SAR4URBAN: SAR for urbanisation monitoring (DUE Innovator III Series)

From the beginning of the years 2000, more than half of the world population live in cities and the overall trend of urbanization is growing at an unprecedented speed. The use of Earth Observations and their integration with other source of information into effective urban planning tools can produce a quantum leap in the capacity of countries to track progress towards and achieving international urban development goals. One of the main sources of information on urban areas that is essential to monitor precisely and with regular periodic updates is the monitoring human settlements. The importance to have up-to-date information on human settlements does not only regard urban areas but also rural and peri-urban areas where most of the un-controlled developments are taking place, hence the urgency to have regular and updated information on the evolution of human settlements worldwide .

The advent of continuous streams of high quality and free of charge satellite observations such as the Sentinels of the European Copernicus program, in combination with the emergence of automated methods for big data processing and image analysis and the democratization of computing costs, have offered unprecedented opportunities to improve our capacities to efficiently monitor the changes and trends in urban development globally.

The SAR4URBAN project developed an innovative approach to automatically extract built-up areas from the joint use of C-band SAR and multi-spectral optical data. The novelty of the method has been the integration of temporal statistics from SAR and optical data into large-scale urban mapping with fully automatic extraction of training samples, machine learning classification and post-classification enhancement. The main output of the SAR4URBAN project has been the World Settlement Footprint (WSF) 2015, the first global map of human settlements generated globally from the joint processing of optical and radar imagery. The WSF 2015 is available at 10m spatial resolution with a global coverage (in urban, peri-urban and rural areas) and is based on the processing of all Sentinel-1 and Landsat-8 imagery acquired in 2014 and 2015.