Rapid web-based crisis information through automation of processes: enhancing first responders’ access to critical geodata

Rapid and straightforward dissemination of information to emergency services is critical for informed decision-making during crises. Earth observation data is a valuable source for obtaining a widespread situational overview, e.g. by delineating the water extent during flood events. In this study, we present a prototype process chain for the rapid delivery of web-based crisis information derived from very high-resolution (VHR) optical imagery, introducing more automation across the stages of satellite-based emergency mapping. Initiated by a monitoring system of relevant warning sources, we leverage the time benefits of pre-tasking VHR satellites by employing a programmatic ordering of satellite data. The water extent is segmented based on a deep learning trained model from which further geodata are derived. The relevant information is disseminated via a web application that enables interactive exploration and rapid updates. Testing the process chain for a simulated flood event demonstrated that the crisis information could be provided within hours after the warned flood event and was positively evaluated by first responders. While automation improves the timeliness of crisis product delivery, we conclude that semi-automation is preferable to full automation, as each crisis is unique and requires expert decisions at critical stages.