Levelplot color key - range and extremes

Question

Is it possible in R to create a color key like the one below? (this one comes from the software Grid Analysis and Display System - Grads).

There are two features that I can't reproduce in R:

1. The sequence is non linear however it is displayed as if
2. Values bigger than 200 are grey / Values smaller than 0 are white

I'm using levelplot from rastervis that plots rasters using the lattice levelplot:

require(raster)
require(rasterVis)

set.seed(200)

X = seq(-40,0,by=1)
Y = seq(-60,-40,by=1)
grid = expand.grid(list(X=X,Y=Y))
Z = rnorm(nrow(grid),mean=10,sd=100)

data = data.frame(grid,Z)
r = rasterFromXYZ(data)
mapTheme <- rasterTheme(region=c('#EEF7FA','#D6F8F7',"#BEDAFF",'#5DA4FF',"#0000FF","#D4F9E2","#00FF7F","#008000","#FFFF00","#FFD27F", "#FFB732"    ,"#EE7600",
                                 "#D53E4F","#FF6A6A"))  
my.at = c(0,1,5,10,15,20,25,30,40,50,75,100,150,200)
myColorkey <- list(at=my.at,
                   space="bottom",
                   labels=list(at=my.at))
p=levelplot(r, par.settings=mapTheme,at = my.at, colorkey=myColorkey,margin=F)
print(p)

The result:

As you can see, both values smaller than 0 and bigger than 200 are white, I've no idea how to set values bigger than or smaller than a certain value to appear as a specific color. Morover, how can I make the space between consecutive thick marks in the color key to have the same size although the intervals are not the same?

Stateless

Answer 1

This is a workaround for equally sized intervals for non linear sequences:

library(raster)
library(rasterVis)

set.seed(200)
X = seq(-40,0,by=1)
Y = seq(-60,-40,by=1)
grid = expand.grid(list(X=X,Y=Y))
Z = rnorm(nrow(grid),mean=10,sd=100)

data = data.frame(grid,Z)
r = rasterFromXYZ(data)
mapTheme <- rasterTheme(region=c('#EEF7FA','#D6F8F7',"#BEDAFF",'#5DA4FF',"#0000FF","#D4F9E2","#00FF7F",
                                 "#008000","#FFFF00","#FFD27F", "#FFB732" ,"#EE7600", "#D53E4F","#FF6A6A"))  

my.at=c(0,1,5,10,15,20,25,30,40,50,75,100,150,200)
my.brks=seq(0, 200, by=15)

myColorkey <- list(at=my.brks, labels=list(at=my.brks, labels=my.at), space="bottom")
p=levelplot(r, par.settings=mapTheme, at=my.at, colorkey=myColorkey, margin=F)
print(p)

This could be a solution for values smaller 0 and greater than 200:

library(raster)
library(rasterVis)

set.seed(200)
X = seq(-40,0,by=1)
Y = seq(-60,-40,by=1)
grid = expand.grid(list(X=X,Y=Y))
Z = rnorm(nrow(grid),mean=10,sd=100)

data = data.frame(grid,Z)
r = rasterFromXYZ(data)
mapTheme <- rasterTheme(region=c('white','#EEF7FA','#D6F8F7',"#BEDAFF",'#5DA4FF',"#0000FF","#D4F9E2","#00FF7F",
                                 "#008000","#FFFF00","#FFD27F", "#FFB732" ,"#EE7600", "#D53E4F","#FF6A6A", "gray"))  
max(values(r))
min(values(r))

my.at=c(min(values(r)), 0,1,5,10,15,20,25,30,40,50,75,100,150,200, max(values(r)))
my.brks=seq(0, 200, by=13)

myColorkey <- list(at=my.brks, labels=list(at=my.brks, labels=c(-276,0,1,5,10,15,20,25,30,40,50,75,100,150,200, 388)), space="bottom")
p=levelplot(r, par.settings=mapTheme, at=my.at, colorkey=myColorkey, margin=F)
print(p)

Your colors are not progressing from light to dark. You can use the RColorBrewer package to fix this.

library(RColorBrewer)
reds = brewer.pal(5, "YlOrRd")
greens = brewer.pal(3, "Greens")
blues = brewer.pal(5, "Blues")
mapTheme <- rasterTheme(region=c('white', blues, greens, reds, "gray"))

Discrete

Thank you so much, this is exactly what I was looking for. I really appreciate the color advice! – Stateless

Answer 2

This is a very helpful workaround. While not addressing question 1, something I found useful for question 2 (adding triangles for values below/above the limits of the colorbar range) can be achieved by adding this:

library(s2dverification)
data_array <- array(Z, dim = c(length(X), length(Y)))
PlotEquiMap(data_array, X, Y,bar_limits=c(0,200),col_inf='white',col_sup='gray')

raster with colorbar

Crabb

Answer 3

Another solution with updates to lattice:

library(raster)
library(rasterVis)

set.seed(200)
X = seq(-40,0,by=1)
Y = seq(-60,-40,by=1)
grid = expand.grid(list(X=X,Y=Y))
Z = rnorm(nrow(grid),mean=10,sd=100)
data = data.frame(grid,Z)
r = rasterFromXYZ(data)

levelplot(r, margin=F, at=c(-Inf, seq(0, 200, 20), Inf),
          colorkey = list(tri.lower = TRUE, tri.upper = TRUE))

As long as you add "-Inf" and "Inf" to your at definition, the option to add triangles to the colorbar is activated.

Crabb