Control of rupture by fault geometry during the 1980 El Asnam (Algeria) earthquake

Abstract

Summary. The El Asnam earthquake of 1980 October 10 (Ms= 7.3) occurred on a segmented reverse fault in the Atlas Mountains of Algeria. This report re-examines the teleseismic data for the main shock and major aftershocks in the light of detailed studies of the surface breaks, geodetic changes and aftershock distribution. The observed thrust fault is split into southern, central and northern segments by distinct offsets and changes in trend. The southern and central segments are each about 12 km long, but the northern segment showed only 3–4 km of surface thrust breaks. However, widespread tensional faulting on ridges in the north-east area, together with focal mechanisms of locally-recorded aftershocks, indicates that a series of imbricate listric thrusts exists to the north of the northern fault segment. Using ISC arrival-time lists, a relative relocation scheme is applied to the main shock and major aftershocks to improve their locations with respect to the mapped faults. The main shock epicentre was at the south-west end of the fault system, in a zone where the thrust has a zig-zag surface trace. By contrast most of the major aftershocks were located further north-east, and were associated with the northern thrust segment and the imbricate thrusts to the north. The largest aftershock (mb= 6.1), which occurred about 3 hr after the main shock, was located to the east of the main thrust fault, beneath the Chelif alluvial basin. First-motion fault plane solutions are presented for the main shock and largest aftershock. The main shock solution indicates a fault plane dipping 54° NW, with strike 040°. The aftershock solution, however, has a more E–W trend (080°), in agreement with the trends of the basin edges near its location. The rupture history of the main shock is investigated by forward modelling of the long-period P-waves. Four subevents are clearly distinguishable. The first three subevents occurred in rapid succession and contributed to the first cycle of the P-wave radiation. These three subevents represent the successive rupturing of the southern, central and northern fault segments. In order to match the observed waveforms, these three ruptures must have different mechanisms and moments: the fault segments have successively shallower dipts from south to north (54°, 45° and 40° respectively), and the central segment had the largest moment release (in agreement with the geodetic data of Ruegg et al.). The delay times between the three sources suggest a rupture velocity of about 3.5 km s−1 and their–nucleation depths were 7 km. The junctions between the fault segments acted as geometric barriers to the rupture propagation, but were broken by its passage. The fourth subevent of the main shock produced the second cycle of the P-wave radiation. It began about 18 s after the first, i.e. several seconds after the end of the third rupture. Nabelek demonstrates that the fourth subevent occurred on a shallow-angle (20°) thrust plane; examination of the locally recorded aftershocks suggests that this plane is a decollement surface underlying the imbricate thrusts of the north-east area. The total seismic moment found from the body-wave modelling is 7.7 × 1026 dyne cm, in good agreement with geodetic and surface-wave studies.