Can Big Data Help Predict Tornadoes?
The recent tornado in Oklahoma is one of those unthinkable events that leave us scratching our collective heads wondering what more could be done to save lives. For those of us whose lives orbit the big data world, we hear a lot of hype about the wonders of all this data. Naturally when an event like this happens, we are left to wonder if big data has the power to provide more warning, and get more people out of harm’s way before the event strikes.
In Oklahoma this week, the residents were given sixteen minutes of warning time before the funnel touched down. National Weather Service spokesman Chris Vaccaro says that the national average is about 14 minutes. These precious minutes are the difference between life and death – so can big data extend them?
The answer is yes, according Kelvin Droegemeier, the VP of Research for the University of Oklahoma, who last month shared his research involving the dynamics and predictability of severe thunderstorms and tornadoes at the Sage Bionetworks Commons Congress.
Posing the million dollar question, will computer models ever be able to predict tornadoes, Droegemeier gives an unequivocal, “yes, sort of.” Using a model of a tornado outbreak in central Oklahoma from 2011, he demonstrated how a computer model that had been developed had very nearly predicted the exact path of two tornadoes that had touched down that year. This numerical forecast was able to build a model which showed the rotation of the storm, and while the model wasn’t perfect (one of the predictions it made was too far to the north), it made these predictions a full hour before these particular funnels touched down.
While it’s not clear from the video why this system is not yet being implemented to prevent tragic outcomes like we saw this week, what is clear is that the data mixed with the high performance computing power does have the potential to make a difference and save more lives.
You can see the clip here: