Facies Modeling using Membership functions

@inproceedings{LabatRM2003,
 abstract = { Stochastic simulation techniques are increasingly used for modelling spatial distribution of categorical attributes(such as rock types) or continuous attributes (such as porosity or permeability). This procedure consists in generating alternative, equally probable 3D realizations, that mimic the heterogeneity expected in the real media and inferred from the available data. For a couple of years, several works have been carried out to develop new methods for building realistic facies models, such as Indicator Kriging or multipoint geostatistic methods. Others methods based on a P-Field have been also developed to account for spatial variability of facies. Most of them do not take into account trends or transition probabilities between one facies and the others. Using Discrete Smooth Interpolation [Mallet(1989)]), the method proposed in this paper interpolates a membership function and is inspired from P-Field ones. P-Field based method consists in first estimating in 3D a local facies distribution and second to generate a random probability field accounting spatial structures. In the presented work, each local facies distribution will be modelled using one membership function, which corresponds to the probability of a given node to belong to a given facies. The estimation of the membership function can also take into account several constraints, such as vertical or horizontal proportion maps, trends or transition probabilities. It is possible to combine the so defined probabilities with classical stochastic simulators in order to generate equiprobable realizations of the facies distribution in the reservoir. Multi P-Fields methods can be used to account for complex, spatial variability of facies (anisotropy which depends of facies type). },
 author = { Labat, Laurent AND Mallet, Jean-Laurent AND Coureaud, B. },
 booktitle = { 23th gOcad Meeting },
 month = { "june" },
 publisher = { ASGA },
 title = { Facies Modeling using Membership functions },
 year = { 2003 }
}