Deep-water siliciclastic systems are of great societal relevance as their deposits in subsurface store hundreds of billions of barrels of hydrocarbons in different sedimentary basins worldwide. The sediment density flows that form these systems, termed turbidity currents, are capable of destroying subsea installations, such as telecommunication cables, due to their high volume and velocity at the ocean bottom. Therefore, understanding what controls these systems is important in order to make predictions of their occurrence and behaviour. Because recent deep-water systems are hundreds to thousands of meters below sea level, direct observation of the flow processes and deposits produced by them is rarely possible. A few flow measurements have been obtained by instruments installed on the ocean floor, but they are commonly destroyed by the violence of the flows. Seismic and sonar imaging from the sea bottom are some of the tools that permit the identification of deep-water deposits and their characteristics in a relatively low-resolution scale. Therefore, most data are based on descriptions of deposits from exhumed systems or through interpretations of seismic reflection data, which might be calibrated with well log and core data from subsurface systems. From these datasets, hypotheses are usually proposed to explain the formation processes of the deposits. Physical laboratory experiments and process-based numerical models are supplementary study approaches that simulate these sedimentary density flows and their deposits, thereby suggesting ties that relate the flow processes to the deposits. These different approaches provide information at different scales of observation. Comparing and integrating them is a useful way to understand better the controlling processes of these deepwater depositional systems. One of the main controls on the depositional architecture of deep-water systems is the relief of the seafloor over which turbidity currents pass, which influences the location of erosion, bypass, and deposition. Seafloor topography can be complicated, and subtle changes may influence flow behaviour and impact the geometry and distribution of the deposits. Stepped submarine slope profiles (sensu Prather, 2003) are one of the most common types of slope-to-basin profiles observed in modern and ancient deep-water systems. Many studies of stepped-slope systems are based on seismic reflection data and a few based on outcrop observations, which focus on the description of the geometries of the deposits. However, no studies on this subject are reported in literature that uses numerical simulations. In this thesis, two natural systems deposited on a stepped-slope were analysed. The first from outcrops of the stratigraphic Units D and E of the Laingsburg-Karoo Basin (South Africa), where it is possible to observe depositional patterns across scale from beds to the depositional architecture in each system. The second case is from a subsurface reservoir system offshore the Campos Basin (Brazil), where the depositional architecture was interpreted through 3D seismic reflection data, integrated with well log and core data. To complement these studies, numerical simulations of multiple flow events using the FanBuilder software (Groenenberg, 2007) were performed using five different bathymetries, varying their slope gradients in one set of simulations, and the degree of confinement in the other. The results of these three data sources were compared with respect to their sand distribution, their geometries, the vertical and lateral stacking patterns, and the connectivity of the sand bodies. Based on these comparisons, conclusions were drawn regarding the geological controls involved in the sediment distribution, from local controls (autogenic processes) to external ones (allogenic processes). The main results of the present study are as follows: • Intraslope lobes are more channelized than basin-floor lobes and their stacking patterns are less influenced by changes in slope gradient; • Their sediment depocentre extends further on the intraslope step with increasing slope gradients; • Lobe compensation is scale-dependent and the shifts between depositional units increase in distance from bed to lobe complex scale; • The controls of the patterns in lateral shifts are mostly related to autogenic processes at the smaller scale and tend to be more allogenically controlled at larger scales; • Slope confinement plays a stronger role than slope gradient in determining whether a system is sand-attached or sand-detached; • The spatial and temporal change from a sand-attached to a sand-detached system depends on allogenic controls such as tectonics; • The opposite change is likely to occur only through depositional processes; • Stepped-slope systems differ from mini-basin fills in several aspects, such as the mud content and the depositional architecture; • The slope gradients control deposition on stepped-slope systems, while in mini-basins this is primarily the slope confinement.@en

The depositional architectures of deep-water turbiditic deposits are strongly influenced by seafloor topography. Slope gradient variations of less than one degree might be sufficient to change the distribution of sands significantly along the basin profile. Stratigraphic units of deep-water sandstones from the Fort Brown Fm. in the Laingsburg depocentre (Karoo Basin, South Africa) are an example of that. Regional mapping and stratigraphic correlation of Units C to F (Van der Merwe et al., 2014) show a change from sand-attached systems in Units C and D to sand-detached systems in Units E and F. The sand-attached systems show a continuity of sands from entrenched slope valleys to basin-floor lobe complexes, while in the sand-detached systems there are widespread sand bypass zones of approximately 10 to 30 km where almost no sand is deposited and erosive features are observed. This is interpreted to reflect the development of a stepped slope profile. Lobe deposits occur before and after the bypass region, but significant differences in depositional architecture are noticed between these lobe deposits. The intraslope lobes are characterized by an aggradational to compensational stacking pattern and a common occurrence of erosive features, while the basin floor lobes show a lateral compensating stacking pattern with less erosive features. In this study, process-based numerical modelling of turbidity currents are performed to test if a stepped slope to basin profile with subtle gradient changes similar to that interpreted for the Laingsburg depocentre during the deposition of Unit E are suitable to generate the sediment distribution pattern observed in the field. Through an iterative modelling workflow we aim to constrain the paleoslope gradient changes using the parameters constrained from outcrop. The study also investigates how flow parameters such as sediment concentration, flow velocity, flow thickness and Froude number behave as a function of different slope gradients. The first promising results of these simulations are presented. @en