Everyone remembers the Mount Carmel Forest Fire.
Beginning just south of Haifa on December 2, 2010, it raced through the hills for 82 hours, killed 44 people, burned down 5 million trees, destroyed hundreds of buildings, and forced 17,000 civilians to evacuate their homes.
It was the deadliest fire in Israeli history.
But what if it could have been prevented? A recent study by a team of researchers at Ben-Gurion University of the Negev suggests that if given appropriate warning, emergency responders could have been more prepared.
“Had officials been notified of the blaze soon after it began, much of the ensuing tragedy might have been avoided,” says Erez Shmueli, a postdoctoral researcher at the Massachusetts Institute of Technology’s Media Lab.
Recently published in the Journal of Statistical Physics, the study features algorithms that can detect the early stages of emergencies like bombings, natural disasters and forest fires. However this isn’t just any data. The study – conducted over an extensive three-year period –used a sample of 12 billion cell phone calls and text messages.
Although Mount Carmel trees were burning by 11:00 a.m., the first evacuation bus didn’t arrive until 45 minutes later. However, local residents noticed the flames long before and called their friends, families and co-workers to warn them. There was a huge spike in cell phone usage in the Mount Carmel area.
The team proposes that if these spikes in cell phone usage were monitored in real time, then the network administrator could have signaled to police of the emergency.
In acquiring the data for the study, analysts searched for cellphone network “hubs;” nodes with high traffic that would likely expose large dips or jumps in communication.
They then detected large increases in activity, or “anomalies” in the hubs, and found that days where the hubs exhibited anomalous cell phone usage were also days when disasters like airplane crashes and bombings occurred.
The problem is that spikes in cell phone usage is also seen during musical events such as concerts and festivals, or sporting events like the Maccabiah games, not just disasters.
“We would need to set the threshold so not every event triggers the algorithm and it only catches spikes that represents ‘big’ events,” said Michael Fire, of the Department of Systems Information Engineering at BGU.
A more controversial problem of the study is the possible invasion of privacy. In lieu of the National Security Agency leaks – of which security contractor Edward Snowden blew the whistle on US methods of tracking electronic communication – any news of cell phone monitoring has the potential to incite a riot.
However, if the “Social Amplifier Method” was ever set into motion, the researchers hope the use of select hubs would prevent everyone’s cell phone usage from being monitored.
“The idea of this research was to target specific, wellconnected users in the network without the need to monitor all users and violate their privacy,” said Fire.
In addition to not monitoring the content of messages, the data doesn’t record phone numbers.
“Yes, we have billions of phone data that contains who called, or text messages and what time. But instead of users’ phone numbers we just have a series of characters [named] ‘called hash strings,’” he continued.
To further emphasize the protection of privacy during the study, Fire used the example of watching a catastrophe occur – even before the news trucks show up, a bystander can see people calling and texting but not know what they are saying. Even if the bystander is too far away to see the disaster itself, a noticeable increase in communication will let the bystander know something is wrong.
The “Social Amplified Method” is that bystander.
In the case of the Mount Carmel Forest Fire, it is possible that onlookers were taking photos from miles away, uploading them to Facebook or Twitter, while anyone with an Internet connection passed them along.
Fire offered that because social media interactions tend to amplify during disasters, the method does not only have to be limited to cell phone usage.
“I believe this method can ultimately work on all types of social networks in which people connect with each other in real time,” Fire said.