The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
elements at different heights along the edges of each flat roof 
segment. Problems resulting from misplaced points will be 
reduced by the use of next-generation LIDAR scanners with a 
higher resolution and precision. Furthermore, new algorithms 
and new 3D software will help to automatize the segmentation 
of roof elements and the removal of single points that differ 
from the general trend of each segment. Sitthole & Vosselman 
(2003) point out that a continuous improvement of the 
reliability of filtering is to be expected. However there will 
always remain ambiguities that cause problems when the data is 
processed (Nardinocchi et al, 2003). 
Generally, areas along roof borders and also roof ridges cause 
problems. For example, calculating the slope leads to undesired 
flat areas here as can be seen in Figure 7. Height information 
makes measure points that are at a similar absolute height and 
are positioned on the opposite sides of the roof ridges fuse 
together and this falsely leads to the identification of flat areas. 
The same happens to the areas along the roof borders as there is 
no delimiting height information outside the buildings’outlines. 
The extent of those wrongly identified areas is interrelated with 
the point density of the supplied LIDAR data. The more points 
are available for an interpolation, the smaller will be the areas 
that are misinterpreted, especially at roof ridges. 
Figure 7. Along roof borders and roof 
ridges areas can mistakenly be identified 
as flat (red patches) 
The modeling of the roof shadows generally leads to good 
results. Most of the shaded areas on flat roofs are correctly 
modeled, as can be seen in Figure 8. The frayed look of the 
LIDAR model here is again due to a lack of resolution, not to 
the processing of the data. The results of the shade modeling for 
sloped roofs are not satisfying according to this field test. Again, 
the reason are the low resolution and the inaccuracy of the 
height information of the LIDAR data. Projecting shadows on 
distorted sloped roofs particularly leads to incorrect results 
because the inaccuracies of the roof models multiply due to the 
projection in the third dimension. However, these problems will 
also be solved as soon as better LIDAR scanner technology is in 
use. 
is no easy way of distinguishing treetops from a roofs 
superstructure, if LIDAR data is used. This affects the 
segmentation of roofparts with overlapping treetops and leads 
to artefacts appearing in the roof models. On the other hand 
stereophotogrammetric digitizing methods are able to correct 
the data coverage by amodal completion of hidden roof parts 
and the generation of simplified tree models as well. To get 
correct results, trees need to be identified as such, their shape 
has to be remodeled with high resolution data and afterwards 
their shadows can be calculated. For processing aerial LIDAR 
data, Matikainen et al (2007) have proposed new approaches to 
distinguishing between different objects. The combined analysis 
of LIDAR data and additional datasets, e.g. aerial colour ortho 
images or satellite imagery representing visible light and/or 
NIR, is promising. Analyzing shape patterns as well as spectral 
analysis can lead to the automized elimination of points in areas 
with an indicated vegetation. 
Figure 8. Modeling the shadow with the 
evaluation data (left) and LIDAR data 
(right) 
This field test demonstrates that the solar potentials of 
buildings’ roofs can be calculated by the use of LIDAR data. 
Technically, the modeling of the azimuthal exposition, the slope 
and the shading can be carried out perfectly if the data sources 
are of sufficient quality. The quality improvement of the data is 
closely related to new filtering algorithms and techniques which 
improve the preprocessing of LIDAR data. In the future “ready 
to use” 3D city models will help to automize the segmentation 
of roofs and other objects for modeling urban solar potentials. 
This will enable local authorities to provide exhaustive 
information about solar potentials within a foreseeable period. 
Even the price per solar factsheet for each building should be 
within the limits of usual official charges or even be offered for 
free. 
REFERENCES 
Ackermann, F., 1999. Airborne laser scanning - present status 
and future expectations. In: ISPRS Journal of Photogrammetry 
and Remote Sensing, 54(2&3), pp. 64-67. 
Brenner, C., 2005. Building reconstruction from images and 
laser scanning. In: International Journal of Applied Earth 
Observation and Geoinformation, 6(3-4), pp. 187-198. 
Brenner, C. & Haala, N., 1999. Towards Virtual Maps: On the 
Production of 3D City Models. Geoinformatics, 5(2), pp. 10-13. 
Chilton, T.D., Jaafar, J. & Priestnall, G., 1999. The use of laser 
scanner data for the extraction of building roof detail using 
standard elevation derived parameters. In: International 
Archives of Photogrammetry and Remote sensing, La Jolla, 
USA, Vol. XXXII, Part 3, pp. 137-143. 
Haala, N., 2005. Laserscanning zur dreidimensionalen 
Erfassung von Stadtgebieten. In: 3D-Geoinformationssysteme, 
Heidelberg, Germany, pp. 26-38. 
Klärle, M. & Ludwig, D., 2005. Standortanalyse für 
Photovoltaikanalagen durch hochauflösende Sensoren in der