The deaths of more than 50 hikers killed in a Japanese volcanic eruption on September 27 left many wondering why officials weren’t more proactive in issuing warnings and keeping sightseers off the mountain.

The deaths occurred when Mount Ontake, a sacred Shinto peak and popular hiking destination roughly TK miles west of Tokyo, erupted suddenly, catching an estimated 250 hikers unaware. The Japan Meteorological Agency did detect tremors in the weeks before the eruption, but they went unannounced, spurring the criticisms, according to The Yomiuri Shimbun.

Although the questions are natural in the wake of such a tragedy, they also highlight misunderstandings about the certainty and extent to which scientists can now forecast eruptions.

“The ability to predict volcanic eruptions is an ambition that volcanologists are far from realizing,” writes University of Hull volcano expert Rebecca Williams in The Conversation. “Magma movement under a volcano will cause volcanic tremor, make the ground rise and fall and release gases such as sulphur dioxide. If these signs are monitored closely, then it may be possible to forecast that an eruption may be imminent.”

Even if those signs are present, Williams goes on to explain, they don’t necessarily guarantee an eruption; scientists rely on data to decode the nuances of each volcano’s unique rumblings. And that’s where Ontake becomes an especially unpredictable case.

Ontake has only erupted three times in recorded history and of those eruptions—in 1979, 1991, and 2007—only one had adequate instrumentation to collect data on magma activity, according to Nature. There are currently 12 seismometers and five GPS stations used to gauge seismic activity and deformations in the earth caused by magma moving underneath.

Although the seismometers did show activity prior to the eruption, the GPS systems didn’t indicate magma movement. Scientists believe the reason for that was because the eruption was phreatic, a sort of steam-fueled explosion caused by groundwater interacting with magma.

So even while the JMA detected tremors before Ontake blew, there wasn’t reason to suspect an impending explosion based on the volcano’s sparse eruption history and absence of magma activity.

“It was generally very quiet,” said Koshun Yamaoka, a professor of volcanology at Nagoya University, told the Wall Street Journal. “The intensity of precursory seismic activities are typically correlated to subsequent eruptions, so nobody thought an eruption of this scale would happen. There's no doubt this was difficult to predict.”

That said, there are improvements that could be made to monitoring at Ontake, experts said. For instance, the addition of instruments that measure gases could have helped agency officials identify fumes released by the phreatic activity. By the same token, having staff physically in the area would also have been a benefit, Yasuyuki Miyake, a volcanology professor at Shinshu University, said.

“It's also important that we have more manned observatories near volcanoes,” Miyake told the Wall Street Journal. “It's easy to accumulate automated data, but it's important that we have people on location who are observing and sensing any changes.”

A nationwide early alert system should also be implemented, Toshikazu Tanada of Japan's National Research Institute for Earth Science and Disaster Prevention told Nature.

“Even if we just start implementing the system one step at a time, we need something that will keep this from happening again,” he said.

Perhaps the best system though, is one in which people realize the limitations of technology, volcano expert Toshitsugu Fujii pointed out in an interview with Reuters. “There is no guarantee of total safety when you're dealing with nature,” he said.