It is well-known that volcanoes are dominated by non-linear processes, and continuous records over long periods of time are required to capture all aspects of volcanic seismicity. Thus, we will create a Parametric Seismic Database for long time series at a set of volcanoes using the vector of features defined above. Therefore, once the best set of features has been defined, FEMALE will transform the original raw seismic data bases into a new parametric seismic database, i.e., the transformation of the original long time series of seismograms to a new set of parametric matrices.

WP2.1. Selection of the test volcanoes: acquisition of long-duration and continuous seismic data

We have selected three reference volcanoes from different eruptive regimes which can be used as models for most of the active volcanoes of the World: effusive (Kilauea), mixed effusive-explosive (Mt. Etna), dome-explosive (Asama). The case of Asama volcano is special, not only because is a very active scenario, but also the original classification of volcanic earthquakes of Minakami20 is from this volcano, and it will be interesting to see how state-of-the-art techniques may change many decades old view. For these volcanoes seismic records are publicly available including both many stations and cover a long time-span (15-20 years). They are also easily transferable to the Cloud services for storage and analysis. The advantage of the Cloud is the fast and universal access for all researchers involved in this work.

The working team will be leader by Prof. Ibáñez with a direct collaboration with Drs. Prudencio, Feriche, García-Yeguas of the Research Team and several members of the Working Team.

WP2.2. Transformation of the 2D seismogram into a parameter matrix representation

. For each seismogram we will select an interval of time (5 minutes long, overlapped 50%) and over this interval the parametric transformation will be computed. For each interval we will create a vector of parameters with dimension D (D being the number of parameters defined in WP1). Each vector will be stacked onto the previous ones, forming a matrix that grows over time. The best manageable option is to create a daily matrix (24 h of record, with 576 rows). For each three component stations this matrix will be created for each component. The procedure will be applied to data from all available seismic stations, covering the entire available time period, and for all selected volcanoes (Figure 5). The final product will be a set of matrices containing characteristics and parameters of each seismic data stream. The conversion from the 2D seismogram space to the embedded matrix space could be exportable as matrices that can be easily indexed. Each matrix will be additionally identified with complementary information such as the volcano, seismic network and station, ground component, starting time of the analysis, duration, sampling rate and characteristic of the instruments, among others. This information can be created in parallel as metadata bases.

The working team will be leader by Prof. Ibáñez with a direct collaboration with Drs. Benitez, Prudencio, Alguacil and Mota of the Research Team and several members of the Working Team.

© 2021 Female

Website designed and developed by Ramón Ruiz