plot-method: Methods for function `plot` in package RandomFields

Description

Plot methods are implemented for simulated random fields (objects of class RFsp), explicit covariance models (objects of class RMmodel), empirical variograms (objects of class RFempVariog) and fitted models (objects of class RFfit). The plot methods not described here are described together with the class itself, for instance, RFfit, RFempVariog RMmodel.

Usage

RFplotSimulation(x, y, MARGIN=c(1,2), MARGIN.slices=NULL, n.slices = if (is.null(MARGIN.slices)) 1 else 10, nmax=6,  plot.variance = !is.null(x@.RFparams$has.variance) && x@.RFparams$has.variance, select.variables, zlim, legend=TRUE, MARGIN.movie = NULL, file=NULL, speed = 0.3, height.pixel=300, width.pixel=height.pixel, ..., plotmethod="image")
RFplotSimulation1D(x, y, nmax=6, plot.variance=!is.null(x@.RFparams$has.variance) && x@.RFparams$has.variance, legend=TRUE, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"plot"(x, y, ...)
"persp"(x, ..., zlab="")
"contour"(x, ...)

Arguments

object of class RFsp or RMmodel; in the latter case, x can be any sophisticated model but it must be either stationary or a variogram model

ignored in most methods

MARGIN

vector of two; two integer values giving the coordinate dimensions w.r.t. which the field or the covariance model is to be plotted; in all other directions, the first index is taken

MARGIN.slices

integer value; if $[space-time-dimension>2]$, MARGIN.slices can specify a third dimension w.r.t. which a sequence of slices is plotted. Currently only works for grids.

n.slices

integer value. The number of slices to be plotted; if n.slices>1, nmax is set to 1. Or n.slices is a vector of 3 elements: start, end, length. Currently only works for grids.

nmax

the maximal number of the x@.RFparams$n iid copies of the field that are to be plotted

MARGIN.movie

integer. If given a sequence of figures is shown for this direction. This option is in an experimental stage. It works only for grids.

file, speed, height.pixel, width.pixel

In case MARGIN.movie and file is given an 'avi' movie is stored using the mencoder command with speed argument speed. As temporary files file__###.png of size width.pixel x height.pixel are created.

...

arguments to be passed to methods; mainly graphical arguments, or further models in case of class 'RMmodel', see Details.

plot.variance

logical, whether variances should be plotted if available

select.variables

vector of integers or list of vectors. The argument is only of interest for multivariate models. Here, length(select.variables) gives the number of pictures shown (excuding the plot for the variances, if applicable). If select.variables is a vector of integers then exactly these components are shown. If select.variables is a list, each element should be a vector up to length $l <= 3$:=""

$l=1$ the component is shown in the usual way

$l=2$ the two components are interpreted as vector and arrows are plotted

$l=3$ the first component is shown as single component; the remaining two component are interpreted as a vector and plotted into the picture of the first component

legend

logical, whether a legend should be plotted

zlim

vector of length 2 with the usual meaning. In case of multivariate random fields, zlim can also be a character wih the value ‘joint’ indicating that all plotted compoments shall have the same zlim OR a matrix with two rows, where the i-th column gives the zlim of the i-th variable OR a list with entries named data and var if a separate zlim for the Kriging variance is to be used.

plotmethod

string or function. Internal.

zlab

character. See persp

Methods

signature(x = "RFspatialGridDataFrame", y = "missing"): Generates nice image plots of simulation results for simulation on a grid and space-time-dimension $>1$. If space-time-dimension $>2$, plots are on 2-dimensional subspaces. Handles multivariate random fields (.RFparams$vdim>1) as well as repeated iid simulations (.RFparams$vdim>n).
signature(x = "RFspatialGridDataFrame", y = "RFspatialPointsDataFrame"): Similar to method for y="missing", but additionally adds the points of y. Requires MARGIN.slices=NULL and all.equal(x@.RFparams, y@.RFparams).
signature(x = "RFspatialGridDataFrame", y = "matrix"): Similar to method for y="missing", but additionally adds the points of y. Requires MARGIN.slices=NULL and all.equal(x@.RFparams, y@.RFparams).
signature(x = "RFspatialPointsDataFrame", y = "RFspatialGridDataFrame"): Throws an error. Probably x and y have been interchanged.
signature(x = "RFspatialPointsDataFrame", y = "missing"): Similar to method for class RFspatialGridDataFrame, but for non-gridded simulation results. Instead of a grid, only existing points are plotted with colors indicating the value of the random field at the respective location. Handles multivariate random fields (.RFparams$vdim>1) as well as repeated iid simulations (.RFparams$vdim>n).
signature(x = "RFspatialPointsDataFrame", y = "RFspatialPointsDataFrame"): Similar to method for y="missing", but additionally adds the points of y. Requires all.equal(x@.RFparams, y@.RFparams).
signature(x = "RFgridDataFrame", y = "missing"): Generates plots of simulation results for space-time-dimension $=1$. Handles different values for the number of repetitions as well as multivariate responses.
signature(x = "RFpointsDataFrame", y = "missing"): Similar to method for class RFgridDataFrame, but for non-gridded data.

Details

Internally, ... are passed to image and plot.default, respectively; if, by default, multiple colors, xlabs or ylabs are used, also vectors of suitable length can be passed as col, xlab and ylab, respectively.

One exception is the use of ... in plot for class 'RMmodel'. Here, further models might be passed. All models must have names starting with model. If '.' is following in the name, the part of the name after the dot is shown in the legend. Otherwise the name is ignored and a standardized name derived from the model definition is shown in the legend. Note that for the first argument a name cannot be specified.

Examples

Run this code

RFoptions(seed=0) ## *ANY* simulation will have the random seed 0; set
##                   RFoptions(seed=NA) to make them all random again



## define the model:
model <- RMtrend(mean=0.5) + # mean
         RMstable(alpha=1, var=4, scale=10) + # see help("RMstable")
                                        ## for additional arguments
         RMnugget(var=1) # nugget


#############################################################
## Plot of covariance structure

plot(model)
plot(model, xlim=c(0, 30))
plot(model, xlim=c(0, 30), fct.type="Variogram")
plot(model, xlim=c(-10, 20), fct.type="Variogram", dim=2)
image(model, xlim=c(-10, 20), fct.type="Variogram")
persp(model, xlim=c(-10, 20), fct.type="Variogram")

#############################################################
## Plot of simulation results

## define the locations:
from <- 0
step <- .1 
len <- 50  # nicer if len=100 %ok
 
x1D <- GridTopology(from, step, len)
x2D <- GridTopology(rep(from, 2), rep(step, 2), rep(len, 2))
x3D <- GridTopology(rep(from, 3), rep(step, 3), rep(len, 3))

## 1-dimensional
sim1D <- RFsimulate(model = model, x=x1D, n=6) 
plot(sim1D, nmax=4)

## 2-dimensional
sim2D <- RFsimulate(model = model, x=x2D, n=6) 
plot(sim2D, nmax=4)
plot(sim2D, nmax=4, col=terrain.colors(64),
main="My simulation", xlab="my_xlab")

## 3-dimensional
model <- RMmatern(nu=1.5, var=4, scale=2)
sim3D <- RFsimulate(model = model, x=x3D) 
plot(sim3D, MARGIN=c(2,3), MARGIN.slices=1, n.slices=4)

 
#############################################################
## empirical variogram plots

x <- seq(0, 10, 0.05)
bin <- seq(from=0, by=.2, to=3)

model <- RMexp()
X <- RFsimulate(x=cbind(x), model=model)
ev1 <- RFempiricalvariogram(data=X, bin=bin)
plot(ev1)

model <- RMexp(Aniso = cbind(c(10,0), c(0,1)))
X <- RFsimulate(x=cbind(x,x), model=model)
ev2 <- RFempiricalvariogram(data=X, bin=bin, phi=3)
plot(ev2, model=list(exp = model))

#############################################################
## plot Fitting results
x <- seq(0, 1, len=21)
 
model <- RMexp(Aniso = cbind(c(10,0), c(0,1)))
X <- RFsimulate(x=cbind(x,x), model=model)
fit <- RFfit(~RMexp(Aniso=diag(c(NA, NA))), data=X, fit.nphi = 2,
             modus="easygoing")
plot(fit) 


#############################################################
## plot Kriging results 
model <- RMwhittle(nu=1.2, scale=2)
n <- 200
x <- runif(n, max=step*len/2) 
y <- runif(n, max=step*len/2) # 200 points in 2 dimensional space
sim <- RFsimulate(model = model, x=x, y=y)

interpolate <- RFinterpolate(model=model, x=x2D, data=sim)
plot(interpolate)
plot(interpolate, sim)


#############################################################
## plotting vector-valued results
model <- RMdivfree(RMgauss(), scale=4)
x <- y <- seq(-10,10, 0.5)
simulated <- RFsimulate(model = model, x=x, y=y, n=1)
plot(simulated)
plot(simulated, select.variables=list(1, 1:3, 4))



#############################################################
## options for the zlim argument
model <- RMdelay(RMstable(alpha=1.9, scale=2), s=c(0, 4)) +
         RMdelay(RMstable(alpha=1.9, scale=2), s=c(4, 0))
simu <- RFsimulate(model, x, y)

plot(simu, zlim=list(data=cbind(c(-6,2), c(-2,1)), var=c(5,6)))
plot(simu, zlim=cbind(c(-6,2), c(-2,1)))
plot(simu, zlim=c(-6,2))
plot(simu, zlim="joint")

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning