INDRA Reporter: Crowdsourcing Quantitative Flooding Data with Impact Classifications to develop real-time flood risk dashboards

Abstract:

Floods continue to cause widespread deaths and economic damages around the world, accounting for roughly one-third of all global geophysical hazards. To manage floods and their aftermath, we need continuous monitoring of ground situation with specific local information regarding the severity of current conditions. Such large amount of observation data at high-resolutions is expensive to obtain in gauged locations and non-existent in vast ungauged areas. Crowdsourcing presents a new opportunity to collect large amounts of useful geo-tagged information in both resolution and coverage. INDRA Reporter has been designed with a mobile-first approach geared towards real-time applications and an emphasis on user-interface/user-experience (UI/UX) to maximize collection of higher fidelity data. Mobile applications and chat bots have been developed and deployed. A custom tool is also being developed for identifying flooding reports in real-time twitter data stream. A prototype dashboard application is under development to intelligently combine the collected data from multiple streams with remote sensing and socio-economic data. Currently, the system is being tested over a limited geographical area as a test-bed. The collected reports are validated against satellite imagery. To the best of our knowledge, no crowdsourcing tool or experiment collects quantitative flooding data with user-supplied impact classifications, which could potentially be very useful in validating hyper-resolution hydrologic models. The API has been kept extensible in order to expand the data collection to other hydrologic and meteorological phenomenon.