Seismicity and Improved Velocity Structure in Kuwait

The Kuwait National Seismic Network (KNSN) began operation in 1997 and consists of nine three-component stations (eight short-period and one broadband) and is operated by the Kuwait Institute for Scientific Research. Although the region is largely believed to be aseismic, considerable local seismicity is recorded by KNSN. Seismic events in Kuwait are clustered in two main groups, one in the south and another in the north. The KNSN station distribution is able to capture the southern cluster within the footprint of the network but the northern cluster is poorly covered. Events tend to occur at depths ranging from the free surface to about 20 km. Events in the northern cluster tend to be deeper than those in south, however this might be an artifact of the station coverage. We analyzed KNSN recordings of nearly 200 local events to improve understanding of seismic events and crustal structure in Kuwait, performing several analyses with increasing complexity. First, we obtained an optimized one-dimensional (1D) velocity model for the entire region using the reported KNSN arrival times and routine locations. The resulting model is consistent with a recently obtained model from the joint inversion of receiver functions and surface wave group velocities. Crustal structure is capped by the thick ({approx} 7 km) sedimentary rocks of the Arabian Platform underlain by normal velocities for stable continental crust. Our new model has a crustal thickness of 44 km, constrained by an independent study of receiver functions and surface wave group velocities by Pasyanos et al (2006). Locations and depths of events after relocation with the new model are broadly consistent with those reported by KISR, although a few events move more than a few kilometers. We then used a double-difference tomography technique (tomoDD) to jointly locate the events and estimate three-dimensional (3D) velocity structure. TomoDD is based on hypoDD relocation algorithm and it makes use of both absolute and relative arrival times. We obtained {approx}1500 absolute P and S arrival times and {approx}3200 P and S wave arrival time differences. Event locations do not change greatly when 3D velocity structure is included. Three-dimensional velocity structure, where resolvable, does not differ greatly from our optimized 1D model, indicating that the improved 1D model is adequate for routine event location. Finally, we calculated moment magnitudes, MW, for nearly 155 events using the coda magnitude technique of Mayeda et al., (2003). The fact that most of the relocated events occur below the known sedimentary structures extending to 7 km suggests that they are tectonic in origin. Shallow events within the sedimentary crust in the (southern) Minagish region may be related to oil field activities, although the current study cannot unambiguously determine the source of current seismicity in Kuwait. The improved velocity model reduces the scatter of travel time residuals relative to the locations reported in the KNSN bulletin and may be used for ground motion prediction and hazard estimate studies in Kuwait