) angle 6j 
indicated 
for each 
A. Rothkirch 
  
  
  
  
  
  
Sample Polarization | ao [-] | a4 [1] | a; xli n k 
Spectralon ss 0.072 0.53 0.048 1.03 | 0.18 
pp 0.095 0.53 0.048 1.03 | 0.18 
sp 0.079 - - - - 
ps 0.064 - - - - 
Red roof tile SS 0.040 0.40 0.038 1.35 | 025 
pp 0.053 0.40 0.038 1.35 |»0.25 
sp 0.036 - - - - 
DS 0.029 = - : 
  
  
  
  
  
  
  
  
Table 1: Coefficents to describe the given samples by the model of Torrance & Sparrow. Note that the diffuse components 
n°? a? ps 
a’, af’, ay and ag” were fitted seperately. 
incident zenith angle is given by the integral of the specular peak over the viewing angles w given by the upper hemisphere 
d 
150) = [ ar 
Q 
With parameters from tab. 1 results are: 
Fig, D. D, fi) G (0;, p. ®) 
cos 0; cos 0, 
  
einai (3) 
pareamalon (0.072 + 0.079): 7. 40.023 = 0,297 
pppectralon: > (0.095 + 0.064) - 7 + 0.004 = 0.504 
pre roof He = (0.040 + 0.086) ++ 0.068 = 0.307 
pheétoct ile = (0.053 + 0.029) - 7 + 0.012 = 0.272 
The difference of the albedos is below 1.5 96 for Spectralon and below 13 % for the "Red roof tile". The contributions 
from the specular albedo to the total albedo A, — J3P*€* / (ao,, m + Qozy * 71) are À, = 5 % and A, = 1 % for the 
Spectralon sample and A, z 28 % and A, = 5 % for the sample "Red roof tile. An accurate reproduction of the specular 
peak is therefore more important for the sample "Red roof tile' than for the Spectralon sample. 
Also noticeable is that for the Spectralon sample the given albedos are close to the albedo determined by a calibration by 
the manufacturer of ppap = 0.511 (Labsphere Inc., 1994). 
4 SUMMARY AND OUTLOOK 
BRDF measurements of urban surface materials made with a linearly polarized HeNe-laser as illumination source were 
presented. Polarization effects in the principal plane were analyzed for a roof tile as well as for a Spectralon sample of 
50% reflectance. Both samples show different reflection properties with respect to the polarization states ss, pp, sp and 
ps. At ss-polarization the samples show a strong specular peak which increases with increasing incident zenith angle 
and which is shifted towards greater viewing zenith angles. At pp-polarization the specular reflection is much weaker as 
compared to ss-polarization. In contrast cross polarization measurements (sp resp. ps) show almost lambertian reflection. 
The BRDF values also differ absolutely with respect to polarization states. 
It was shown that the reflectance from our surface materials can be described by the model of Torrance & Sparrow (Tor- 
rance et al., 1967). The reflection of the samples can be interpreted as a combination of a diffuse component (internal 
scattering) and a component that is due to specular reflection of statistically distributed surface facets. The model re- 
produces the specular behavior of the samples. Cross polarization measurements of the samples 'Spectralon' and "Red 
roof tile' can be described by a lambertian assumption within deviations of 6.1 96 on average. Absolute Differences of the 
cross- and like-polarized diffuse components are about 10 % at s-polarized incident light and about 39 % at p-polarized in- 
cident light on average. Differences of the total albedo for s- and p-polarized incident light are within 1.5 % for Spectralon 
and below 13 % for the "Red roof tile". 
Acknowledgements 
We would like to thank our colleagues of our working group and the coworkers of the Dep. for Computer Science for their 
Support and fruitful discussions. The measurement campaign was performed in collaboration with the Joint Research 
Center (JRC), Ispra, Italy. We acknowledge the contributions particularly of B. Hosgood and G. Andreoli from JRC. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 783