Fast generation of microclimate weather data for building simulation under heat island using map capturing and clustering technique

Abstract

In this research, Urban Weather Generator (UWG) is utilized for UHI modeling in a center city area of Shenzhen based on urban morphologic data collected by an information-collection framework. The framework uses map capturing technique and unsupervised clustering for building type classification to gather input parameters for UHI model. The developed building type clustering technique, which adopted principal component analysis and Gaussian Mixture Model, is shown to have potential application in district-scale UHI modeling. Together with a differential-evolution-based calibration procedure, the proposed framework is implemented to the UWG modeling for the selected urban tissue. After calibration, it is found that the weighted sum of coefficient of variance for temperature and relative humidity has been reduced from 7.45 % to around 5.5 % compared with the uncalibrated model. The simulation results shows that the annual mean UHI index and max UHI index in center city Shenzhen is found to be 0.50 °C and 5.9 °C. After applying UHI effect to the typical meteorological year (TMY) data, the cooling degree day and heating degree day change by +12.60 % and −11.92 % compared with the original TMY. The developed method is computationally efficient and the morphed TMY data is a more accurate reflection on local urban microclimate.