Using the International Monitoring System to Validate SALSA3D: A Global 3D P-Velocity Model of the Crust and Mantle for Improved Seismic Event Location

We demonstrate the ability of SALSA3D (SAndia LoS Alamos 3D) version 2.1, a global 3D P-wave velocity model of the Earth's crust and mantle, to reduce mislocations compared to those derived from standard 1D and 2-2.5D models, for a set of realizations using only IMS stations - an example of a sparse network whose locations depend heavily on the velocity model employed - and a carefully chosen group of globally-distributed ground truth (GT) events. Our model is derived from the latest version of the GT catalog of P/Pn travel-time picks assembled by Los Alamos National Laboratory. The model uses the GeoTess triangular tessellation system described by Ballard et al. (2009; www.sandia.gov/geotess), which incorporates variable resolution both laterally and radially. For our starting model, we use a simplified version of the NNSA Unified model in Eurasia and the Crust 2.0 model elsewhere. Damping reduces velocity adjustments so that ray path changes between iterations are small. We obtain proper model smoothness via progressive grid refinement using the diagonal of the model resolution matrix to determine where the data warrant such a refinement. Our approach provides more consistent and continuous areas of refinement, producing a smooth, multi-resolution model with node density appropriate to both ray coverage and the velocity gradients required by the data. This scheme is computationally expensive, so we use a distributed computing framework of ~400 processors. The global IMS network consists of approximately 150 primary and auxiliary stations, forming a pre-defined, sparse network with which to locate seismic events. We compare the travel-time prediction and location capabilities of SALSA3D to standard 1D and 2/2.5D models via location tests on a global event set with GT of 5 km or better. Using Pn and P picks from IMS stations only, we generate different realizations of station distributions, yielding a range of azimuthal coverage and ratios of teleseismic to regional arrivals, with which we test the accuracy and precision of relocation. We test using the full 3D covariance matrix of the current model to calculate path-dependent travel time uncertainty, rather than applying standard, 1D, distance-dependent uncertainty. SALSA3D reduces mislocation over the standard 1D ak135 model regardless of Pn to P ratio, with the most pronounced improvement at higher azimuthal gaps. SALSA3D also reduces mislocation compared to the combined RSTT/ak135 model (2.5D - RSTT for regional phases), with minimal improvement over RSTT when only regional Pn phases are used to compute locations.