Researcher Builds Better System for Monitoring Volcanoes
WenZhan Song hopes a new generation of smart sensors he is developing will allow scientists to monitor volcano activity in real time, and give people more time to reach safety before a lethal eruption.
Song, an associate professor of computer science at Georgia State University, got interested in volcano research while living in Washington state. The infamous Mount St. Helens was practically in his backyard.
When the volcano blew in 1980, it became the most destructive eruption in U.S. history, killing 57 people and destroying more than 200 homes. Ash clogged sewage systems, damaged cars and buildings, and temporarily shut down air traffic over the Northwest. Mudslides tore downhill, traveling nearly 50 miles from their source. In the end, damage to the timber and agriculture industries and to the area’s infrastructure was estimated at more than $1 billion.
“Volcano hazards directly involve people,” Song said. “If we could study and understand them, we would potentially be able to predict an eruption earlier, before it happens.”
Song is the principal investigator for a project known as VolcanoSRI (Seismic Realtime Imaging).
The idea is to harness the power of sensors, small computer-like devices, to gather and crunch data faster and in more detail.
Song isn’t the first to use wireless sensors to study volcanoes, but his project is important because it involves what scientists call distributed tomography imaging. This approach uses algorithms – essentially a formula to crunch data – and distributes that data among the sensors to perform real-time calculations within the network, he said.
“The devices can talk to each other and form a mesh network,” Song said. “Each of those stations is detecting the arriving earthquake signals, which penetrate through the magma areas.”
The effort, supported by a $1.8 million grant from the National Science Foundation, will use the data gathered by the sensors to produce a real-time, high-resolution image of a volcano system. The resulting image goes beyond 3-D because the images change as new data comes in, creating what scientists call a 4-D image.
10 Most Deadly Volcano Eruptions
|Deaths||Volcano||Year||Cause of Death|
|29,025||Mt. Pelee, Martinique||1902||Ash flows|
|14,300||Unzen, Japan||1792||Volcano collapse, tsunami|
|3,500||Vesuvius, Italy||1631||Mudflows, lava flows|
|3,360||Vesuvius, Italy||79||Ash flows and falls|
Song hopes the system will allow researchers to develop a deeper understanding of how volcanoes work, and to improve early warning systems for volcanoes and other earth hazards.
He and his team will travel to South America in 2015 to put their technology to the test, placing 500 seismometer sensors, also known as geophones, around an active volcano named Cotopaxi. The volcano, in the Andes Mountains, south of Quito, Ecuador, erupted in 1877, sending mudflows that traveled for 60 miles and killing 1,000 people.
The research team, which includes members from University of North Carolina and Michigan State University, plans to run the sensor network for up to three months. They chose a South American volcano because volcanoes in the United States tend to be less active.
In the past, data had to be fed into a centralized database and it could take months to organize and analyze.
“Here, the computing is done among the 500 sensor stations in real-time,” Song said. “The network of sensor nodes is basically a supercomputer cluster. The data will be processed within this wireless network and the 3-D image will be sent out in real time.”
The approach taken in the VolcanoSRI project could also have important applications for oil field exploration, biomedical health monitoring and structural monitoring of bridges and tall buildings. Song said the system can also be used to find oil under the ocean, using hydrophones, which are sensors that can be used underwater.
In many fields, such as oil exploration, using sensors has traditionally required cables weighing tons and possibly stretching for miles. Months-long delays in transporting and crunching data added to the cost. But a system of smart sensors could simplify the hunt for new oil fields.
“Smart sensors have many very important implications,” Song said. “The main instrumentation is the same.”