PREDICTION OF EPICENTER AND SHORT-TERM PRECURSORS OF THE JUNE 17 EARTHQUAKE

For various reasons it was expected that the next SISZ earthquake would occur in the epicentral area of the June 17 earthquake. This was stated most clearly in 1988 as follows: "... there are strong indications that the next large earthquake of size approaching 7 in this zone will take place near longitude 20.3$^{\circ}$-20.4$^{\circ}$W" (Stefánsson and Halldórsson 1988), i.e. on the EW oriented SISZ at around 64$^{\circ}$. This was based on a lack of strain release in historical earthquakes since the year 1700 in a narrow area (Halldórsson 1987; Stefánsson and Halldórsson 1988). A similar but less pronounced a gap was also indicated for a narrow area around 20.7$^{\circ}$W. Some years later it was pointed out that these gaps coincided with a long-term concentration of microearthquake activity in the seismic zone (Stefánsson et al. 1993). It was never stated definitely if the microearthquake activity was expressing aseismic strain release or if it was reflecting high stresses in preparation of a large earthquake, although the latter was assumed to be more likely. The question has apparently now been answered by nature. As there is a tendency in the seismic history for earthquake sequences to start east of the center of the zone and to trigger earthquakes farther west, it was also expected that any subsequent large earthquake would most probably occur to the west. On the basis of historical intensitites and known earthquake faults, it was also expected that the fault planes of the earthquakes would have NS strike.

While the June 17 earthquake occurred in the area identified as the probable location of the next large earthquake in the SISZ, no short-term precursory signals were recognized before it occurred. In hindsight, however, several changes can be recognized, changes that were possibly premonitory. These include:

1) Several microearthquakes (ML=0-1) clustered at depth along the fault of the impending earthquake, weeks and days before its occurrence (Figure 5). This was a significant change compared to the less clustered microearthquake activity that had been observed during the previous 10 years.

2) In a geothermal borehole at Flúðir 10 km to the north of the NS striking fault plane (Figure 5), short-term water level drop, of not less than 5 m, was alarmed 24 hours before the earthquake. Unfortuneately this signal is not seen in the preserved recorded trace averaging water level on 15 minutes intervals. During a 5 years of continuous operation this was the first such an alarm (Björnsson et al. 2001). Most probably the alarm was a short-lived low pressure pulse related to the preparatory phase of the earthquake.

Both these changes were possibly related to relocation of strains towards the impending rupture, possibly related to dislocation in the deeper and ductile part of the seismic zone. Also mechanism of slow rupture initiation, as described by Dietrich 1986, and Roy and Marone 1996, may apply in this case.

Several other observed changes, listed below, prior to the earthquake may possibly also be related to a crustal process leading to it:

1) The volcano Hekla (Figure 1), 30-35 km east of the epicenter, has been anomalously active since 1970, last erupting at the end of February 2000. An eruption with similar mechanism occurred in 1991. In that case a flurry of small earthquakes followed in the SISZ during days and weeks (Stefánsson et al. 1993). There are also earlier examples of this effect which probably is caused by transfer of strain energy along a seismically active zone where local spots are close to fracture criticality, even if the area is not ready for a large earthquake. Thus it was considered noticeable that small earthquakes did not occur in the zone following the February 2000 eruption. After the large earthquakes occurred in the zone it is tempting to suggest that this lack of earthquakes is comparable to periods of quiescence sometimes reported before large earthquakes. It remains to be studied what is the mechanism of stress transfer in the region and how this may explain the flurry of small earthquakes frequently observed as well as the lack of it this time. Better understanding of this may lead to the application of such a quiescence as a precursor.

2) A slight but persisting increase in seismicity was observed in the SISZ area early in 1995-1996 (Guðmundsson et al. 2001). On a long-term scale the seismic rate in the SISZ was slightly increasing since that time. This may have been linked to dyke intrusions in Vatnajökull area, 150 km towards northeast in the eastern volcanic zone, preceding the large Vatnajökull eruption in October 1996. Such a mechanism was proposed (Stuart Crampin 2000, personal communication; Volti and Crampin 2003) to explain increased stress in the area indicated by shear-wave splitting. There were also marked and possibly related changes in seismic rates at different parts of the zone, especially during the last 1-2 years before the earthquake. Automatic fault plane solutions of microearthquakes showed anomalous variations during 3 months before the earthquakes, 20 km to the west of these, and inside the SISZ (Stefánsson et al. 2000).

3) Radon anomalies, i.e. anomalously low values and positive spikes were observed in radon from geothermal water wells in the area during 1-5 months before the earthquakes (Einarsson et al. 2003).

4) Anomalous strain signals were observed in May and June at borehole strainmeter stations, 3 and 20 km respectively from the epicenter of the June 17 earthquake.

5) There was some increase in shear-wave splitting time in microearthquakes of local origin, a few weeks before the earthquake (Stuart Crampin 2000, personal communication; Volti and Crampin 2003).

These observations of possible premonitory changes have been reported by scientists at the Icelandic Meteorological Office, the National Energy Authority, the Science Institute, University of Iceland and the University of Edinburgh. They are now the subject of further studies and modelling to try to understand how they might be related to the earthquakes.

 

Figure 5: The area near the June 17 earthquake. The epicenter is shown by a green star and the rupture is indicated with a red NS striking line. The NS striking black lines indicate old earthquake faults. Red dots show microearthquakes (36 in number, ML=-1 to 1), during a period of 17 days before the earthquake. Hydrological changes were observed at the village Flúðir.

=