ISSN:
1365-2478
Source:
Blackwell Publishing Journal Backfiles 1879-2005
Topics:
Geosciences
,
Physics
Notes:
The theme of the 2003 EAGE/SEG imaging workshop concerned the contrast between different philosophies of ‘model building’: whether an explicit, user-determined model should be imposed throughout the processing, with user updates at each step; or alternatively, whether user intervention should be kept to a minimum so as to avoid preconceived bias, and instead to allow the data itself to guide some heuristic process to converge to an optimal solution.Here we consider a North Sea study where our initial approach was to build the subsurface model using interpreted horizons as a guide to the velocity update. This is common practice in the North Sea, where the geology ‘lends itself’ to a layer-based model representation. In other words, we encourage preconceived bias, as we consider it to be a meaningful geological constraint on the solution.However, in this instance we had a thick chalk sequence, wherein the vertical compaction gradient changed subtly, in a way not readily discernible from the seismic reflection data. As a consequence, imposing the explicit top and bottom chalk horizons, with an intervening vertical compaction gradient (of the form v(x, y, z) =v0(x, y) +k(x, y).z), led to a misrepresentation of the subsurface.To address this issue, a gridded model building approach was also tried. This relied on dense continuous automatic picking of residual moveout in common-reflection point gathers at each iteration of the model update, followed by gridded tomography, resulting in a smoothly varying velocity field which was able to reveal the underlying local changes within the chalk.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1111/j.1365-2478.2004.00441.x
Permalink