Description: Volcanic island flank collapses have the potential to trigger devastating tsunamis threatening coastal communities and infrastructure. The 1888 sector collapse of Ritter Island, Papua New Guinea (in the following called Ritter) is the most voluminous volcanic island flank collapse in historic times. The associated tsunami had run-up heights of more than 20 m on the neighboring islands and reached settlements 600 km away from its source. This event provides an opportunity to advance our understanding of volcanic landslide-tsunami hazards. Here, we present a detailed reconstruction of the 1888 Ritter sector collapse based on high-resolution 2D and 3D seismic and bathymetric data covering the failed volcanic edifice and the associated mass-movement deposits. The 3D seismic data reveal that the catastrophic collapse of Ritter occurred in two phases: (1) Ritter was first affected by deep-seated, gradual spreading over a long time period, which is manifest in pronounced compressional deformation within the volcanic edifice and the adjacent seafloor sediments. A scoria cone at the foot of Ritter acted as a buttress, influencing the displacement and deformation of the western flank of the volcano and causing shearing within the volcanic edifice. (2) During the final, catastrophic phase of the collapse, about 2.4 km³ of Ritter disintegrated almost entirely and travelled as a highly energetic mass flow, which incised the underlying sediment. The irregular topography west of Ritter is a product of both compressional deformation and erosion. A crater-like depression underlying the recent volcanic cone and eyewitness accounts suggest that an explosion may have accompanied the catastrophic collapse. Our findings demonstrate that volcanic sector collapses may transform from slow gravitational deformation to catastrophic collapse. Understanding the processes involved in such a transformation is crucial for assessing the hazard potential of other volcanoes with slowly deforming flanks such as Mt. Etna or Kilauea.

Description: Highlights • Compilation of rifting events in the Neoproterozoic • Analysis of continental arc, continental rift and connectedness of continental lithosphere for the last 1 Ga • Two stage supercontinent cycle may better explain changes in the connectedness of continental lithosphere • Extraversion and introversion models of successive supercontinents occur on different timescales Abstract The extent of continental rifts and subduction zones through deep geological time provides insights into the mechanisms behind supercontinent cycles and the long term evolution of the mantle. However, previous compilations have stopped short of mapping the locations of rifts and subduction zones continuously since the Neoproterozoic and within a self-consistent plate kinematic framework. Using recently published plate models with continuously closing boundaries for the Neoproterozoic and Phanerozoic, we estimate how rift and peri-continental subduction length vary from 1 Ga to present and test hypotheses pertaining to the supercontinent cycle and supercontinent breakup. We extract measures of continental perimeter-to-area ratio as a proxy for the existence of a supercontinent, where during times of supercontinent existence the perimeter-to-area ratio should be low, and during assembly and dispersal it should be high. The amalgamation of Gondwana is clearly represented by changes in the length of peri-continental subduction and the breakup of Rodinia and Pangea by changes in rift lengths. The assembly of Pangea is not clearly defined using plate boundary lengths, likely because its formation resulted from the collision of only two large continents. Instead the assembly of Gondwana (ca. 520 Ma) marks the most prominent change in arc length and perimeter-to-area ratio during the last billion years suggesting that Gondwana during the Early Palaeozoic could explicitly be considered part of a Phanerozoic supercontinent. Consequently, the traditional understanding of the supercontinent cycle, in terms of supercontinent existence for short periods of time before dispersal and re-accretion, may be inadequate to fully describe the cycle. Instead, either a two-stage supercontinent cycle could be a more appropriate concept, or alternatively the time period of 1 to 0 Ga has to be considered as being dominated by supercontinent existence, with brief periods of dispersal and amalgamation.