Stratigraphic variability due to uncertainty in model boundary conditions: A case-study of the New Jersey Shelf over the last 40,000 years

Abstract

Prediction of stratigraphy by numerical sedimentary models is influenced by a series of geological boundary conditions: paleo-bathymetry, sediment supply, sea level, ocean storm climate, and tectonics. These boundary conditions evolve non-linearly over time and have considerable uncertainty. This study explores the impact of the uncertainty of complex boundary conditions on stratigraphic prediction. Two numerical models are sequentially run: the climate-driven hydrological model, HydroTrend, generates a time series of discharge and sediment load at the river mouth, which then feed into the process-based stratigraphic model, SedFlux-2D. SedFlux disperses sediment in the fluvio-marine domain; the bedload fluxes drive floodplain and mouthbar deposition, while suspended sediment is dispersed in the ocean through hypopycnal plumes. Seafloor sediment is reworked by ocean storms, failures and subsequent transport as sediment gravity flows. SedFlux records the thickness of the deposited sediment and the grain size preserved over time along a longitudinal profile. One ‘base-case’ simulation and 20 sensitivity tests of the New Jersey shallow margin over the last 40,000 yr serve to explore the impact of uncertain boundary conditions. Initially, the shelf is exposed and Arctic conditions prevail in the drainage area. Paleo temperature and precipitation estimates differ significantly between different Community Climate System Model realizations, impacting river sediment flux predictions. Subsequently, high sediment supply from the rivers draining the melting Laurentide Ice Sheet has a dominant impact on the shelf stratigraphy. Drainage basin characteristics are strongly controlled by the hard to define details of the Laurentide Ice Sheet dynamics. Over the last 10,000 yr the shelf becomes sediment-starved and storm-dominated, but the paleo storm conditions are uncertain and thus require a wide range of values. Shallow seismic data have been used to validate the large-scale deposited thicknesses of the SedFlux prediction. The ‘base-case’ SedFlux prediction for the sedimentary architecture of the New Jersey Shelf is consistent with the acoustically observed distribution of major sediment wedges, due to the strong forcing of the relatively well-constrained sea-level changes. The range in thickness predictions over the sensitivity tests is well within the lateral variations reconstructed from the observed seismic data. SedFlux predictions and a data set of 654 seafloor samples taken on the New Jersey Shelf between 50 and 150 m water depth shows a subtle fining towards deeper water. SedFlux systematically overpredicts the component of fine sediments at the seafloor, likely because the model does not predict shell hash. Paleo-storm climate is found to have a high uncertainty and strongly impacts the predicted grain size distributions. We advocate a modeling approach which couples a series of sensitivity experiments to a predicted “base-case” experiment. In that way the stratigraphic variability caused by uncertainty in the boundary conditions is evident for later users.