ALOS-2 disaster mapping algorithms and their implementation in an operational system

Abstract:
Increasing human and economic losses from urban disasters demand synthetic aperture radar (SAR), which has allweather, day-and-night observation capability. We have developed fully automated algorithms for detecting flood extents and earthquake-induced building damage using SAR data from Advanced Land Observing Satellite-2 (ALOS-2). The algorithms rapidly process ALOS-2 and ancillary data (flood simulation, hazard map, and other geographical information) and provide damage information less than an hour after the input data are entered. The validation results showed that the accuracy of the estimated flood extent is about 70% in average and depends on the incidence angle of SAR observation. The accuracy of the damaged building detection was about 70% for buildings with a footprint larger than 200 m². We implemented these algorithms in an operational disaster response system and launched a one-stop operation that automatically processes ALOS-2 data after emergency observation and provides the damage information to the end-users.