D42 A map of Coulomb failure stress changes indicating areas of increased hazard – NVI

We have analysed all campaign and continuous GPS data collected on Reykjanes Peninsula and in the South Iceland Seismic Zone from 1992-2004, using the GAMIT/GLOBK software (King and Bock, 2003; Herring, 2003), in the ITRF2000 reference frame. We estimated the average velocity field for the time period 1992-2000 (Árnadóttir et al., 2005), prior to the June 2000 earthquake sequence in the south Iceland seismic zone (Stefánsson et al., 2003). We also processed all continuous GPS data collected in Iceland (ISGPS data) and generated a velocity field for stations in Iceland in the ITRF2000 reference frame. This work is described in Geirsson et al. (2005). In addition to the plate spreading, the GPS velocity field is affected by inflation at Hengill (Sigmundsson et al., 1997; Feigl et al., 2000) and Hekla volcanoes.

We use the non-linear optimization algorithm (D40) to model the pre-seismic velocity field, to explain the plate boundary deformation (Árnadóttir et al., 2005). Our preferred model is a 3D dislocation and point source model with left-lateral slip along the plate boundary on the Reykjanes Peninsula and below the SISZ, and opening across the Western and Eastern Volcanic Zones. We can not uniquely determine all our model parameters from the GPS data. In particular, there is a trade-off between the locking depth and the deep slip rate. For constant deep slip rates in the range 16-20 mm/yr, we find a 5-7 km locking depth on the Reykjanes Peninsula, and 8-11 km depth below the SISZ. These locking depths show reasonable agreement with the thickness of the seismogenic layer in southwest Iceland, which appears to vary between 8 and 10 km.

Using our preferred plate boundary model we can estimate the rate of Coulomb failure stress change during the pre-seismic time interval, using standard methods (e.g. Harris, 1998). Figure 1 shows that the inflation in Hengill and left-lateral slip below 7 km depth along the plate boundary on Reykjanes Peninsula combine to increase the Coulomb failure stress on N-S, right-lateral strike-slip faults on the Reykjanes Peninsula, such as ruptured in the triggered earthquakes on June 17, 2000. The Hengill inflation appears to have acted to decrease the Coulomb failure stress on N-S faults east of Hengill, possibly explaining why the June 2000 earthquake sequence did not progress further west following the June 21, 2000 main shock.

Fig. 1: The rate of Coulomb failure stress change at 5 km depth, on N-S right-lateral strike-slip faults, caused by plate boundary deformation (Árnadóttir et al., 2005).  Warm colors (yellow to red) indicate an increase in Coulomb failure stress. Small changes in Coulomb failure stress (0.001 MPa) have been shown to correlate with areas of increased seismicity following a mainshock (e.g. Harris, 1998). The model assumes left lateral slip of 20 mm/yr along the Reykjanes Peninsula and the SISZ, below 7 and 11 km depth, respectively. The model also includes opening across the Western and Eastern Volcanic Zones as well as inflation at Hengill and Hekla volcanoes and subsidence at Svartsengi. The diamonds denote earthquake locations, where the June 17 and 21 main shocks are labelled J17 and J21 respectively, the numbers 1, 2, 3 refer to the June 17 triggered earthquakes. The M=5 1998 earthquakes in the Hengill and Ölfus are labelled J4 (June 4) and N13 (November 13).

Conlusion:

The velocity field during the pre-seismic time interval has been used to generate a 3D kinematic model for the plate boundary in SW Iceland. The model was then used to estimate the rate of Coulomb failure stress changes due to the crustal deformation. The analysis provides a map of stress changes. Integrated analysis of stress changes and changes in the seismicity rate is important when evaluating future seismic hazard in the area.

A general algorithm has been completed and implemented for joint inversion of InSAR and GPS data in South Iceland (Deliverable D40). It is described in details in a scientific publication (Pedersen et al., 2003). The algorithm was further tested and implemented in a paper by Pagli et al. (2003) using InSAR data to constrain triggered fault slip on the Reykjanes Peninsula in 2003, and for joint inversion if InSAR and GPS data on Reykjanes Peninsula (Árnadóttir et al., 2004). The algorithm has also been used for magmatic sources in Iceland (Pedersen and Sigmundsson, 2004; Pedersen and Sigmundsson, 2005; Pagli et al, 2005). It was developed further to model the plate boundary in SW Iceland and calculate Coulomb failure stress changes for the plate boundary model (Árnadóttir et al., 2005).