GNGTS 2018 - 37° Convegno Nazionale

748 GNGTS 2018 S essione 3.3 CORRELATION STRUCTURE OF SUBMARINE LANDSLIDE DEPOSITS FROM SEISMIC REFLECTION IMAGES J. Ford, P. Cance, A. Camerlenghi Istituto Nazionale di Oceanografia e di Geofisica Sperimentale - OGS, Trieste, Italy Extended summary. We propose a method to extract subsurface correlation lengths from “chaotic” seismic reflection images of submarine landslide deposits. The method inverts for the second-order image statistics (2-D autocorrelation function) of the chaotic zone by forward modelling an ensemble of stochastic models of slide heterogeneity. This builds on previous work in the hydrological community to characterise scale lengths in heterogeneous near-surface deposits using ground-penetrating radar reflection images (Irving et al. , 2009). We generalise this technique to better account for uncertainties introduced by a complex overburden, scattering and seismic processing and migration. The geohazard from slope failure is significant but currently poorly understood. Many studies use geophysical images of ancient submarine landslide deposits to inform present day hazard to seafloor infrastructure and to coastal populations from tsunami. Key parameters for such studies are the degree and type of internal deformation from sliding. This helps to constrain slide dynamics – for example slide velocity and the degree of frontal confinement – important for proper numerical modelling of slide runout and tsunami generation. Reflection images of submarine landslides often show a characteristic “chaotic” texture (disordered, incoherent reflections) because the scale of structures inside the slide are around or below the seismic resolution (Chopra and Marfurt, 2016). This is in part due to strong stratal disruption caused by deformation during sliding, which acts in particular to reduce lateral scale lengths with respect to unfailed sediments. The lack of coherent reflections makes traditional seismic interpretation techniques such as horizon picking difficult or impossible (Fig. 1). Here the subsurface velocity structure of submarine landslide deposits is described by two components: 1) a deterministic “background” component, with slowly varying velocity and density and 2) a stochastic component, with relatively small velocity pertubation from the background. We assume that the stochastic component can approximately represent the heterogeneity inside submarine landslide deposits from stratal disruption. For this study we also assume that the heterogeneity approximates a random medium with exponential, anisotropic Fig. 1 - Seismic profile extracted from a 3-D seismic volume acquired in Nankai Trough, offshore Japan. This profile intersects an exceptionally thick submarine landslide deposit (Strasser et al. , 2011). The “chaotic” seismic texture of the slide deposits prevents confident interpretation of internal structure by horizon picking.