If your community floods, what do you do? If you are like many people, as soon as you are safe you tweet. If you are an active citizen scientist, you may also post your pics to MyCoast, an ap designed to collect reports of flooding and subsequent impacts.
According to the U.S. National Oceanic and Atmospheric Administration (NOAA): "In 2016, global sea level was 3.2 inches (82 mm) above the 1993 average—the highest annual average in the satellite record (1993-present)."
That might not sound like a lot, but for many communities built on the edge, the reduced efficiency of water drainage dominoes until water backs up into significant floods. And with sea levels rising further, the peak events today will become the average of the future.
But current sources of flood data - such as insurance reports, satellite monitoring, local sensor networks or eyewitness reports - cannot fuel hyper-resolution mathematical models capable of supporting detailed risk analysis, flood control planning, and early warning systems.
Dr Roger Wang and colleagues from the University of Dundee’s School of Science and Engineering have shown how data deciphered by using AI tools on tweets and MyCoast images correlates with traditional measures of flood severity such as quantity of precipitation or road closures. Dr. Wang explains that,
"We were particularly interested in the increased incidence of what we call sunny day flooding – flooding that occurs in the absence of any extreme weather event due to the mean sea level being higher."
While the AI assessment of the images needs improvement to help differentiate flood zones from photos of coastal waters or other normal wet areas, the paper concludes that the "big data" available via these social media avenues could enhance the monitoring networks and enhance the granularity of data available for improving computer models of urban flood events.
The technique will be especially useful if global sea level rise makes the existing flood maps obsolete faster than scientists can rewrite them.
The full paper, Hyper-resolution monitoring of urban flooding with social media and crowdsourcing data, appears in Computers & Geosciences, Volume 111, February 2018, Pages 139-147.