Seismic evidence for fluids in fault zones on top of the subducting Cocos Plate beneath Costa Rica

In the 2005 TICOCAVA explosion seismology study in Costa Rica, we observed crustal turning waves with a dominant frequency of ∼10 Hz on a linear array of short-period seismometers from the Pacific Ocean to the Caribbean Sea. On one of the shot records, from Shot 21 in the backarc of the Cordillera Central, we also observed two seismic phases with an unusually high dominant frequency (∼20 Hz). These two phases were recorded in the forearc region of central Costa Rica and arrived ∼7 s apart and 30–40 s after the detonation of Shot 21. We considered the possibility that these secondary arrivals were produced by a local earthquake that may have happened during the active-source seismic experiment. Such high-frequency phases following Shot 21 were not recorded after Shots 22, 23 and 24, all in the backarc of Costa Rica, which might suggest that they were produced by some other source. However, earthquake dislocation models cannot produce seismic waves of such high frequency with significant amplitude. In addition, we would have expected to see more arrivals from such an earthquake on other seismic stations in central Costa Rica. We therefore investigate whether the high-frequency arrivals may be the result of a deep seismic reflection from the subducting Cocos Plate. The timing of these phases is consistent with a shear wave from Shot 21 that was reflected as a compressional (S×P) and a shear (S×S) wave at the top of the subducting Cocos slab between 35 and 55 km depth. The shift in dominant frequency from ∼10 Hz in the downgoing seismic wave to ∼20 Hz in the reflected waves requires a particular seismic structure at the interface between the subducting slab and the forearc mantle to produce a substantial increase in reflection coefficients with frequency. The spectral amplitude characteristics of the S×P and S×S phases from Shot 21 are consistent with a very high V
p
/V
s
ratio of 6 in ∼5 m thick, slab-parallel layers. This result suggests that a system of thin shear zones near the plate interface beneath the forearc is occupied by hydrous fluids under near-lithostatic conditions. The overpressured shear zone probably takes up fluids from the downgoing slab, and it may control the lower limit of the seismogenic zone.