Unveiling hidden Holocene explosive activity at Teide volcano through Data-Driven tephra correlation

Reconstructing the eruptive history of active volcanic complexes is essential for carrying out a proper long-term volcanic hazard assessment. This reconstruction relies on the accurate correlation of volcanic deposits (e.g. tephra layers), which is often hampered by compositional overlaps between units. Traditional methods (e.g. comparison of samples using harker diagrams of two elements) often struggle with the high-dimensional nature of geochemical data. On the other hand, data-driven approaches (e.g. cluster analysis) offer a robust solution by using statistical algorithms to objectively identify subtle geochemical signatures.

The lack of extensive and homogeneous datasets prevents using data-driven approaches for tephra correlation studies, as it is the case of the Teide-Pico Viejo (T-PV) volcanic complex (Tenerife, Canary Islands). Despite posing a significant threat to a densely populated island, T-PV's eruptive dynamics remain only partially constrained: current models attribute high-explosivity (sub-plinian) events primarily to satellite felsic domes (e.g., Montaña Blanca, Pico Cabras), whereas activity at the main Teide stratocone is widely assumed to be limited to moderate-intensity, violent Strombolian eruptions. However, stratigraphic constraints on Teide’s northern flank remain limited, as the source vents for a significant number of tephra deposits have not been identified yet.

To link these unidentified deposits to their source vent, we constructed a geochemical dataset of 74 samples that integrates reference material (lavas and tephras with source vents confirmed by geological mapping) with the target unidentified tephra layers. We analyzed the complete suite for whole-rock geochemistry and micro-analytical phases (phenocrysts and glasses) in some selected samples. We then developed a two-stage data-driven workflow: first, we performed an agglomerative Hierarchical Cluster Analysis (HCA) on whole-rock major and trace elements to group the unknown tephras with chemically affine reference lava and/or tephra samples. Second, these potential correlations were validated through petrographic assessment and a second HCA on the electron probe microanalyses (EPMA) of glass and minerals. The accuracy of our methodology is evidenced by the correct attribution of reference lava-tephra pairs from known eruptions (e.g. Montaña Blanca and Montaña Majúa).

When applied to the unidentified deposits, this method enabled the correlation of three previously unidentified tephra deposits: one with Montaña Reventada eruption, and two with the central Teide vent. Both Teide-sourced eruptions, dated to <10 ka, reveal that the Teide stratocone has hosted significant sub-plinian activity during the Holocene. This finding suggests that the explosivity of the central vent has been historically underestimated, necessitating a re-evaluation of volcanic hazard assessment for the island.

This research was partially funded by E.G., grant EVE (DG ECHO H2020 Ref. 826292), the Intramural CSIC grant MAPCAN (Ref. 202130E083), and Sub-Project 1 ‘Canary Islands, destiny of Volcanoes’, funded by PROMOTUR SA through Next Generation EU funds, PRTR. 2024krQ00nnn. OD was supported by an FPU grant (FPU18/02572) and a complementary mobility grant (EST19/00297) from the Ministry of Universities of Spain.