The Norris Geyser Basin has spoken to Phil Dawson after he figured out how to use voice recognition software to decode "noisy" monitoring data gathered in 2003.
"It's very exciting science," said Hank Heasler, geologist for Yellowstone National Park, where the geyser basin is located. "Who would have thought that voice recognition software could be applied to this kind of problem?"
Voice recognition software has made a splash most recently in the new iPhone from Apple. Using the Siri application, phone users can ask their phone a question, ask it to call someone or even take dictation.
Although the algorithms behind voice recognition have been around for a while, the technology has most recently made great strides because of the rapid improvement in the speed of computer processing.
Dawson, a geophysicist with the U.S. Geological Survey in California, said he happened by chance upon the discovery that voice recognition could be applied to seismic data. A visiting researcher from Spain who specialized in voice recognition needed a desk and the USGS just happened to have a spare that she could use. Dawson and Carmen Benítez got to talking one day and decided to apply her skills to decoding data collected from the Norris Geyser Basin.
"All sorts of folks have been looking at the data since 2004," Heasler said. "Some seismologists thought the signal was way too busy. This shows the value of having a new viewpoint to look at a problem."
Hydrothermal movement in geysers like those found in Yellowstone, and magma movement in volcanoes like Kilauea in Hawaii, are similar in that both contain bubbles that make the fluids pulse, Dawson explained. When those bubbles bump into solid earth, sensitive seismometers can record the event.
"The cool thing about volcanoes is very often there are repeated signals, indicating that the source process is the same," Dawson said. "What that means is, in the context of speech recognition, it's very easy to see them as a sentence or a song."
The data that were collected by seismometers set up in the Norris Geyser Basin back in 2003 had a lot of "noise," or random variation, Heasler said. The voice recognition software that Dawson built could sift through the noise to find the basin's particular "words."
"It's very common in volcanic systems that you see the same 'word' over and over again," Dawson said. "It became a tool where you can recognize events happening."
The power of the tool is that it may someday be capable of monitoring geysers or volcanoes for unusual activity.
"We want to be able to do efficient monitoring so that when things begin to get active, we can protect people," Dawson said.
The goal is to also build monitors that would understand all of the different "words" that volcanoes and geysers speak, rather than one that just looks at a specific volcano or geyser, he added.
Heasler said the data collected by Dawson's analysis were known -- that steam bubbles rise and collapse, becoming more frequent at certain times of the day. What interested Heasler is that data Dawson collected were from an area near Porkchop Geyser where such activity had not been noted by Yellowstone researchers before.
"It suggested to me additional areas I should be looking for this kind of signal on the ground surface and if there are small accelerometers I could buy and deploy," Heasler said. "
The seismic data were collected when about 40 percent of the Norris Geyser Basin was temporarily closed for visitor safety in 2003 when some thermal features in the back of the basin changed character. Water in a stream boiled away and some ground reached the temperature of steam.
"To me, this study very clearly illustrates how much more we have to learn about this system," Heasler said. "Nature is a very complex, fascinating system and this study helps point that out."
Please Wait…