spatstat (version 1.64-1)

plot.ppp: plot a Spatial Point Pattern


Plot a two-dimensional spatial point pattern


# S3 method for ppp
plot(x, main, …, clipwin=NULL,
                    chars=NULL, cols=NULL,
                    use.marks=TRUE, which.marks=NULL,
                    add=FALSE, type=c("p","n"),
                    leg.side=c("left", "bottom", "top", "right"),
                    symap=NULL, maxsize=NULL, meansize=NULL, markscale=NULL,
                    show.window=show.all, show.all=!add, do.plot=TRUE,



The spatial point pattern to be plotted. An object of class "ppp", or data which can be converted into this format by as.ppp().


text to be displayed as a title above the plot.

extra arguments that will be passed to the plotting functions plot.default, points and/or symbols. Not all arguments will be recognised.


Optional. A window (object of class "owin"). Only this subset of the image will be displayed.


plotting character(s) used to plot points.


the colour(s) used to plot points.


logical flag; if TRUE, plot points using a different plotting symbol for each mark; if FALSE, only the locations of the points will be plotted, using points().


Index determining which column of marks to use, if the marks of x are a data frame. A character or integer vector identifying one or more columns of marks. If add=FALSE then the default is to plot all columns of marks, in a series of separate plots. If add=TRUE then only one column of marks can be plotted, and the default is which.marks=1 indicating the first column of marks.


logical flag; if TRUE, just the points are plotted, over the existing plot. A new plot is not created, and the window is not plotted.


Type of plot: either "p" or "n". If type="p" (the default), both the points and the observation window are plotted. If type="n", only the window is plotted.


Logical value indicating whether to add a legend showing the mapping between mark values and graphical symbols (for a marked point pattern).


Position of legend relative to main plot.


List of additional arguments passed to plot.symbolmap or symbolmap to control the legend. In addition to arguments documented under plot.symbolmap, and graphical arguments recognised by symbolmap, the list may also include the argument sep giving the separation between the main plot and the legend, or sep.frac giving the separation as a fraction of the relevant dimension (width or height) of the main plot.


Optional. The graphical symbol map to be applied to the marks. An object of class "symbolmap"; see symbolmap.


Maximum physical size of the circles/squares plotted when x is a marked point pattern with numerical marks. Incompatible with meansize and markscale. Ignored if symap is given.


Average physical size of the circles/squares plotted when x is a marked point pattern with numerical marks. Incompatible with maxsize and markscale. Ignored if symap is given.


physical scale factor determining the sizes of the circles/squares plotted when x is a marked point pattern with numerical marks. Mark value will be multiplied by markscale to determine physical size. Incompatible with maxsize and meansize. Ignored if symap is given.


Fraction between 0 and 1. When x is a marked point pattern with numerical marks, zap is the smallest mark value (expressed as a fraction of the maximum possible mark) that will be plotted. Any points which have marks smaller in absolute value than zap * max(abs(marks(x))) will not be plotted.


Logical value indicating whether to plot the observation window of x.


Logical value indicating whether to plot everything including the main title and the observation window of x.


Logical value determining whether to actually perform the plotting.


Logical value giving permission to display multiple plots.


(Invisible) object of class "symbolmap" giving the correspondence between mark values and plotting characters.

Removing White Space Around The Plot

A frequently-asked question is: How do I remove the white space around the plot? Currently plot.ppp uses the base graphics system of R, so the space around the plot is controlled by parameters to par. To reduce the white space, change the parameter mar. Typically, par(mar=rep(0.5, 4)) is adequate, if there are no annotations or titles outside the window.

Drawing coordinate axes and axis labels

Coordinate axes and axis labels are not drawn, by default. To draw coordinate axes, set axes=TRUE. To draw axis labels, set ann=TRUE and give values to the arguments xlab and ylab. See the Examples. Only the default style of axis is supported; for more control over the placement and style of axes, use the graphics commands axis and mtext.


This is the plot method for point pattern datasets (of class "ppp", see ppp.object).

First the observation window Window(x) is plotted (if show.window=TRUE). Then the points themselves are plotted, in a fashion that depends on their marks, as follows.

unmarked point pattern:

If the point pattern does not have marks, or if use.marks = FALSE, then the locations of all points will be plotted using a single plot character

multitype point pattern:

If x$marks is a factor, then each level of the factor is represented by a different plot character.

continuous marks:

If x$marks is a numeric vector, the marks are rescaled to the unit interval and each point is represented by a circle with diameter proportional to the rescaled mark (if the value is positive) or a square with side length proportional to the absolute value of the rescaled mark (if the value is negative).

other kinds of marks:

If x$marks is neither numeric nor a factor, then each possible mark will be represented by a different plotting character. The default is to represent the \(i\)th smallest mark value by points(..., pch=i).

If there are several columns of marks, and if which.marks is missing or NULL, then

  • if add=FALSE and multiplot=TRUE the default is to plot all columns of marks, in a series of separate plots, placed side-by-side. The plotting is coordinated by plot.listof, which calls plot.ppp to make each of the individual plots.

  • Otherwise, only one column of marks can be plotted, and the default is which.marks=1 indicating the first column of marks.

Plotting of the window Window(x) is performed by plot.owin. This plot may be modified through the ... arguments. In particular the extra argument border determines the colour of the window, if the window is not a binary mask.

Plotting of the points themselves is performed by the function points, except for the case of continuous marks, where it is performed by symbols. Their plotting behaviour may be modified through the ... arguments.

The argument chars determines the plotting character or characters used to display the points (in all cases except for the case of continuous marks). For an unmarked point pattern, this should be a single integer or character determining a plotting character (see par("pch")). For a multitype point pattern, chars should be a vector of integers or characters, of the same length as levels(x$marks), and then the \(i\)th level or type will be plotted using character chars[i].

If chars is absent, but there is an extra argument pch, then this will determine the plotting character for all points.

The argument cols determines the colour or colours used to display the points. For an unmarked point pattern, cols should be a character string determining a colour. For a multitype point pattern, cols should be a character vector, of the same length as levels(marks(x)): that is, there is one colour for each possible mark value. The \(i\)th level or type will be plotted using colour cols[i]. For a point pattern with continuous marks, cols can be either a character string or a character vector specifying colour values: the range of mark values will be mapped to the specified colours.

If cols is absent, the colours used to plot the points may be determined by the extra argument fg (for multitype point patterns) or the extra argument col (for all other cases). Note that specifying col will also apply this colour to the window itself.

The default colour for the points is a semi-transparent grey, if this is supported by the plot device. This behaviour can be suppressed (so that the default colour is non-transparent) by setting spatstat.options(transparent=FALSE).

The arguments maxsize, meansize and markscale incompatible. They control the physical size of the circles and squares which represent the marks in a point pattern with continuous marks. The size of a circle is defined as its diameter; the size of a square is its side length. If markscale is given, then a mark value of m is plotted as a circle of diameter m * markscale (if m is positive) or a square of side abs(m) * markscale (if m is negative). If maxsize is given, then the largest mark in absolute value, mmax=max(abs(marks(x))), will be scaled to have physical size maxsize. If meansize is given, then the average absolute mark value, mmean=mean(abs(marks(x))), will be scaled to have physical size meansize.

The user can set the default values of these plotting parameters using spatstat.options("par.points").

To zoom in (to view only a subset of the point pattern at higher magnification), use the graphical arguments xlim and ylim to specify the rectangular field of view.

The value returned by this plot function is an object of class "symbolmap" representing the mapping from mark values to graphical symbols. See symbolmap. It can be used to make a suitable legend, or to ensure that two plots use the same graphics map.

See Also

ppp.object, plot, par, points, text.ppp, plot.owin, symbols.

See also the command iplot in the spatstat.gui package.


Run this code

   plot(cells, pch=16)

   # make the plotting symbols larger (for publication at reduced scale)
   plot(cells, cex=2)

   # set it in spatstat.options
   oldopt <- spatstat.options(par.points=list(cex=2))

   # multitype 

   # marked by a real number

   # just plot the points
   plot(longleaf, use.marks=FALSE)
   plot(unmark(longleaf)) # equivalent

   # point pattern with multiple marks
   plot(finpines, which.marks="height")

   # controlling COLOURS of points
   plot(cells, cols="blue")
   plot(lansing, cols=c("black", "yellow", "green", 
   plot(longleaf, fg="blue")

   # make window purple
   plot(lansing, border="purple")
   # make everything purple
   plot(lansing, border="purple", cols="purple", col.main="purple",
   # controlling PLOT CHARACTERS for multitype pattern
   plot(lansing, chars = 11:16)
   plot(lansing, chars = c("o","h","m",".","o","o"))

   ## multitype pattern mapped to symbols
   plot(amacrine, shape=c("circles", "squares"), size=0.04)
   plot(amacrine, shape="arrows", direction=c(0,90), size=0.07)

   ## plot trees as trees!
   plot(lansing, shape="arrows", direction=90, cols=1:6)

   # controlling MARK SCALE for pattern with numeric marks
   plot(longleaf, markscale=0.1)
   plot(longleaf, maxsize=5)
   plot(longleaf, meansize=2)

   # draw circles of diameter equal to nearest neighbour distance
   plot(cells %mark% nndist(cells), markscale=1, legend=FALSE)

   # inspecting the symbol map
   v <- plot(amacrine)

   ## variable colours ('cols' not 'col')
   plot(longleaf, cols=function(x) ifelse(x < 30, "red", "black"))

   ## re-using the same mark scale
   a <- plot(longleaf)
   juveniles <- longleaf[marks(longleaf) < 30]
   plot(juveniles, symap=a)

   ## numerical marks mapped to symbols of fixed size with variable colour
   ra <- range(marks(longleaf))
   colmap <- colourmap(terrain.colors(20), range=ra)
   ## filled plot characters are the codes 21-25
   ## fill colour is indicated by 'bg'
   sy <- symbolmap(pch=21, bg=colmap, range=ra)
   plot(longleaf, symap=sy)

   ## or more compactly..
   plot(longleaf, bg=terrain.colors(20), pch=21, cex=1)

   ## clipping
   B <- owin(c(4810, 5190), c(4180, 4430))
   plot(B, add=TRUE, border="red")
   plot(humberside, clipwin=B, main="Humberside (clipped)")

   ## coordinate axes and labels
   plot(humberside, axes=TRUE)
   plot(humberside,            ann=TRUE, xlab="Easting", ylab="Northing")
   plot(humberside, axes=TRUE, ann=TRUE, xlab="Easting", ylab="Northing")
# }

Run the code above in your browser using DataLab