Classification of 4D seismic data - An overview

Inversion of 4D seismic is a non-unique process. Hence the inversion needs to be constrained by additional information - typical well-logs - to be uni que. Classification is an attractive inversion procedure that allows the us er to adapt the constraints to the problem at hand. An overview of tailor m ade approach to 4D seismic inversion is presented. The repeatability betwee n time lapses are quantified to assess the potential uncertainty of the pro cess. Fluid substitution modelling is performed to establish the relationsh ip between the elastic observables attributes and the desired reservoir flo w parameters. A training step allows the user to adapt the constraints guid ed by the well-logs or the fluid substitution modelling. A novel set of ort hogonal seismic attributes capable to extract all information from thin tim e-windows are use to represent the observations. Classification is applied to the observation by applying the established relationship at the training step to the whole dataset.