jump to: DNA13 | PNW10 | DNA10 | DNA09 | DNA07 | DNA home

S-velocity model of the Cascadia Subduction Zone using Ambient Seismic Noise Tomography


Model highlights: PNW10-S combines the state of the art ambient noise tomography method with time tested analyst selection to ensure the highest quality data is used as input to the model inversion. Incorporating spatially and temporally long paths with this manual selection step allows recoverable structure from the surface to ~120km depth. This model focuses on the US Pacific Northwest to address a series of questions relating to variations in arc volcanism, seismicity, tremor activity, and the relation to subduction complex structure. In this model we image lateral discontinuities of the subducting slabs in line with offshore transform faults in the Juan de Fuca plate. This suggests zones of weakness in the subducting plate creating pathways for upwelling material through the plate. This observation along the continuation of the Blanco transform fault explains the high heat production observed in the southern Cascades as arising from an upper mantle source. Variations within the crustal structure align with long term tremor segmentation boundaries suggesting the structure of the overriding plate has an effect on tremor frequency.

N-S cross-section throught the Cascadia forearc. See Fig 7 Porritt et al 2011

Model specifications:

  • Parameters: PNW10-S is a Vsv model derived from fundamental mode Rayleigh waves on ambient seismic noise cross correlations
  • Coverage: Cross correlations span the entire US, with a handful of stations in Canada and Mexico. The inversion solves for relative structure from 33N to 49.5N latitude and 126W to 110W longitude, but we extract the well resolved section from 38.2N to 49N and 126W to 119W due to its data density.
  • Data source: While our dataset focuses on two Flexible Array Experiment (FACES and Mendocino) more than 1200 stations where used from the USArray transportable array, regional seismic networks, and temporary seismic deployments.
  • Data type: Phase velocities measured from inter-station paths with a Frequency-Time Analysis procedure to prepare period dependent maps.
  • Inversion: Phase velocities are converted to perturbations to the 1D WUS model of Pollitz, 2009 with a crustal thickness determined by prior receiver function analysis (Audet et al., 2010, Levander et al., 2007). The relative phase velocities are then inverted with damped LSQR.
  • Resolution: Horizontal resolution varies with spatial coverage and can perhpas be best estimated around the minimum station spacing of ~50km. Vertical resolution varies with depth as the shallower regions are more densely sampled and can thus be considered better resolved, but we estimate an overall vertical resolution of ~10km. See the published manuscript for more detailed resolution analysis.

Porritt, R. W., Allen, R. M, Boyarko, D. C., and Brudzinski, M. R., Investigation of Cascadia Segmentation with Ambient Noise Tomography. Earth and Planet. Sci. Lett., 309, 67-76. doi: 10.1016/j.epsl.2011.06.026, 2011. download

Along strike variation in the characteristics of subduction zone processes has been observed throughout the Cascadia Subduction Zone through analysis of arc magmas and the distribution of seismicity. We investigate links between these observations and subduction zone structure by imaging three-dimensional lithospheric scale shear velocity with ambient noise tomography (ANT). The crustal portion of the model is well resolved through typical ANT processing techniques. We expand the methodology to use longer period phase velocities in order to recover structure to ~120km depth. The resulting model, PNW10-S, represents structural information in terms of relative shear velocity in the crust and uppermost mantle. Crustal structure mirrors surface geology to ~10 km depth and then transitions to a structure that is dominated by the subducting slab. The subducting slab and overriding crust appear segmented into three parts with boundaries near 43°N and 46°N. This three-way structural segmentation is aligned with the variation in recurrence of episodic tremor and slip along the subduction zone (Brudzinski and Allen, 2007). Upper to middle crustal boundaries between the Klamath Mountains and Siletzia Terrane (43N) and between the Crescent Formation and Olympic Peninsula (47N) are also coincident with locations of increased occurrence of tremors raising the question of whether there is a link between the intensity of tremor activity and shallow (less than 10km) crustal structure. The slab-segment boundary at 43N is a stronger feature than the northern segment boundary at 46N and appears to be the continuation of the Blanco Fracture Zone separating the Gorda segment of the plate from the rest of the Juan de Fuca plate. The southern half of the arc system, south of 45N, shows lower velocities from the surface to ~80 km depth relative to the northern portion of the arc. We propose this is due to clockwise plate rotation, which causes extension in the south, and results in increased melting. Along the arc, four broad low-velocity features are also imaged just below the Moho and centered at 42N, 44N, 47N, and 49N. We interpret these as ponding of melt just below the crust where differentiation can occur before further ascent through the crust.


The following have contributed to the development of the DNA models: Richard Allen (UC Berkeley) Shu-Huei Hung (National Taiwan University) Mathias Obrebski (UC Berkeley) Robert Porritt (UC Berkeley) Fred Pollitz (USGS) Mei Xue (formerly UC Berkeley, now Tongji University)

Support for this project is provided by the Earthscope program of the National Science Foundation, and the University of California, Berkeley. The 3D visualization has been facilitated by the Keck Caves of the University of California, Davis.

This page is maintained by Richard Allen