Distinguished Lecturer, Spring 2010

Over the past two decades, BP has developed workflows to make predictions about lithology and fluid from seismic amplitudes. Their current form, which has stabilized in recent years, is based around a few fairly simply steps: two-term AVO coordinate rotations, spectral shaping derived from a power-law Earth spectrum model, and map-based detuning and calibration. These are underpinned by careful conditioning of the seismic and by close integration with petrophysical rock property analysis.

These processes have evolved to meet a number of criteria: (1) robustness and repeatability, (2) avoiding parallel workflows by integrating within existing practices, and (3) allowing most steps to be carried out by nonspecialist geophysicists.

Achieving robustness is also a prerequisite for making uncertainty estimates, which has many potential benefits: ranking the sources of uncertainty; validating our prediction models; providing a framework for integrating with other data types; and helping to understand the theoretical limits of seismic prediction.

In this talk, I’ll outline the concepts behind the main steps in the seismic characterization workflow and discuss approaches to uncertainty estimation.