Crust and Upper Mantle Velocity Structure of the New Madrid Seismic Zone

Wednesday, 17 December 2014
Cecilia Anyango Nyamwandha, Christine Ann Powell and Charles A Langston, Center for Earthquake Research and Information, Memphis, TN, United States
Detailed P wave velocity (Vp) and S wave velocity models (Vs) and Vp/Vs ratios for the crust and upper mantle associated with the New Madrid Seismic Zone (NMSZ) are presented. The specific study region spans latitude 34 to 39.5 degrees north and longitude 87 to 93 degrees west and extends to a depth of at least 500 km. The density of data from three networks – The Cooperative New Madrid Seismic Network (CNMSN) operated by CERI, the Earthscope transportable array (TA), and the FlexArray (FA) Northern Embayment Lithospheric Embayment (NELE) project stations – provides us with the opportunity to derive detailed velocity models for this region. We use arrival times from local and regional earthquakes and travel time residuals from teleseismic earthquakes recorded by the three networks from September 2011 to date. The teleseismic body wave arrival times are measured using an Automated and Interactive Measurement of Body Wave Arrival Times (AIMBAT) package (Lou et al., 2012). We perform a joint local and teleseismic inversion (Zhao et al.,1994) to determine the velocity structure. For the local events, the hypocenters are relocated iteratively in the inversion process using an efficient 3-D ray tracing technique. We image a significant low velocity anomaly in the upper mantle with a concentration at about 200 - 300 km depth and it is a consistent feature in both the Vp and Vs tomography results. Checkerboard tests show that the spatial resolution is high in the upper mantle especially for the Vp model. The spatial resolution in the crust is fairly high for most of the study area except at the edges and the southeastern part, which can be attributed to diminished local earthquake activity. We perform synthetic tests to isolate smearing effects and further confirm the features in the tomographic images. Vp/Vs ratios are determined for the portions of the model with highest resolution. Preliminary results indicate that significant Vp/Vs ratio variations are present only at depths shallower than 200 km.