Lithology and fluid prediction from amplitude versus offset (AVO) seismic data

Abstract

AbstractSeismic reflection data as used in the oil industry is acquired and processed as multitrace data with source‐receiver offsets from a few hundred metres (short offset) to several kilometres (long offset). This set of data is referred to as ‘pre‐stack’. The traces are processed by velocity analysis, migration and stacking to yield a data volume of traces with ‘zero‐offset’. The signal‐to‐noise enhancement resulting from this approach is very significant. However, reflection amplitude changes in the pre‐stack domain may also be analysed to yield enhanced rock physics parameter estimates. Pre‐stack seismic data is widely used to predict lithology, reservoir quality and fluid distribution in exploration and production studies. Amplitude versus offset (AVO) data, especially anomalous signals, have been used for decades as indicators of hydrocarbon saturation and favourable reservoir development. Recently, enhanced quantification of these types of measurement, using seismic inversion techniques in the pre‐stack domain, have significantly enhanced the utility of such measurements. Using these techniques, for example, probability of the occurrence of hydrocarbons throughout the seismic data can be estimated, and as a consequence the many pre‐stack volumes acquired in a three‐dimensional (3D) can be survey, reduced to a single, more interpretable volume. The possibilities of 4D time lapse observation extend the measurements to changes in fluid content (and pressure) with time, and with obvious benefits in establishing the accuracy of dynamic reservoir models and improvements in field development planning. As an illustration, recent results from the Nelson Field (UK North Sea), are presented where we show the method by which probability volumes for oil sands may be calculated. The oil–sand probability volumes for three 3D seismic datasets acquired in 1990, 1997 and 2000 are compared and production effects in these data are demonstrated.