How to fill in the contour fully using stat_contour

Question

I am looking for ways to fully fill in the contour generated by ggplot2\'s stat_contour. The current result is like this:

# Generate data
library(ggplot2)
li         


        

Answer 1

As @tonytonov has suggested this thread, the transparent areas can be deleted by closing the polygons.

# check x and y grid
minValue<-sapply(volcano3d,min)
maxValue<-sapply(volcano3d,max)
arbitaryValue=min(volcano3d$z-10)

test1<-data.frame(x=minValue[1]-1,y=minValue[2]:maxValue[2],z=arbitaryValue)
test2<-data.frame(x=minValue[1]:maxValue[1],y=minValue[2]-1,z=arbitaryValue)
test3<-data.frame(x=maxValue[1]+1,y=minValue[2]:maxValue[2],z=arbitaryValue)
test4<-data.frame(x=minValue[1]:maxValue[1],y=maxValue[2]+1,z=arbitaryValue)
test<-rbind(test1,test2,test3,test4)

vol<-rbind(volcano3d,test)

w <- ggplot(vol, aes(x, y, z = z))
w + stat_contour(geom="polygon", aes(fill=..level..)) # better

# Doesn't work when trying to get rid of unwanted space
w + stat_contour(geom="polygon", aes(fill=..level..))+
  scale_x_continuous(limits=c(min(volcano3d$x),max(volcano3d$x)), expand=c(0,0))+ # set x limits
  scale_y_continuous(limits=c(min(volcano3d$y),max(volcano3d$y)), expand=c(0,0))  # set y limits

# work here!
w + stat_contour(geom="polygon", aes(fill=..level..))+
coord_cartesian(xlim=c(min(volcano3d$x),max(volcano3d$x)),
                ylim=c(min(volcano3d$y),max(volcano3d$y)))

The problem remained with this tweak is finding methods aside from trial and error to determine the arbitaryValue.

[edit from here]

Just a quick update to show how I am determining the arbitaryValue without having to guess for every datasets.

BINS<-50
BINWIDTH<-(diff(range(volcano3d$z))/BINS) # reference from ggplot2 code
arbitaryValue=min(volcano3d$z)-BINWIDTH*1.5

This seems to work well for the dataset I am working on now. Not sure if applicable with others. Also, note that the fact that I set BINS value here requires that I will have to use bins=BINS in stat_contour.

Answer 2

Thanks for @chengvt's answer. I sometimes needs this technique, so I made a generalized function().

test_f <- function(df) {
  colname <- names(df)
  names(df) <- c("x", "y", "z")
  Range <- as.data.frame(sapply(df, range))
  Dim <- as.data.frame(t(sapply(df, function(x) length(unique(x)))))
  arb_z = Range$z[1] - diff(Range$z)/20
  df2 <- rbind(df,
               expand.grid(x = c(Range$x[1] - diff(Range$x)/20, Range$x[2] + diff(Range$x)/20), 
                           y = seq(Range$y[1], Range$y[2], length = Dim$y), z = arb_z),
               expand.grid(x = seq(Range$x[1], Range$x[2], length = Dim$x),
                           y = c(Range$y[1] - diff(Range$y)/20, Range$y[2] + diff(Range$y)/20), z = arb_z))
  g <- ggplot(df2, aes(x, y, z = z)) + labs(x = colname[1], y = colname[2], fill = colname[3]) + 
    stat_contour(geom="polygon", aes(fill=..level..)) + 
    coord_cartesian(xlim=c(Range$x), ylim=c(Range$y), expand = F)
  return(g)
}

library(ggplot2); library(reshape2)
volcano3d <- melt(volcano)
names(volcano3d) <- c("xxx", "yyy", "zzz")
test_f(volcano3d) + scale_fill_gradientn(colours = terrain.colors(10))