Inversion of an open-source forward stratigraphic model: a case study from the Bengal Fan

Inverting forward models to constrain output against data is critical for predicting and understanding sedimentary systems. We present an open-source workflow using SciPy functions to invert a Python Badlands model with spatially variable tectonic and rainfall parameters.

Tectonic uplift and rainfall maps are input to model surface uplift, erosion, transport and sediment deposition across the model grid. A SciPy function optimises against an objective function to calculate error and invert the model to find the optimal tectonic and rainfall parameters required to reproduce key aspects of the observed strata.

The workflow has four steps:

1. Collate paleoDEM’s, paleogeographic, thermochronological and climatic data to create initial topography, tectonic uplift and rainfall input maps

2. Run the Badlands forward stratigraphic model to calculate evolving topography and basin-fill strata

3. Use an objective function to calculate error from the comparison of model output and observed stratal properties

4. Use SciPy to iteratively optimise parameters through steps 2 and 3 to minimize model error

This approach generates a useful series of best-fit models for the Bengal Fan source-to-sink system. The nature of the objective function reveals important aspects of the model behaviour and indicates that climatic factors increasing erosion and sediment yields were the main forcing of Miocene deep-water sedimentation in the Bengal Fan.