Kriging can be used as an interpolation technique to estimate the value of a parameter at unsampled locations. The interpolation depends on the existence of spatial dependency in the space where the parameter has been sampled. This space can be of any finite dimensional size, but is typically restricted to the 3D spatial (XYZ) or 4D spatio-temporal (XYZ-time) space. Here we will use kriging to interpolate between observed continuous data values in multivariate attribute space.
We will show how we make use of this approach to perform kriging interpolation in a model parameter space spanned by seismic attributes. To the extent that the seismic attributes reflects the underground geology the interpolation will be based on similarity of geology rather than distance in spatial space.
In an example from a chalk section of the South Arne Field in the North Sea, we use the method to estimate the distribution of porosity, using seismic attributes as guide. The results are very encouraging as the estimated maps follows known geological features very well. The porosity estimate is supported by subsequent drilling. achieve a high degree of linking to cited sources that appear on the Web.