Logical depth modeling of a reservoir layer with the minimum available data-integration geostatistical methods and seismic attributes

Abstract

For rational depth modeling of a prominent reservoir layer in north of Iran (Gorgan plain, Chelekan top), geostatistical methods were proposed to use with the minimum available data. This data consisted of ten wells, five 2D seismic lines (three vertical lines perpendicular to two horizontal ones) which covers the area, and one small 3D seismic area, which was applied solely for evaluation of findings and optimizing our choices. Because the expansion of this area was limited as opposed to region aimed for modeling. Hence, for a reasonable geostatistical modeling, an appropriate secondary variable (soft data) was crucial. Initially, the reservoir layer should be pursued in five seismic lines with a suitable seismic attribute and achieved its time model (TWT) all over the Gorgan plain due to existing a few number of lines, linear form of data set (located on the seismic lines) and the smoothing effect of kriging, the estimate and average simulated realizations (E-type) could not give acceptable results in time modeling of the layer based on merely five seismic lines. Therefore, one of 100 realizations related to sequential quassian simulation (SGS) selected as the best secondary data after probing their correlation and similarity with the real 3D seismic data and obtaining a proper correlation coefficient. Moreover, this realization revealed the best correlation with the depth amounts of 10 wells, reproducing geostatistical and statistical parameters of input data. For this reason, it was utilized as secondary data in kriging with an external drift method (KED). Having been applied it, the smoothing effect was diminished dramatically in comparison with one variable model and consequences of final modeling, investigation of uncertainty and estimate error prior to using secondary data and after that, all of them signified the final model was much more reasonable than initial one (without secondary data).