It is easy predicting when a volcano will erupt once one sees a few hundred or a few thousand minor
seismic events - a swarm of Richter 1s and 2s. This signals magma is moving. The more events and the
wider the area, the bigger the eruption and usually the sooner. In terms of predicting when a volcano or volcanoes in a region might erupt months or years in the future, things are not so clear. However, there is a fair statistical relationship. With some tedious covariance analysis one can derive a rule of thumb that the bigger major earthquake ( 7.5 or greater) one has, the more likely and the sooner a
linked volcano will erupt. Puyahue in Chile is an example of that.
"The answer is blowin' in the wind" - noted meteorologist B. Dylan
Our software caught a cluster of small seismic events (high Richter 4s and some 5s) near the
poorly-defined Ethiopian-Eritrean border. The area is part of the African rift, so one should expect
earthquakes - just not that many in that short a time. Our models suggested a volcano named Dubbi
might erupt. It had been active in 1861 (and maybe twice since then) and, despite being small, its lava
had reached the Red Sea. Odds of an eruption looked slim, and the area is not exactly an air transport
corridor. Yes, if marine shipping, especially oil tankers, were interrupted one hardly needs a computing cloud to figure out we'd have a whole different situation. We've asked several contacts in the shipping business if LNG (Liquid Natural Gas) vessels can be safely sailed when hot ash is drifting down. Sadly, for a lot of reasons (poverty, civil war, not much activity, horrible climate ...) there was no modern seismological measuring gear in the area.
A little later, we were surprised, but not concerned that Dubbi had erupted explosively. Early reports
indicated an ash cloud larger and higher than we expected, but we had ignored Dylan's "advice" and had not checked what the winds of East Africa were doing.
BREAKING NEWS UPDATE