Developing a global building morphology dataset for urban hydroclimate simulation

As prevalent infrastructures in cities, buildings play a vital role in urban hydroclimate simulation with their effects being parameterized through their 3D morphology. In order to achieve a comprehensive understanding of 3D building morphology distribution, significant efforts have been devoted to the generation of global-scale 3D building morphology datasets. However, existing works typically produce such information at a coarse resolution (around 1000 m) using either machine-learning-based regression with medium-resolution open-source satellite images or simple empirical rules with high-resolution but restricted satellite images, which limits their effectiveness in providing accurate and refined mappings for public usage and academic research. Considering this bottleneck, this work establishes a practical solution leveraging open-source Sentinel-1/2 imagery with cloud computing infrastructures and deep-learning-based models to create a dataset named GLAMOUR, which estimates the average building height and footprint at a resolution of 100 m across global urbanized areas. Based on the results of accuracy assessment in 18 cities worldwide, the GLAMOUR dataset achieves the median root mean square errors of 7.5 m for building height and 0.14 for building footprint estimation, with corresponding standard deviations of 4.0 m and 0.02, achieving better overall performance compared to existing datasets with similar spatial coverage. As a publicly accessible and reproducible dataset, GLAMOUR exemplifies the workflow of annual global monitoring of 3D building structures in urbanized areas and offers rasterized 3D building structures on 100 m grids for the year 2020 in GeoTiff format, which can be seamlessly integrated with open-source GIS software and urban hydroclimate models. With the help of GLAMOUR, researchers and policymakers can easily derive refined building-related physical parameters such as surface roughness and zero-plane displacement for urban hydroclimate simulation in their desired areas worldwide. It will facilitate a global-scale assessment of urban hydroclimate disaster risks, paving the way for holistic climate adaptation strategies.