Volcanic island sector collapses have produced some of the most voluminous mass movements on Earth and have the potential to trigger devastating tsunamis. In the marine environment, landslide deposits offshore the flanks of volcanic islands often consist of a mixture of volcanic material and incorporated seafloor sediments. The interaction of the initial volcanic failure and the substrate can be highly complex and have an impact on both the total landslide deposit volume and its emplacement velocity, which are important parameters during tsunami generation and need to be correctly assessed in numerical landslide-tsunami simulations. Here, we present a 2D seismic analysis of two previously unknown, overlapping volcanic landslide deposits north-west of the island of Sakar (Papua New Guinea) in the Bismarck Sea. The deposits are separated by a package of well-stratified sediment. Despite both originating from the same source, with the same broad movement direction, and having similar deposit volumes (~15.5–26 km3), the interaction of these landslides with the seafloor is markedly different. High-resolution seismic reflection data show that the lower, older deposit comprises a proximal, chaotic, volcanic debris avalanche component and a distal, frontally confined component of deformed pre-existing well-bedded seafloor sediment. We infer that deformation of the seafloor sediment unit was caused by interaction of the initial volcanic debris avalanche with the substrate. The deformed sediment unit shows various compressional structures, including thrusting and folding, over a downslope distance of more than 20 km, generating >27% of shortening over a 5 km distance at the deposit's toe. The volume of the deformed sediments is almost the same as the driving debris avalanche deposit. In contrast, the upper, younger landslide deposit does not show evidence for substrate incorporation or deformation. Instead, the landslide is a structurally simpler deposit, formed by a debris avalanche that spread freely along the contemporaneous seafloor (i.e., the top boundary of the intervening sediment unit that now separates this younger landslide from the older deposit). Our observations show that the physical characteristics of the substrate on which a landslide is emplaced control the amount of seafloor incorporation, the potential for secondary seafloor failure, and the total landslide runout far more than the nature of the original slide material or other characteristics of the source region. Our results indicate the importance of accounting for substrate interaction when evaluating submarine landslide deposits, which is often only evident from internal imaging rather than surface morphological features. If substrate incorporation or deformation is extensive, then treating landslide deposits as a single entity substantially overestimates the volume of the initial failure, which is much more important for tsunami generation than secondary sediment failure.
Read full abstract