← Back to Gallery

How Is Differential Shannon Entropy Related to Volcanic Processes?

Previous studies show that Shannon Entropy (SE) calculated on continuous seismic records can be used to track unrest and eruptive activity at active volcanoes (Rey-Devesa et al., 2023 a, b). SE usually remains constant, and decays for a short-term period before an eruption occurs, reaching minimum values just in the run-up to a paroxysm. This has been observed in a large set of different volcanoes around the globe, although it remains unclear how SE relates to physical processes and how it can be used for improving routine forecasting strategies.

SE, or Differential Shannon Entropy (DSE) when the source is continuous, is used in information theory to understand the uncertainty of a source of information (Shannon, 1948). The uncertainty is related to the amount of information required for describing the next element in a sequence, or how predictable a sequence is. Thus, a very probable event is translated as a low value of uncertainty and DSE. In this context, we can interpret the decay in the DSE of the seismic signal as an indicator of a reorganization happening in the volcanic system, approaching an eruptive outcome. However, in the real-time monitoring scenario, the performance of the feature needs some fine-tunning, since signals usually contain different types of noise (wind, tides, and storms) that can also generate a decay in the DSE.

In this work, we propose a synthetic analysis of the DSE, studying how it varies with the type and characteristics of noise. The goal is to identify if DSE can detect changes in the systematic noise. In addition, we apply the conclusions to a real-life scenario, the eruption of Shishaldin in 2023, associating every decay of the DSE in one year record to different sources or external processes.

Preliminary results show that more predictable noises imply lower values of DSE. We also see that the amplitude of the decay and the frequency band containing the energy of the signal are relevant features to associate the decay with their sources.

A better understanding of DSE and its evolution is crucial for improving the quality of the DSE as a volcanic eruption monitoring tool.

Session: Multidisciplinary Approaches for Volcanic Eruption Forecasting [Poster Session]

Type: Poster

Room: Exhibit Hall

Date: 5/2/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Janire Prudencio

Student Presenter: No

Additional Authors