Making sense of the “firehose” of flood maps during disasters: the TC Amphan success story

Abstract:

Oftentimes, many flood maps or related information, such as nightlight or damage proxy maps, shared to assist during disasters have been produced by different entities using the same satellite source image but different mapping algorithms, making each map different. This situation is confusing and in addition, many entities, including operational services, do not provide maps during the entirety of the event but stop after one or two satellite images over one or two days. In order to efficiently respond to flood disasters, WFP is developing its own fully automated flood alerting and monitoring system. A prime use case demonstrating WFP’s capabilities and reach in flood mapping is Tropical Cyclone Amphan. Starting around 20^th May, the first Sentinel-1 SAR flood image and associated flood map was received, quickly followed by many entities sending in flood maps from the same source processed with different algorithms, thus resulting in different flood map delineations. Other flood maps received were processed from JAXA PALSAR, TerraSAR-X and many other S-1 images over the event duration up to May 30th^th. WFP took all these flood maps and merged them into a single final flood map for decision making, expressing both spatial and temporal flood dynamics (flood rising and receding conditions) as well as uncertainties associated with processing the same satellite image with different algorithms. This final map was shared with the CO and also FAO to assist them with their mandated response actions. WFP used this opportunity and coordinated with FAO a field data collection campaign for validation as well as a request to Planet Labs for obtaining their high-resolution optical images as part of this validation exercise. Other maps received included damage proxy maps from JPL’s ARIA mapping system, maps of before and after nightlight impacts from NASA and event maps as well as historical flood maps of maximum observed flood extent from the Global Flood Observatory DFO at CU Boulder. During the event, WFP also successfully demonstrated the full use of the WASDI web platform with its flood mapping app HASARD (LIST), obtaining, for the first time, flood maps from each individual S-1 SAR scene available during the length of the event, thus allowing monitoring in real-time and in a fully automated way, spatial and temporal dynamics of flooding.