par(mfrow) or layout. To plot just a palette,
e.g. to get an x-y plot with points coloured according to a palette, use
drawPalette and then draw the main diagram.
imagep(x, y, z, xlim, ylim, zlim, zclip = FALSE, flipy = FALSE, xlab = "", ylab = "", zlab = "", zlabPosition = c("top", "side"), decimate = TRUE, breaks, col, colormap, labels = NULL, at = NULL, drawContours = FALSE, drawPalette = TRUE, drawTriangles = FALSE, tformat, drawTimeRange = getOption("oceDrawTimeRange"), filledContour = FALSE, missingColor = NULL, mgp = getOption("oceMgp"), mar, mai.palette, xaxs = "i", yaxs = "i", asp = NA, cex = par("cex"), adorn = NULL, axes = TRUE, main = "", axisPalette, add = FALSE, debug = getOption("oceDebug"), ...)z are defined. In this case, both x and y
must be supplied and, within each, the values must be finite and
distinct; if values are out of order, they (and z) will be
transformed to put them in order.
ordered in a matching way). Mode 2.
If z is provided but not x and y, then the latter
are constructed to
indicate the indices of the matrix, in contrast
to the range of 0 to 1, as is the case for image.
Mode 3. If
x is a list, its components x$x and x$y are used
for x and y, respectively. If the list has component
z this is used for z. (NOTE: these arguments are meant to
mimic those of image, which explains the same description
here.) Mode 4. There are also some special cases, e.g. if x is a
topographic object such as can be created with read.topo
or as.topo, then longitude and latitude are used for
axes, and topographic height is drawn.image, which explains the same
description here.)zlim may take several forms. First, it may be a pair
of numbers that specify the limits for the colour scale. Second,
it could be the string "histogram", to yield a flattened
histogram (i.e. to increase contrast). Third, it could be the
string "symmetric", to yield limits that are symmetric
about zero, which can be helpful in drawing velocity fields,
for which a zero value has a particular meaning (in which case,
a good colour scheme might be col=oceColorsTwo).zlim. This only works if zlim is
provided. Clipped regions will be coloured with missingColor.
Thus, clipping an image is somewhat analogous to clipping in an xy
plot, with clipped data being ignored, which in an image means to be be
coloured with missingColor.TRUE indicating that the image
should be flipped top to bottom (e.g. to produce a profile image
for a downward-looking acoustic-doppler profile).decimate=FALSE then every grid cell in the matrix will
be represented by a pixel in the image. Case 2 (the default).
If decimate=TRUE, then decimation will be done
in the horizontal or vertical direction (or both) if the length of the
corresponding edge of the z matrix exceeds 800. (This also creates
a warning message.) The decimation
factor is computed as the integet just below the ratio of z dimension
to 400. Thus, no decimation is done if the dimension is less than 800,
but if the dimension s between 800 and 1199, only every second grid
point is mapped to a pixel in the image. Case 3.
If decimate is an integer, then that z is subsampled
at seq.int(1L, dim(z)[1], by=decimate) (as is x), and
the same is done for the y direction. Case 4. If
decimate is a vector of two integers, the first is used for
the first index of z, and the second is used for the second
index.pretty, in determining clean break points.oce.colorsJet for a rainbow.colormap. If
provided, then colormap$breaks and colormap$col take
precedence over the present arguments breaks and col.
(All of the other contents of colormap are ignored, though.)at).labels.drawPalette=TRUE, a palette is drawn at the right-hand side of the
main image. If drawPalette=FALSE, no palette is drawn, and the
right-hand side of the plot has a thin margin. If
drawPalette="space", then no palette is drawn, but space is put in
the right-hand margin to occupy the region in which the palette would
have been drawn. This last form is useful for producing stacked plots
with uniform left and right margins, but with palettes on only some of
the images.drawPalette.oce.plot.ts, for
plot types that call that function. (See strptime for the
format used.)x axis is a
time. If TRUE, then an indication of the time range of the
data (not the axis) is indicated at the top-left margin of the
graph. This is useful because the labels on time axes only indicate
hours if the range is less than a day, etc.NULL for transparent (to see this, try setting
par("bg")<-"red").par(mgp), and
also for par(mar), computed from this. The default is
tighter than the R default, in order to use more space for the
data and less for the axes.par("mar"). If not
given, a reasonable value is calculated based on whether xlab and
ylab are empty strings.mai value used for the palette. Use with care."i") or not; see
par("xaxs").xaxs but for y axis.plot.default. If
x inherits from topo-class and asp=NA (the
default) then asp is redefined to be the reciprocal of the
mean latitude in x, as a way to reduce geographical distortion.
Otherwise, if asp is not NA, then it is used directly.par("cex").expression or vector of expressions that contain
R code that is to be executed immediately after each panel of the plot.
If the number of expressions matches the number of panels, then the
expressions are used for the corresponding panels; otherwise, the
expression list is extended to match the number of panels (i.e.
to obtain length(which) elements). Note that adorn
is a dangerous argument, because if the expressions contained
therein set up local storage, there is a chance of entirely
disrupting the plotting. For this reason, adorn was
marked as defunct in June 2016, and will be removed entirely
after the July CRAN release. Users with existing code that uses
adorn should simply plot the panels individually, and
use conventional R functions, e.g. lines etc.,
after each panel, to achieve the desired effect. (See
oce-defunct for notes on other deprecated or
defunct oce features.)TRUE to get axes on the main image.axis(), passed to
drawPalette.FALSE indicates that a new plot is to be created.
However, if add is TRUE, the idea is to add an image (but not
its palette or its axes) to an existing plot. Clearly, then, arguments
such xlim are to be ignored. Indeed, if add=TRUE, the only
arguments examined are x (which must be a vector; the mode of providing
a matrix or oce object does not work), y, z,
decimate, plus either colormap or
both breaks and col.xat and yat,
values that can be used by oce.grid to add a grid to the
plot.
imagep is called. The problem is not isolated to imagep;
it occurs with other packages, and a web
search reveals repeated bug reports submitted to RStudio.
The problem seems to come and go, as RStudio evolves. In the
imagep case, things worked properly for
RStudio version 0.99.451 (released late in 2015), but not
for version 0.99.878 (released early
in 2016). A bug report was sent to RStudio in
January 2016, with a minimal example that boiled the issue
down to a few lines of basic R code (not using imagep
or even oce).
Although communications with RStudio gave
reason for optimism, the problem persisted in version 0.99.892,
released March 4. New versions of RStudio will be checked as they
come out, with status updates here.
Pending an RStudio solution, users can avoid the error
simply by opening
a new (and separate) plotting window with dev.new.
In doing so, they may find that this is preferable generally,
given the limitations of one-window interfaces.filled.contour except that with imagep it is possible to
set the layout outside the function, which enables the creation
of plots with many image-palette panels. Note that the contour lines may not
coincide with the colour transitions, in the case of coarse images.Note that this does not use layout or any of the other screen
splitting methods. It simply manipulates margins, and draws two plots
together. This lets users employ their favourite layout schemes.
NOTE: imagep is an analogue of image, and from that
it borrows a the convention that the number of rows in the matrix corresponds to
to x axis, not the y axis. (Actually, image permits
the length of x to match either nrow(z) or 1+nrow(z), but
here only the first is permitted.)
drawPalette, and is used by plot,adp-method,
plot,landsat-method, and other image-generating functions.
library(oce)
# 1. simplest use
imagep(volcano)
# 2. something oceanographic (internal-wave speed)
h <- seq(0, 50, length.out=100)
drho <- seq(1, 3, length.out=200)
speed <- outer(h, drho, function(drho, h) sqrt(9.8 * drho * h / 1024))
imagep(h, drho, speed, xlab="Equivalent depth [m]",
ylab=expression(paste(Delta*rho, " [kg/m^3]")),
zlab="Internal-wave speed [m/s]")
# 3. fancy labelling on atan() function
x <- seq(0, 1, 0.01)
y <- seq(0, 1, 0.01)
angle <- outer(x,y,function(x,y) atan2(y,x))
imagep(x, y, angle, filledContour=TRUE, breaks=c(0, pi/4, pi/2),
col=c("lightgray", "darkgray"),
at=c(0, pi/4, pi/2),
labels=c(0, expression(pi/4), expression(pi/2)))
# 4. a colormap case
data(topoWorld)
cm <- colormap(name="gmt_globe")
imagep(topoWorld, colormap=cm)
Run the code above in your browser using DataLab