Volcanic or tectonic swarms? How Shannon Entropy could solve the riddle

Distinguishing volcanic from tectonic seismicity remains a critical challenge for volcano monitoring and hazard assessment. Traditional approaches often rely on spectral or amplitude-based criteria, which can be ambiguous during seismic swarm activity. Here, we explore the potential of Shannon entropy as a robust discriminator of seismic signal complexity, focusing on its ability to identify seismo-volcanic signals and, specifically, the onset of volcanic tremor.

We apply Shannon entropy to band-pass filtered seismic data (1–16 Hz) using 10-minute sliding windows. This metric captures the degree of predictability in the signal: low entropy indicates a more coherent, structured waveform, while high entropy reflects greater randomness. By tracking these changes over time, entropy provides a dynamic measure of signal organization. Three case studies illustrate the method: the 2011 submarine eruption of Tagoro volcano (El Hierro, Canary Islands), the 2021 subaerial Tajogaite eruption (La Palma, Canary Islands), and the 2025 magmatic unrest at Santorini (Greece).

In all these cases, entropy exhibits a decay coinciding with the onset of seismo-volcanic activity. Remarkably, a precursor pattern emerges: a gradual decrease in entropy preceding the main drop, suggesting early changes in seismic dynamics preceding the onset of genuine volcanic tremor.

Our findings highlight Shannon entropy as a simple yet powerful tool for real-time monitoring. By capturing changes in complexity in seismic signals, this metric provides an additional layer of information that complements conventional spectral analyses. The detection of precursor entropy decay could enhance early-warning capabilities, reducing uncertainty in distinguishing volcanic from tectonic processes during swarm activity.