International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999 
the Bavarian Alps. Wet, marshy soils exist at the borders to the 
tertiary aged hills. 
2.2. Dataset 
The MOMS-2P mode D dataset investigated was acquired in an 
ascending track during data take 08FE on June 25 1998. Scene 
22 (ID: To8FeMDC01SO22L0) was taken at 9.30h MET with a 
path azimuth of around 73° and a sun azimuth of 131° and sun 
elevation of 57°. The atmosphere was cloudless. The 
management maps of the experimental Farm Diimast and the 
forest management maps have been used for verification. 
2.3. Software 
All the image analysis and interpretation processes were 
performed with the PCI Easi/Pace V6.1 software. 
3. METHODS 
The largest differences between forward and backward looking 
stereo images can be expected at similar path and sun azimuths 
and low sun elevations. The Mexico scene investigated by 
Schneider et al. (1999) was very close to these optimum 
conditions. The illumination-to-sensor geometry for the scene 
evaluated in this paper is a „worst case“ example. The path and 
sun azimuth differences of 58° and 122°, respectively, and the 
high sun elevation of 57° (Figure 4) are such that only minor 
anisotropy effects can be expected. The goal of our 
investigations was therefore to check the limits of the 
anisotropy approach under such unfavourable conditions. The 
low contrast of the „ratio“ image (band 6 / band 7) in Figure 5 
shows the small differences in backscatter intensity of these two 
bands. 
Fig. 4. Illumination-to-sensor geometry during MOMS-2P 
mode D data take 08FE on June 25 th 1998, 9.30h MET. 
3.1. Radiometric calibration 
For the retrieval of bio- and geo-physical parameters, the 
accurate radiometric correction is of decisive importance. Both, 
radiometric calibration of the sensor, as well as the correction of 
atmospheric attenuation, are to be performed before data 
analysis. 
Fig. 5. Ratio between forward (6) and backward (7) looking 
stereo bands showing the low contrast due to the 
unfavourable illumination-to-sensor geometry. 
MOMS-2P stereo data are observations on the Earth's surface at 
three different times from three different viewing directions. 
Compared to sensors with across-track stereo, the time 
difference between forward and backward looking data 
acquisition of the same scene is only 40 seconds, corresponding 
to an orbit distance of ca. 300 km and a base/height ratio of 
about 0.8. Thus, the radiometric similarity of the images is high 
and their correlation easy. Using these stereo images, a digital 
elevation model (DEM) can be calculated (Lehner and Komus, 
1995). A DEM allows the derivation of exposure and slope, 
parameters that are needed for the correction of terrain relief 
effects. 
Sensor calibration. In the case of the MOMS-2P sensor, there 
are 8 different CCD sensors, which need to be calibrated. Each 
Fairchild linear CCD has 6000 sensor elements, each with 10 
pm x 10 pm size. The output signal is digitised to 8 bit. To 
adjust the output signal to the scene brightness, an electronic 
gain setting with 8 steps can be used. 
For sun calibration, the protection cover of the optics is used 
(Figure 6). The inner side of this cover is coated with grey 
spectralon material, which is a diffuse reflector. 
Fig. 6. MOMS-2P with tilted cover for sun calibration. 
For calibration, the cover was tilted by 20 degrees. During 
calibration, the MIR-station flew in a so-called inertial attitude, 
in which the angle between the normal vector of the cover plate 
and the sun remained constant. In this attitude, the optical axis 
was not looking vertical down to the earth. The inner coating of 
the cover plate with spectralon was originally foreseen for 
controlling the stability of the laboratory calibration. The data 
that are gathered during the in-flight sun calibration 
measurements have the same format like the real images taken 
from the earth surface. For the purpose of these investigations, 
calibration data from two subsequent days have been used (27 th 
and 28 th January 1997). Both datasets were acquired during a 
sunset. This means that during a time period of about 20