399
ANALYSIS OF THE SOLAR POTENTIAL OF ROOFS
BY USING OFFICIAL LIDAR DATA
R. Kassner 3 ’*, W. Koppe b , T. Schüttenberg b , G. Bareth b
a solarkarte.com, Cologne, Germany - info@solarkarte.com
b GIS & Remote Sensing Group, Department of Geography, University of Cologne, Germany - g.bareth@uni-koeln.de
Commission IV, WgS IV/3
KEY WORDS: Image Processing, DSM, DEM, LIDAR, Texture Analysis, Feature Detection, Image Understanding, Geography
ABSTRACT:
This paper focuses on a field test that locates roof areas with a high solar potential and predicts the solar “harvest” per m 2 . The test
analyzes 2.5D LIDAR data provided by official surveying and mapping sources. The primary LIDAR data is prepared by masking
the roofs’ contours and afterwards filtering the point cloud by a threshold value. The remaining LIDAR data, which represents the
buildings’ roofs, is analyzed according to the slope, the azimuthal exposition and shaded roof areas. The quality assessment of the
derived roof areas is carried out by means of a 3D dataset which is semiautomatically acquired from panchromatic
stereophotogrammetric aerial photographs.
1. INTRODUCTION
Both a common awareness of the need to reduce C0 2 emissions
and the rapidly rising energy costs have induced a growing
demand for sustainable energies. Roof-mounted solar heating
and photovoltaic systems are not only important technologies to
decrease the emissions of carbon dioxide caused by domestic
fuel consumption, but they also help saving energy and
financial costs. Therefore, today the worldwide use of solar
systems is increasing. Private investors as well as local
authorities have a rising interest in identifying roof areas which
are suitable for mounting solar systems.
Commonly LIDAR data is used to generate digital elevation
models (DEM) (Kraus & Pfeifer, 1998; Kraus, 2001). Filtering
and segmenting the LIDAR data leads to the recognition and
the modeling of single objects, such as e.g. building extraction
and reconstruction, and, on a larger scale, to 3D city modeling
(Brenner, 2005; Ackermann, 1999; Chilton et al, 1999). As a
consequence this technology and knowledge is also used for the
analysis of solar potentials of roofs (Vogtle et al, 2005a; Vogtle
& Tovari, 2005b; Klarle & Ludwig, 2005).
Nowadays the cost of LIDAR data is comparatively low and the
data is often provided by local authorities or state governments.
An extensive automized use of airborne LIDAR data for
locating roof areas suitable for mounting solar systems is to be
expected in the near future.
2. TEST AREA & DATA SOURCES
2.1 Test Area
This field test is based on data referring to 13 buildings within
the urban campus of the University of Cologne, Germany. The
project was conducted in the framework of the CampusGIS-
project where all required data for this study have been
available. Due to its location on the fluvial planes of the river
Rhine in the Cologne Bight, the morphoglogy of the campus
area can generally be described as flat. These selected buildings
comprise a variety of flat and sloped roofs, different
complexities of the superstructure and different amounts of
sections.
2.2 LIDAR data
For the analysis of the roofs, two primary irregular digital
elevation models (DEM) of the campus area are used (Figure 1).
The 2.5D point clouds of the digital terrain model (DTM)
represent terrain surfaces at ground level, which means that
vegetation and buildings are excluded. The digital surface
model (DSM) provides height information on vegetation and
buildings’ surfaces. According to the specification given by the
official surveying and mapping sources, the former
“Landesvermessungsamt Nordrhein-Westfalen”, the average
point distance of the DTM is 1-5 m, whereas the distance of the
DSM points is 1-2 m. The height accuracy is specified with +/-
50 cm for the DTM and +/- 30 cm for the DSM.
Figure 1. Irregular point clouds of the
LIDAR data (left: DSM; right: DTM)
2.3 Stereophotogrammetric Evaluation
The quality assessment of the results of the LIDAR data
analysis is carried out by analyzing panchromatic