Skip to content

Probabilistic forecast of Vulcanian explosions at Sakurajima volcano using statistical features of ground deformation

Kyoka Ishii 1, Masato Iguchi2

  • Affiliations: 1Aso Volcanological Laboratory, Institute for Geothermal Sciences, Graduate School of Science, Kyoto University, Japan; 2Crisis Management Bureau, Kagoshima City, Japan

  • Presentation type: Talk

  • Presentation time: Monday 10:45 - 11:00, Room S160

  • Programme No: 2.4.8

  • Theme 2 > Session 4

Abstract

High-resolution ground deformation monitoring is useful for forecasting volcanic explosions because pre-eruptive deformation due to magma intrusion has been observed at many volcanoes. However, it is still challenging to precisely predict the timing and size of the imminent explosions. Therefore, the probabilistic approach effectively forecasts explosions and quantifies the risk of volcanic hazards considering various uncertainties. We focused on the short-term forecast of Vulcanian explosions at Sakurajima volcano, Japan. The volcanic activity of the Sakurajima volcano is characterized by frequent Vulcanian explosions, accompanied by inflation-deflation patterns of extensometer records. We made a large database of the volume change and durations of precursory ground inflation related to explosions at Showa and Minami-dake craters in 2009--2020. By using this database, optimal stochastic models for their distributions were estimated. As a result, the log-logistic distribution is more appropriate than Weibull and exponential distributions. The model parameters of the log-logistic distribution temporally fluctuated, reflecting volcanic activity, especially in increasing the magma supply from a deep region. We also proposed a methodology to calculate the probabilities of the likely timing and size of the subsequent explosion using strain monitoring and the estimated stochastic model. Although validation with real cases and some refinements are still needed before a forecast system can be practically implemented, the basic concepts we propose would also pave the way to the short-term forecasts of other volcanoes.