From January 2019 to the beginning of January 2020 the shallow and the deep reservoirs of Fernandina inflated at a constant rate of 2.15 ± 0.03 x106 m3/year and 9.15 ± 0.52x106 m3/year, respectively (Fig. 3a and S2). Inflow rates of the shallow reservoir are similar to that inferred after the June 2018 eruptions, from September 2018 to the beginning of January 2019, while the rates of the deep reservoir are slightly (38.7%) lower (Galetto et al., 2023b). This suggests that magma inflow rates in both reservoirs have not changed significantly during the inter-eruptive period among the two eruptions.
Modelling results show that syn-eruptive deformation is mainly related to the southward propagation of magma (Fig. 3b). According to the depths and the geometries of the three inflating RD sources, this propagation started below the caldera area as a sub-horizontal intrusion, probably from the deeper portion of the shallow reservoir (~ 1.79 km below the surface). Then, below the upper flanks of Fernandina, the intrusion turns upward and twists, becoming a slightly inclined (~ 31°) intrusion with a different strike angle (Fig. 2h,k, and 3b). The dike changes its strike and dip angle again, becoming sub-vertical, below the lower, gently sloping, flank. The depth and geometry of this latter dike suggest that a submarine eruption might be occurred in the escarpment below the south flank of Fernandina (Fig. 3b). The geometry of the modelled dikes is consistent with the model proposed by Bagnardi et al., (2013) for radial dikes at Galápagos. In addition, similar topographic-related changes in the strike, or in the strike and in the dip angles, of the radial dike have been observed also in other volcanoes, like Bárðarbunga (Iceland) and Cerro Azul (Galápagos), suggesting a control of the topography on the propagating radial dike (Heimisson et al., 2015; Sigmundsson et al., 2015; Corbi et al., 2015; Urbani et al., 2017; Galetto et al. 2020). The modelled radial dike is located below the same flank of the 2005 and 2009 radial eruptions, but it is deeper than the dikes associated to these two eruptions (Jónsson et al., 1999; Bagnardi et al., 2013). This could explain why the 1995 and the 2009 eruptions were triggered by mid-low dip angle radial dikes that caused eruptions located in the steep upper flank and/or at the transition with the gently sloping lower flank (Jónsson et al., 1999; Bagnardi et al., 2013). On the contrary, the 2020 radial dike, being deeper, achieved to propagate also below the lower flank, becoming subvertical.
The syn-eruptive subsidence recorded nearby the caldera area has been mainly modelled with the deflation of the Yang’s source, which represents the deep magma reservoir (Bagnardi & Amelung, 2012). This is the modelled source that lost the higher amount of volume, similar to what inferred for the previous four eruptions at Fernandina (Chadwick et al., 2011; Bagnardi et al., 2013; Galetto et al., 2023b). The other deflating source in the caldera area is a vertical dike that represents the conduit connecting the deep with the shallow reservoir (Galetto et al., 2023b). The deflation of both the deep reservoir and the conduit suggests a rapid magma transfer from the deep reservoir to the dikes, passing through the conduit and, perhaps, the shallow reservoir. The lack of coherence in the caldera area, where most of the deformation associated with the shallow reservoir usually focalized (Chadwick et al., 2011; Bagnardi & Amelung, 2012; Galetto et al., 2023b), does not allow to determine if the shallow reservoir inflated or deflated during this eruption.
The distal RD deflating source has been used to obtain a better-fit solution and might represent a distal portion of the deep magma reservoir, suggesting that the magma was drained from wide portions of the deep reservoir.
None of the recorded deformation matches in location with the circumferential eruptive fissures placed on the upper east flank (Fig. 1), probably because the deformation associated to all the modelled sources masks the one connected with the circumferential eruption. However, the modelled sub-horizontal RD source below the SE caldera area, interpreted as the incipient propagation of the radial dike, might also represent the incipient propagation of the circumferential dike that triggered the eruption, due to the geometry and position of this RD source (Chadwick et al., 2011; Galetto et al., 2023b). In this scenario, part of the magma propagated upward from the north edge of this RD source feeding the eruptive fissures, while most of the magma is drained southward from the horizontal sill, promoting the propagation of the radial dike (Fig. 3b). This could explain why the eruption lasted for 9 hours only (Bernard et al., 2022). This interpretation is also supported by the intruded and the erupted volumes. Indeed, also without considering the volume of the sub-horizontal source that could have been shared by both the radial and the circumferential dike, the modelled volume of the two distal RD inflating sources is of 30.16 ± 7.16 x106 m3. The exact erupted volume is unknown, but Vasconez et al., (2022) estimated that lavas erupted in 2020 covered a surface of 1.63 km2. By assuming a constant average thickness equal to that of the previous five eruptions occurred at Fernandina (between 1 and 8.5 m; Vasconez et al., 2018) the resulting erupted bulk volume would be between 1.63 and 13.9 x106 m3 (1.2–10.4 x106 m3 DRE by assuming 25% of void according with Vasconez et al., 2018). Thus, the volume intruded in the distal portion of the radial dike is from two to ten times larger than that erupted, suggesting that the volume lost by the deflating sources was mainly drained in the radial dike rather than being erupted through in the circumferential dike.