Construct a regionalized compositional data container to be used for Gaussian-based geostatistics: variogram modelling, cokriging and simulation.
make.gmCompositionalGaussianSpatialModel(
data,
coords = attr(data, "coords"),
V = "ilr",
prefix = NULL,
model = NULL,
beta = model$beta,
formula = model$formula,
ng = NULL,
nmax = ng$nmax,
nmin = ng$nmin,
omax = ng$omax,
maxdist = ng$maxdist,
force = ng$force
)
either a compositions::acomp()
compositional data set, or else a sp::SpatialPointsDataFrame()
containing it
the coordinates of the sampling locations, if no SpatialPointsDataFrame was provided
optionally, a matrix of logcontrasts, or else one of the following strings: "alr", "ilr" or "clr"; to produce a plot of the empirical variogram in the corresponding representation; default to variation-variograms
the desired prefix name for the logratio variables, if this is wished to be forced; otherwise derived from V
a variogram model, of any relevant class
(see formula
) the coefficients of dependence of the mean of the random field, if these are known; e.g. if formula=~1
constant mean,
and the mean is indeed known, beta
would be a compositional mean; seldom used directly
a formula without left-hand-side term, e.g. ~1
or ~Easting+Northing
, specifying what do we know of the
dependence of the mean of the random field; this follows the same ideas than in gstat::gstat()
optional neighborhood information, typically created with KrigingNeighbourhood()
optional, neighborhood description: maximum number of data points per cokriging system
optional, neighborhood description: minimum number of data points per cokriging system
optional, neighborhood description: maximum number of data points per cokriging system per quadrant/octant
optional, neighborhood description: maximum radius of the search neighborhood
optional logical, neighborhood description: if not nmin
points are found inside maxdist
radius,
keep searching. This and all preceding arguments for neighborhood definition are borrowed from gstat::gstat()
A "gmSpatialModel" object with all information provided appropriately structured. See '>gmSpatialModel.
SequentialSimulation()
, TurningBands()
or CholeskyDecomposition()
for specifying the exact
simulation method and its parameters, predict.gmSpatialModel()
for running predictions or simulations
Other gmSpatialModel:
as.gmSpatialModel()
,
gmSpatialModel-class
,
make.gmCompositionalMPSSpatialModel()
,
make.gmMultivariateGaussianSpatialModel()
,
predict.gmSpatialModel()
# NOT RUN {
data("jura", package="gstat")
X = jura.pred[1:20,1:2]
Zc = compositions::acomp(jura.pred[1:20,7:13])
make.gmCompositionalGaussianSpatialModel(data=Zc, coords=X, V="alr")
# }
Run the code above in your browser using DataLab