The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
14 
is recorded at a different sensor viewing angle (nominally listed 
at -/+55°, -/+36°, 0°, (Barnsley et al., 2004)). A total of 16 
multi-angular data acquistions between winter 2003 - 2007 have 
been performed over the SNP site and nine acquisitions over 
VOR, respectively, the latter mainly acquired during the 
growing seasons of 2005 and 2006. We have set up a full pre 
processing scheme for CHRIS/PROBA data for geometric and 
atmospheric processing over mountainous terrain (Kneubiihler 
et al., 2005), which is a pre-requisite for subsequent spectro- 
directional data analyses. 
2.2 Swiss National Park 
The study site covered by CHRIS/PROBA data is located near 
Ofenpass within the Swiss National Park (10° 14’ E, 46° 40’ N). 
The Ofenpass represents an inner-alpine valley at an average 
altitude of about 1900 m a.s.l. with an annual precipitation of 
900-1100 mm. Embedded in this environment are dry, boreal 
type subalpine forests. The south-facing floor of the Ofenpass 
valley is considered as the core test site (Figure 1). It has long 
been subject to ecological studies (Kotz et al., 2004) and is 
described extensively in Schaepman et al. (2005). The forests 
are largely dominated by mountain pine (Pinus montana ssp. 
arborea) and some stone pine (Pinus cembra) as a second tree 
species. 
I 
L 
Smf’ ItaiP 
Figure 1: Subset of the geometrically and atmospherically 
corrected CHRIS/PROBA nadir scene acquired over the Swiss 
National Park study site on 17 February 2004. The Ofenpass 
valley stretches from left to right in the central part of the image. 
2.3 Vordemwald 
The study site Vordemwald (7° 53’ E, 47° 16’ N) is located on 
the Swiss Plateau in central Switzerland (Figure 2). The hilly 
area is dominated by agricultural fields in the lower parts (450- 
500 m a.s.l.) and forests mainly on the hilltops (elevations up to 
700 m a.s.l.) Agriculture concentrates on barley, wheat, maize, 
sugar beet and pasture land (Kneubiihler et al., 2006). The 
forest canopy is composed of a mixture of needle-leaf and 
broadleaf species, dominated usually by European beech 
(Fagus sylvatica L.) and Norway spruce (Picea abies L.). In 
total, nine different species can be found belonging to two plant 
functional groups (coniferous (evergreen) and broadleaf 
(deciduous) species). 3 
Figure 2: Subset of the geometrically and atmospherically 
corrected CHRIS/PROBA nadir scene acquired over the study 
site Vordemwald on the Swiss Plateau on 26 May 2005. 
2.4 Geometric Processing 
Since both the SNP and VOR study sites are located in either 
high mountainous, rugged or at least hilly terrain, a parametric 
approach for geometric correction of each data set of a 
CHRIS/PROBA acquisition scenario (up to five viewing angles) 
was applied. This approach is based on a three-dimensional (3- 
D) physical model (Toutin, 2004) which is implemented in the 
commercially available image processing software 
PCI/Geomatica (PCI Geomatics, 2006). The method allows us 
to achieve a high geometric accuracy with resulting root mean 
square errors (RMSE) derived from GCPs at 0.46-0.79 pixels 
along track and 0.39-0.73 pixels across track when using a 
digital surface model (DSM) (© swisstopo) with 2 m resolution. 
2.5 Atmospheric Processing 
Atmospheric correction of the CHRIS radiance data was 
performed using ATCOR-2/3 (Richter, 1998) which is based on 
MODTRAN-4. ATCOR-3 is suitable for atmospheric correction 
of sensor data acquired over rugged terrain. The software has 
recently been adapted to include the option to process tilted 
sensors by accounting for varying path lengths through the 
atmosphere and varying transmittance. Validation showed a 
good agreement between atmospherically corrected 
CHRIS/PROBA data and spectral ground measurements with 
deviations within ±1 stdev of the ground measurements of 
homogenous targets for most bands. The CHRIS/PROBA data 
are geo-corrected Hemispherical-Directional-Reflectance- 
Factor (HDRF) data. 
3. CASE STUDIES 
3.1 Canopy Structure and Heterogeneity Assessment 
Multi-angular observations of the reflectance anisotropy have 
proven to be diagnostic for structural surface properties, which