LESSENGER, MARGARET A., and TIMOTHY A. CROSS, Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO

ABSTRACT: A Stratigraphic Inverse Simulation Model

The new generation of numerical process-response stratigraphic models simulate and predict stratigraphic architecture and facies distributions to varying degrees of accuracy. Matching model output with real-world stratigraphic data often is accomplished by multiple trial-and-error runs of the forward model until the comparison is considered close. Another approach toward determining the magnitudes and frequencies of process variables that operated to produce observed stratigraphy is inversion of real or modeled stratigraphic data.

We are applying geologically constrained nonlinear numerical optimization, using a modified Gauss-Newton technique, to a two-dimensional siliciclastic stratigraphic forward model developed at CSM. The inversion technique minimizes weighted differences between observed and synthetic data, guided by a geologically constrained sensitivity matrix. The forward model calculates distributions of time-significant surfaces and facies between surfaces, and therefore areas and geometries of facies tracts within time-bounded sequences. The model creates accommodation space with tectonic subsidence, eustatic change, and flexural compensation of sediment and water loads. Sediment is then eroded and distributed within this space, constrained by a subaerial and sub sea base level, and a maximum erosi n rate.

Using this forward model, the inverse technique extracts values for the model parameters for eustasy, tectonic subsidence, sediment supply, erosion rates, depositional topography, and lithospheric rigidity from a limited stratigraphic data set. For this inverse model, stratigraphic data consist of thicknesses of three facies tracts (fluvial/coastal plain, shoreface, and shelf tracts) correlated within a time-stratigraphic framework at different geographic positions within a basin.

Results of inversion on simulated data suggest that inversion of stratigraphic data is possible. Convergence towards a true solution is fast, and solution uncertainties for process parameters are less than a priori uncertainties. Inversion results indicate that model responses are nonlinear, even though input equations are linear. Results also indicate that stratigraphic responses are nonunique, and that the nonunique interaction between eustasy, tectonics, sediment supply, and depositional topography may be quantitatively assessed.

AAPG Search and Discovery Article #90987©1993 AAPG Annual Convention, New Orleans, Louisiana, April 25-28, 1993.