This project is motivated by one simple, but inconvenient, reality: Almost all volcanic eruptions are preceded by earthquake swarms, but most earthquake swarms at volcanoes are not followed by eruptions. Though seismology remains one of the primary tools for tracking and understanding volcanic behavior, this fact is responsible for more false eruption forecasts than any other. Non-eruptive earthquake swarms at volcanoes have caused countless false alarms, economic impacts, and community responses. By addressing eruptive and non-eruptive earthquake swarm characteristics, this project attacks one of the fundamental challenges in volcano science. We will improve capabilities for interpreting the numerous sources of volcanic earthquake swarms by characterizing them with simple, but quantitatively repeatable, parameters. Funds will be used almost exclusively to support a post doctoral researcher.
Objective 1: Quantify the defining characteristics of volcanic earthquake swarm mechanisms. Comparisons between volcanic swarms have historically been challenging because each swarm has been studied with a custom technique. We plan to reanalyze swarms from a wide suite of volcanoes using a uniform methodology. Beginning with the original continuous seismic records, automated techniques will be used to create new earthquake catalogs that span the swarms of interest. Using the new catalogs, a suite of quantitative parameters will be derived based on rates, magnitudes, and waveform characteristics. Principal component analysis of the parameter space, paired with traditional and Bayesian hypothesis testing, will be used to identify the defining characteristics of each swarm mechanism. As a byproduct of this research, earthquake swarm catalogs will be made openly available on the web as they are produced. For most of the swarms proposed here, the new catalogs will contain vastly more earthquakes than anything currently available.
Objective 2: Assess the eruption forecasting potential of multiplet, or repeating, earthquakes. Because multiplets imply a long-lived non-destructive source, they are particularly promising for forecasting eruptive behavior. Building on swarm characteristics derived in objective 1, swarms with significant multiplet sequences will be identified. From these multiplets many of the same parameters derived for the swarms will be calculated. Additional measures are available for multiplets that exploit their waveform similarity for source migration inferred from high-precision travel times, coda wave interferometry, and precise relative magnitudes. By distilling these phenomena to simple parameters the same parametric techniques will be applied as in objective 1. By examining sequences from numerous volcanic settings, the defining characteristics of each source type will be isolated. While this analysis will span many types of multiplets, the motivation is to assess the features that most frequently portend eruption.
