Microearthquake analysis at local seismic networks in Iceland and Sweden and earthquake precursors

Abstract

Microearthquakes represent a flow of information from the seismogenic crust up to the surface if they are detected and recorded. If the detection threshold is lowered two magnitudes the amount of information is increased by a factor of about 40 as all microearthquakes essentially give the same amount of information. This is the basic idea behind the local seismic networks in Iceland and Sweden. The large number of microearthquakes requires automatic routine analysis. For these networks the routine automatic analysis includes phase detection, event detection and location, wave form analysis, fault plane solution, and estimation of fault radius and dynamic source parameters. The Icelandic network has since 1990 detected and recorded over 180,000 microearthquakes which all have been automatically analysed and inverted for source mechanisms. In addition two M S >6.4 earthquakes occurred in June 2000 within the well covered area in SW Iceland. A number of different precursors have been observed. Foreshocks before all M L >4.9 earthquakes within the network are found and their behaviour is similar to observations from shallow strike slip earthquakes in other parts of the world. Premonitary swarms are also observed at the epicenters for several years. Two new precursors which are related to recent theoretical achievements were also observed. One was the statistical ditribution of the microearthquake fault radii. They showed in all cases remarkable changes including high values prior to the earthquakes. This is in agreement with the idea of each fracture and loading system to have a minimum fault radius. Changes in loading rate may then influence the fault radii. Another related precursor is the observation of increased percentage of low stress drop events before the earthquakes. This may possibly be related to seismic activity at the edges of larger locked patches (stronger asperities) and/or to increasing loading rate. The combination of the foreshock observations and the fault radius variations together with the old idea of a limited number of microearthquakes involved in an asperity breaking gave a promising precursory concept: the domino pattern. The essence of this pattern is to have an increased rate of microearthquakes within a small volume followed by a lower rate (quiescence) and ending with an increasing rate. The number of events in such a pattern was found to be 30–60 within a volume of 4–6 km diameter. Such pattern could be detected by measuring the fit to the theoretical activity rate for an asperity breaking model. Based on these observations a simple earthquake warning algorithm is described and applied to all Icelandic data for the years 1990–2001. If the warning threshold is set to produce warnings 0.25–3 days before the six M L >5.0 earthquakes and within 6 km of their epicentres the number of false alarms was 54 meaning a probability for the warning to be correct of 10%. Most of the false warnings are at correct places but months or years too early. This retrospective study illustrates the high value of the microearthquake information when the detection threshold is about M L =0. The Icelandic data will be used for future more detailed studies of the earthquake nucleation processes hopefully leading to high quality earthquake warning algorithms.