2008 Joint Meeting of The Geological Society of America, Soil Science Society of America, American Society of Agronomy, Crop Science Society of America, Gulf Coast Association of Geological Societies with the Gulf Coast Section of SEPM

Paper No. 9
Presentation Time: 3:30 PM

Improving Statistical Predictions of Subseismic Faults

SOLUM, John G., Bellaire Technology Center, Shell International Exploration and Production, 3737 Bellaire Blvd, Houston, TX 77025, NARUK, S.J., Bellaire Technology Center, Shell International E&P, 3737 Bellaire Blvd, Houston, TX 77005 and DULA, W.F., Shell International E&P, Rijswijk, Netherlands, F.Dula@shell.com

Faults that are too small to resolve seismically can dramatically influence subsurface fluid flow by acting as seals or baffles to cross-fault flow. In order to more fully understand the hydrologic properties of a faulted region it is therefore of the utmost importance to be able to characterize the subseismic fault population in that region. This is commonly done by using larger, seismically resolvable fault populations to define a power-law size distribution (i.e., fault size distributions are assumed to be fractal), and assuming that this trends holds true for subseismic fault populations. The point at which the mapped fault population deviates from this trend line marks the point at which the fault population begins to be undersampled, with the degree of undersampling increasing as fault size decreases. However, the fault populations presented in this study do not necessarily follow a power law size distribution at all length scales, rather at some they are better fit by an exponential distribution, which results in far fewer predicted subseismic faults. This deviation occurs when fault growth is affected by mechanical stratigraphy, and since a reservoir interval may be a mechanical unit, this finding has significant implications for defining reservoir compartments. In addition, it appears that predictions of the number and size of subseismic faults is highly survey dependent, which high and low-resolution seismic surveys of the same area yielding dramatically different predictions. This occurs because discrete faults separated by ramps are lumped into a single fault on low-resolution surveys, changing the fractal dimension of the fault distribution that is used to predict the subseismic fault population.