Earthquake prediction at the South Icelandic Fracture Zone

Fissure from the the first of the 2000 Icelandic earthquakes. Photograph by Ross Beyer.

Volumetric strainmeters are used around the Globe to try to predict earthquakes. One of the main reasons for the proliferation of them is to be found in Iceland. Before we go to the latest events in the Southern Icelandic Fracture Zone (SIFZ) we should start with the classic example from prior to the M6.6 earthquake on the 17^th of June in 2000.

The 2000 Big Bump at Saurbaer strainmeter. Image by Ragnar Stefánsson.

The spike above is known in its highly scientific term as the “Big Bump”, it was recorded as starting on the 28^th of May and ended on the 1^st of June. It is believed to have been caused by a pocket of fluids being squashed upwards by the mounting strain at the fault line. Notice that the big bump lowered the crustal strain as it ended, something that probably helped release the earthquake 16 days later.

Icelanders are Icelanders so they are pretty cool with their natural disasters.
An active transform fault that goes right through the town of Hveragerdi, South Iceland. In one place it lies under the town’s shopping centre. Image from earthquake.is

The big bump was recorded by the Saurbaer strainmeter that to my knowledge today is defunct, or not publicly available. This means that we sadly can’t compare the readings for the latest earthquakes with that same station.

Another thing is that at the onset of the earthquakes there was a signal recorded on all strainmeters around Hekla. The signal was a negative strain signal 2 seconds prior to the first break of the earthquake and can be seen as the first part of the P-wave.

There are though at least 3 problems with these readings. The first problem is that even if you see something like the big bump on a strainmeter it does not imply really that an earthquake must occur, nor does it in any way state when it will occur.

Second problem is that the reason for the big bump was a fluid pocket being relocated upwards. So it only works if there is such a pocket inside the fault line.

The third problem is with the 2-second warning, for most points and purposes that is way too late to make a difference.

May 8^th 2014 Earthquake

I will here just relate the facts, and leave the theorizing to the relevant geologists, I do though find it intriguing enough to write about it.

As you can see in the plot below a medium fast increase in strain started about half an hour before the initial break of the M4.2 earthquake at the Búrfell Strainmeter, as the initial break occurred the strain rapidly fell at such a speed that I for a few minutes thought that Hekla would erupt since the Hekla strainmeter reacted by a strain increase.

Image by the Icelandic Met Office.

Image by the Icelandic Met Office.

No big bump was recorded on any strainmeter, but there is an intriguing “small bump” starting at ten in the morning on the 8^th that might be related. It could though equally well be related to activity at Herdubreid or out at Geirfugladrangur.

But, to the best of my knowledge no strain increase was recorded in the same way during the 2000 earthquakes in Iceland, so I wrote it off as a one off fluke.

At ten o’clock the next morning yet another medium rapid strain increase started and this time it ended with a M3.6 earthquake at Geirfugladrangur. The strain fall part did though not occur this time around; instead the strainmeter resumed the previous trajectory that was interrupted by the strain increase. This surprised me quite a lot since I would have bet that such an earthquake was both to small and too far away to greatly influence the Búrfell strainmeter, and to the best of my knowledge no such influence had been recorded before. I was therefore watching for another one.

After effects of the earthquake.

I did not have to wait for long; at twelve o’clock another one occurred. This time I had to check for a corresponding earthquake. I found it a whopping 260 kilometers due west of Geirfugl, and even though it was a M3.9 to M4.4 earthquake it should not have been registered as well as it was.

Now I was sitting waiting for it to happen again, but since M3 or larger earthquakes are fairly far apart I did expect to have to wait for quite some time. As it happened I did not have to wait for long, at seven in the evening the same day yet another M3.4 earthquake happened at Geirfugladrangur. This time nothing showed on the strainmeter. Nor did it show during a M3.2 earthquake north of Eiriksjökull on the 10th or during the M3.6 Eldey earthquake the day after that.

This left me fairly stumped and out of ideas on what was going on and with a lot of questions that I have no good answer for. Why did 3 earthquakes in a row come with a strain increase ending with initial break? How could such a far away earthquake affect strain at such a large distance? What had changed at Búrfell since 2000 to give a readable signal from the 3 earthquakes? And why did the last 3 of the earthquakes not show up on the strainmeter?

Final thoughts

In the end this all proves how hard it is to find a way to predict earthquakes. You think you are on to something, but the same pre-signal rarely if ever comes twice. And in the end you are left with more questions than you started with.

In the end I think we will be able to predict many earthquakes and I even think I know a way to get to some, or even most of them. Question in the end is probably cost. The solution I think is the most viable is both large in scale and does not come cheap. I am sorry that I can’t write what I think would be a way forward, I will happily write what it is after I have filed the patent application.

CARL

Reference

Ragnar Stefánsson: Advances in Earthquake Prediction: Research and Risk Mitigation