Virtues and vices of turbidite bed thickness statistics : data sets from four confined to semi-confined basins of Central-Northern Apennines (Italy) compared and contrasted

The variability in turbidite bed thickness from a wellbore along with the (unknown) bed shape are important factors in modelling hydrocarbon reservoirs deposited in deep water. Over the last two decades, a number of paper has shown how parameters of statistical distribution of turbidite bed thickness can be useful for discriminating depositional settings, different rheology of parent flows as well as the effect of basin topography on sedimentation. Statistical distributions in the literature are diverse (truncated Gaussian, lognormal, exponential and power law) and variously interpreted as resulting from imprinting of depositional/erosional processes on the input signal (i.e. distribution of parent flow magnitude). However, other than the long time known ‘thin bed problem’, in outcrop studies there is no assessment of how different data collection methods affect statistics of turbidite bed thickness, which instead is crucial for avoiding misinterpretations. Critical factors include: i) minimum significant number of beds/stratigraphic thickness with reference to type (e.g. channel-levee, depositional lobe etc.) and hierarchy of the investigated turbidite body; ii) single vs. multiple correlative sections; iii) position of the sampling line/area with respect to topographic features (e.g. slope break, bounding slopes in confined situations etc.) of the basin. Four different data sets from as many turbidite units (namely, the Cengio-Castelnuovo of Tertiary Piedmont Basin and the Marnoso Arenacea, Laga and Cellino Fm. of the Apennine Foreland Basin system) from Central-Northern Apennines (Italy) are compared and contrasted in this study. In agreement with current models in the literature, examples of influence from primary depositional processes and basin confinement on bed thickness statistics are presented and reviewed critically. To do this, we break down the populations of each case study into sub-categories tied to depositional processes (e.g. bed types) and sub-data sets with different degree of flow confinement and location with respect to confining topographic feature. Furthermore, we explore the bias of different data collection procedures on statistical distribution of bed thickness by comparing distributions from single and multiple correlative logs as well as from units of different hierarchical scale and internal architecture. Lastly, we address the universal applicability of any type of statistical distribution to thickness of turbidite beds.