[Sgems-users] SISIM categorial (problem with assign hard data to simulation grid) More facies than expected

[Jean B. <jea...@st...>]

Dear all SGeMS users,

Here is a very confusing *problem *I met *with *SGeMS and *SISIM 
categorial* (apparently inherent to SGeMS, never mentioned anywhere on 
the web)

What I would like to do : a Sequential Indicator Simulation of 
geological facies (categorial) conditionned upon synthetic data.

The inherent problem : realization of 3 facies instead of 2 as expected 
(*always one more than expected in each case*).

First I simulated two random facies in Matlab (50 data points) in a grid 
100 x 100, unit spacing. Then I run SISIM conditionned upon my hard data 
in SGeMS to get one realization with 2 facies. In property item (Hard 
data parameters) I selected facies data, not indicators (am I right?). 
In interval data, I selected nothing. The problem comes when I select 
'assign hard data to simulation grid' item. I get a realization with 3 
facies instead of 2,i.e. one facies = 0, another one = 1 and the last 
one = 2. The latter is restricted to the data point of facies 2. The 
same problem also happens if I select my two indicators in interval data 
and my facies in property. However, if I select one indicator in 
property instead of facies, I get two facies as expected, but 
conditionning occurs on one indicator only.

*I noticed the same problem in two examples*: one I found*in *a document 
intitled : *"S-GeMS Tutorial Notes", 26 june 2007, Geoff Bohling from 
Kansas Geological Survey*. Here, the author simulates 5 facies, but I 
can see 6 facies in the realization (data and parameters are provided at 
http://people.ku.edu/~gbohling/BoiseGeostat. Here is the document 
http://people.ku.edu/~gbohling/BoiseGeostat/SGeMSTutorial.pdf. Also, *in 
"Geostatistical Earth Modeling Software: User's Manual", Nicolas Remy , 
2004* : the example illustrated at Figure 3.8 using Stanford V data set 
and the same parameters. Again, result seems similar but if we zoom on 
it (in SGeMS), there are more facies than expected.

Any help would be greatly appreciated

Thanks in advance

Regards

Jean Beaujean
Ulg Belgium