Informal urban growth monitoring in earthquake-prone areas using SAR satellite images

In recent decades, Peru’s primary mode of urban growth has been the informal occupation of bare lands. These urban areas are characterized by their lack of essential services, such as electricity and water. With the lack of suitable land for urban development, informal urban has grown into unsafe areas against earthquakes. Due to the lack of resources in developing countries, detecting recent informal occupations in unsafe areas cannot be performed, which is an important task for relocation purposes. This article reports the performance of machine learning applied in synthetic aperture radar (SAR) satellite images for the early detection of informal settlements in hazardous areas. The methodology uses a set of temporally SAR images of a specific area, binary pixel classification, and post-processing techniques to improve the prediction performance. Two informal occupations that occurred in the districts of Chorrillos and Villa el Salvador, Lima, Peru, in April 2021 were used as experimental evaluation. A set of SAR images of the constellation Sentinel-1 was used with a resolution of 10m. The results show that time series analysis of SAR images can identify recent informal occupations. However, the geometrical distortions in SAR images reduce the accuracy of the spatial extent of the occupations. We conclude that SAR images are a valuable source for a sustainable informal urban growth monitoring system.