Anisotropy, repeating earthquakes, and seismicity associated with the 2008 eruption of Okmok volcano, Alaska

Abstract

[1] We use shear wave splitting (SWS) analysis and double-difference relocation to examine temporal variations in seismic properties prior to and accompanying magmatic activity associated with the 2008 eruption of Okmok volcano, Alaska. Using bispectrum cross-correlation, a multiplet of 25 earthquakes is identified spanning five years leading up to the eruption, each event having first motions compatible with a normal fault striking NE-SW. Cross-correlation differential times are used to relocate earthquakes occurring between January 2003 and February 2009. The bulk of the seismicity prior to the onset of the eruption on 12 July 2008 occurred southwest of the caldera beneath a geothermal field. Earthquakes associated with the onset of the eruption occurred beneath the northern portion of the caldera and started as deep as 13 km. Subsequent earthquakes occurred predominantly at 3 km depth, coinciding with the depth at which the magma body has been modeled using geodetic data. Automated SWS analysis of the Okmok catalog reveals radial polarization outside the caldera and a northwest-southeast polarization within. We interpret these polarizations in terms of a magma reservoir near the center of the caldera, which we model with a Mogi point source. SWS analysis using the same input processing parameters for each event in the multiplet reveals no temporal changes in anisotropy over the duration of the multiplet, suggesting either a short-term or small increase in stress just before the eruption that was not detected by GPS, or eruption triggering by a mechanism other than a change of stress in the system.

Automatic measurement of shear wave splitting and applications to time varying anisotropy at Mount Ruapehu volcano, New Zealand

“…We also developed an automatic grading technique that evaluates the measurement on the basis of the clustering characteristics instead of simply choosing the best cluster. We have previously applied the method on earthquakes recorded at volcanoes in Japan , Alaska [Johnson et al, 2010] and Montserrat (D. C. Roman et al, Forward modeling of seismic anisotropy images volcanic conduit inflation months prior to eruption, submitted to Journal of Geophysical Research, 2010), and we found time variations that we interpret to be caused by deformation caused by magmatic movement. Here we present a detailed analysis of the method and we apply it to data analyzed previously around Mount Ruapehu volcano, from 1994Ruapehu volcano, from , 1998Ruapehu volcano, from and 2002, in which temporal changes in anisotropy have been reported in association with volcanic activity [Gerst and Savage, 2004;Miller and Savage, 2001].…”

“…So far it has been tested on other data sets on Mount Asama volcano and three other volcanoes in Japan , Okmok, Alaska [Johnson et al, 2010] and at Soufriere Hills volcano in Montserrat (Roman et al, submitted manuscript, 2010). But we have also made some preliminary studies in nonvolcanic areas of New Zealand and have found it to work equally well.…”

“…The "all" or "unculled" group (no quality criteria applied beyond the original Teanby et al methodology) has 212 measurements, reducing to 160 when the null criteria and the D quality measurements from clustering are left out (noNnoD). The AB group, which are the quality that were used in previous publications [Johnson et al, 2010;Savage et al, 2010], have 145 measurements and when only the top "A" quality measurements and the "Eng8" criteria are applied it reduces to 75, leaving out almost 2/3 of the measurements. Interpreting this must be tempered by the fact that we chose the "Eng8" criterion in order to get roughly the same number of measurements as were in the A and AB manual measurements in an earlier version of the codes.…”

“…Here we search the parameter space in units of 1°a nd a time unit depending on the scale of the problem for all fast polarizations and for delay times from 0.0 to a maximum value (tlagmax), 1.0 s in this study. For some studies in which there were many local earthquakes close to a volcano, we used a maximum of 0.4 s for the local events [Johnson et al, 2010;Savage et al, 2008Savage et al, , 2010. For SKS measurements, 4 to 6 s may be used [Greve et al, 2008;Savage et al, 2007].…”

“…This allows us to examine the frequency dependence of the results, and in averaging the parameters, it ensures that the measurements that are most stable with frequency contribute most to the final measurement. Earlier versions of the code used only the SNR itself rather than the SNR-bandwidth product, and were more prone to cycle skipping, relying more heavily on later grading to weed out poor measurements [Johnson et al, 2010;Savage et al, 2010].…”