SGSim Input File

To invoke SGSim, the two keywords # parameter group and # split units have to be invoked in the Input File.

The name of the SGSim input file is specified in # parameter group and could be for example

 sgsim1.par

The necessary form of the file is as follows (change only the input values at the beginning of the lines):

 Parameters for SGSIM
 ********************
 
 START OF PARAMETERS:\
 well_frac_data.gslib           \file with data
 1  2  3  4  0  0              \  columns for X,Y,Z,vr,wt,sec.var.
 -1.0e21       1.0e21             \  trimming limits
 1                             \transform the data (0=no, 1=yes)
 sgsim.trn                     \  file for output trans table
 1                             \  consider ref. dist (0=no, 1=yes)
 logperm.dat                \  file with ref. dist distribution
 1  0                          \  columns for vr and wt
 -14.3    -11.7                   \  zmin,zmax(tail extrapolation)
 1       -14.3                   \  lower tail option, parameter
 1       -11.7                   \  upper tail option, parameter
 0                             \debugging level: 0,1,2,3
 sgsim.dbg                     \file for debugging output
 sgsim.out                     \file for simulation output
 1                            \number of realizations to generate
 20     0.0     50.0             \nx,xmn,xsiz
 20     0.0     50.0             \ny,ymn,ysiz
 1      0.0     25.0              \nz,zmn,zsiz
 22222                         \random number seed\
 0     12                       \min and max original data for sim
 12                            \number of simulated nodes to use
 1                             \assign data to nodes (0=no, 1=yes)
 0     3                       \multiple grid search (0=no, 1=yes),num
 0                             \maximum data per octant (0=not used)
 500.0  500.0  20.0              \maximum search radii (hmax,hmin,vert)
 0.0   0.0   0.0              \angles for search ellipsoid
 1     0.60   1.0              \ktype: 0=SK,1=OK,2=LVM,3=EXDR,4=COLC
 ../data/ydata.dat             \  file with LVM, EXDR, or COLC variable
 4                             \  column for secondary variable
 1    0.00                      \nst, nugget effect
 1    1.00  0.0   0.0   0.0     \it,cc,ang1,ang2,ang3
 500.0  500.0  10.0     \a_hmax, a_hmin, a_vert

Detailed documentation can be found in the book "Geostatistical software library and user's guide" by Deutsch and Journel.

Reference distribution

The reference distribution is saved in a file called for example logperm.dat, specified in the seventh parameter line of the SGSim Input file.

 Target Histogram
 1
 log permeability
 -1.11538441e+01 
 -1.07167960e+01 
 -1.03286878e+01 
 -8.71080153e+00 
 -9.11528149e+00 
 -1.11749435e+01 
 -8.98253827e+00 
 -1.01372979e+01 
 -1.05315274e+01 
 -1.01426285e+01 
 -1.06024830e+01 
 -1.05620722e+01 
 -9.75515120e+00 
 -9.79548276e+00 
 -9.79140379e+00 
 -1.01642514e+01
 ...

Most important: Leave the first line as it is, the second line should contain only the 1 (this is the number of variables, in our case always just one), the third line is a comment.

SGSIM information

Found at http://www.statios.com/Help/sgsim.html

GSLIB Help Page: SGSIM

Sequential Gaussian simulation program

Parameters

• datafl: the input data in a simplified Geo-EAS formatted file. If this file does not exist then an unconditional simulation will be generated.
• icolx, icoly, icolvr, icolwt and icolsec: the column numbers for the x,y and z coordinates, the variable to be simulated, the declustering weight, and the secondary variable (e.g., for external drift if used). One or two of the coordinate column numbers can be set to zero which indicates that the simulation is 2-D or 1-D. For equal weighting, set icolwt to zero.
• tmin and tmax: all values strictly less than tmin and strictly greater than tmax are ignored.
• itrans: if set to 0 then no transformation will be performed; the variable is assumed already standard normal (the simulation results will also be left unchanged). If itrans=1, transformations are performed.
• transfl: output file for the transformation table if transformation is required (igauss=0).
• ismooth: if set to 0, then the data histogram, possibly with declustering weights is used for transformation, if set to 1, then the data are transformed according to the values in another file (perhaps from histogram smoothing).
• smthfl: file with the values to use for transformation to normal scores (if ismooth is set to 0).
• icolvr and icolwt: columns in smthfl for the variable and the declustering weight (set to 1 and 2 if smthfl is the output from histsmth).
• zmin and zmax the minimum and maximum allowable data values. These are used in the back transformation procedure.
• ltail and ltpar specify the back transformation implementation in the lower tail of the distribution: ltail=1 implements linear interpolation to the lower limit zmin, and ltail=2 implements power model interpolation, with w=ltpar, to the lower limit zmin.
• The middle class interpolation is linear.
• utail and utpar specify the back transformation implementation in the upper tail of the distribution: utail=1 implements linear interpolation to the upper limit zmax, utail=2 implements power model interpolation, with w=utpar, to the upper limit zmax, and utail=4 implements hyperbolic model extrapolation with w=utpar. The hyperbolic tail extrapolation is limited by zmax.
• idbg: an integer debugging level between 0 and 3. The larger the debugging level the more information written out.
• dbgfl: the file for the debugging output.
• outfl: the output grid is written to this file. The output file will contain the results, cycling fastest on x then y then z then simulation by simulation.
• nsim: the number of simulations to generate.
• nx, xmn, xsiz: definition of the grid system (x axis).
• ny, ymn, ysiz: definition of the grid system (y axis).
• nz, zmn, zsiz: definition of the grid system (z axis).
• seed: random number seed (a large odd integer).
• ndmin and ndmax: the minimum and maximum number of original data that should be used to simulate a grid node. If there are fewer than ndmin data points the node is not simulated.
• ncnode: the maximum number of previously simulated nodes to use for the simulation of another node.
• sstrat: if set to 0, the data and previously simulated grid nodes are searched separately: the data are searched with a super block search and the previously simulated nodes are searched with a spiral search (see section II.4). If set to 1, the data are relocated to grid nodes and a spiral search is used and the parameters ndmin and ndmax are not considered.
• multgrid: a multiple grid simulation will be performed if this is set to 1 (otherwise a standard spiral search for previously simulated nodes is considered).
• nmult: the number of multiple grid refinements to consider (used only if multgrid is set to 1).
• noct: the number of original data to use per octant. If this parameter is set less than or equal to 0, then it is not used; otherwise, it overrides the ndmax parameter and the data is partitioned into octants and the closest noct data in each octant is retained for the simulation of a grid node.
• radius_hmax, radius_hmin and radius_vert: the search radii in the maximum horizontal direction, minimum horizontal direction, and vertical direction (see angles below).
• sang1, sang2 and sang3: the angle parameters that describe the orientation of the search ellipsoid. See the discussion on anisotropy specification associated with Figure II.4.
• ktype: the kriging type (0 = simple kriging, 1 = ordinary kriging, 2 = simple kriging with a locally varying mean, 3 = kriging with an external drift, or 4 = collocated cokriging with one secondary variable) used throughout the loop over all nodes. SK is required by theory; only in cases where the number of original data found in the neighborhood is large enough can OK be used without the risk of spreading data values beyond their range of influence
• rho: correlation coefficient to use for collocated cokriging (used only if ktype = 4).
• secfl: the file for the locally varying mean, the external drift variable, or the secondary variable for collocated cokriging (the secondary variable must be gridded at the same resolution as the model being constructed by sgsim).
• nst and c0: the number of semivariogram structures and the isotropic nugget constant.
• For each of the nst nested structures one must define it, the type of structure; cc, the c parameter; ang1,ang2,ang3, the angles defining the geometric anisotropy; aa_hmax, the maximum horizontal range; aa_hmin, the minimum horizontal range; and aa_vert, the vertical range.

Application notes

• This program requires standard normal data and writes standard normal simulated values. Normal score transform and back transform are to be performed outside of this program
• Recall that the power model is not a legitimate model for a multiGaussian phenomenon and it is not allowed in sgsim
• The semivariogram model is that of the normal scores. The kriging variance is directly interpreted as the variance of the conditional distribution; consequently, the nugget constant c0 and c (sill) parameters should add to 1.0.