Researchers from Pennsylvania State University (Penn State), the University of Wisconsin, and elsewhere analyzed 2013 flooding in Colorado and found 150,000 tweets from people affected by the disaster.
The researchers processed the tweets with an existing tool called CarbonScanner, which analyzes tweet hashtags and matches their locations onto a map, and found "clusters of posts," implying significant damage in those areas.
The researchers also examined more than 22,000 images using a machine-learning algorithm, which automatically analyzes thousands of images. The algorithm enabled the researchers to quickly identify pixels in images that contained water. The raw imagery was obtained through satellites, Twitter, Flickr, the Civil Air Patrol, unmanned aerial vehicles, and other sources, according to the researchers.
"We looked at a set of images and manually selected areas that we knew had water and areas that had no water" in writing the algorithms, says Penn State researcher Elena Sava. The researchers fed the information into their algorithm, which enabled it to learn what was and was not water.
The names of rivers and streets in the tweets, combined with remarks related to how the individual tweeter could not get home, were clues indicating flooding.
From Network World
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA