Skip to main content

EOS-exploration oriented seismic modelling and inversion. Wave propagation and scattering in random media


Development of concepts for a realistic parameterization of geological features as random media applicable to reservoir seismology, the investigation of scattering effects for wave propagation in random media and the development of methods of estimating media parameters from the seismic wavefields.

The objective of the project is to extract from seismic data information about the physical properties of a reservoir region which is below the normal limits of seismic resolution. The key step is to determine the most effective parameterizations of sedimentary environments by statistical means. This will be achieved by numerical modelling of random media and by the analysis of field data. It will be necessary to identify the features of seismic waveforms that carry information about the scattering behaviour of the media through which they have travelled, to investigate how discriminatory those features are, and to devise ways of calibrating the resulting interpretation of the features. The study will establish how well attributes estimated from seismic wavefields can be used to distinguish the scattering properties of different target lithologies. The degree of discrimination between lithologies depends upon how much the observed distributions of the attributes overlap. Pattern recognition techniques will e developed to relate observed patterns of seismic character to sub-surface geology. The final product will be a map indicating variation in scattering properties over the imaged locations of the wavefield. Images of the earth parameters will also be obtained by tomographic procedures.

Work on the numerical modelling of the complex structures will include the development of efficient algorithms to reduce the time required for these calculations. Computer codes will be developed for decomposing seismic data into coherent and incoherent components using spatial prediction operators designed from the spectra of the data and Karhunen-Loeve decomposition using a spatially sliding window. Tomographic imaging will include the estimation of two-dimensional images of anisotropy, investigation into the effectiveness of seismic attenuation imaging and the validity of back projecting a variety of seismic attributes. Pattern recognition software will include a supervised classifier of seismic data based on linear discriminants. Measures will be provided of the expected classification errors for the discriminant and of whether the input of further attributes produces a significant reduction in classification error.


South Kensington Campus
United Kingdom

Participants (2)

United Kingdom
Malet Street
WC1E 7HX London
Kaiserstrasse 12
76128 Karlsruhe