Item

Abstract

Changes in seismic complexity could reveal the dynamic changes inside a volcanic system prior to an eruption, which can be utilized to improve eruption forecasting. Permutation Entropy (PE) is a robust method to quantify the complexity of a time series, by calculating it directly using the amplitude. Examples of successful PE applications are eruptions forecasting at Strokkur Geyser, Iceland, the 1996 eruption of Gjálp, Iceland, and the eruptions of Shinmoedake volcano, Japan, in 2011, 2017, and 2018. While PE could show temporal changes prior to an eruption, these features are not always prominent. To improve this method, we calculated not only PE by using seismic amplitudes, but also Phase Permutation Entropy (PPE) by using seismic instantaneous phases. We calculated both PE and PPE for seismic station FLUR from January 2014 to December 2015, which covers the eruption period of Holuhraun in Iceland. During the unrest period, both PE and PPE show changes in entropy, but their patterns are not always the same. While PE shows a strong increase which marks the onset time of dyke propagation two weeks before the main eruption, it does not show a clear transition toward the main eruption. In contrast, PPE only shows a gradual change when the dyke started migrating. Jowever, it displays strong changes at the possible subglacial eruptions on 23 August, the small eruption on 29 August, and the main eruption on 31 August 2014. This proves the capability of PE and PPE to contribute to the framework of eruption forecasting.