1-1-1-1-1-1-1-1-1-1-1-1-1-1 From: "Flavio Cannavò" <flavio.cannavo@xxxxxxx> Dear colleagues, Here you can find the first competition on Volcanic Eruption Prediction by machine learning, host by kaggle.com: https://www.kaggle.com/c/predict-volcanic-eruptions-ingv-oe The eruption prediction problem is crucial for volcanic risk mitigation in sensible areas (such as human settlements). Indeed, volcanic eruptions could lead to important loss of life, property, and disruption of human activities and could compromise flight safety. Just one unforeseen eruption can result in tens of thousands of lives lost. If scientists could reliably predict when a volcano will next erupt, safety actions could be more timely and the damage mitigated. The science of volcanic eruption prediction has significantly advanced over the past decades. Many studies indicate that signs of unrest preceding an eruption would be detectable in seismic signals for different time scales (from minutes to months). In solving the short-term prediction problem, some types of observed seismic activity (e.g. very long period events, long period events, volcanic tremor, etc.) are, at the moment, considered reliable precursors of volcanic activity since mostly associated with variations in the fluid-conduit pressure field. Unfortunately, patterns of seismic events are difficult to interpret. In very active volcanoes, current approaches predict eruptions some minutes in advance, but they usually fail at longer-term predictions. Nevertheless, it is believed that seismic waves can carry more information than is known. In this aim, a deep learning approach on massive data could help scientists in exploring currently hidden features in seismic data. Enter Italy's Istituto Nazionale di Geofisica e Vulcanologia (INGV), with its focus on geophysics and volcanology. The INGV's main objective is to contribute to the understanding of the Earth's system while mitigating the associated risks. Tasked with the 24-hour monitoring of seismicity and active volcano activity across the country, the INGV seeks to find the earliest detectable precursors that provide information about the timing of future volcanic eruptions. In this competition, using your data science skills, you will try to improve the current weak predictive capabilities by extracting the maximum information encoded in the signals that the Earth sends us. You'll analyze a large geophysical dataset collected by sensors (seismometers) deployed on active volcanoes to predict when a volcano's next eruption will occur. If successful, your algorithms will identify signatures in seismic waveforms that characterize the development of an eruption. In the data set, you will find segments of signals recorded at 10 deployed permanent seismic stations on an active volcano. Each segment has associated time-to-eruption information which represents the time that was passed from the end of the segment to the beginning of the next eruption. We hope you will enjoy the challenge and, whether you win or you donâ??t, you will contribute to the advancement of our knowledge on volcanoes. We wish you the best of luck. Flavio Cannavò (INGV-OE, Italy) Isaak Kavasidis (University of Catania, Italy) Andrea Cannata (University of Catania, Italy) Concetto Spampinato (University of Catania, Italy) Sohier Dane (Kaggle Staff, Seattle, US) Maggie Demkin (Kaggle Staff, Seattle, US) 1-1-1-1-1-1-1-1-1-1-1-1-1-1 ============================================================== Volcano Listserv is a collaborative venture among Arizona State University (ASU), Portland State University (PSU), the Global Volcanism Program (GVP) of the Smithsonian Institution's National Museum of Natural History, and the International Association for Volcanology and Chemistry of the Earth's Interior (IAVCEI). ASU - http://www.asu.edu/ PSU - http://pdx.edu/ GVP - http://www.volcano.si.edu/ IAVCEI - https://www.iavceivolcano.org/ To unsubscribe from the volcano list, send the message: signoff volcano to: listserv@xxxxxxx, or write to: volcano-request@xxxxxxx. To contribute to the volcano list, send your message to: volcano@xxxxxxx. Please do not send attachments. ============================================================== ------------------------------ End of Volcano Digest - 16 Oct 2020 to 19 Oct 2020 (#2020-101) **************************************************************
