A systematic method to develop three dimensional geometry models of buildings for urban building energy modeling

Three dimensional (3D) geometric models of buildings are foundational for urban building energy modeling. A complete 3D geometric model contains building-relevant information like building footprint, height and Window-to-Wall Ratio (WWR). Existing methods creating these models have certain limitations, such as unavailability of Geographic Information System (GIS) databases and Light Detection and Ranging (LiDAR) data for many cities, and restricted flying space for Unmanned Aerial Vehicles (UAV). To tackle these issues, this study has developed a systematic method developing 3D geometric models, with 1) building footprint acquired from combination of two internet maps, namely Baidu Map and OpenStreetMap; 2) building height estimated from the number of storeys for residential buildings and determined using the building vertical edge method for non-residential buildings, and 3) building WWR calculated from buildings’ elevation images using an Artificial Intelligence (AI). The validation work revealed that more than 85 % of acquired building footprints had absolute relative errors less than 10 %, and this percentage was 87 %, 74 % and 75 %, for height of residential buildings, height of non-residential buildings and WWR, respectively. To demonstrate the application of the method, a newly developed urban district in Nanjing, China was used as a case study.