Volcanic edifice collapse is a gravitational process; therefore, mass is a key variable, which raises the following question: if mass is critical, then why, and how, do small volcanic ocean islands (small mass) undergo full flank collapse (full flank plus summit), and sometimes catastrophically? The Azores volcanic province is a target of particular interest to investigate such topics, because, unlike large intra-plate volcanic ocean islands developed on deep abyssal plains (e.g. in Hawaii, the western Canary, Reunion, or Tahiti), most of the Azores islands have a modest total height (less than 2500 m from the seafloor). An edifice height of 2500 m was considered as a threshold value by Mitchell (2003), because major gravitational destabilization episodes were only rarely observed for edifices with sizes below that value. However, since the first flank collapse recognized in Pico Island by Woodhall (1974), more than twelve other flank collapses with individual volume on the order of a few cubic kilometres have been recognized in the Azores from on and offshore studies since then (Hildenbrand et al., 2012a, b, 2018; Marques et al., 2013b, 2018, 2020a, 2021; Costa et al., 2014, 2015; Sibrant et al., 2014, 2015a, b, 2016; Weiß et al., 2016). This makes the archipelago a very special case to further investigate why and how small volcanic islands experience full flank failures.
The Santa Maria volcanic edifice is < 2500 m tall, and lied on a tectonically active setting between ca. 8 and 2.8 Ma ago. From onshore geological, geomorphological, structural and geochronological studies, two SE-directed full flank collapses have been distinguished (Marques et al., 2013b, 2020a; Sibrant et al., 2015a). Therefore, Santa Maria is the ideal setting to argue in favour of full flank collapses in small volcanic islands, probably triggered by tectonic shaking. For the sake of comparison, Santa Maria is ca. 170 times smaller than the Big Island in Hawaii, and thus ideal to address the effect of small mass on flank collapse. Additionally, the Santa Maria edifice does not have steep flanks (< 10º on average), therefore the effect of slope magnitude is not considered here.
If (1) the early Santa Maria seamount is < 8 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 marine sediments accumulated over a period of ca 30 Ma. Based on seismic reflection and refraction data, Batista et al. (2022) calculated a maximum sediment thickness of 1500 m between São Miguel and Santa Maria. Beier et al. (2022) estimated the thickness of sediments on 30 Ma basaltic crust to be about 650–700 m, assuming interval velocities of 1.8-2 km/s for a two-way travel time of ca. 0.7 s for the sediments. Hence, Santa Maria sits on a pile of sediments of similar thickness.
What are the effects of such a basal sedimentary pile on the evolution of a small volcanic ocean island, especially regarding edifice spreading and flank collapse? Is a soft base capable of detachment, thus facilitating full flank collapse? Here we use numerical modelling to address this problem. We distinguish between full and partial flank collapse because the mechanics of the two processes is different, as indicated by the numerical simulations. We further use new on and offshore data (bathymetry and reflection seismic profile) to track potential collapse structures and deposits generated by the flank collapses, which may provide additional insight into landslide mechanisms and debris-avalanche propagation.
The new onshore data, the mapping of Strombolian cones, seems of little importance at first sight, and that is probably the reason why they have never been mapped. However, they are critical for the recognition of flank collapses because most of them sit on a concave surface, which is not the typical situation in a shield volcano. The eastern half of Santa Maria shows a pronounced curved topographic surface concave to the east, which, at first sight, might be interpreted as the result of recent erosion. However, it is not before we map and date tens of 3.8–3.6 Ma old Strombolian cones on this concave surface that we realize that this surface is not the result of long-term erosional processes, and it cannot be recent, it has to be older than ca. 3.8 Ma. That is why it is so critical to add these new data to Santa Maria’s dataset, and help understand the evolution of the island.