Detailed analysis of focal mechanisms of earthquakes and recent faulting in the central part of the SISZ (work carried out with Gudmundsson and Rögnvaldsson)

$\\$First, a seismotectonic analysis was carried out in the Vördufell mountain considered as a test area (64.05$^{\circ }$-64.12$^{\circ }$N and 20.5$^{\circ }$-20.6$^{\circ }$W). Good outcrops permit geological study of the recent faults which affect the quaternary lava pile, and focal mechanisms of shallow earthquakes are available. Among the earthquakes recorded in southern Iceland by the SIL network with determinations of earthquakes mechanisms, a dataset of 68 earthquakes, occurring in this area from 1991 to 1995, was used. 18 data corresponding to quarry blasts were rejected, and 50 focal mechanisms of natural earthquakes were selected.

The variety of these focal mechanisms of shallow earthquakes shows that the whole set cannot be accounted for by a single tectonic stress state. Numerous mechanical analyses were done, based either on geometrical considerations or on stress-slip consistency. Within the range of acceptable misfits (defined as a function of both the assumptions about stress-slip relationships and the uncertainties of data), a separation of two main groups of data (and related stress regimes) accounts for the whole dataset. First, using the P- and T-dihedra method, the general consistency within each group is highlighted. The largest subset includes 30 strike-slip, 4 reverse and 4 normal mechanisms. It is consistent with NW-SE compression and NE-SW extension, in agreement with left-lateral shear along E-W trends and right lateral strike-slip on N-S trending faults. The smallest subset includes 8 strike-slip, 1 reverse and 3 normal mechanisms. Its indicates NE-SW extension and NW-SE compression, with more dispersion than for the main group. Second, numerical inverse methods were used in order to compute the average stress tensors which best fit the observed fault plane solutions. Two main methods were used: the 4-D search (R4DT-R4DS), and the direct inversion method (INVD). Contrary to the P- and T-dihedra method, these methods require a choice among nodal planes. Because of its arbitrary character in geological terms, the choice of the nodal plane which best fits an average stress tensor was not adopted as an unique criterion. The geological study in the field and from aerial photographs allowed identification of the orientations of faults, fractures and other zones of weakness at both the scale of outcrops and that of the Vördufell mountain. A comparison between fault plane solutions of earthquakes and fault slip data observed in outcrops was carried out. Combining these three criteria resulted in the final selection. In terms of numerical estimators considered alone, one may simply choose the best fitting fault plane solution for each earthquake. This was not done because, dealing with shallow earthquakes, reasonable choices between nodal planes imply consideration of the geological structure.

For each subset, the orientation of stress axes, the ratios of principal stress differences and the misfit estimators depend relatively little on the method adopted. The results are quite significant for the main subset. For the secondary subset, the misfits are larger despite its smaller size, which indicates inhomogeneity. Weighting data according to the quality of individual determinations did not result in significant improvement which suggests that mechanisms with small weights are quite acceptable. The main difference between the subsets mainly results from a kind of permutation between extreme stress axes. The direction of the maximum stress average N60$^{\circ }$E for the main subset and N120$^{\circ }$E for the subsidiary one, while the directions of the minimum stress average average N150$^{\circ }$E and N40$^{\circ }$E, respectively. The main subset reflects regional tectonic mechanisms, whereas the secondary subset, mechanically less consistent, principally reflects fault rebound and local accommodation.

Before examining earthquake data, a surprising result of geological studies of fault slip data in the field was the identification of two opposite tectonic regimes, respectively characterized by a NW-SE extension (the major one, principally including strike-slip and normal faults) and a NW-SE compression (the minor one, principally including strike-slip and few reverse faults). We point out first that for most of geological and geophysical data independently collected, similar tectonic regimes dominated by NW-SE maximum stress and NE-SW minimum stress were identified, and second that both these studies revealed permutation of extreme stress axes for the remaining data.

We conclude that the Vördufell area is dominated by NW-SE extension, principally accommodated by strike-slip and normal faulting, in agreement with the general behaviour of the SISZ. Local stress permutations, however, play a major role, resulting in subsets of conflicting mechanisms for both the present-day shallow earthquakes and the quaternary fault movements.