Anticipating volcanic eruption relies on having quick access to standardized historical monitoring data and statistical tools for forecasting. In FY2018, the development of the WOVOdat project focused on four main tasks:
Data population. We continuously populate the database with historical volcano unrest data from various sources. In collaboration with Cornell University we added thermal and InSAR data covering around 47 volcanoes, and with the help of 2 ASE student interns we compile classical unrest episodes of St. Helens for 1980-1983 and 2004-2005 eruptions. Currently we have incorporated about 47% of targeted worldwide unrest data, covering 135 volcanoes with more than 921 unrest episodes.
Database & interactive web interface. We create a robust online-interface allowing users to efficiently submitting data, explore and use the database through data query and visualisation tools. We completed the visualisation tool for “temporal evolution of unrest”, which allows us to interactively plot multi-parameter monitoring data together with eruption phases and alert level changes in chronological order.
Data mining and analytical tools. These tools are created to facilitate advanced analysis of integrated worldwide unrest data in WOVOdat, using classical statistical and recently developed machine learning techniques, to better understand the volcanic processes behind unrest and to improve eruption forecasts. In FY2018 we developed various interactive tools using machine learning algorithm with open source software (python and Jupyter notebook); e.g. using K-means clustering to find spatio-temporal precursory seismic pattern of Kilauea (Hawaii, USA), detecting temporal precursory pattern of 1983-2014 Merapi (Indonesia) time series data using Symbolic Aggregate Approximation (SAX). The results were been presented in COV-10 conference in Napoli (Italy) and will soon be submitted to a peer-review journal.
Capacity building and community engagement. WOVOdat constantly promotes and supports the development of volcano monitoring database at various volcano observatories (CVGHM, PHIVOLCS, and RVO-Papua New Guinea), regional open data initiatives (EPOS, EUROVOLC, and JVDN-Japan), and actively engages with the international volcano community as a major partner of Global Volcano Model (GVM) and are involved in UNISDR’s Global Assessment Report of 2019 (GAR19).
- Earth Observatory of Singapore
2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019
In November 2018 we organized an international workshop on “Optimizing the use of volcano monitoring database to anticipate unrest” in Yogyakarta (Indonesia). The participants came from worldwide volcano observatories and universities: CVGHM (Indonesia), UGM (Indonesia), PHIVOLCS (Philippines), RVO (Papua New Guinea), SERNAGEOMIN (Chile), IPGP (France), INGV (Italy), NIED (Japan), ERI (Japan), MVO (Montserrat), USGS (USA), and EOS-NTU (Singapore)