Image classification of urbanization processes require reference data either for training or for validation. However, there is scarcity of reference labeled datasets specifically to detect urban areas. In this talk, I will present three approaches for collection of open-source reference data that mark built-up areas.

Open data and publicly available satellite imagery are transforming how cities can prepare for disaster and climate risks and strengthen urban resilience. In this talk, we present a data-fusion approach that utilizes free satellite imagery (Sentinel-1 and 2) to identify flood prone areas in developing cities. The approach, which is fully automated, relies on SAR (Sentinel-1) and optical data (Sentinel-2) to identify and map urban areas that have been prone to flooding in the last 5 years.