HR: 18:05h
AN: V44A-08    [Abstracts]
TI: Automated Detection of Volcanic Thermal Anomalies: Detailed Analysis of the 2004 - 2005 Mt. Etna, Italy Eruption
AU: * Steffke, A M
EM: steffke@higp.hawaii.edu
AF: HIGP/SOEST, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822, United States
AU: Harris, A
EM: harris@higp.hawaii.edu
AF: HIGP/SOEST, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822, United States
AU: Garbeil, H
EM: harold@higp.hawaii.edu
AF: HIGP/SOEST, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822, United States
AU: Wright, R
EM: wright@higp.hawaii.edu
AF: HIGP/SOEST, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822, United States
AU: Dehn, J
EM: jdehn@gi.alaska.edu
AF: Alaska Volcano Observatory, Geophysical Institute, University of Alaska, Fairbanks, 903 Koyukuk Drive, Fairbanks, AK 99775, United States
AB: Use of thermal infrared satellite data to detect, characterize and track volcanic thermal emissions is an appealing method for monitoring volcanoes for a number of reasons. It provides a synoptic perspective, with satellites sensors such as AVHRR and MODIS allowing global coverage at-least 4 times/day. At the same time, direct reception of calibrated digital data in a standard and stable format allows automation, enabling near-real time analysis of many volcanoes over large regions, including volcanoes where other geophysical instruments are not deployed. In addition, extracted thermal data can be use to convert to heat and volume flux estimates/time series. The development of an automated algorithm to detect volcanic thermal anomalies using thermal satellite data was first attempted over a decade ago (VAST). Subsequently several attempts have been made to create an effective way to automatically detect thermal anomalies at volcanoes using such high-temporal resolution satellite data (e.g. Okmok, MODVOLC and RAT). The underlying motivation has been to allow automated, routine and timely hot spot detection for volcanic monitoring purposes. In this study we review four algorithms that have been implemented to date, specifically: VAST, Okmok, MODVOLC and RAT. To test how VAST and MODVOLC performed we tested them on the 2004 - 2005 effusive eruption of Mount Etna (Sicily, Italy). These results were then compared with manually detected and picked thermal anomalies. Each algorithm is designed for different purposes, thus they perform differently. MODVOLC, for example, must run efficiently, up to 4 times a day, on a full global data set. Thus the number of algorithm steps are minimal and the detection threshold is high, meaning that the incidence of false positives are low, but so too is its sensitivity. In contrast, VAST is designed to run on a single volcano and has the added advantage of some user input. Thus, a greater incidence of false positives occurs, but more subtle anomalies are detected.
DE: 8419 Volcano monitoring (7280)
DE: 8485 Remote sensing of volcanoes
SC: Volcanology, Geochemistry, and Petrology [V]
MN: 2007 Joint Assembly