No description available.

This project investigates the seismic response to facies variation by modelling facies variation itself, using two different modelling techniques, and then by modelling the seismic response to this variation. This study looks at a new set of attributes, examines their potential both as standard seismic attributes (a qualitative approach), and uses geostatistical analysis to further develop the ability of these attributes to differentiate the seismic response to facies variation. Sedpak, a basin modelling package was used to try to create facies models as a basis for further geophysical modelling. A case study using data from the Beryl area was unsuccessful at trying to create facies models due largely to the limited amount of input data and the scale of the models being attempted. Although an impressive package, Sedpak is most useful when modelling at a basin scale. In order to study the seismic response of sedimentary facies variation simple, deterministic models were set up using the geophysical modelling package, Gxii. An established methodology for analysing seismic data is the study of seismic attributes. The study considers some autocorrelation and power spectrum-derived functions previously described in the literature (Sinvhal and Sinvhal, 1992), and treats them as seismic attributes. Initial analysis of these new attributes in 2D showed them to be successful at detecting the presence of channels within seismic data. On the basis of this, a multivariate study was carried out. Results of this analysis show these attributes to have the potential to detect the presence of channels within seismic data. A suite of computer programs were developed to calculate 3D volumes of the new attributes, and to produce colour sections through the attribute volumes. The volumes were granted using a moving time window and calculating attribute values down through the data volume. Evaluation of the colour sections themselves to illustrate facies changes was disappointing. It is apparent that more work is needed to evaluate the window length over which the attributes are calculated.