Source characterization of two Reykjanes Ridge earthquakes: Surface waves and moment tensors; P waveforms and nonorthogonal nodal planes

Abstract

Well‐constrained fault plane solutions from P wave first motions for mid‐ocean ridge normal faulting earthquakes usually require nonorthogonal nodal planes. Local structural effects and/or departures from a double‐couple source mechanism have been invoked to explain this phenomenon. In order to obtain an independent determination of the source mechanisms for the April 24, 1970, and April 3, 1972, events on the southern Reykjanes Ridge, we invert the Rayleigh wave radiation pattern to obtain the source moment tensor. The moment tensor formulation should be particularly well suited to this problem because it is not restricted a priori to a double‐couple source mechanism. A potential drawback of the technique, however, is the requirement that phase velocities along the earthquake‐station paths be known very accurately in order to obtain the source phase from the observed phase, and an objective of this study was to determine whether a regionalized phase velocity model compiled from published dispersion curves is adequate. The results of the moment tensor inversion for both events indicate shallow normal faulting with the tension axis approximately horizontal and perpendicular to the local strike of the ridge. Apparent departures from a pure double‐couple source seem to result from errors in the data and the poor resolution of the Mxz and Myz components of the moment tensor for shallow sources. After performing the inversion under a series of increasingly more stringent constraints we conclude that the data for both events are compatible with pure double‐couple sources with moments of 4.8 and 7.5 × 1024 dyn cm, respectively. We then show that interference between P, pP, and sP due to shallowness of the source can account for the observed nonorthogonality and match the observed P waveforms for the April 3, 1972, event with theoretical seismograms calculated for a shear fault whose orientation is consistent with the surface wave solution. The best fit to the data is obtained for a long, narrow fault (13 km by 3 km), with rupture initiating near the seafloor. The moment indicated by the P waves is 7.5 × 1024 dyn cm. These source parameters give an average displacement of about 60 cm and a stress drop of 30–60 bars, taking into account various uncertainties. Although we might expect attentuation to be high in the mid‐ocean ridge environment, the average attenuation required to fit the teleseismic data is not higher than normal (t* = 1 s). The P waves from the April 24, 1970, earthquake were too small to be suitable for quantitative modeling by synthetic seismograms but are qualitatively consistent with a shallow fault model similar to that for the larger event. We conclude that the faulting process described by these two earthquake mechanisms is directly related to the formation of rift valley topography.