1-D Velocity Model for Use by the SANSN in Earthquake Location

Abstract

Reliable local earthquake locations depend on many factors, a major one of which is the velocity model. Locating earthquakes using unreliable models contributes in part to the uncertainties of active fault mapping and unexplained scatter of s eismic locations. Given that we strive to continually improve our locating abilities, it is necessary to always improve on the model, when possible. The South African National Seismograph Network (SANSN) has been using a model (Saunders et al. 2006) from work published by Wright et al. (2002) and refined by Brandt (2005). Even though there have been many crust and mantle structure studies in South Africa ( e.g. , Nguuri et al. 2001; Simon et al. 2002; Wright et al. 2002; James et al. 2004), the models obtained are too complicated for use in seismic event locations, especially when using the earthquake analysis software SEISAN (Havskov and Ottemoller 2008). The HYPOCENTER program (Lienert 1994), which is the location program in SEISAN, cannot use models with low-velocity layers, because these usually introduce instabilities. Modification/refinement of the results of one of these studies (Wright et al. 2002) by Brandt (2005) produced a model that the SANSN is currently using in earthquake locations. This model can therefore be used as a starting point in conducting local earthquake tomography to obtain a more realiable model. Our study aims to compute a 1-D velocity model for South Africa by inverting P -wave travel times recorded by SANSN. Ideally, SANSN should use different models for the different major tectonic units in South Africa, i.e. , Kaapvaal craton, Namaqualand, and Cape Fold belt. However, available data dictates what can be done, and for now we can only produce a 1-D model whose velocities are approximately equal to the average velocity of the 3-D structure within the same …