In this thesis, we implement a workflow for rooftop solar potential analysis. The objective is a robust, reproducible, and applicable workflow that shortens the time required for irradiance modeling and roof segmentation, using open-source libraries and open geospatial data. The theoretical part introduces input datasets from airborne laser scanning (LiDAR) and the realestate cadastre. We explain the model for computing global, diffuse, and direct solar irradiance, using monthly atmospheric parameters (Linke turbidity factor, surface albedo, and clearness index). We also present the development of the roof-segmentation procedure and the geometric generalization of roof segments. To identify buildings already equipped with photovoltaic systems, we incorporate a machine-learning model. The previously disparate tools are merged into a single Python script that executes the full analysis. Using a Ljubljana case study, we demonstrate the workflow and validate the results against reference data and comparative irradiance estimates. The solution is reproducible via Docker. The contribution is an integrated, automated, and applicable workflow for rooftop solar potential estimation that reduces data-preparation time and increases decision transparency in photovoltaic system planning.
|
