Volcanic edifice collapse is a gravitational process; therefore, mass is a key variable, which raises the following question: if mass is critical, then why do small volcanic ocean islands (small mass) collapse, and sometimes catastrophically? We use Santa Maria, a small volcanic ocean island in the Azores ca. 170 times smaller than the Big Island in Hawaii, to address this problem. If (1) the early Santa Maria seamount is ca. 10 Ma (6 Ma above sea level), and (2) the oceanic crust below Santa Maria is ca. 40 Ma old (anomaly 18-20, cf. Luís and Miranda, 2008), then Santa Maria sits on an oceanic crust covered by ca. 30 Ma of marine sediments. If pelagic sediments have deposited around Santa Maria like elsewhere in the deep ocean in and around the Azores, at a rate of a ca. 30 mm/ka (Searle, 1977; Vlag et al., 2004)(no other islands existed at the time to produce volcaniclastic sediments), then Santa Maria would sit on a pile of sediments around 1,000 m thick (cf. Searle, 1977), which corresponds to about one third of the height of the volcanic edifice in Santa Maria (currently 2,500 m below sea level and 500 m above). What are the effects of such basement sediments on the evolution of a small volcanic ocean island, especially regarding edifice spreading and flank collapse? Is a soft basal ring capable of partial detachment, thus facilitating collapse of a full flank, including summit? Here we use numerical modelling to address this problem.
The Azores islands sit on a particular geodynamic setting, the Azores Triple Junction (ATJ, Fig. 1), where the North America, Eurasia and Nubia plates meet. Although not experiencing high magnitude earthquakes (M < 7.0), the Azores are frequently shaken by earthquakes along the main tectonic structures, i.e. the Mid-Atlantic Rift and the diffuse boundary between the Eu and Nu plates. Although with moderate magnitude, some earthquakes can be very destructive when they occur at shallow depth. Therefore, we must consider seismicity as a possible trigger of flank collapses.
The evolution of subaerial Santa Maria (Sibrant et al., 2015a; Ramalho et al., 2017; Marques et al., 2020; see Fig. 1B, C) started ca. 6 Ma ago with a basaltic subaerial shield volcano (Old Volcanic Complex), which collapsed to the east at ca. 5.1 Ma. A new basaltic shield volcano (Young Volcanic Complex) grew on the collapse scar, therefore initially submarine and later subaerial as it grew out of the water, which collapsed to the east at ca. 3.8 Ma. Immediately following the collapse, numerous Strombolian cones grew unconformable on the eastern concave scar facing east, from ca. 3.7 to 2.8 Ma, thus rapidly filling the collapse scar. Similarly to other volcanic ocean islands (e.g. Hildenbrand et al., 2003; Ramalho et al., 2013; Marques et al., 2020b), Santa Maria experienced considerable vertical motions (Marques et al., 2020a), witnessed by submarine lava flows currently outcropping at ca. 200 m above sea level. The topography of Santa Maria (Fig. 2A) shows a well-defined divide separating the contrasting topography between E (rugged and concave to the E) and W (structural surface gently dipping to the W, similarly to the lava flows of the two volcanic complexes) Santa Maria.
Work on flank collapses has concentrated on relatively large volcanic ocean islands, and all modelling work (analogue and numerical) has considered a full detachment of the volcano’s base (Borgia, 1994); Merle & Borgia, 1996; van Wyk de Vries & Borgia, 1996; van Wyk de Vries & Matela, 1998). This does not seem realistic to us, because the core of volcanic edifices typically has a high geothermal gradient and high pressure that should transform underlying soft sediment into hard metamorphic rock. However, the effects of pressure and temperature on the rheology of sediments underlying the island should depend on island size. Therefore, in the present work, we used a small edifice and low viscosity rheology in a ring totally or partially occupying the base of the volcanic edifice.
Since 2012 many large landslides in the Azores have been reported by the MEGAHazards team (PI F.O Marques) based on extensive on and offshore observation and data (Hildenbrand et al., 2012, 2018; Marques et al., 2013b, 2018, 2020, 2021; Costa et al., 2014, 2015; Sibrant et al., 2014, 2015a, 2015b, 2016). Regarding Santa Maria, Marques et al. (2013b, 2020) and Sibrant et al. (2015a) recognized two flank collapses at around 5.1 and 3.8 Ma using only onshore observation and data. The main argument is that the summits and eastern flanks of the two main shield volcanoes are missing (only the western flank is well preserved; cf. Fig. 2B), which should thus comprise two debris deposits at the ocean bottom, which are conspicuous in the new high-resolution bathymetry presented here.
Given the above state-of-the-art, the novelty in the present study is two-fold: (1) the high-resolution bathymetry showing the distribution and flow of the deposits produced by the flank collapses; (2) numerical simulations showing how small islands can collapse.
New field data
Despite their age (ca. 2.8-3.6 Ma), most Strombolian cones in Santa Maria have their original shape well recognizable, and can thus be spotted on the 10 m resolution DEM. Confirmation on the field was carried out by observing the typical volcanic stratigraphy (alternating pyroclastic layers and lava flows dipping radially to a central crater), sometimes also with outcropping necks with massive lava. We detected ca. 130 cones, of which only 9 occur W of the divide (cf. Fig. S1). 120 is a minimum number of cones E of the divide, because the terrain is in many cases very difficult to access or densely vegetated with no outcrops. We can distinguish two main types of cones, according to their dimensions (Fig. S2): large cones (diameter at the base > 500 m; Fig. S2), and small cones (diameter at the base <200 m).
New high-resolution bathymetry
Santa Maria lies very close to the junction between the East Azores Fracture Zone (EAFZ), the scarp of which is taller than 1000 m, and the Terceira Rift. The bathymetry around Santa Maria shows the following main features (Figs. 3 and S3): (1) Santa Maria is only conical in the NW quadrant, elsewhere the contours are either straight or concave outwards; (2) the Terceira Rift lies only ca. 25 km east of Santa Maria; (3) south of Santa Maria, the EAFZ is filled with thick sediments; (4) the EAFZ scarp is blanketed by thick sediments ca. 50 km ESE of Santa Maria; (5) these sediments comprise a hummocky terrain with lobes convex outwards (to ESE); (6) a tongue of sinuous sediment lies inside the EAFZ, indicating flow of fine debris into the EAFZ and over a thick older deposit; (7) SSW of Santa Maria, sediments partially blanket the tall scarp of the EAFZ (Figs. 4 and S6); (8) inside the EAFZ SSW of Santa Maria (Figs. 4 and S6), the thick sediment front is conspicuous, with younger lobes overlying an older and thicker deposit; (9) topographic profiles along critical directions show the shape of the topography where we infer the existence (convex upward) or absence (concave upward) of debris deposits (Figs. S4 and S5).
Given that the Terceira Rift is younger than the Santa Maria’s flank collapses, part of the deposits SE of Santa Maria are currently inside the Terceira Rift, whose bounding fault in the S has displaced the deposits into the rift.
Numerical modelling
Our main goal was to investigate the effects of partial or total detachment at the base of the volcanic edifice, which differs from previous work (e.g. Borgia, 1994); Merle & Borgia, 1996; van Wyk de Vries & Borgia, 1996; van Wyk de Vries & Matela, 1998), where a fully detached base was used. Our argument for this critical difference is that pressure (edifice load) and temperature (geothermal gradient) increase towards the centre of the base of the edifice, which can transform soft sediments (detachment) into hard medium- to high-grade metamorphic rocks (strong coupling with volcanic rocks) depending on the pressure and temperature conditions, which in turn depend on edifice size and geothermal gradient within the edifice.
The numerical results for a viscosity of 1022 Pa s and a friction angle of 15º (Fig. 5) show that: (1) the island collapses only where there is basal decollement (lefthand side of the edifice), otherwise there is no collapse (righthand side); (2) a main detachment forms where most of the outward motion is accommodated; (3) strain within the collapsing edifice is accommodated by a number of small synthetic and antithetic faults; (4) when the soft layer occupies the whole radius, the entire flank, summit and part of the opposite flank collapse; (5) when the soft layer only partly occupies the base, only partial flank collapse is observed within relevant time.