Monte Carlo Inversion Applied to Reaction-Transport Modeling of Methane Hydrate in Continental Margin Sediments

Abstract:

Reaction-transport modeling has been widely used to calculate methane hydrate content in continental margin sediments. Modeling accounts for sediment accumulation and compaction, pore water advection, and diffusion of dissolved methane. Methane is sourced in situ by microbial degradation of organic matter and/or is generated at depth and transported upward by pore water advection. Results are typically calculated as a function of depth only and at steady state. Modeling has successfully reproduced the depth distribution of methane hydrate observed in drill sites and has been applied to estimate the amount of hydrate in the global continental margin reservoir.

A shortcoming of current approaches is that several key input parameters (e.g., sedimentation rate) are typically not well known a priori, while other parameters that control the solution (e.g., the depth distribution of microbial methane generation) are not uniquely determined by fitting observations. This study builds on earlier work by setting up a Monte Carlo inversion where the variability of uncertain model parameters is constrained by prior information and by the requirement of fitting observed methane hydrate contents. Monte Carlo sampling requires a fast reaction-transport modeling routine. The calculation used here obtains solutions in different depth intervals: the unknowns are the dissolved methane concentration where it is less than solubility, and hydrate content (or gas bubble volume fraction below the base of hydrate stability) where methane comes out of solution. The depths of the interfaces between these intervals are determined by requiring an equal mass flux of methane above and below each interface.

Rather than computing a single best-fit solution, the Monte Carlo inversion iteratively samples sets of input parameters that fit the depth distribution of methane hydrate observed at a drill site. This strategy allows for investigating trade-offs between modeling parameters, e.g., between upward methane transport by pore water advection and in situ microbial methane generation. The inversion can also be used to determine the depth distribution of microbial methane generation and to quantify its uncertainty.