This thesis presents the development and application of a numerical inverse and forward model
of stratigraphy applied to shallow-marine wave-dominated sedimentary systems. The approach
links a “process-based” forward model of stratigraphy (i.e. BARSIM, developed by J.E.A.
Storms, University of Delft) to a fully non-linear stochastic inverse scheme. The inverse
problem has been formulated using a Bayesian framework in order to sample the full range of
uncertainty and explicitly build in prior knowledge. The methodology combines Reversible
Jump Markov chain Monte Carlo and Simulated Tempering algorithms which are able to deal
with variable dimensional inverse problems and multi-modal posterior probability
distributions, respectively. The numerical scheme requires the construction of a likelihood
function to quantify the agreement between simulated and observed data (e.g. sediment ages
and thicknesses, grain-size distributions).
Prior to real case study applications, the method has been successfully validated on different
scenarios built from synthetic data, in which the impact of data distribution, quantity and
quality on the uncertainty of the inferred environmental parameters were investigated. The
numerical scheme has then been applied to two case studies: the outcrop-constrained Lower
Cretaceous “Standardville” parasequence of the Aberdeen Member of the Blackhawk
Formation (Boock Cliffs, Utah, U.S.A.) and the Emsian sub-surface data of South Algeria. The
inverse modelling scheme successfully reproduced stratigraphic architecture in both cases,
within the constraints of the input data quality. The inferences of the relative sea level,
sediment supply and wave regime histories contribute to the understanding of mechanisms that
produced the observed stratigraphy. Of equal importance, the inverse results allowed complete
characterisation of uncertainties in these forcing parameters and in the stratigraphic
architecture developed in between data constraints. These results suggest that the inverse
model may ultimately provide a process-based geological complement to standard
geostatistical tools for the static characterization of hydrocarbon reservoirs.