Quantitative Seismic Amplitude Analysis

Abstract

The Seismic Value Chain quantifies the cyclic interaction between seismic acquisition, imaging and reservoir characterization. Modern seismic innovation to address the global imbalance in hydrocarbon supply and demand requires such cyclic interaction of both feed-forward and feed-back processes. Currently, the seismic value chain paradigm is in a feed-forward mode. Modern seismic data now have the potential to yield the best images in terms of spatial resolution, amplitude accuracy, and increased illumination in terms of offset and azimuth. Today’s challenge lies with reservoir characterisation. An immediate requirement is extracting quantitative rock properties information from these improved data-sets and images, to move from a, geophysically based, smooth elastic characterisation of reservoirs towards a geologically accessible, blocky layer-based rock properties parameterisation. Currently reservoir characterization does not fully exploit the wide-angle information present in seismic data. This is primarily due to the fact that the current paradigm for analysing reflectivities is at odds with the assumptions made for standard seismic data processing. The current practice of using single interface models to calculate reflectivities in a long-offset layered earth is inconsistent with the assumed time invariant convolution data model for seismic. The interplay of the modelling and inversion enables a better seismic characterisation of the reservoir by moving away from traditional band-limited, smooth, elastic attributes and towards obtaining high resolution, blocky, rock properties that correlate better to well measurements. A new layered earth forward model is developed that preserves linearity at large ray-parameter and handles kinematic wave-field effects at their proper scale. This full linearisation of the elastic property contrasts for successive layers means partitioning the compressional wave and shear wave velocity fields into two fundamental scales: a kinematic scale that governs wave-field propagation effects and a dynamic scale that governs wave-field scattering effects. The proposed layered-earth forward model recognizes the physics of seismic wave propagation, in addition to wave-field scattering, and allows for a more complete exploitation of the information available in the pre-critical seismic amplitudes. Following the extension to a layered earth, in so far as to be physically meaningful, a second step in quantitative seismic amplitude analysis is taken. Leveraging the knowledge gained from the forward analysis, a methodology to extract quantitative layer properties from the acquired seismic data is developed and creates a framework which furthers the use of seismic data in quantitative hydrocarbon reservoir characterization and management. Data kernels built with these new forward models combines with the imposition of non-quadratic regularisation (in the vertical direction) on the least-squares solution to deliver reliable broadband reservoir rock properties estimates from pre-stack seismic amplitudes. The analysis goes on to show that converted wave inversion delivers better estimates of parameters that suffer in the compressional wave inversion alone. Furthermore, considering both wave-modes simultaneously improves both the problem conditioning and the parameter estimates. The final thrust of the inversion analysis takes the theoretical inversion development and applies it to create a novel and practical method to infer sparse, high-resolution/well-resolved, rock properties from pre-stack compressional wave seismic amplitude data. Evolving from current industry standard techniques for linear inversion, the method is an extension of the solution to the classical damped least-squares problem. The final deliverable is a pre-conditioned conjugate gradient algorithm that performs broad-band, minimum structure, least-squares inversion of pre-stack seismic field data. The outcome of the inversion demonstrates that the proposed method successfully identifies the reservoir and its properties at a significantly higher resolution than is currently available with most standard, commercial, techniques and that the results are more geologically plausible. The overall message is that sparse, well-resolved, seismic reservoir characterization is possible from field acquired pre-stack seismic amplitude data.