oa Co-Krigged Porosity Modeling Exhibits Better Results than Conventional Regression Analysis and Multiattribute Transform Porosity Models

Reservoir heterogeneity characterization is always a real challenge for the sub-surface professionals.<br>Although there is no direct way to assess the true heterogeneity, still certain models can imitate the<br>important features of variability. The spatial distribution of reservoir properties can be determined by<br>stepping through a workflow which starts where standard workstation seismic and geologic<br>interpretation ends. In order to obtain the most accurate and detailed results, one must design a<br>multidisciplinary workflow that quantitatively integrates all the relevant sub surface data. This paper<br>demonstrates the enhanced results of regression analysis and the multi-attribute transforms which are<br>used for porosity prediction in one of the areas in Middle Indus Basin. The co-kriging method used in<br>geostatistics has been applied to derive a combined effect of both the techniques. The dataset used for<br>this study consists of the available well data including VSP & the petrophysical logs, a 3D seismic<br>volume consisting both reflectivity & Inversion data for attribute extraction. A conventional regression<br>analysis using the single polynomial function incorporating the AI & the well porosities were used to<br>extrapolate the average porosities away from the known control points. We then applied the multiattribute<br>transform using various seismic attributes and the well data. A cross-validation of porosity<br>with the significant seismic attributes was done through neural networking. The results were then<br>applied to derive initial porosity map. Both the results were integrated using co-kriging approach which<br>involved creation & comparison of different variograms to get the enhanced version of porosity model.<br>The co-kriged porosity maps showed a better delineation of good porosity zones as compared to initial<br>porosity maps.