Simulation of solar radiation on metropolitan building surfaces: A novel and flexible research framework

The current use of the Geographic Information System (GIS) and Parametric Environmental Simulation (PES) to calculate the solar radiation on building surfaces cannot be effectively applied to large-scale scenes. To overcome the scale and dimensionality issues when estimating solar radiation on large-scale building surfaces, an ingenious solar radiation simulation framework, which can extend the depth of study and expand the number of possible calculations, is proposed. First, the framework is used to investigate the performance and error of different clustering methods used for urban splitting. K-means displays the best performance, and the MAPE is 0.89%. In addition, Monte Carlo integration (MCI) is applied to explore a microelement reconstruction method for the complex contours of building roofs. NSGA-II is used to derive the optimal parameters and reconstruction accuracy for the reconstruction process. Notably, R2 reaches 0.98, and the MAPE is only 0.25%. The framework is applied to evaluate the radiation type, spatial and temporal distributions, and the overall occlusion rate (RT) of clusters of over 600,000 buildings in Shanghai. The results indicate that approximately 80% of the samples in Shanghai display an RT in a mid-high to high range, and less than 3.5% of the samples display an low RT that is low. Based on this study, we can refine assessments of solar radiation on the surfaces of buildings at the city scale and even the national scale and thus provide a reference for the utilization of solar energy resources in areas where solar resources are not abundant.