Large-eddy simulations of pedestrian-level ventilation for assessing a satellite-based approach to urban geometry generation

Abstract

Realistic digital elevations of urban areas are required in urban studies but are not always available. There is a need to extract urban information from satellite images that can be used for, but not limited to, studies of the urban wind environment. This study evaluates urban geometries, including building heights and building footprints, extracted from various satellite images by large-eddy simulations for air ventilation assessment (AVA). The result shows that building heights extracted from TerraSAR-X synthetic aperture radar (SAR) images and the fused results of SAR and WorldView-2 optical (stereo) images are suitable for AVA. Better performance in representing tall buildings, rather than low buildings, is found to be more important for AVA purposes. Moreover, the performance of building geometries retrieved from fused satellite images with and without real building footprints is comparable, which suggests that building footprints extracted from stereo images are reliable.