3D laser scanning and open source GIS for solar potential assessment

There is an increasing demand among home owners and operators of commercial and public facilities to reduce energy costs, the reliance on declining fossil fuel resources and the impact on the environment. Therefore, cost effective sustainable energy production such as biomass, wind power and solar energy must play a major role in the energy supply chain. The latter is in the focus of this thesis. Solar thermal and photovoltaic conversion systems can be mounted on both, roof planes and building facades in order to produce heat and electricity and thus to increase renewable energy supply. The amount of the incoming solar energy on building areas can be assessed by using e.g. solar radiation models that are implemented in Geographical Information Systems (GIS). However, this requires detailed information about the three dimensional representation of the buildings of interest as well as of their surrounding objects that may cast shadows in order to find suitable areas for the installation of solar based renewable energy technologies. In recent years Light Detection and Ranging (LiDAR), also referred to as laser scanning, has established as a standard technology for highly accurate three dimensional data acquisition of object surfaces. The geometrical information of the scanned surface is stored as 3D point cloud (xyz-triples). Each point is tagged with additional information such as strength of backscatter (intensity), timestamp and scan angle. Most solar radiation models that are implemented in standard GIS software operate on 2.5D raster data and cannot be combined with 3D point cloud data. Hence, an aggregation and simplification of the 3D point cloud to 2.5D raster cells is required leading to an irreversible loss of the third dimension. This procedure reduces the amount of point cloud data significantly and processing such simple data models is less time consuming because the complexity of the third dimension has not to be considered anymore. Although 3D shapes such as roof overhangs or building parts covered by e.g. vegetation are not represented in 2.5D raster data, they have been used for building detection and solar radiation modeling in many cases. This thesis aims to utilize the highest degree of information - the third dimension - of the laser scanning point cloud for both the detection of planar areas of buildings and solar radiation modeling. The algorithms and workflows developed in the framework of this thesis are implemented in Open Source GRASS and SAGA GIS allowing the integration of own modules and the use of existing visualization and spatial analysis tools to interpret and further process the results. Methods are demonstrated generating 2D GIS-ready information in the form of vector polygons and vector lines of the detected objects and their properties (e.g. area, total amount of the incoming solar energy). This offers the possibility for 'normal' GIS users such as spatial planners to analyze and further process the spatial information in standard GIS environments. In the first part of this thesis approaches for the detection and segmentation, respectively, of roof planes using 3D point cloud data acquired by Airborne Laser Scanning (ALS) are presented. The detected roof planes are used as input for point cloud-based solar radiation modeling. Shadows of nearby objects are considered by modeling the 3D horizon of each point being reflected from a roof plane within the original laser scanning point cloud. As solar thermal and photovoltaic conversion systems can also be mounted on building facades the developed point cloud based solar radiation model is transferred on building walls extracted from Mobile Laser Scanning (MLS) data in order to assess their solar potential. The developed algorithms are completely executed within the computer's main memory and thus are not suitable for large study areas because the huge amount of point cloud data produced by LiDAR technology cannot be processed at once. In the second part of this thesis GIS-based workflows are developed allowing an application of the developed algorithms on large LiDAR datasets. Suggestions for both area-wide point cloud based solar radiation modeling and seamless area-wide roof plane segmentation are provided.The thesis concludes that point cloud based solar radiation modeling considers the strong local gradients of the incoming solar energy in a more realistic way due to a better consideration of 3D shapes such as roof overhangs or overhanging vegetation. However, the developed algorithms and workflows might be highly valuable for a variety of research areas and applications and are not limited to the point cloud based assessment of the incoming solar energy of building areas. Future research should focus on the integration of the developed methods in further applications. This might contribute to raise state of the art research activities in spatial science to a higher level. (author)