Page 80 - MSc_thesis_R A Kil

bodies is the result. Very small isolated patches are not representative for the depositional environment,
with exception of few scour fills. Furthermore, the variograms were not reliable enough to use the SIS
algorithm with a high degree of confidence. To conclude, the SIS method has not been used to model the
Favignana calcarenite.

7.6.3 Truncated Gaussian simulation

Like the SIS method, the truncated Gaussian simulation (TGS) uses an algorithm provided by the GSLIB.
The same input parameters can be given. The difference between the two methods is the handling
of a transition between facies. TGS handles this in a more natural way, making it suitable for e.g.
carbonate depositional systems and shallow marine environments. The algorithm itself is based on a
normal distribution of the facies data.
The TGS model (figure D.8) shows little improvement when compared to the SIS model discussed before.
The amount of isolated patches is reduced due to the introductions of more natural transition between
facies, but there is still no option to include information that is geologically meaningful. Despite the
transition algorithm, especially at the boundary of two main facies regions still noisy grid cells with
mainly tabular and through cross-lamination exist. Again, TGS appears not to be a suitable algorithm
for a model of a fieldwork area.

7.6.4 Indicator kriging

Indicator kriging is the only deterministic facies modelling algorithm in Petrel R . The algorithm discussed
previously are all based on stochastic methods, which use a random seed for interpolation. Indicator
kriging is a way to use a kriging method on discrete properties like facies. Main input for this algorithm
is the variogram, introducing weaknesses like sensitivity to dominant facies, few data points, or bad
variogram quality.

61