Accuracy of interpolation for 3D contour mapping from 2D seismic sections

Abstract

This research examines the accuracy of contour maps sub-sampled from a 3D seismic survey. A 3D seismic dataset was interpreted using LANDMARK?s ``Seisworks-3D' software to build three contour surfaces of stratigraphic horizons at various depths and structural complexities. These horizons were exported and imported to a Geographic Information System (GIS), resampled at various inline and crossline, and point spacings, and interpolated to create 3D surface grids from these subsamples of the horizons to simulate interpolation from 2D seismic lines. In the first set of experiments, for both transect and point data structures, map error decreased as a power function of sample size. This systematic increase in error as sample size decreases allows prediction of the accuracy of interpolation according to sample size and distribution of the data. This relationship facilitates estimation of errors for seismic data interpreters picking a sub-sample of sections for a particular stratigraphic horizon, determination of receiver and line spacings for sufficient survey accuracy for least economic outlay. Another application of this relationship is to determine the accuracy of interpolative contour mapping on a series of parallel 2D seismic lines according to their spacing. In a second set of experiments on the point data of a set sample size, map error increased with the local structural complexity of strata sampled. This relationship would allow prediction of the relative precision expected in areas of varying complexity. These findings corroborate earlier work on topographic maps and indicate that similar trade offs between map accuracy and both sample size and surface complexity apply to geologic blocks.