Multi-Station Trace Data from four Chilean Volcanoes: Nevados de Chillán Volcanic Complex, Villarrica, Laguna del Maule, and Puyehue-Cordón Caulle.

This dataset contains 24.493 seismic events recorded by Observatorio Vulcanológico Andes Sur (OVDAS, SERNAGEOMIN) of four Chilean volcanoes: Nevados de Chillán Volcanic Complex (NVChVC), Villarrica (VCA), Laguna del Maule (LDM), and Puyehue-Cordón Caulle (CAU). Each event is stored as a NumPy array (.npy) of shape 14x8192. All arrays correspond to a 81.92-second (8192 samples @100Hz) window containing a single seismic event (or part of it, for longer events).  A 10-hour multistation continuous signal recorded at NVChVC is also included, this trace contains only the 8 channel seismograms and reference events are detailed in the accompanying .csv file.

The first 8 rows of each array contain data from 1 to 8 seismic stations, with stations that didn't record data filled with zero-valued arrays. The remaining 6 rows represent the true labels of the event at each sample in time:

• Row 9 (Class 0) represents the background class (1 for no event, 0 when an event is present).
• Rows 10 to 14 represent the five seismic event classes:
  • Class 1: Volcano-Tectonic (VT) Events
  • Class 2: Long-Period (LP) Events
  • Class 3: Tremor (TR) Events
  • Class 4: Avalanche (AV) Events
  • Class 5: Ice Quake (IC) Events

The dataset is broken down as follows:

• NVChVC: 3068 VT, 1892 LP, 2360 TR, 805 AV, and 977 IC events, recorded from January 2017 to December 2022.
• VCA: 1516 VT events, recorded from September 2012 to June 2023.
• LDM: 6663 VT events, recorded from April 2012 to July 2023.
• CAU: 2298 VT, 2081 LP, and 2833 TR events, recorded from April 2010 to June 2017.
• Continuous 10-hour NVChVC trace: 38 VT, 55 LP, 28 TR, 23 AV, and 51 IC events (205 total).

All seismic signals were recorded at 100 Hz. Preprocessing steps include applying a bandpass filter between 1 and 15 Hz and normalizing amplitude by dividing each signal channel by the single maximum amplitude across the 8 channels.

The data was curated and manually reclassified for the article "A Framework for Real-Time Volcano-Seismic Event Recognition Based on Multi-Station Seismograms and Semantic Segmentation", and can be explored through codes available at: https://github.com/camilo-espinosa/volcano-seismic-segmentation.

The seismic data used in this research were recorded by the Observatorio Vulcanológico de los Andes del Sur (OVDAS, https://rnvv.sernageomin.cl/observatorio-volcanologico-de-los-andes-del-sur/), part of the Servicio Nacional de Geología y Minería (SERNAGEOMIN, https://rnvv.sernageomin.cl/), Chile. Raw seismic signals for the Villarrica, Laguna del Maule, and Puyehue-Cordón Caulle volcanoes were obtained through a public information request process(https://www.consejotransparencia.cl/solicitud-informacionpublica/). The specific request codes are:

• Laguna del Maule (2012-2023): AS004T0005608, AS004T0004571, AS004T0004553, AS004T0006050.
• Puyehue-Cordón Caulle (2010-2017): AS004T0005484.
• Villarrica (2010-2024): AS004T0004268, AS004T0005733, AS0004T0006292, AS004T0006637.

Data from Nevados del Chillán were provided through a cooperation agreement between Universidad de La Frontera and OVDAS.