simpgs {RGeostats}R Documentation

Plurigaussian Conditional Simulations

Description

Plurigaussian conditional simulations

Usage

simpgs(dbin = NA, dbout, dbprop = NA,
       rule = rule.input(), model1 = NA, model2 = NA, 
       neigh=neigh.input(), props = NA, nostat=NA, 
       flag.gaus = FALSE, flag.prop=FALSE, flag.check=FALSE,
       flag.show = FALSE, seed = 232131, nbsimu = 1, nbtuba = 100,
       ngburn = 10, ngiter = 100, ngint=5, percent=0.1,  
       flag.spde=FALSE, triswitch="nqQ", gext=NA, 
       verbose=FALSE, accept.fun = NA, accept.nloop.max = 10,
       accept.verbose=TRUE, radix = "SimPGS", modify.target = db.locmod())

Arguments

dbin

The db-class structure containing the data file. If not provided, non-conditional simulations are performed

dbout

The db-class structure which will contain the simulation outputs

dbprop

The db-class where the proportion variables must be located in the non-stationary case. See details in rule.check.

rule

The rule-class structure which defines the Lithotype rule

model1

The model-class structure containing the Model information for the first underlying gaussian random function. This model must be provided if the first gaussian random function is involved in the Lithotype rule. For more information check check.model.

model2

The model-class structure containing the Model information for the second underlying gaussian random function. This model must be provided if the second gaussian random function is involved in the Lithotype rule. For more information check check.model.

neigh

The neigh-class structure containing the Neighborhood information. Only the Unique or Bench neighborhoods are authorized.

props

Array giving the (constant) proportions of the different facies involved in the Lithotype rule. See details in rule.check.

nostat

List of non-stationary parameters. For details see model.param.define.

flag.gaus

When FALSE, the simulation outcome is coded into facies. When TRUE, the (conditional) simulation is kept in the gaussian scale

flag.prop

When TRUE, the procedure returns the proportion of facies, rather than the simulations

flag.check

When TRUE (and if dbin is defined), the facies of conditioning data is compared to the facies of the closest grid node.

flag.show

When TRUE (and if dbin is defined), the grid node which coincides with a data is represented with the data facies (with a negative sign).

seed

Seed used for the generation of random numbers. When 0, the seed is not initialized.

nbsimu

Number of simulations

nbtuba

Number of turning bands

ngburn

Number of iterations performed for bootstrap

ngiter

Maximum number of iterations for calculating the conditional expectation

ngint

Number of iterations inside the Gibbs sampler iterative algorithm (only for SPDE technology)

percent

Amount of nugget effect added to the model if this initial model only contains Gaussian structures. If set to zero, the initial model is kept unchanged. This amount is defined as a percentage of the global variance.

flag.spde

When TRUE, the simulation is carried out using SPDE technique. Otherwise the Turning Bands method is used.

triswitch

Command line for the internal triangulation step. For more information see meshing.

gext

The 'dbout' may be dilated by gext. This argument designates an array, with its dimension equal to the dimension of the space and which contains the extension in each direction. If not defined, the 'dbout' is not dilated and the simulated results may suffer some edge effect problems.

verbose

Verbose flag

accept.fun

An acceptation function. See details for more information.

accept.nloop.max

Maximum number of iterations before 'nbsimu' acceptable simulation outcomes is reached.

accept.verbose

When TRUE, the activity of the acceptation function is echoed.

radix

Radix of the name given to the simulation outcomes stored in the output Db.

modify.target

Decides whether or not the newly created variables will have their locator defined or not. For more information, see db.locmod.

Value

The data Db where the simulations outcomes have been added.

The acceptation function must have the following argument:

The function must return a list containing the following items:

When an acceptation function is provided, the simulation procedure is iterated until 'nbsimu' accepted simulations are stored. This iterative procedure is iterated until one of the following criteria is fulfilled:

Examples

# Acceptation function: an outcome is valid if the two samples designated #
# by their grid indices ('ranks') belong to the same connected component  #

accept <- function(db,iatt) 
{
                                        # Locations of wells in the grid #
  ranks = c(5021, 5081)

                                        # Perform the Connected components #
  a = db
  a = db.locate(a,iatt,"z")
  a = morpho(a,0.5,1.5,oper="cc")
  cc.rank = a$natt

                                        # Do all wells belong to same CC #
  ok = TRUE
  if (length(unique(a[ranks,cc.rank])) > 1) ok = FALSE
  if (ok) {
    vect = ifelse(a[,cc.rank] != a[ranks[1],cc.rank],0,1)
    db[,iatt] = ifelse(is.na(vect),0,vect)
  }
  res = list(ok=ok, db=db)
  res
}

# Perform a conditional simulation of facies #
nbsimu = 25
model  = model.create(vartype="Cubic",range=30)
rule   = rule.create(c("S","F1","F2"))
x1     = c(20,80,50)
x2     = c(50,50,50)
z1     = c(1,1,2)
data=db.create(x1=x1,x2=x2,z1=z1)
neigh = neigh.create(type=0)

grid = db.create(nx=c(100,100))
grid = simpgs(data,grid,rule=rule,model1=model,neigh=neigh,
       accept.fun=accept,nbsimu=nbsimu,nbtuba=1000)
grid = db.locerase(grid,"z")
grid = db.compare(grid,names="SimPGS*",fun="mean")
plot(grid,title="Connectivity probability",pos.legend=1)
plot(data,add=TRUE,name.post=1,pch=19,col=c("yellow","red")[z1])

rm(grid,model,rule,data,neigh,nbsimu,accept,x1,x2,z1)
[Package RGeostats version 14.0.10 Index]