How A.I. captured a volcano’s altering lava lake

Synthetic intelligence onboard NASA’s Earth Observing 1 (EO-1) spacecraft assisted in imaging an eruption at Erta’Ale volcano, Ethiopia, from an altitude of 438 miles (705 kilometers). The remark was scheduled autonomously by way of the Volcano Sensor Net, which was alerted to this new exercise by knowledge from one other spacecraft. Credit score: NASA/JPL/EO-1 Mission/GSFC/Ashley Davies

Certainly one of our planet’s few uncovered lava lakes is altering, and synthetic intelligence helps NASA perceive how.

On January 21, a fissure opened on the prime of Ethiopia’s Erta Ale volcano—one of many few on the planet with an lively lava lake in its caldera. Volcanologists despatched out requests for NASA’s Earth Observing 1 (EO-1) spacecraft to picture the eruption, which was giant sufficient to start reshaping the volcano’s summit.

Because it turned out, that spacecraft was already busy accumulating knowledge of the lava lake. Alerted by a detection from one other satellite tv for pc, a man-made intelligence (A.I.) system had ordered it to take a look at the volcano. By the point scientists wanted these pictures, they have been already processed and on the bottom.

It is a becoming capstone to the A.I.’s mission. That software program, referred to as the Autonomous Sciencecraft Experiment (ASE), has guided the actions of EO-1 for greater than 12 years, serving to researchers research pure disasters across the globe. ASE will conclude its operations this month, when EO-1’s mission involves an finish. ASE leaves behind a legacy that means nice potential for A.I. in future area exploration.


Apart from the current eruption, ASE helped scientists research an Icelandic volcano as ash plumes grounded flights throughout Europe in 2010. It additionally tracked catastrophic flooding in Thailand. The software program minimize the turnaround time for knowledge from weeks to only days, as customers might put in requests in actual-time.

ASE was developed by NASA’s Jet Propulsion Laboratory in Pasadena, California, and uploaded in 2003 to EO-1, an earth science satellite tv for pc managed by Goddard Area Flight Middle in Greenbelt, Maryland. The software program directed EO-1 to alert researchers each time it detected occasions of scientific curiosity, and autonomously tasked the spacecraft to take photographs throughout subsequent orbital passes.

Moreover, it manages a “sensor net,” a community of different satellites and floor sensors that each one “speak” to at least one different, serving to to prioritize which occasions to give attention to.
“It is a milestone in A.I. software,” stated Steve Chien, principal investigator of ASE and head of the Synthetic Intelligence Group at JPL. “We have been supposed to do that for six months, and we have been so profitable that we did it for greater than 12 years.”

The software program sometimes notified researchers inside ninety minutes of detecting an occasion. It then downlinked knowledge and re-tasked EO-1 inside a number of hours—a course of that beforehand took weeks when scientists and operations groups on the bottom needed to coordinate.

A.I. can free a spacecraft to behave first, inside rigorously programmed parameters, permitting it to seize useful science knowledge that may in any other case be misplaced, stated Ashley Davies, lead scientist for ASE and a volcanologist at JPL.

“It is placing some scientific smarts onboard a spacecraft,” Davies stated.

The current eruption of Erta Ale highlights the velocity and impression of area A.I. When a 1.9 mile-lengthy (three kilometer) fissure opened in late January, it brought about elements of the caldera to break down—precisely the type of quick-shifting occasion that’s onerous to seize knowledge on until you are awaiting it.

Luckily, the JPL sensor-net has a large attain. It is comprised of different satellites apart from EO-1, and even on-the-floor sensors. When a type of different satellites picked up speedy temperature modifications on the volcano’s summit, that is when it pinged EO-1, which started planning to picture the location.

“We caught this occasion on the good time, throughout an early, creating part of the eruption,” Davies stated. Now he and different scientists had a a lot better sense of how the discharge of lava is evolving over time. “This merely would not have occurred with out the Volcano Sensor Net.”

Each Chien and Davies agreed that autonomy has monumental potential relating to learning occasions removed from Earth, the place huge distances make it inconceivable to know what’s occurring till the occasion has already handed. For instance, A.I. might make it a lot simpler to seize these dynamic moments when a comet passes by or volcanoes start erupting on a distant moon.

The above submit is reprinted from NASA


Geology IN

This entry was posted in Geology and tagged , , , , , . Bookmark the permalink.

Leave a Reply