A novel 3D-geographic information system and deep learning integrated approach for high-accuracy building rooftop solar energy potential characterization of high-density cities

Abstract

This paper presents a novel 3D-GIS and deep learning integrated approach for high-accuracy rooftop solar energy potential characterization. Rooftop solar potential distribution is evaluated based on 3D-GIS-based irradiance modeling to consider adjacent building shading effects, and also based on available area identified by a deep learning technique. The casestudy results showed that the individual buidling solar potential reductions varied from 13.4% to 74.5%. Further analysis showed that simple addition of shading-induced reductions and availability-induced reductions tends to overestimate the acutal reduction by up to 16%. This study reveals the mechnisms why such effects should be jointly considered.