Understanding Urban Temperature Differences through 2D/3D Urban Morphology; In Context of the Belt & Road Initiative

Abstract
This study explores the global phenomenon of urbanization, often caused by initiatives such as the Belt and Road Initiative (BRI). The BRI is an global infrastructure development effort initiated by the Chinese aimed at improving connectivity and collaboration through strategic investments in other countries and cities which contribute to development and urbanization. As involved cities develop and expand both horizontally and vertically, concerns arise regarding Urban Temperature Differences (UTD), including issues like Urban Heat Islands (UHI) and their associated environmental and health impacts. Despite the acknowledged significance of understanding the relationship between urban morphology and UHIs, a research gap exists in exploring this relationship and integrating 3D indicators into analyses. Focusing on Shymkent and Kyzylorda in Kazakhstan, cities which are selected to analyse the impact of the BRI, the study uses road-based building block segmentation and high-resolution Worldview-1 stereo imagery to generate 3D building models, supplemented by OpenStreetMap (OSM) building footprint data. Urban morphology indicators (UMIs), including building height, density, and coverage, aggregated at the building block level, are calculated to analyse their impact on UTD. Using the eXtreme Gradient Boosting (XGBoost) Machine Learning Model with SHAP-based interpretability, the research analyses the influence of UMIs on UTD intensity. The findings of this study challenge the traditional UHI standard, revealing an Urban Cooling Island (UCI) effect in the semi-arid climate of the studied cities. The study also highlights the impact of both 2D and 3D urban morphology on UTD intensity, particularly, Mean Sky View Factor (M-SVF) and Building Coverage Ratio (BCR) at the building block level manifest as important features in influencing UTD intensity. The results further emphasize the unexpected significance of standard deviation over mean aggregations for specific indicators. This stresses the importance of variability of urban morphology within building blocks. Connecting all results with the BRI underscores differences in urban growth and UTD intensities between larger and smaller cities, which highlights the need for targeted policies within the realm of global infrastructure initiatives.