VLP Source Inversion and Evaluation of Error Analysis Techniques at Fuego Volcano, Guatemala

Abstract:

In January of 2012, our team occupied 10 sites around Fuego volcano with broadband seismometers, two of which were collocated with infrasound microphone arrays and tilt-meters (see Figure 1 for full deployment details). Our radial coverage around Fuego during the 2012 campaign satisfies conditions outlined by Dawson et al. [2011] for good network coverage. Very-long-period (VLP) events that accompany small-scale explosions were classified by waveform and eruption style. We located these VLP event families which have been persistent at Fuego since at least 2008 through inversion in the same manner employed by Lyons and Waite [2011] with improved radial coverage in our network. We compare results for source inversions performed with independent tilt data against inversions incorporating tilt data extracted from the broadband.

The current best-practice method for choosing an optimum solution for inversion results is based on each solution’s residual error, the relevance of free parameters used in the model, and the physical significance of the source mechanism. Error analysis was performed through a boot strapping in order to explore the source location uncertainty and significance of components of the moment tensor. The significance of the number of free parameters has mostly been evaluated by calculating Akaike’s Information Criterion (AIC), but little has been done to evaluate the sensitivity of AIC or other criteria (i.e. Bayesian Information Criterion) to the number of model parameters. We compare solutions as chosen by these alternate methods with more standard techniques for our real data set as well through the use of synthetic data and make recommendations as to best practices.

Figure 1: a) Map of 2012 station network: stations highlighted in red were collocated with infrasound arrays. b) Location of Fuego within Guatemala and view of the complex from the west with different eruptive centers labeled. c) Operational times for each of the stations and cameras.