The triggering of volcanic eruptions represents the ultimate stage of a series of non-linear physico-chemical processes. Although the conditions that make a volcano ready to enter in eruption are still elusive, deciphering those that participate in the destabilization of metastable systems could provide new insights on the processes leading to eruption triggering. Knowledge of how volcanic systems work has been mainly gained from studies on individual volcanoes first based on specific methodologies and more and more on multi-method approaches (e.g., McNutt, 1996; Newhall and Punongbayan, 1996; Battaglia et al., 2006; Dzurisin, 2006; Sigmundsson et al., 2015; Gudmundsson et al., 2016). Despite of the high variability of volcanoes illustrated in particular by their eruptive activity, internal structure, or chemical composition, common processes drive volcanic activity (Cashman and Biggs, 2014). Studies on a large number of volcanoes have allowed to provide more constraints on global processes controlling their dynamics and evolution as for instance, the magnitude-frequency relationship of large explosive eruptions (e.g., Deligne et al., 2010), the link between deformation and eruption (e.g., Biggs et al., 2014), as well as the role of external forcing on the eruption triggering such as tectonic earthquakes (Seropian et al., 2021), the sea level and climate variations (e.g., Laskar et al., 2004; Mason et al., 2004; Kutterolf et al., 2013; Satow et al., 2021) and Earth tides (Mauk and Jonhston, 1973).

Actually, the variations of the sea level, climate and Earth tides are all related, taking part of a global dynamics as pieces of a giant puzzle. The Liouville-Euler equations link the redistribution of mass at Earth’s surface, and in particular that of the fluid envelopes to the movement of the Earth’s rotation axis through a system of linear equations and excitation functions (Lambeck, 2005; Lopes et al., 2021). These movements take place at different time scales ranging from a few days to at least 18.6 years, as revealed by the study of the length-of-day variations, a global parameter whose fluctuations on short-term capture the mass redistribution on Earth through their impact on Earth’s rotation velocity (Le Mouël et al., 2019). The revolution of the Earth around the Sun and the variations of the Earth’s rotation axis that are associated with seasons and Milankovitch cycles modulate this force and mass movements and interfere with those induced by plate tectonics, ocean and atmospheric currents (e.g., Lambeck, 2005; Le Mouël et al., 2021a,b; Zaccagnino et al., 2020).

The idea of tidal influence on volcanic activity has persisted over the last century (e.g., Jaggar et al., 1924; McNutt and Beavan, 1987; Emter 1997; Sottili et al., 2021). Statistical approaches have shown that the onset of volcanic eruptions at global scale (e.g., Mauk and Johnston, 1973) as well as at local scale (e.g., Johnston and Mauk, 1972; Filson et al., 1973; Golombek and Carr, 1978; Dzurizin, 1980; Martin and Rose, 1981) could be related to the Moon and in particular, its alignment with the Earth and the Sun, known as the fortnightly cycle. Similar observations of preferred periods for eruption onsets have been made with ocean tides for submarine, coastal and insular volcanic systems (e.g., McNutt and Beavan, 1987; McNutt, 1999; Mason et al., 2004; Tolstoy, 2015). In coastal environment, the tidal deformation coupled with the ocean loading, which corresponds to the movement of surface water mass, induce variations in the compressional stresses that can promote squeezing out magma (e.g., McNutt and Beavan, 1987; McNutt, 1999). At time scales of hundreds of thousands years, pronounced sea level variations related to ice melting and more generally long-period climatic and astronomical variations have also been associated with pulses of volcanic activity, e.g., Mason et al. (2004); Andrew and Gudmundsson (2007); Pagli and Sigmundsson (2008); Kutterolf et al. (2013); Watt et al. (2013); Satow et al. (2021); Sottili et al. (2021).

The influence of solid Earth and ocean tides has also been suggested on shorter time scales through periodic behavior detected in various physical parameters measured at volcanoes, e.g., seismic tremor, earthquake rate, gas fluxes. Actually, tidal constituents have been identified at volcanoes independently of their phases of activity including quiescence, unrest and eruptions (e.g., Jaggar et al., 1924; McNutt and Beavan, 1981, 1984; Shimozuru, 1987; Leandro and Alvarado 1988; Rydelek et al., 1988; Caltabanio et al., 1994; Kasahara et al., 2001; Custódio et al., 2003; López et al., 2006; Roult et al., 2012; Sottili and Palladino, 2012; Yakiwara et al., 2013; Bredemeyer and Hesteen, 2014; Conde et al., 2014; Contadakis et al., 2017; Dinger et al., 2018; Girona et al., 2018; Patrick et al., 2019; Dumont et al., 2020, 2021; İçhedef et al., 2020; Miguelsanz et al., 2021). Most of these parameters express the transfer of fluids, either melt or/and hydrothermal fluids, from deeper layers towards and at the Earth’s surface. At the quiescent and hydrothermally active caldera of Campi Flegrei (Italy), studies have demonstrated that tidal modulation can take place according to a complex coupling mechanism that affects both crustal structure and hydrothermal fluids (e.g., Bottiglieri et al., 2010; De Lauro et al., 2012, 2013, 2018; Petrosino et al., 2018, 2020; Ricco et al., 2019; Caputo et al., 2020; Cusano et al., 2021; Petrosino and Dumont, 2022). At erupting volcanoes, increase in explosions or seismic activity as well as second-order variations of geophysical time-series were also attributed to tides (e.g., Mauk and Kienle, 1973; Golombek and Carr, 1978; Martin and Rose, 1981; Jentzsch et al., 2001; Williams-Jones et al., 2001; Dumont et al., 2020, 2021).

Despite all these observations, not all volcanoes show a sensitivity to solid Earth and ocean tides, and the cause-effect relationship is still elusive (e.g., Sparks, 1981; Neuberg, 2000). Actually, past studies have mainly focused either on short or very-long time scales, from hours to a few years on one hand, or over several kyr on the other hand. However, the interaction between tidally-generated movements and volcanic systems is expected to occur on all time scales on which this external forcing acts. Therefore, should we not expect a possible response of volcanoes at all these time-scales?

We address these questions by exploring the link between ocean tides and volcanoes from a global viewpoint over a ∼130 years time interval. Oceans as all fluid and solid envelopes on Earth respond to external forcing induced by lunisolar tidal potential. They participate in a significant redistribution of mass through ocean loading, and therefore stresses at Earth’s surface over different time scales. Considering that most volcanoes lie near, within or beneath the oceans (Figure 1), we can raise the question of whether the tidally-induced ocean movements and therefore stresses can participate in changing the dynamical systems and so the destabilization of volcanoes on decadal timescales. This is what we tackle with this paper, by analyzing the common periodicities which may exist in the number of worldwide volcanic eruptions and the global mean sea level using the Singular Spectrum Analysis (SSA) technique.

Our analysis, performed for both the whole eruption catalog and VEI≥2 eruptions, shows the existence of consistent multi-decadal periodicities not only for low and non-explosive eruptions but also for explosive-dominated phenomena as already suggested for sulfur rich tropical eruptions (Ammann and Nadeau, 2003), and despite a catalog completeness which has kept improving over the 20th century (Siebert et al., 2015). Although some of the uncertainties are significant, very comparable periodicities were found in the GMSL. These different periodicities form the larger contributions in each time-series representing more than 90% of the original signals, once considered together with the trend (Supplementary Figures S2, S3, S4). Moreover, a specific link between these multi-decadal components extracted in the NVE and GMSL is observed (in phase, phase shifted/opposition). All together, these results lead us to consider that these multi-decadal periodicities in the NVE and GMSL time-series are related.

Most volcanic eruptions considered in this study, e.g. ∼70%, occur on volcanoes located either on islands or sea floor, indicating a prime link of spatial order between volcanic systems and oceans, knowing that the submarine eruptions are by far underestimated (Figures 1C,D). Moreover, for volcanoes situated on continents or islands, about 54% of them are located within 50 km to the coast (Figure 1D, Supplementary Figure S1). Water, which is ubiquitous on Earth, is not only present at Earth’s surface but also as groundwater, penetrating beneath the islands and far into the continents. Water also accumulates within intraplate regions (e.g., Gleeson et al., 2016; Craig et al., 2017; MacMillan et al., 2019; Jasechko et al., 2020). All fluid layers redistribute water masses from the microscopic scale through pore-fluid pressures to the global scale via their loading, by oscillating at periods ranging from few hours to multi-decades in response to tidal forces and pole motion (e.g., Lambeck, 2005; MacMillan et al., 2019; Le Mouël et al., 2021b; Lopes et al., 2021). These two different ways by which water can interfere with dynamical systems have not only been suggested for volcanoes, but also for fault systems where tidal stresses may enhance the long-term tectonic loads (Thomas et al., 2009; Neves et al., 2015; Ide et al., 2016). Considering that both faults and volcanoes share some commonalities in their driving processes, we believe that taking into account their response to external forcing could improve understanding of how this interplay works.

Past studies spotted that long periods (>6 months) of the hydrological cycle can influence eruption onsets at volcanoes located nearby the coast (e.g., McNutt and Beavan, 1987; McNutt, 1999). This idea seems to be confirmed by our study and the work of Mason et al. (2004) based on worldwide catalog of eruptions who showed a correlation between eruption rates and annual sea level variations. Besides, it is well known that bathymetry and boundary effects associated with coastlines have a major impact on the local amplitudes of semi-diurnal, spring, and neap tides (Supplementary Figure S5). Likewise, sea level variations over longer time scales, like the ones highlighted in this study, have different amplitudes and larger loading depending on geographic location. As long periods of hydrological loading redistribute more water masses than shorter periods, it could explain why long periods and multi-decadal periodicities would have more impact on unstable and dynamical systems.

Detecting these decadal periodicities in both time-series suggests as well that they relate to a global phenomenon. Although the way these processes take place and interfere is still puzzling, we note that four similar multi-decadal periods were detected in the fluctuations of the pole motion, being of 21.6, 31.5, 57.3 and 92.2 years (e.g., Lopes et al., 2021). Yet, any movement of the Earth’s rotation axis is accompanied by mass movements according to Liouville-Euler equations (Lambeck, 2005; Lopes et al., 2021). This may be illustrated by the late Pleistocene deglaciation, the concomitant sea level changes and a secular drift present in the pole motion (Nakiboglu and Lambeck, 1980). This is what we observe through the common periodicities extracted in the GMSL and the pole motion (Lopes et al., 2021). We interpret the presence of these decadal periodicities in the NVE as a consequence of this global mass redistribution operating at similar time scales. The water masses transported over these decadal periods induce pressure changes on crust that adds to the local stresses, leading to a slow destabilization of magma plumbing systems. In fact, the decadal cycles extracted in the NVE do not properly trigger one eruption every ∼50 or ∼20 years as evidenced by the amplitude of the waveforms extracted by SSA (Figure 2), although they participate to the triggering process. Complementary approaches will be necessary for better understanding this link, as for instance, numerical modeling to investigate how tidal stresses can destabilize magma chambers and fault systems (McNutt and Beaven, 1987; Ide et al., 2016; Jonhson et al., 2017; Scholz, 2019). Moreover, the impact of the fluctuations of the Earth’s rotation axis on volcanic eruptions was also suggested locally and regionally (Kutterolf et al., 2013; Lambert and Sottili, 2019).

Our study leads us to suggest that movements affecting globally our planet over a wide range of time scales, e.g., lunisolar gravitational forces, pole motion, may contribute to the dynamics of volcanic systems. These considerations and results raise also some more questions as for instance, are volcanoes sensitive to all or just some of these periodicities and what may explain these differences? Could this forcing be viewed as a background variation participating in the regular dynamics of magma plumbing systems? We think that these decadal variations together with short and very long periodicities deserve to be more explored and investigated as they might reveal global conditions and space-time patterns favorable to eruption triggering.

These results represent an avenue to explore how the mass redistribution acting at global scale participates in the processes leading to the triggering of volcanic eruptions as already suggested by Mason et al. (2004) or Sottili et al. (2021). Deciphering the link between global and local scale is critical to understand how external forcing acts on different time scales and how it can disrupt dynamical systems like volcanoes. We think that assessing these processes at multiple scales will allow to provide more constraints on the parameters and conditions that make volcanoes sensitive to small environmental changes (e.g. Canon-Tapia, 2014; Caricchi et al., 2021).

Singular Spectrum Analysis, in addition to be a robust time-series analysis technique that has been applied to a large variety of Earth sciences data, allows to extract trends and periodic components without an a priori model, which has some advantages when analyzing signals integrated at various time scales such as tides and whose modulation may vary over time as in case of seismic tremor in a highly heterogeneous medium (e.g., Petrosino and Dumont, 2022). Similar analyses could provide new insights on studied processes, including cause-effect relationships between parameters as well as what makes some volcanoes sensitive to quasi-permanent external forcing and why (e.g., Dumont et al., 2021). All active volcanoes do not respond to solid Earth and ocean tides (e.g., Neuberg, 2000) and that sensitivity may be related to inherent properties of volcanoes, their location on Earth or the plate boundaries conditions (e.g., Mauk and Johnston, 1973; Dzurisin, 1980; Dumont et al., 2021). Many studies have also reported an increase of the sensitivity to tidal forcing when plumbing systems are approaching a critical state coinciding for some volcanoes to pre-eruptive periods (e.g., Filson et al., 1973; Berrino and Corrado, 1991; De Lauro et al., 2013; Girona et al., 2018; Miguelsanz et al., 2021). More systematics observations are required to unravel these critical conditions.

Technique plays an important role in the road to findings, but data, as all the ingredients in a recipe, play an equally fundamental role. The data availability and quality are of great importance to advance in the comprehension of the interplay between external and internal processes at volcanoes. The role of volcano observatories and initiatives such WOVOdat (Newhall et al., 2017) or the Global Volcanism Program (2013) have been primordial to tackle these questions and many others. Revisiting past volcanic activity together with recent volcanic events, from global to local scale using time-series analysis, may help to advance in this direction.