Task 4: Methods for monitoring of stable/unstable fault movements

Start: October 1996 (month 8)
End: February 1998 (month 24)
Responsible partner: UUPP.DGEO
Cooperative partner: IMOR.DG

The work in this field has been concentrated in developing tools for handling the large amounts of data produced by the PRENLAB network. Examples will be given in a number of figures from the TFZ transform fault north of Iceland. The activity from August 1, 1997 onwards will be used in this demonstration. This analysis is based primarily on a multievent location algorithm [] and fault plane solution inversion using spectral amplitudes and first motion directions [72,].

The activity during August 1-9, 1997, is shown in Figure 8. Figures 9 and 10 shows fitting of planes to different grouping of these events. One should note that accurate locations are produced although the closest station is more than 20 km from this rather shallow activity. The magnitudes of these events are mostly in the range M_L=0.5-1.5. The estimated relative location uncertainties are about 10-30 m.

  

Figure: The left map is an overview showing the stations as diamonds and the events of the swarms on August 1-9, 1997, are shown as circles with diameter 500 m. The closest station in the analysis is about 22-28 km from the events. The events have been located by multievent analysis and the blow-up to the right gives a clearer picture of the epicenters. The diameter of the event circles in the right map is 200 m. Note that the multievent location gives accurate locations although the ratio closest station distance to depth is about 3.

  

Figure 9: The left map shows the again the epicenters while the right part is a depth view along the strike of the plane fitting the hypocenter locations of the events. The surface interception of the best fitting plane is shown in the left map. The RMS deviation from the plane is 88 m. This is larger than the estimated uncertainty of the relative locations of these events. The sizes of the event circles are 200 m and the depth scale of the right part is in kilometers.

  

Figure 10: This is a slight blow-up of the central densest group of events. The sizes of the event circles are here 100 m. The RMS deviation from the best fitting plane is 12 m which is in agreement with the estimated uncertainties of the relative locations of the events. This illustrates that rather accurate locations are achieved even with only few stations and with no station very close to the epicenters.

The consistency between the multievent locations and the routine fault plane solutions of the SIL network is illustrated in Figure 11 which shows the best fitting fault planes from

  

Figure: This figure is the same as Figure 10 but here the events are shown as disks. The radii are the estimated fault radius of each event. Among all fault plane solutions (FPS) having an acceptable fit to the observed spectral amplitudes and first motion directions the one having one of its two possible fault planes closest to the orientation of plane fitting the hypocenter locations has been chosen. For each event the disk shows the orientation of that FPS plane. The median deviation from the hypocentral plane is which is reasonable as the grid step in the FPS analysis is . This figure indicates a reasonable consistency between the multievent locations and the FPS inversion.

the results of the fault plane solutions. A reasonable conclusion is that the plane marked in the figure well approximates the slipping fracture.

Figure 12 shows how the microearthquakes are distributed over the fault area. The sizes of the earthquakes are estimated from the corner frequencies. One can see that the seismicly slipping parts do not cover the whole active area and also that some parts slips seismicly several times during the two days of this activity. In addition the peak slips (not shown in this figure) varies between 0.03 and 3 mm which gives a remarkable different size of total slip over the area during this concentrated activity.

The group in Figures 10, 11, and 12 is very close to other events, see Figure 13. To get some indications about the reality of the slight location differences the distributions of their dynamic parameters (seismic moment M₀and peak slip sl) are shown in Figure 14. There is a clear difference in b-value and also in the distribution of the peak slip. This supports the indication by the multievent location that the additional events do not belong to the same fracture as the starting group.

  

Figure: This figure shows the same disks as Figure 11 but now the depth view to the right is normal to the strike of the hypocentral plane. The slight marks at the peripheries of the circles show the slip direction of the bedrock of the visible side of the disks. This slip direction pertain for each event to the chosen acceptable FPS. Interesting aspects of this figure is that parts of the active area has no detected seismic event and that parts of the area slip seismicly several times during the two days of this activity.

  

Figure: This figure shows the group of the previous figures with small circles while the larger circles show the other close events. The sizes of the larger circles are 200 m. An obvious question is if the small location differences are real.

The activity along this fault has continued after August 9, 1997, and the main features of this activity are summarized in Figure 15 which shows the locations of all major swarms after August 1. The figures close to the central parts of the swarm fault areas indicate their relative times in days. One can see that later activity tends to occur in parts of the faults that are close to immediate previous major faults. The simplest guess is that the swarm activity is caused by an increased velocity of the stable slip of that part of the fault. There seems to be a complicated episodic migration of the stable slip along the fault.

The recent state- and rate-dependent models for fault stability may be valuable tools in the analysis and in understanding of of fault behaviour as illustrated in these figures.

  

Figure: The distributions of two dynamic parameters of the two groups of events, the seismic moment M₀, and the peak slip sl. In the right part the distributions are shown in a log-log plot. The small circle group contains 124 events and the great circle group contains 47 events. For each group the right curve shows the distribution of log(M₀) and the left group shows the distribution of log(sl). For the log(M₀) distribution the horizontal scale is approximately the local magnitude M_L. For log(sl) a value 1 means an estimated peak slip of 1 mm, 0 means a peak slip of 0.1 mm, etc. The two groups have different b-values (slope of log(M₀)) and also different slip distributions. This gives some support to the different hypocenter locations.

  

Figure: The locations of swarm activity along the fault after August 1, 1997. To each major swarm a plane containing most of the events has been fitted. In the depth view to the right (normal to the strike) the figure close to the central part of each plane gives the day number of the swarm activity (typically they last for a week or more). The first swarm is marked by day 1. The swarms marked 1 and 6 are based on multievent locations and for these two groups the individual event locations are shown. The remaining groups are based on single event locations which explains there wider depth distributions. The activity migrates in such a way that calm areas close to active areas are likely areas of next activity. Note that plane marked 13 overlapping plane marked 6 ``avoids'' the most active part of plane marked 6 in agreement with this statement.