Crustal seismic velocity models of Texas

Our investigation distinguishes six distinct geologic regions within Texas and determines a preferred one-dimensional (1D) crustal structure for each. These models, which consist of flat layers of varying thicknesses and constant P and S-wave velocities in each layer, represent the best average crustal velocity structure. Our investigation is motivated by TexNet, a new statewide seismograph network, which will need more accurate regional crustal models to better locate earthquakes throughout the state. We test previously published models as well as newly generated models. The data used to develop the new models include previously used velocity models, geologic cross sections, refraction and reflection studies, sonic logs, receiver function results and any other geophysical survey information available for the specific regions. We test the accuracy of the various regional models by relocating earthquakes with Hypoinverse1.40 (HYPO1.40). The earthquake catalogs vary by region but meet standard criteria for quantity and quality of phases recorded. We relocate each set of regional earthquakes with all previously published and newly generated models and determine the preferred model by lowest RMS (root mean square) residuals, i.e., the differences in recorded and modeled travel times. To understand which layers most significantly affect observed travel times, we perform source-to-station ray tracing for available regional earthquakes with magnitudes larger than M2.6. We also use the arrival data to plot Wadati diagrams and find the regional Vp/Vs ratio, which is applied to the preferred P-velocity model to determine a preferred S-velocity model. Our data allow us to determine new preferred velocity models for four of the six regions (East Texas, Fort Worth Basin, Panhandle, and West Texas) and confirm a previously published model for two regions (Central and Gulf Coastal Plain). Central Texas does not have enough earthquake data or geophysical studies to determine a new model, so we suggest the continued use of the Mitchell and Landisman (1971) velocity model until new seismic data is available. The velocity model published by Cram, Jr. (1961) is the preferred model for regional earthquake location for the Gulf Coastal Plain.