New insight into the evolution of large-volume turbidity currents: comparison of turbidite shape and previous modelling results

The Marnoso Arenacea Formation provides the most extensive correlation of individual flow deposits (beds) yet documented in an ancient turbidite system. These correlations provide unusually detailed constraints on bed shape, which is used to deduce flow evolution and assess the validity of numerical and laboratory models. Bed volumes have an approximately log-normal frequency distribution; a small number of flows dominated sediment supply to this non-channelized basin plain. Turbidite sandstone within small-volume (·7 km3) beds thins downflow in an approximately exponential fashion. This shape is a property of spatially depletive flows, and has been reproduced by previous mathematical models and laboratory experiments. Sandstone intervals in larger-volume (0·7–7 km3) beds have a broad thickness maximum in their proximal part. Grain-size trends within this broad thickness maximum indicate spatially near-uniform flow for distances of ∼30 km, although the flow was temporally unsteady. Previous mathematical models and laboratory experiments have not reproduced this type of deposit shape. This may be because models fail to simulate the way in which near bed sediment concentration tends towards a constant value (saturates) in powerful flows. Alternatively, the discrepancy may be the result of relatively high ratios of flow thickness and sediment settling velocity in submarine flows, together with very gradual changes in sea-floor gradient. Intra-bed erosion, temporally varying discharge, and reworking of suspension fallout as bedload could also help to explain the discrepancy in deposit shape. Most large-volume beds contain an internal erosion surface underlain by inversely graded sandstone, recording waxing and waning flow. It has been inferred previously that these characteristics are diagnostic of turbidites generated by hyperpycnal flood discharge. These turbidites are too voluminous to have been formed by hyperpycnal flows, unless such flows are capable of eroding cubic kilometres of sea-floor sediment. It is more likely that these flows originated from submarine slope failure. Two beds comprise multiple sandstone intervals separated only by turbidite mudstone. These features suggest that the submarine slope failures occurred as either a waxing and waning event, or in a number of stages.