Most current reservoir-characterization workflows are based on classic amplitude-variation-with-offset (AVO) inversion techniques. Although these methods have generally served us well over the years, here we examine full-waveform inversion (FWI) as an alternative tool for higher-resolution reservoir characterization. An important step in developing reservoir-oriented FWI is the implementation of facies-based rock physics constraints adapted from the classic methods. We show that such constraints can be incorporated into FWI by adding appropriately designed regularization terms to the objective function. The advantages of the proposed algorithm are demonstrated on both isotropic and VTI (transversely isotropic with a vertical symmetry axis) models with pronounced lateral and vertical heterogeneity. The inversion results are explained using the theoretical radiation patterns produced by perturbations in the medium parameters.
The current reservoir characterization workflows are based on ‘classic’ seismic inversion techniques. The term ‘classic’ refers to methods that make use of the amplitude variation of reflected waves with offset or angle (AVO/AVA) to invert for elastic or reservoir properties in deterministic or stochastic fashion (Russell, 1988). Also, these techniques are typically applied to migrated data and often based on 1D convolutional modelling. Nonetheless, such classic AVO-based techniques have been used to obtain geologically plausible models by constraining the inversion using well log data. These types of constraints are often included in the form of linear rock-physics relationships between the elastic properties (e.g., between acoustic impedance and density). This approach, however, inevitably imposes the same linear trend for all facies in the model, which is often violated for different lithologies (e.g., shales and sands). To address this issue, a natural solution (which is more difficult to implement) is to impose rock-physics constraints for each facies type, as shown in Gunning et al. (2013).
With the objective to extend full-waveform inversion to reservoir characterization, we introduced a practical approach to add facies-based rock-physics constraints through regularization terms. The method was tested on synthetic isotropic and anisotropic elastic models with lateral heterogeneity. The results show the benefits of this new approach in resolving fine structural details. Interestingly, including the anisotropy may be an advantage (depending on the availability of tight constraints) in a facies-based approach as it allows more degrees of freedom to classify the facies and constrain the inversion. The algorithm is currently being applied to more complex models and will be tested on field data.
