Combining point cloud and surface methods for modeling partial shading impacts of trees on urban solar irradiance

Although trees and urban vegetation have a significant influence on solar irradiation in the built environment, their impact on daylight and energy consumption is often not considered in building performance and urban environment simulation studies. This paper presents a novel method for comprehensive solar irradiance assessment that considers the dynamic partial shading impacts from trees. The proposed method takes urban point clouds as input and consists of three subsequent steps: (a) DGCNN-based segmentation, (b) fusion model generation, (c) matrix-based irradiance calculation. The method is validated by comparing model outputs with field measurement data, and an inter-model comparison with eleven state-of-the-art tree shading modeling approaches. Analyses carried out on daily and long-term basis show that the proposed fusion model can significantly reduce simulation errors compared to alternative approaches, while limiting the required input data to a minimum. The primary source of uncertainty stems from mismatches between tree morphology in the fusion model and reality, attributable to phenological growth and seasonal variations.