701 
2- MEASUREMENT AND PROCESSING 
The measurements analyzed in this paper have been acquired during a joint POLDER/AVIRIS campaign that 
took place in June 1991 (Baret et al„ 1992). The area of interest is located in southern France, close to the river 
Rhone delta, in the so called “La Crau” area. The zone is mostly cultivated with a dominance of rice, wheat and 
sunflower. Sorghum, grass and orchards are also present. Although the campaign lasted several days, from June 
18th to June 24th, June 19th was selected here for its rather low aerosol optical thickness of 0.094 at 550 nm. 
This value allows us to apply a crude atmospheric correction on the measurements where only single scattering 
is accounted for. 
During the campaign, the POLDER instrument was flown on the ARAT aircraft operated by IGN with a 
typical flight altitude of 3000 m. The flight pattern is shown in Fig. 1. It is composed of five flight lines parallel 
to the sun direction , and one perpendicular flight line. 
The POLDER instrument optical design is based on the concept of a CCD matrix, a rotating filter wheel that 
carries spectral filters and polarizers, and a wide field of view lens. On the filter wheel, 9 slots allow 3 spectral 
measurements, each of them for three polarization directions. During the campaign, the POLDER filter wheel 
was set up with spectral filters at 550, 650 and 850 nm, each 40 nm wide. The bidimensional CCD matrix 
permits a view of the area in one instantaneous shot. A typical POLDER field of view is shown in Fig. 1. The 
POLDER field of view is 51° along track and 43° crosstrack. The field of view on the ground is, therefore, 
about twice the instrument altitude, or 7.4x5.6 km 2 . The CCD matrix is composed of 384x288 pixels, which 
yields a spatial resolution of about 20 meters. The time lag between two successive shots is 10 seconds which, 
according to the aircraft flight speed corresponds to 0.7 km. Thus, there is an overlap of 90% between two 
successive images, and a target on the surface can be seen in 10 POLDER images corresponding to 10 different 
viewing directions. Similarly, the crosstrack image size is much larger than the distance between the various 
parallel flight lines. There is, therefore, an overlap between the measurements of those flight lines and a surface 
target can be observed from the corresponding viewing directions. 
Data processing is as follows: The CCD spectral measurements are converted to radiance using calibration 
coefficients (determined in the laboratory using an integrating sphere) and then to reflectance. A crude 
atmospheric correction is applied to the measurements: 
— 
^mes ^Ray 
( 1 ) 
where R cor is the reflectance corrected for atmospheric effects, R mes is the measured reflectance, R aer is the 
estimated aerosol reflectance (according to ground-based measurements of aerosol optical thickness), R R ay is 
the reflectance resulting from molecular scattering below the aircraft, and r atrri is the atmospheric transmitance. 
In this simple correction, we only account for single scattering and we neglect the coupling between surface 
reflectance and aerosol scattering. Moreover, we assume the aerosol layer to be homogeneously distributed 
below the aircraft, and we make use of a theoretical scattering phase function. Although these are strong 
hypothesis, we have confidence in our corrected reflectances because of the low atmospheric optical thickness. 
The reflectance presented in this paper are always corrected reflectances. 
One make then use of the position and orientation parameters given by a GPS system and a gyroscopic central 
unit connected to the POLDER instrument. Each POLDER image is registered on a surface grid of 10x10 km. 
shown in Fig. 1, with a spatial resolution of 100 meters. This resolution was chosen because preliminary tests 
showed that it is the maximal registration error in our measurements. For each grid point, the spectral 
reflectance is stored together with the solar and viewing angles. The number of directional measurements 
associated to a given grid point depends on its position in the study area. It reaches a maximum of 50 at the 
zone center. 
3- ANGULAR SIGNATURES OF SELECTED TARGETS 
Fig. 2 to 5 are a representation of the reflectance directional signature for selected surface targets. In these 
polar diagrams, the radius corresponds to the viewing zenith angle. The polar angle is the viewing azimuth 
relative to the sun direction. The backscattering (azimuth = 0°) is on the right side of the diagram. The polar 
representations are limited to viewing zenith angles up to 60° since this is the practical limit of airborne 
POLDER. Each point corresponds to one directional measurement and the figure next to it gives the measured 
reflectance in percent. The lines are a result of an isocontour processing on the data points. The data points that 
are aligned in the polar plot correspond to successive measurements of a given flight line. 
Visual analysis of the figures gives us confidence in the data quality: 
• The successive measurements from a given flight line show a monotonous variation. 
• The isolines are relatively symetric with respect to the principal plane (the 0-180° line on the figures). Such 
symetry is expected if the surface does not have a favoured direction.