Comparing SAR X and L bands to map the urban extent in a metropolis of South America. The potential of multitemporal data

Monitoring city boundaries and extents is a
mandatory task so as to achieve the 11th Goal of sustainable development agenda 2030. This work aims at exploring the potential of the SAR dataset recorder by the SIASGE constellation for mapping urban areas and their extent. In particular, urban maps have been produced by applying the Urban EXTent algorithm to Very High Resolution (VHR) Cosmo-SkyMed first generation (CSK and CSG, respectively) in X-band and SAOCOM (SAO) data in L-band. Urban extractions based on multitemporal SAR data and based on a single date have also been compared. SAR-based urban maps have been finally analyzed against existing land cover maps produced by multi-spectral data. All three SAR derived maps presented high accuracy, with CSG being the best one, and CSK and SAO slightly less performant. Urban maps derived from multitemporal SAR information presented higher accuracy than those based on a single date and, at the same time, resulted coherent with land cover maps produced by traditional multi- spectral data classification. The VHR CSG and CSK, with high computational efforts, well support urban extent mapping at very fine local scale, while SAO data provide a good support for mapping urban areas at a regional or country scale.