This is an animation of a migrated P to S converted wave image volume for the STEEP area.  The right panel is a view from above and the left panel is a view from the west.  The volume shown is the top 200 km of the Earth in true spherical geometry.  The coastline is plotted as blue lines, seismic stations are plotted as orange spheres, and a subset of earthquakes inside this volume are shown is yellow dots.  The vertical red bars are positioned at the location of volcanoes in the Wrangells and run from the surface to a depth of 100 km (the approximate depth of calc-alkaline volcanics).  The image volume is not in cardinal directions.  The x-axis points E 20 S (azimuth 110 degrees) and the center of the volume is at 60.5 N and 142.8 W.  The animation is a slice through this volume that moves from the western to eastern side of this volume.  Two features are noteworthy:  (1) the northernly dip of structures is consistent with subduction of lower crust of the Yakatat Block; and (2) the fine structure of the conversions show a discontinuity across the northern edge of the Yakatat Block, which is approximately in the center of this volume. Our current working hypothesis is that much of this fine structure is sedimentary multiples that are projected into artificial structure at depth.  Evidence in support of this hypothesis is seen below. 

This is a complementary impage to the P to S conversion volume displayed above. Here we see the S to P conversion image, which is somewhat the mirror image of the P to S conversion result.  That is, here we image precursors to teleseismic S that have the odd property that they arrive before S with the lead time scaling to increasing depth. This volume was also produced by a process called CCP stacking in which the data are binned and stacked at piecing points computed at a depth of 90 km.  The result is then converted to true geometry by using a vertical ray path integration.  The P result above, in contrast, was produced by a full 3D migration procedure.   All reference features (coastline, earthquakes, stations, and the volcanoes) are as described above. There are two features of these data compared to the P data that are important to realize.   First, these data naturally have a lower frequency content which maps to lower resolution.  This is an inevitable consequence of higher attenuation of S relative to P in the mantle. Second, because S to P conversions arrive BEFORE the teleseismic S phase they are nearly immune to problems of crustal multiples that strongly contaminate the P result above.  This image shows the subduction process unambiguously without the complications of interfering multiples seen in the P data.

Here we compare the P and S wavefield image results side by side to show what is and is not reliable. This is a useful example of cross validation of imaging results from completely different data.  The left shows the S to P conversion volume and the right shows the P to S conversion volume.  The animations are exactly the same as the left frames of the two movies above but we do not show the map views to make the resolution of the display sufficient to be useful.  This movie shows clearly that the P to S data are strongly impacted by multiples, but the higher resolution of the P data helps provide a higher resolution estimate of crustal thickness and depth to the top of the subducting slab.  Together these results yield an important scientific result.  There has been significant debate in the past about the subduction geometry of the lithosphere linked to the Yakatat Block.  These results have shown clearly that the subducting slab is continuous with Pacific lithosphere in central Alaska. Low angle subduction in central Alaska curls downward to a somewhat steeper dip to produce the Wrangell volcanics.  These images show the dip of the structure is consistent with the expected depth of generation of calc-alkaline volcanis.