HR: 1340h
AN: V33A-0675    [Abstracts]
TI: KrakMon: Seismic signals recorded at Krakatau Volcano, Indonesia
AU: * Ibs-von Seht, M
EM: m.ibs@bgr.de
AF: Federal Institute for Geosciences and Natural Resources, Stilleweg 2, Hannover, 30655 Germany
AU: Hoffmann-Rothe, A
EM: a.hoffmann-rothe@bgr.de
AF: Federal Institute for Geosciences and Natural Resources, Stilleweg 2, Hannover, 30655 Germany
AU: Kniess, R
EM: r.kniess@bgr.de
AF: Federal Institute for Geosciences and Natural Resources, Stilleweg 2, Hannover, 30655 Germany
AB: A recently installed multi-parameter monitoring system on the Krakatau volcanic island complex located in the Sunda Strait (Indonesia) provides continuous broadband records of seismic data. We present here an overview of the different signal types identified and first results of an automated detection and classification procedure for volcano-seismic events recorded at Krakatau volcano. In comparison to seismic signals known from other volcanoes, an unusually high portion of high-frequency content is observed in the spectra of the Krakatau signals. This observation applies to short-term volcano-tectonic (VT) events as well as to continuous tremor signals: most VT events show significant energy at frequencies above 30Hz, harmonic signals last almost continuously for days and show spectral peaks at distinct frequencies well above15Hz. The automated detection and classification procedure bases a spectrogram analysis of volcano-seismic signals using a straight-forward pattern recognition approach: a suitable threshold operator generates a binary representation of the spectrogram which is processed by a contour finding algorithm. The resulting contour-polygons define regions in the spectrogram containing significant spectral energy and their shapes reveal information about the respective volcano-seismic signals. By the extraction of stable shape-describing properties from the polygons and their statistical analysis it is attempted to identify different classes of signal types. A comparison of the resulting signal types with those determined visually by the operator can improve classification schemes for volcano-seismic signals and contribute to defining the activity status of Krakatau and other volcanoes.
DE: 7280 Volcano seismology (8419)
DE: 8419 Volcano monitoring (7280)
DE: 8488 Volcanic hazards and risks
DE: 9340 Indian Ocean
SC: Volcanology, Geochemistry, Petrology [V]
MN: Fall Meeting 2005