Shallow velocity imaging of an active volcano

Monday, 15 December 2014
Bill Fry1, Lauriane Chardot2 and Arthur D Jolly1, (1)GNS Science-Institute of Geological and Nuclear Sciences Ltd, Lower Hutt, New Zealand, (2)University of Canterbury/GNS Science, Taupo, New Zealand
We use a linear array of temporary seismometers to derive a shear-wave velocity model of the upper ~1000m of the crater area of White Island, an active volcano in New Zealand. We use noise interferometry to generate dispersion curves and invert these dispersion curves to obtain a layered 1D model. By exploiting the varying interstation distances along the array, we are able to define a strong shallow impedance contrast in the upper 10 meters as well as a depth to 'effective' bedrock at about 100m. We limit the bandwidth of the measured dispersion using a 2-wave cycle approximation and construct a composite dispersion curve. We then invert the dispersion curves with two separate inversion algorithms in an effort to test the validity of using this broadband approach for monitoring active volcanoes. The first method is a non-linear approach and is useful when an a-priori starting model is poorly known or if a velocity inversion is likely. Unfortunately, this type of non-linear inversion is more sensitive to small perturbations in the recovered Green's Functions, which may be due to non-equipartitioning of the wavefield as well as to velocity changes. The second is a linearized and damped LSQR approach which we envision will be more useful for routine monitoring in situations in which the starting model is well defined. In this case, selective regularization can be used to stablize moving time-window inversion. Lastly, our results will be used as input for hydrothermal fluid flow modelling conducted in a concurrent study.