Application of Seismic Data and Sequence Stratigraphy for Constraining a Stochastic Model of Calcite Cementation

Abstract Calcite cemented sandstone layers are common heterogeneities in shallow marine reservoirs where they cause significant flow barriers. Field analogue studies indicate that the most laterally persistent cemented layers are associated with sequence stratigraphic bounding surfaces where the cementation forms a complex network of beds concentrated around both maximum flooding surfaces and sequence boundaries. Although individual calcite cemented beds are thin, they are associated with very high acoustic impedances and concentration of these thin, high impedance beds around bounding surfaces produces a significant seismic response which can be used to provide information on the spatial distribution of flow barriers. This information along with a sequence stratigraphic interpretation of the reservoir and geometrical data extracted from field analogues provides the input for an integrated reservoir description of calcite cemented barriers using a stochastic modelling approach. The stochastic model is a non-stationary indicator field with an external trend in the indicator proportions. Variograms are used to control the geometry of individual cemented beds. Sequence stratigraphy and seismic data are used to define spatial trends in the proportion of cemented beds; sequence stratigraphy provides the vertical constraints, and seismic data provide lateral constraints. The modelling approach has been tested using data from the TOGI area of the Troll gas field in the Norwegian North Sea. A controlled (verifiable) test has been carried out using synthetic seismic data based on a realistic, heterogeneous, 3D acoustic model of TOGI. These data have been inverted using a constrained sparse-spike seismic inversion technique. The test demonstrated that the prototype modelling procedure functioned satisfactorily. The integration of data from several sources contributed to the generation of realistic heterogeneity models constrained by a maximum amount of information. A test using real seismic data from TOGI has been initiated. The results of the sparse-spike inversion are similar to those obtained in the synthetic test. These inverted data are presently being depth converted and prepared for use in stochastic heterogeneity modelling.