Subsurface modelers normally work with inadequate data. For example, given poor seismic imagery and a single 1D borehole, how can a plausible model of reservoir connectivity within a lobe environment be conditioned? Stochastic approaches are commonly adopted, but do a poor job in honoring geologic complexity. Analog approaches offer potential, but how can we be confident that the chosen analog corresponds to the sub-surface reality? Here we describe a database approach, in which large numbers of documented case studies on both modern and ancient systems are digitized to a common standard and uploaded into a relational database. The database stores information on system controls, architectural hierarchy and facies development. It can be interrogated to assess variability in system expression under different combinations of boundary condition, and the results used to refine subsurface characterization.
The database is still growing, but already demonstrates potential. For example, through incorporation of both outcrop and sub-surface data sets it confirms the linear relationship documented in the literature between submarine channel width and thicknesses, but extends it across multiple orders of magnitude with high confidence. The database can also be used to quantify the geometric relationships between spatially related units. Thus, depending on boundary conditions, levee bodies range from being 1 to 20 times wider than adjacent genetically related channel bodies (average 4.5). Outputs such as the proportion of mud within architectural elements can be evaluated to constrain net-to-gross ranges and their intra-element spatial variability. For example, in sand-rich systems intra channel net-to-gross can be shown to decrease from 0.65 to 0.5 from channel axes to fringes. Lobe architecture at different scales can be constrained based upon evaluation of the likely lateral and dip positioning of a target area, and first order estimates of likely stacking patterns made, based upon appropriately chosen subsets of architectural analogs.
Subsurface data will likely always remain a scarce commodity. But the compound analysis of many analog case studies offers a complementary approach to traditional methods of addressing this shortcoming. Development of more precise system scenarios based upon the interrogation of large data sets is set to become a useful new tool in deep-marine reservoir characterization and associated model conditioning.
AAPG Datapages/Search and Discovery Article #90323 ©2018 AAPG Annual Convention and Exhibition, Salt Lake City, Utah, May 20-23, 2018