Time-varying seismic velocity in New Zealand's volcanic regions: Comparisons between shear wave splitting and surface wave noise correlations

Abstract

Volcanic eruptions are difficult to predict and new methods are being pursued to try to add to the available techniques. Recently, changes in seismic wave properties before volcanic eruptions have been observed and proposed to be used to monitor and predict volcanic eruptions. These changing properties are interpreted as caused by changes in fluid-filled cracks, which respond to changes in stress conditions and to fluid movements. We use and compare two techniques for measuring seismic wave speeds and their variation with direction (anisotropy) in the Tongariro volcanic region. Using one technique, called “shear wave splitting” or “seismic birefringence”, measured on seismic waves from nearby earthquakes, we find that there was an increase in anisotropy during the 1995/1996 Mt. Ruapehu eruption sequence and at the Tongariro geothermal area after early 2001.

In contrast, anisotropy decreased during the 2006/2007 period around the times of the small 2006 and 2007 eruptions. Fast directions of anisotropy rotated after the 1995/1996 eruptions and during the time of decreasing anisotropy in 2006/2007. We attribute the changes to increasing cracks during the large eruptions of 1995/1996, to a regional movement of fluids associated with the 2006 and 2007 eruptions, and to a local disruption in the geothermal field after 2001. We also implement a new computer algorithm to extract seismic waves from background seismic noise collected around Mt. Ruapehu. We use these waves to compute the isotropic and azimuthally dependent seismic velocity of the volcano and its surroundings. We find time-variable results that can constrain models of the evolution of the 2006 eruption. We compare these results to the above discussed shear wave splitting measurements of anisotropy. These techniques provide substantial steps toward our ultimate goal of eruption early-warning based on near-real time seismic tomography.

---

Technical Abstract

We used data from GeoNet and past portable deployments of seismometers to determine seismic anisotropy from shear wave splitting in the Tongariro region and to compare it to isotropic and anisotropic analysis of surface waves generated by continually occurring seismic noise. Seismicity generated from the Erua earthquake cluster (a consistently active area of seismicity about 20 km to the west of Mount Ruapehu) over the last 12 years was used to study seismic anisotropy in the Ruapehu region. In particular, we searched for changes associated with two minor phreatic eruptions on the 4th of October 2006 and the 25th of September 2007. The seismicity rate, magnitude of completeness, focal mechanisms and b‐value of the cluster were also examined to investigate whether the characteristics of the seismicity changed over the duration of the study. The hypocenters were relocated, which revealed a westward dip in the shallow seismicity. Shear wave splitting revealed a decrease in delay time in the 2006–2007 period and a significant variation in the fast shear wave polarization in the same time period. The b‐value also increased significantly from 1.0 ± 0.2 in 2004 to a peak of 1.8 ± 0.2 in 2007, but no other parameters were found to vary significantly over this time period. We attribute these changes to an increase in pore‐fluid pressure in the Erua region due to fluid movement and suggest that this fluid movement may be associated with the eruptions in 2006 and 2007.

We applied a simplified two-dimensional tomographic inversion of recorded delay times of shear wave splitting and a spatial averaging of fast direction of anisotropy to data from temporary seismic deployments in the Tongariro Volcanic Centre in order to identify regions of changing seismic anisotropy. We observed a region of strong anisotropy (>0.025 s/km greater than the surrounding area) centered on Mt. Ruapehu in 1995, the time of a major magmatic eruption. This is interpreted to be due to an increase in fluid-filled fractures during the eruption. We also observed strong anisotropy (~0.018 s/km greater than the surrounding area) and a change in fast direction (~80°) at Mt. Tongariro in 2008 and examined the temporal evolution of this anomaly using clusters of earthquakes and permanent seismic stations in operation since 2004. This anomaly is attributed to a change in the geothermal system. A pronounced and unchanging feature was observed at Waiouru, even when the source and receiver locations differ. We therefore conclude that the transient features of strong anisotropy associated with volcanic and geothermal activity detected with this method are also robust.

Under funding from EQC grant 10/600, we have implemented an algorithm to invert ambient noise data for the isotropic velocity of Rayleigh waves. We also use this algorithm to solve for the azimuthal variance in wave speed. Since Rayleigh wave velocity depends on shear wave speeds, its analysis is complementary to the analysis of shear-wave splitting. An important side product of our method is the ability to model static Vs and Vs changes over very short time scales, making an advance toward the ultimate goal of near-real time monitoring based on seismic tomography. We applied this method to model changes in anisotropy generated during the 2006 eruption.