Flash floods are quick and deadly. At least 13 people died when a surge hit parts of Saudi Arabia in May, 2013. Two years earlier, 123 were killed when thunderstorms dumped rain over arid land to the east of the Red Sea port of Jeddah, hitting the city with no warning. A drone monitoring system that tracks floods in real time would sound the alarm before the water hits.
Existing forecasting models are good at predicting roughly when an area might experience the right mix of conditions to create a flash flood, but they can’t say precisely when or where a flood will strike. Christian Claudel at the King Abdullah University of Science and Technology outside Jeddah is working on a drone system that could give such cities between 30 minutes and 2 hours of warning, as well as predicting the flood’s path.
The goal is to launch a swarm of about 10 drones to automatically monitor a potential flash flood. The drones will drop disposable wireless sensors across the region at risk. If the sensors meet floodwater they will be carried away on the current, sending out a simple signal that the drones can track. The drones relay the sensors’ changing positions back to a central database, which builds up a model of floodwater flow.
“The sensors are made of printed circuits on paper, which reduces the cost of the device,” says Claudel. The printed sensors act as location tags, pinging a unique ID number to the drones over a short range. This is much cheaper than using dedicated sensors with their own communications systems, which may never be recovered.
Kristen Rasmussen, who studies flash flooding at the University of Washington in Seattle, says it would be really useful to combine the drone system with existing models that take account of terrain and how the water moves. “That could be very powerful,” she says. But she warns that the social aspect of prediction is equally important: getting the message out and persuading people to pay attention to it.
Source: Hal Hodson, New Scientist