Three-dimensional Reduced-Complexity Simulation of Fluvio-Deltaic Clastic Stratigraphy

A novel reduced-complexity approach to 3D forward modeling of siliciclastic stratigraphy is presented for the simulation of erosion, transport, and sedimentation in continental, transitional, and marine depositional domains. The numerical model is based on defining centerlines that connect sediment input points to the shoreline. For each centerline, erosional and depositional surfaces bound depositional domains, and sand and mud proportions are assigned to each domain. The position of each depositional surface follows a set of geologic rules and ensures mass balance with sediment input. The numerical model is tested by simulating the basin-fill architecture of the XES02 laboratory experiment run at the University of Minnesota, which generates stratigraphy mimicking a passive-margin basin fill. Automatic calibration is used to test multiple combinations of uncertain model input parameters to find those that produce scenarios consistent with the experimental stratigraphy. Calibrated models accurately reproduce shoreline and mass-balance centroid migration, marine sediment proportions, and shoreline trajectories measured in the XES02 experiment. The models also provide reasonable approximations of sand distribution. Of interest, falling shoreline trajectories in the experiment and the calibrated models develop coeval to topset aggradation or topset incision depending on rate of base-level fall. The results reported herein validate the numerical approach for simulating sand distribution in an experimental basin, representing a first step towards the application of the numerical model in fluvio-deltaic settings. For field applications, analogue data can be used to independently constrain the geometry of surfaces bounding depositional domains and their composition. The simplicity of the numerical model then enables multiple realizations by quickly varying uncertain boundary conditions, allowing probabilistic predictions in settings with limited data constraints.