The occurrence of the North Yemen earthquake of 13 December 1982 provides an important opportunity to study source properties of earthquakes in the southern Arabian Peninsula. Although moderate in size (mb 6.0), this earthquake is the largest to have occurred in the region since the deployment of the Global Digital Seismograph Network. P waves recorded at teleseismic distances can be obtained from stations of this network and of the Graefenberg array that are flat to displacement and velocity in the frequency range from 0.01 to 5.0 Hz. The broad bandwidth of the data is necessary for deriving a dynamic description of the earthquake as well as the static properties of the source. In contrast, inversions using only long-period data provide much less insight into the complexity of rupture. By fitting the displacement pulse shapes with synthetics, we find that the Yemen earthquake was a complex rupture consisting of two events that occurred about 3.0 sec apart. The depth of both hypocenters is estimated to be 7 km. The focal mechanisms of the events each have a nodal plane that strikes 340° and dips 60° to the east. However, the slip directions of the focal mechanisms are in slightly different directions, 230° and 300°, respectively. The total moment of the earthquake is 3.0 × 1025 dyne-cm, with the second event having three times the moment of the initial event. By inverting P-wave arrival time differences at each station, we find the second hypocenter is located 7.5 km from the initial hypocenter in a direction that is nearly colinear with the NNW-striking nodal plane of the focal mechanisms of each event. The trend of extensional ground cracks in the epicentral region and the predominant strike of composite focal mechanisms of locally recorded aftershocks are also in a NNW direction. From the consistency of these observations, we infer that the east-dipping NNW-striking nodal planes are the fault planes. The static stress drop of the first and second events are 44 and 31 bars, respectively. Inverting the energy flux in the velocity waveforms, we find the energy radiated by the earthquake is 4.8 × 1020 dyne-cm. This implies an average apparent stress of 4 bars.
At different times during the aftershock sequence, seismicity tended to cluster about one or another of two different planes which, although having roughly the same strike, had conjugate dips. One of these planes, delineated late in the aftershock sequence, coincides with the inferred fault plane for the main shock. The complexity of the main shock and of the aftershock sequence suggests that strain accumulated on a system of strongly interdependent faults. As the region became critically loaded, the effect of a major rupture was to critically stress faults adjacent to the initial nucleation. The second event of the main shock represents the rapid release of stress from the fault system, while the ensuing aftershock sequence represents the gradual release of stress.