Arrange odd number of plots using rasterVis and gridExtra

Question

I am trying to plot a panel with seven rasters using the levelplot function of the rasterVis package, combined with gridExtra's grid.arrange.

I almost get what I need by using the following code:

# load required packages
library(rasterVis)
library(gridExtra)

# load sample raster
f <- system.file("external/test.grd", package="raster")
r <- raster(f)

# create plots
p1 <- levelplot(r, xlab=NULL, ylab=NULL, margin=FALSE)
p2 <- levelplot(r*2, xlab=NULL, ylab=NULL, margin=FALSE,colorkey=FALSE)

# put plots in list
p.list <- list(p1,p2,p2,p2,p2,p2,p2)

# create layout
lay <- rbind(c(1,1,1),
             c(2,3,4),
             c(5,6,7))

# arrange plots
grid.arrange(grobs=p.list, layout_matrix=lay)

which yields this figure:

However, there are some things I still need to improve:

1. How to decrease the blank space between the plots in the bottom rows?
2. How to add a single, combined legend for the six bottom rasters, preferentially put to the bottom of the figure?

Is this possible to achieve using rasterVis and gridExtra? Is there any other approach that can be used?

Answer 1

The white space is a combination of lattice margin settings, but also the plots having a fixed aspect ratio (they can't be too close unless the device itself has a compatible aspect ratio).

With regards the legend, you could use draw.colorkey(), but from what I can tell you need to manually ensure that the colours match by passing them explicitly to both plots and the key.

# load required packages
library(rasterVis)
library(gridExtra)

# load sample raster
f <- system.file("external/test.grd", package="raster")
r <- raster(f)

my_theme <- rasterTheme(region = blues9)

# create plots
p1 <- levelplot(r, xlab=NULL, ylab=NULL, margin=FALSE, par.settings = my_theme)
leg <- p1$legend$right$args$key
p1$legend <- list()
p2 <- levelplot(r*2, xlab=NULL, ylab=NULL, margin=FALSE,colorkey=FALSE, par.settings = my_theme)

# put plots in list
p.list <- list(p1,p2,p2,p2,p2,p2,p2)

# create layout
lay <- rbind(c(NA,1,NA),
             c(2,3,4),
             c(5,6,7),
             c(8,8,8))

leg$col <- my_theme$regions$col
legGrob <- draw.colorkey(key = leg, vp = grid::viewport(height=0.5))
# arrange plots
grid.arrange(grobs=c(p.list, list(legGrob)), layout_matrix=lay,
             vp = grid::viewport(width=0.7,height=1))

(needless to say, facetting seems the better option by a wide margin)