Reducing the prediction error of seismic-phase travel times leads directly to
improvement in earthquake location accuracy. One-dimensional (1D) velocity models
are
most commonly used to calculate seismic phase travel times because computer codes
are readily available and the computations can be done on inexpensive hardware. Travel
time predictions based on 1D models are accurate to within 1 to 2 seconds at teleseimic
distance, but complex crust and upper mantle structure can triple prediction errors
at
regional distance. Increased travel time prediction error at regional distance is
particularly prevalent in regions like Central America, where subduction tectonics
results in large lateral variations in seismic velocity and crustal thickness. The
Regional
Seismic Travel Time (RSTT) method (Myers et al., 2010) was specifically developed
to
improve travel time prediction accuracy by accounting for 3D seismic velocity structure.
The RSTT model is inherently global in extent, but the accuracy of travel time prediction
is dependent on the accuracy of structure along each ray path. Many studies make use
of the well-instrumented Nicoya Peninsula in Northern Costa Rica and the underlying
seismogenic zone to image the structure of the region. In this study we update the
RSTT
3D velocity model in northern and Central Costa Rica using published studies of velocity
structure (DeShon et al., 2006 and Arroyo et al., 2009). Travel times for the updated
model are compared to observed travel times for well-constrained earthquakes. We
relocate the earthquakes using only regional data to measure the improvement in
location that can be achieved with the updated model. This work was performed under
the auspices of the U.S. Department of Energy by Lawrence Livermore National
Laboratory under Contract DE-AC52-07NA27344.
