Structure of the subduction zone beneath the Wellington region, New Zealand , from passive seismic recordings

The Seismic Array on the HiKurangi Experiment I consisted of 50 2-Hz seismometers deployed in a two-dimensional array and ten broadband seismometers deployed in a line above the Hikuangi Subduction Zone throughout the Wellington/Wairarapa region of New Zealand. Wellington is the capital and second largest city in New Zealand. Continuous signals were recorded between November 2009 and March 2010 on the short period sensors and up to 18 months on the broadband sensor. These stations densified the GeoNet network of two broadband and 11 1-Hz seismometers. Airgun shots and earthquakes were extracted for analysis. The E-W line was also occupied at several times with a high-density array of geophones deployed to record airgun shots and explosives. Here we summarize the results of preliminary analysis of earthquakes and seismic noise. Receiver function images of the plate boundary reveal similar structures to the results of active source analysis, suggesting that at long wavelength the S velocity and P velocity change at the same boundaries. A low velocity layer at the top of the plate and the within-slab Moho is well imaged. Deeper features are less clearly imaged but, like the controlled source reflectors, suggest that some converters are dipping in the same direction as the slab and some in the opposite direction. We use SKS phases recorded on the broadband array and permanent stations in the eastern part of our study area to investigate the deep anisotropic structure. Preliminary SKS splitting measurements display NE/SW fast polarization azimuths sub-parallel to Hikurangi trench and the predominant upper plate fault strike. Delay times of these splitting measurements range from 1.30 - 4.9 s (+/- 0.46 s) and SKS phases with large periods ( > 12 s) tend to show higher delay times ( > 2 s). Shear wave splitting on local earthquakes with magnitude greater than 4 yield mostly NE-SW polarization azimuths, consistent with previously determined local and SKS anisotropy at nearby stations. Smaller earthquakes yield more scattered fast directions. Delay times average 0.18 s and range from 0.02 to 0.7 seconds from earthquakes that extend from 5 to 180 km depth. However, delay times do not increase with depth, suggesting that most waveforms are re-split in the upper crust. Stacked seismic noise cross correlation functions obtained from broadband and short period stations alike exhibit coherent signals to distances of 80 km. Packets of coherent energy travel with speeds of between 5.2 km/s and 1.3 km/s, with the fastest group coherent to 60 km on the 2-stations. This analysis suggests that such high-frequency seismometers can usefully record surface waves with periods much longer than the natural period of the seismometers. We therefore expect to be able to use these data to determine surface wave velocity models of the uppermost 5 km of the crust, which will complement the controlled source profiles in the region.