A framework of urban thermal environmental planning approaches for climate change adaptation with static and dynamic factors

Abstract

Why should the various spatial conditions constituting the detailed urban area be considered in climate change research? This study examines the spatial factors that is related to urban thermal environment, considering how they can contribute to thermally resilient urban planning. Climate change has intensified due to rapid urbanization worldwide, and the intensity and frequency of abnormal climate phenomena within cities are increasing. In order to respond to urban heat-related disasters the process of estimating the distribution of temperature and heat vulnerability within a city must be fundamentally preceded. A city is composed of various elements (e.g. environment, infrastructure, people), and the local temperature of the city may vary depending on the interaction between these elements and the thermal environment. Accordingly, various existing studies have attempted to explore the relationship between the degree of urbanization and urban heat. Furthermore, various urban indicators influencing the thermal environment and heat vulnerability of cities have been frequently explored. However, this study derived several problems to be solved. First, the index expressing the urbanization that has been used in the existing research was generally a two-dimensional concept that mainly reflected the degree of the built-up area. Since urban geometry reflects spatial properties encompassing volume beyond a flat concept on the ground, it is necessary to develop a diversified comprehensive index representing urbanization. Second, in many existing studies, time-series data-based studies for an entire city were conducted. It is also important to explore in spatial units because even within a city, the temperature and the degree of vulnerability may be different in adjacent regions. Third, prior studies have generally focused on indicators of static sectors related to the environment (e.g. built-up area, green area) and infrastructure (e.g. buildings, roads). Since dynamic sectors related to human behavior (e.g. population mobility, traffic mobility) are also a major cause of anthropogenic heat release in cities, it is necessary to introduce indicators with more diverse characteristics in thermal environment research. Although various studies attempted to explore thermal environment or heat vulnerability with urban indicators, attempts to explore urban heat or heat vulnerability by comprehensively considering static and dynamic spatial indicators that are diversely distributed in a detailed spatial unit within a city were rarely conducted. Therefore, this study proposes a novel conceptual framework to comprehensively explore urban indicators with static and dynamic characteristics to contribute as a reference to a thermal environment planning. The main findings and implications of this dissertation are as follows. First, this study proposed a framework for developing comprehensive urbanization rate (CUR) index that implies three-dimensional concept of urban physical and environmental factors. We confirmed that the proposed CUR index is correlated with urban heat. In addition, a spatial unit Deep Neural Network (DNN)-based temperature estimation model is developed using the developed CUR index. It is expected that this finding will become a framework to support decision-making when estimating varying heat in a local area by analyzing the quantitative distribution of physical and environmental factors within a city. Second, this study developed a Random Forest (RF) based heat-related mortality estimation model. By interpreting the model with SHAP method, indicators related to the vulnerable population (i.e. young and elderly, aging rate) was derived as key factors contributing to the process of heat-related mortality estimation. Through in-depth analysis of the model, it was suggested that the distribution of the vulnerable population has a more dominant effect on the risk of spatial unit heat in the city than the climatic condition. Third, by focusing on the mobility sector, this study developed a heat estimation model that includes factors in a broader domain. By confirming that the mobility factors are correlated with urban heat through spatial data analysis, we derived feasibility of dynamic domain in urban heat estimation. The comprehensive heat vulnerability analytic framework of this dissertation can contribute in part to the decision-making process when planning an effective urban thermal environment in terms of mitigating heat-related disasters and effectively managing risk of public health sector.