The key assumption of flat horizontal layers does not honor the complexity often present in the geometry of subsurface layers in areas of great oil exploration or production interest. The survey designer usually ignores this discrepancy, however, partly because of mistrust of the available subsurface information and partly because of fear that exploiting that information may lead to ineffective logistics or may bias the results. Maintaining the assumption of flat layers is thus seen as a way to streamline the design process and to guarantee that the design is conservative with respect to our possibly inaccurate knowledge of the subsurface.
Survey designers often choose the source-receiver geometry from among the few standard geometries available (parallel, orthogonal, slanted, zig-zag) on the basis of uniformity of offset and azimuth in the subsurface bins Galbraith (1994). Wavefield sampling Vermeer (1998) may also be a consideration. In some cases, a 3-D subsurface model obtained from existing 2-D or 3-D data, well logs or geological plausibility is used to compute illumination maps of the reflectors of interest via forward modeling with various candidate geometries Cain et al. (1998); Carcione et al. (1999). The geometry that provides the least distortion in illumination is chosen as the best design, after maybe tweaking it manually to fine tune its illumination response.
The methodology I propose for the optimization of the survey design poses the selection of the survey parameters as an optimization process that allows the parameters to vary spatially in response to changes in the subsurface. In a previous report Alvarez (2002b) I described the basic idea behind the method and illustrated it with a very simple 2-D synthetic model. In this paper I will illustrate the method in 3-D using a subsurface model I created to simulate a land survey. The emphasis will be on the description of the inversion to compute the parameters. I will show that a standard acquisition geometry will either sacrifice the offset coverage of the shallow part of the target horizon or require a large number of sources, which negatively impacts the cost of the survey. For the sake of simplicity, target depth was the only parameter I allowed to influence the spatial change of the geometry. Three geometries were computed according to the depth of the target reflector. Each geometry was locally optimized for uniformity of subsurface illumination. The optimum geometry, being more flexible, relaxes the acquisition effort without compromising the shallow part of the target reflector.