Revealing the hidden seismicity of Tenerife (Canary Islands) through machine learning and cross-correlation

In recent years, the development of techniques based on integrated machine learning techniques with traditional methods like cross-correlation has dramatically improved the capability of detection, location and characterisation of seismic events.

Machine learning-based techniques have been applied to improve earthquake catalogues, aiding seismicity analysis in different geodynamic contexts. These techniques are especially valuable in volcanic and geothermal contexts, where volcano-tectonic earthquakes and long-period events have low magnitude, often preventing a successful manual analysis. Since 2016, the island of Tenerife has been affected by increased seismicity and gas emissions from the crater of Teide volcano, the most prominent feature of the island of Tenerife. We applied the software Qseek (https://github.com/pyrocko/qseek) to enhance the seismic catalogue based on manual detections in Tenerife.

This reanalysis of Tenerife's seismic dataset revealed intense seismicity related to episodes of magmatic fluid injection into the upper crust. Subsequently, high-resolution relocation techniques based on the software GrowClust (https://github.com/dttrugman/GrowClust) imaged the spatial pattern of the hypocenters, highlighting sources of magmatic fluid injection into the hydrothermal system of Tenerife and the subsequent response of the upper crust to this disturbance.