A robust approach to global urban area extent extraction using ASAR Wide Swath Mode data

This paper stems for a research aimed at showing the potentials of ASAR Wide Swath data to extract human settlement extent at the global level. Exploiting the features characterizing human settlements in SAR images, and considering the coarse spatial resolution of the ASAR Wide Swath mode, partially compensated by the relatively large number of images acquired on the same area during one year´s time, the developed technique shows that it is possible to delineate human settlements in different parts of the world with an overall accuracy better than 80%, and with commission and omission errors that are smaller than those available in current Globcover data sets, exploiting optical (MERIS) data at similar spatial resolution and more complex classification algorithms.