The maps look great thanks for this.

.
Color scale too wide? Set your scale limits with argument mv: “plt.map(trends2, mv=.25)”:
.

.
Want smoother colors? Set the number of divisions on each side of zero with argument divs: “plt.map(trends2, mv=.25, divs=24)”:
.

.
Still not smooth enough? Try “plt.map(trends2, mv=.25, divs=1000)”:
.

.
Want the color scale bigger or you want to move it? The x and y arguments set the upper left corner, and the sl argument scales the box. You can scale the text with the st argument: “plt.map(trends2, mv=.25, x=-.5, y=-.3, sl=1.7, st=1.7)”:
.

.
For plotting dimensionless quantities – like eigenvectors – it’s sometimes convenient to normalize them so features clearly stand out. Just set argument norm to TRUE and the plot will normalize for you (without disrupting your data): “plt.map(trends2, norm=TRUE, divs=1000)”:
.

.
Now let’s say you want to highlight a certain range in the graph – like from -0.7 to -0.4. Just use argument hlite: “plt.map(trends2, norm=TRUE, divs=1000, hlite=c(-.7, -.4))”:
.

.
Don’t like that color? Try a different one with hcol: “plt.map(trends2, norm=TRUE, divs=1000, hlite=c(-.7, -.4), hcol=”yellow”)”:
.

.
The arguments for the function plt.map are:
.
dat: Data to be plotted
coord: Coordinates (lon/lat)
mv: Limit for the color scale
divs: Number of colors between zero and +mv (and, also, number of colors between -mv and zero)
labels: Vector of UNITS, MAIN TITLE, SUB TITLE
norm: If TRUE, normalize the data before plotting
hlite: Value range to highlight
hcol: Color to use for highlighting
mbx: If TRUE, draws a box around the plot
mxs: If TRUE, places x-y labels on the plot
x, y: The upper-left corner of the legend
sl: The scaling factor for the legend box
st: The scaling factor for the legend text
sp: The scaling factor for plotted points
s.main: The scaling factor for the main title
s.sub: The scaling factor for the subtitle
s.axis: The scaling factor for the axis labels
.
Here’s the plt.map function:

plt.map=function(dat, coord=sat.coord, mv=1, divs=9, labels=map.lbls, norm=FALSE, hlite=NA, hcol=”#99FF33″, mbx=TRUE, mxs=FALSE, x=-.3, y=-.35, sl=1, st=1, sp=1, s.main=1, s.sub=1, s.axis=1) {

library(mapproj)

### Set up Antarctic projection and lat/lon lines
map(“world”, proj=”azequidistant”, orient=c(-90, 0, 0), ylim=c(-90, -60))
a=seq(-180, 150, 30)
for(i in 1:length(a)) {lines(mapproject(list(x=rep(a[i], 181), y=c(-90:90))), col=4, lty=2)}
a=c(-180,0)
for(i in 1:length(a)) {lines(mapproject(list(x=rep(a[i], 181), y=c(-90:90))), col=4, lty=1)}
a=seq(-50,-80,-10)
for(i in 1:4) {lines(mapproject(list(y=rep(a[i], 72), x=seq(-177.5, 177.5,5))), col=4, lty=2)}

### Make the color scale
mapcolors=get.colors(maxval, divs, hlite, hcol)

### Normalize the data to a maximum of +/- 1?
nm=ifelse(norm==TRUE, max(abs(as.vector(dat))), 1)

### Assign the color scale
temp=(divs+2)+(round((divs+1)*(dat/(mv*nm))))
temp=ifelse(templength(mapcolors), mapcolors[length(mapcolors)], mapcolors[temp])

### Plot points
xy=mapproject(coord[, 1], coord[, 2])
points(xy$x, xy$y, pch=15, cex=sp, col=temp)

### Overlay map
map(“world”, proj=”azequidistant”, orient=c(-90, 0, 0), ylim=c(-90, -60), fill=FALSE, add=TRUE)

### Set up legend and labels
legends=get.legend(x, y, sl, st, sp, h=hlite, mv, ml=map.lbls, mbx, mxs, mc=mapcolors, divs, s.main, s.sub, s.axis)
}

.
It first calls the function get.colors to make the color map. I broke this out as a separate function because I use it to generate color maps for other purposes as well:

get.colors=function(mv, divs, hlite, hcol) {

### Make color scale
cold.colors=matrix(nrow=divs); hot.colors=matrix(nrow=divs)
for(i in 1:divs) {
lum.ratio=25*(i/divs)
c1=ifelse(lum.ratio<15, 255-(17*lum.ratio), 0)
c2=ifelse(lum.ratio0, 255-(12*lum.ratio), 0)
h2=ifelse(255-(25*lum.ratio)>0, 255-(25*lum.ratio), 0)
cold.colors[divs+1-i]=rgb(red=c1, green=c1, blue=c2, maxColorValue=255)
hot.colors[i]=rgb(red=c2, green=h1, blue=h2, maxColorValue=255)
}
neut.col=ifelse(divs<11, “#EEEEEE”, “#FFFFFF”)
mapcolors=c(“darkmagenta”, as.vector(cold.colors), neut.col, as.vector(hot.colors), “black”)
if(!is.na(hlite[1])) {
highlmin=(divs+2)+(round((divs+1)*(hlite[1]/(mv))))
highlmax=(divs+2)+(round((divs+1)*(hlite[2]/(mv))))
if(!is.na(highlmax)) {mapcolors[highlmin:highlmax]=hcol} else {mapcolors[highlmin]=hcol}
}
mapcolors
}

.
It generates the legend by calling a separate function called get.legend. I did it this way to be able to build color scale legends for things other than Antarctica.

get.legend=function(x, y, sl, st, sp, h, mv, ml, mbx, mxs, mc, divs, s.main, s.sub, s.axis) {
rect(xleft=x, xright=x+(.6*sl), ytop=y, ybottom=y-(.080*sl), col=”white”)
points(x=x+(.3*sl), y=y-(.023*sl), col=mc[2+divs], pch=15, cex=1.3*sl)
for(i in 1:divs) {
points(x=x+(.3*sl)-(.2/(divs+1))*i*sl, y=y-(.023*sl), col=mc[divs+2-i], pch=15, cex=1.3*sl)
points(x=x+(.3*sl)+(.2/(divs+1))*i*sl, y=y-(.023*sl), col=mc[divs+2+i], pch=15, cex=1.3*sl)
}
points(x=x+(.1*sl), y=y-(.023*sl), col=mc[1], pch=15, cex=1.3*sl)
points(x=x+(.5*sl), y=y-(.023*sl), col=mc[length(mc)], pch=15, cex=1.3*sl)
text(x=x+(.3*sl), y=y-(.055*sl), ml[1], cex=st)
text(x=x+(.085*sl), y=y-(.055*sl), -mv, cex=st)
text(x=x+(.51*sl), y=y-(.055*sl), mv, cex=st)
rtext=ifelse(!is.na(h[2]), paste(h[1], “to”, h[2]), h[1])
xtext=ifelse(!is.na(h[1]), paste(rtext, ml[1], “Highlighted”), NA)
title(main=ml[2], sub=ml[3], xlab=xtext, cex.main=s.main, cex.sub=s.sub, cex.axis=s.axis)
if(mbx==TRUE) {box()}
if(mxs==TRUE) {map.axes()}
}

.
Anyway, since several of us seem to be plotting similar things but need to change color scales conveniently, I figured I’d throw this out there in case someone else finds it useful.
.

I just finished a short post on the satellite vs ground trends showing a warming trend in the sat data as compared to the surface stations.

http://noconsensus.wordpress.com/2009/04/14/warming-the-raw-sat-data/

I’ve got a request to do a reconstruction without the peninsula data which I’ll do and then I’d like to try your data in some of the reconstructions.

In the meantime I’ve been creating new reconstructions based as much as possible on surface station data. This reconstruction is based on a closest station algorithm and uses no satellite data at all.

http://noconsensus.wordpress.com/2009/04/12/closest-station-antarctic-reconstruction/

The trend drops all the way down to 0.052 C/decade. Everywhere we look the trend comes out less than reported.

I bet it’s consistent with the models though.