Citizen scientists assess hurricanes from the safety of their computers
It’s been said that every cloud has a silver lining. As Superstorm Sandy was thrashing the eastern United States in late October, hundreds of citizen scientists from around the world were contributing to a UNC Asheville project to analyze the intensity of tropical cyclones across history.
Chris Hennon, associate professor of atmospheric science, is leading a project to analyze trends in tropical cyclones over a 30-year period, using the web-based Cyclone Center, the first citizen-science project to analyze these storms. Along with collaborators Ken Knapp of the National Climatic Data Center, and Paula Hennon, Carl Schreck, and Scott Stevens of the Cooperative Institute for Climate and Satellites, Hennon hopes to involve participants on campus and in the greater Asheville area and beyond.
Hennon, currently in his eighth year on the faculty at UNC Asheville, says the project could allow scientists to resolve a set of critical disagreements about trends in tropical cyclones. Consider the storms occurring over the last three decades, Hennon argues. One study of tropical cyclones in the west Pacific suggests that storm intensity has increased over that period, while another says storm intensity has declined; yet a third shows no trend in either direction.
"Obviously, all three can’t be right,” Hennon told listeners at a Nov. 1 campus talk on the project. “We will provide data that can be used to reconcile these disparate records of historical tropical cyclone intensity. Eventually we can get to questions such as, how has climate affected the intensity of tropical cyclones,” he said.
We are turning a complicated algorithm—used exclusively by experts for over 30 years—into a tool that anyone can apply." –Chris Hennon
“One of our objectives with this project is to make people more aware of tropical cyclones in general—why do they form, why do they move where they move,” Hennon said, adding that the prospect of more intense hurricanes in a time of climate change is a primary motivation for this study. “Our best estimate is that hurricanes in general will be stronger, and that we will see an increase in the frequency of the strongest hurricanes.”
The central problem this project hopes to address lies in the fact that hurricanes are presently measured and tracked by a number of different international agencies, which don’t always provide the same picture of the same storm. “Each agency believes in their own data…but there is always bias that can be introduced,” Hennon explained. Here’s where crowd sourcing can help, he said. “The power of our greater number of evaluators should help remove that bias.”
The project is now recruiting volunteers who estimate the strength of the cyclones by assessing satellite images presented online at CycloneCenter.org. “We have nearly 300,000 hurricane images from around the world,” said Hennon. “By collaborating with the public, we hope to perform more than a million classifications in the coming months, something that would take a team of analysts over a decade to accomplish. We’re hoping to do it in a year or less.”
Computers working alone aren’t the solution, he said, because patterns in storm imagery are best recognized by the human eye. “Humans are better at pattern recognition than even the best computer,” Hennon told listeners at last week’s seminar. “Citizen scientists will help provide a unified picture of how tropical cyclones have been changing in the last thirty years.”
Hennon says the project has already made progress. Cyclone Center went live in mid-September, and has since logged over 95,000 individual classifications by more than 2,000 people around the world.
Visitors to the project web page are presented with color satellite images of storms, selected at random from around the world and across time, and answer a series of questions for each. “We developed very simple questions anyone can answer,” said Hennon. Does the storm have an organized shape? Does it have an eye? If so, what colors surround the eye—indicating cloud temperature? At the project’s end, the investigators will collect all the responses and use statistics to get an intensity rating for each storm. Since numerous people will examine the same image, information about uncertainty will also be available.
The project may be most interesting for how it reverses the usual paradigm in which citizens look to scientists to provide all the answers about natural phenomena. In this case, said Hennon, the scientific community is asking citizens to help discern the true nature of tropical storm behavior.
Could rogue users corrupt the process? That prospect is minimized by the sheer volume of images, Hennon said, each of which will be evaluated by 30 different observers. The vast majority are expected to be serious users, he continued. And you don’t have to worry about getting the “wrong” answer on any particular case. “Even experts don’t always know what the ‘right’ answer is. Anybody can do this; we want everyone to participate.”
Interestingly, the project has met with some resistance from the tropical meteorological community. "We are turning a complicated algorithm—one that has been used exclusively by experts for over 30 years—into a tool that anyone can apply," said Hennon. "Some don't believe it can be useful in the hands of laypeople." But the investigators are plowing ahead. "This is how science proceeds sometimes,” Hennon added. “You have a good idea, so you just have to see if it works."
“At the very least, we’re going to have 30 unique opinions of tropical cyclone intensity and structure for each image, and that will be useful for something.”
Want to try your hand at tropical storm classification? Log on to Chris Hennon’s tropical cyclone project at cyclonecenter.org.