Estimation of anthropogenic heat from buildings based on various data sources in Singapore

Heat released from the energy consumption in buildings (QB) is an important component of anthropogenic heat, which is a major contribution to the urban heat island (UHI) phenomenon. However, it is still a challenge to integrate all the available data to improve the estimation of QB due to inconsistences among multiple data sources. This paper presents a solution to estimate the anthropogenic heat from buildings in Singapore by classifying the buildings into the electricity consumption sectors based on a random forest classification model using multiple data sources. The classification model achieves a cross-validation accuracy score of 95% by using building use type, land use type of land parcel where the building stands, location and area as predictors. Based on the classification result, the annual average QB from the commercial buildings (CS), Housing and Development Board (HDB) apartments, private apartments and condominiums (AC), and landed properties (LP) were estimated to be 12.1, 4.4, 3.2 and 1.1 W m−2 on a 200 m-by-200 m grid, respectively. QB was approximately 9% of the net all-wave radiation in Singapore in 2015. Our approach can serve as a useful tool for integrating datasets from different sources with inconsistent categorizations, and our results can benefit urban planning as well as urban climate modeling at both microscale and mesoscale.