Entropy-based seismic monitoring of the Tajogaite volcano, La Palma
-
AARÓN ÁLVAREZ HERNÁNDEZ [3] ; Luca D'Auria [3] ; Iván Cabrera-Pérez [1] ; Jean Soubestre [1] ; Carmen Benítez [2] ; Jesús M. Ibáñez [2] ; Pablo Rey-Devesa [2] ; Ligdamis Gutiérrez [2] ; Janire Prudencio [2] ; Víctor Ortega-Ramos [3] ; Rubén García-Hernández [3] ; David Martínez van Dorth [3] ; Nemesio M. Pérez [3]
-
[1] Université de Genève
Université de Genève
Genève, Suiza
-
[2] Universidad de Granada
Universidad de Granada
Granada, España
-
[3] Instituto Volcanológico de Canarias (INVOLCAN)
-
- Localización: Cosmológica, ISSN 2792-7423, Nº. Extra 1 (Conferencia Internacional: Erupción del Tajogaite (Los Llanos de Aridane, noviembre de 2025)), 2025, págs. 241-242
- Idioma: inglés
- Enlaces
-
Resumen
Seismological techniques are widely used to detect volcanic unrest and forecast eruptions. Among recent advances, Shannon Entropy (SE) analysis has proven effective in volcano monitoring. This method is based on the premise that, as an eruption approaches, coherent seismic sources begin to dominate the wavefield, reducing the signal’s entropy. In contrast, during dormant periods, ambient noise prevails, resulting in higher entropy values. SE is typically applied to a single component per station and has shown promising results in diverse volcanic settings, such as Colima (Mexico) and Cumbre Vieja (Canary Islands) (Rey-Devesa et al., 2023a, b, c). In this study, we propose an extension of this approach using cross-entropy (CE) between station pairs (Shannon, 1948; Shannon & Weaver, 1949). CE not only detects coherent signals but also enables source localization through simple Gaussian interpolation of CE values. We applied this method to seismic data from the permanent network, Red Sísmica Canaria, of the Instituto Volcanológico de Canarias (INVOLCAN) on La Palma, prior to the 2021 Tajogaite eruption. A marked decrease in both SE and CE values was observed weeks before the eruption. The CE-based source location remained temporally and spatially stable, aligning with the area where the eruption eventually began. Moreover, stronger anomalies were detected in the hours leading up to the eruption. These early signals could have significantly supported evacuation decisions. Therefore, we suggest that combining SE and CE analyses offers a valuable tool for the automatic seismic monitoring of active volcanoes.
