Details of seismic refraction modeling. The seismic refraction data was modeled interactively (Song and ten Brink, 2004) using forward and inverse travel-time ray tracing routines (Zelt and Smith, 1992). The preferred model is that which minimizes the travel time difference between observed and calculated arrivals. Figure S1 shows the observed (in black) and calculated (in red) travel times for each shot. The lack of calculated arrivals propagating to the right from shot (b), is due to the fact that two-point ray tracing algorithms are only valid for smooth velocity structures (Cerveny, 1987). The velocity structure near this shotpoint, being located <1 km west of the Dead Sea Transform Fault, has a sharp lateral velocity discontinuity. Model runs, which produce good fit to these arrivals east of the shotpoints, required modifications to smooth the near-surface velocity structure (layers 1 and 2) in the vicinity of the shotpoint. Figure S2 shows the ray coverage of the model. Note that the sediments and upper crust are well covered by both diving waves and wide-angle reflections, but the lower crust is covered mostly by wide-angle reflections. This implies good lateral resolution of lower crust velocity, but poor vertical resolution in that layer. Therefore, data coverage is satisfactory to resolve the presence or absence of lateral variations in P wave velocity in the lower crust under the Dead Sea basin, but is not satisfactory to resolve the overall velocity gradient (6.8-7.0 km/s) of this layer. References: Cerveny, V., Ray-tracing algorithms in three-dimensional laterally-varying layered structures, D. Reidel, Norwell, MA, 1987. Song, J., and ten Brink, U.S., RayGUI2.0 - A graphical user interface for interactive forward and inversion ray tracing, USGS Open-file report 2004-1426, 2005. Zelt, C.A. and Smith, R.B., Seismic traveltime inversion for 2-D crustal velocity structure, Geophys. J. Int., 108, 16-34, 1992 Acknowledgements: We thank D. Lizarraldi for providing software to calculate to the bottoming points. Table S1. Goodness of fit for various arrivals in the best-fit model. Phase RMS(s) Chi Squared Number of Points 11 0.155 2.426 107 21 0.072 0.520 238 22 0.074 0.550 181 31 0.088 0.779 529 32 0.126 1.589 613 41 0.197 3.891 594 42 0.162 2.621 1068 51 0.154 2.390 262 52 0.142 2.012 1150 53 0.133 1.779 499 Table S2. Goodness of fit for various arrivals in a model with a constant Moho slope between the two edges of the model (Run11). Phase RMS(s) Chi Squared Number of Points 51 0.121 1.477 276 52 0.344 11.838 1152 53 0.355 12.621 490 Table S3. Goodness of fit for various arrivals in a model with a constant slope of upper-lower crust interface between model km 40 and 175 (Run09a). Phase RMS(s) Chi Squared Number of Points 41 0.205 4.199 497 42 0.187 3.484 1034 51 0.211 4.487 309 Table S4. Goodness of fit for various arrivals in a model with a Moho step 1.4 km high and 5.6 km wide (dashed yellow line in Figure 2) (Run06e). Phase RMS(s) Chi Squared Number of Points 51 0.154 2.390 262 52 0.142 2.012 1150 53 0.133 1.779 499 Table S5. Goodness of fit for various arrivals in models with a Mantle wedge rising to different levels into the lower crust. Run06d - 5.7 km high and 2.3 km wide at the top: Phase RMS(s) Chi Squared Number of Points 51 0.155 2.414 260 52 0.203 4.127 945 53 0.367 13.488 505 Run06c - 4.3 km high and 7.75 km wide at the top. Phase RMS(s) Chi Squared Number of Points 51 0.150 2.262 280 52 0.247 6.109 1028 53 0.182 3.333 409 Run06cB - 2 km high and 7.75 km wide at the top (dashed pink line in Figure 2). Phase RMS(s) Chi Squared Number of Points 51 0.150 2.262 280 52 0.209 4.354 1082 53 0.153 2.347 521 Run06cA - 1 km high and 7.75 km wide at the top. Phase RMS(s) Chi Squared Number of Points 51 0.153 2.348 257 52 0.166 2.752 1092 53 0.142 2.032 520 Table S6. Goodness of fit for various arrivals in models with flexural uplift of the entire crust east of the Dead Sea Transform. Run06b - 1.3 km maximum upward deflection. Phase RMS(s) Chi Squared Number of Points 51 0.153 2.340 269 52 0.149 2.218 1180 53 0.157 2.467 508 Run06g - 2 km maximum upward deflection Phase RMS(s) Chi Squared Number of Points 51 0.153 2.347 268 52 0.154 2.363 1089 53 0.160 2.560 510 Run06h - 2.5 km maximum upward deflection (dashed white line in Figure 2). Phase RMS(s) Chi Squared Number of Points 51 0.153 2.349 268 52 0.161 2.586 1072 53 0.165 2.713 518 Run06i - 3 km maximum upward deflection. Phase RMS(s) Chi Squared Number of Points 51 0.153 2.357 267 52 0.207 4.269 1095 53 0.191 3.673 523 Run06j - 3.5 km maximum upward deflection. Phase RMS(s) Chi Squared Number of Points 51 0.154 2.365 266 52 0.214 4.571 1076 53 0.202 4.097 522 Table S7. Goodness of fit for various arrivals in models to verify the existence of a lower velocity structure extending to a depth of 18 km beneath the Dead Sea Basin. Run06l - Make layer 4 velocity under the basin similar to that in the surrounding areas. Phase RMS(s) Chi Squared Number of Points 41 0.252 6.383 562 42 0.190 3.621 1086 Run06k - Same as Run06l and make layer 3-4 interface flat under the basin. Phase RMS(s) Chi Squared Number of Points 41 0.276 7.634 598 42 0.198 3.940 1088 Table S8. Goodness of fit for various arrivals in velocity anomaly extends into the lower crust. Run08A 1.9 km deep and 12 km wide depression of the 4-5 layer interface beneath the DSB area. Phase RMS(s) Chi Squared Number of Points 41 0.203 4.138 524 42 0.157 2.469 1015 51 0.218 4.767 231 Run08B 3 km deep and 40 km wide depression of the 4-5 layer interface beneath the DSB area. Phase RMS(s) Chi Squared Number of Points 41 0.202 4.104 522 42 0.147 2.154 994 51 0.178 3.179 309