very high spatial
resolution. The physical process underlying the emitted flux in
thermal infrared is very complex. The recent history of the
landscape is present in the simulated image. The accuracy of the
models of the physical processes and their interactions should
be high in order to obtain a good quality. Consequently, a new
methodology has been devised to design a simulator of
landscape described by a 3-D representation.
For an efficient implementation of the 3-D properties, the
concept of element was defined. It permits to describe in an
accurate manner the objects of the landscape with respect to the
physical processes and their variations in time and interactions.
The elements are homogeneous entities with respect to
geometry, material and temporal evolution of physical
phenomena. The architecture of the simulator was adapted to
this concept of element and to the specificity of the simulation
in the thermal infrared range. The simulator is made of four
