Multi-scale simulation of fluvio-deltaic and shallow marine stratigraphy

We present a parameterization of fluvio-deltaic drainage network evolution and alluvial architecture in a basin-scale 2-DH model. The model setup is capable of producing convergent and divergent channel networks. Major elements are the alluvial-ridge aggradation and the coupled overbank deposition, the dimension and style of the channel belt and the sub-grid stratigraphic expression. Avulsions are allowed to develop out of randomly instigated crevasses. Channel stability is modelled one dimensionally by calculating the flow and sediment transport at prospective avulsion nodes. The ultimate fate of crevasses (failed avulsion, successful avulsion, stable bifurcation) depends on the ratio of cross-valley and in-channel gradients in the local neighbourhood of the grid cell under consideration and on the amount and distribution of the suspended sediment load in the water column. The sub-grid parameterization yields implicit knowledge of the alluvial architecture, which may be analysed stochastically. Stochastic realisations of the alluvial architecture allow us to investigate the relationship between basin-fill architecture and small-scale alluvial architecture, which is likely to improve geological reservoir modelling of these notoriously complex deposits. Modelling results under conditions of time-invariant forcing indicate significant quasi-cyclic autogenic behaviour of the fluvio-deltaic system. Changes in avulsion frequency are correlated with the number and length of distributary channels, which are in turn related to alternating phases of progradational and aggradational delta development. The resulting parasequences may be difficult to distinguish from their allogenically induced counterparts. Introduction In numerical models on the continental-margin scale, sediment transport in fluviodeltaic channel networks is typically calculated by the combination of a method for routing water across the grid to determine local discharge, and a sediment-transport routine which calculates how much sediment is actually transported (Howard, 1994; Granjeon & Joseph, 1999; Clevis et al., 2003). The channel network itself is not resolved, and consequently, such large-scale models cannot be directly compared to