Urban local surface temperature prediction using the urban gray-green space landscape and vegetation indices

Abstract

With the continuous rise of global temperature, the urban thermal environment is facing severe challenges. The planning of urban buildings and urban green space (UGS) has a major impact on urban thermal environment. In this study, vegetation and landscape indices were taken as independent variables, and land surface temperature was taken as dependent variable, the gray-green space was used to predict local land surface temperature. The importance of each index was evaluated to explore the importance of gray-green space in mitigating the heat island effect and improving the thermal comfort. The results show that: 1) The characteristics of gray-green space can be used to predict the surface temperature, and the Gradient Boosting Decision Tree (GBRT) model has the best prediction effect and higher accuracy. 2) The building material has a significant effect on the surface temperature, so consideration of the gray space allows a more accurate prediction effect of building material on the local surface temperature in cities. 3) In the course of predicting surface temperature, PLAND _ UGS is the most influential index, which is the proportion of UGS patches in landscape area. Overall, vegetation and landscape indices can predict local surface temperature. The findings can help city planners in designing better urban construction plans, so as to provide more effectively thermal comfort.