Sensitivity analysis of a forward stratigraphic model to study carbonate production in the Oligo-Miocene Maldives carbonate platform

Forward stratigraphic models quantify geological processes over time to reproduce the stratigraphic architecture of a sedimentary body. Due to the great number of interacting input parameters and their uncertainties, their individual impact on the stratigraphy can be obscured. A sensitivity analysis of such models can identify key factors in the stratigraphic evolution of real-world examples.

The isolated Maldives carbonate platform offers an ideal case study to disentangle the complex web of stratigraphic controls. The stratigraphic architecture of the western margin is well-documented by seismic and well data, providing a robust framework for model calibration. The platform evolution from the latest Oligocene to the Middle Miocene correlates remarkably well with eustatic sea-level changes, suggesting eustasy as the primary control. Nevertheless, the studied time interval is also notorious for climatic changes affecting carbonate production, and the development of drowning sequences around the globe. Previous modelling efforts have illustrated the great uncertainty of carbonate production rates and its significant impact on stratigraphic architecture. In case of the Maldives carbonate platform, we investigate this uncertainty through automated multi-realizations of a forward stratigraphic model combined with a sensitivity analysis.

Firstly, a reference model was designed with a selection of input parameters that reflect the most relevant geological processes to the studied section. This includes the evolving basin conditions (such as eustasy, subsidence and hydrodynamics), and the production and transport of carbonate sediments. For our model purpose, we subdivided the tropical carbonate factory into smaller-scale sediment classes with distinct biological, ecological and physical properties. Each class is characterized by spatial production constraints that interact with the simulated basin conditions over time. The maximum production potentials of the sediment classes remain poorly constrained and are therefore used as calibration variables. Calibration of the reference model exposed the uncertainty and significant impact of these parameters on the platform stratigraphy.

Subsequently, we conducted a sensitivity analysis on the carbonate production rates to quantitatively assess their individual and combined effect. Due to the different biotic nature of the sediment classes, the carbonate production rates can be assumed to fluctuate independent, parallel or inversely, depending on the root cause of the change. We investigated this further by generating multiple automated realizations for all scenarios. The results enable us to speculate on the cause of the carbonate sediment fluctuations and the nature of associated environmental changes across the studied time interval.