I came across a really cool tweet a few days ago containing an animated GIF demonstrating rise in global temperatures since 1880:

Visualizing the Warmest August in 136 Years https://t.co/PSjA2NfYem Data via @NASAGISS #NASA #climate pic.twitter.com/DIk8z7Zodj

— NASA Earth (@NASAEarth) September 12, 2016

Link to the blog is here.

I wanted to see if I could recreate this GIF in R, and below are some instructions on how to get close.

The first step is to try and see if I can replicate the original, final image.  The code below comes close enough for this demo, but strangely I am approx. 1 degree C out.  I’ve left a comment on the original blog to ask why – it’s not clear how the exact figures are calculated.

Anyway, here is how to get close:

# download data
d <- read.table("http://data.giss.nasa.gov/gistemp/tabledata_v3/GLB.Ts.txt", 
                skip=7, header=TRUE, nrows=143, colClasses = "character")

# clean data
d <- d[-grep("Year", d$Year),]
rownames(d) <- d[,1]
d <- d[,-1]
d <- d[,-13:-19]
d[d=="****"] <- NA

# convert to numeric and celsius
for (i in 1:12) {
	d[,i] <- as.numeric(d[,i]) / 100
}	

# download seasonal adjustments
s <- read.table("http://data.giss.nasa.gov/gistemp/faq/merra2_seas_anom.txt", 
                skip=3, header=TRUE, colClasses = "character")
sa <- as.numeric(s$seas_anom)

# create colours from blue through to red
colours <- colorRampPalette(c("grey","blue","red"))(nrow(d))

# create initial plot
mmin <- -3
mmax <- 3
par(mar=c(2,2,2,0))
plot(1:12, d[1,]+sa, type="l", ylim=c(mmin,mmax), 
     yaxt="n", xaxt="n", bty="n")

# add axes
axis(side=1, at=1:12, 
     labels=c("Jan","Feb","Mar",
              "Apr","May","Jun",
              "Jul","Aug","Sep",
              "Oct","Nov","Dec"), 
     lwd=0, lwd.ticks=1, cex.axis=0.8)

axis(side=2, at=-3:3, labels=-3:3, 
     lwd=0, lwd.ticks=1, las=2, cex.axis=0.8)

# add title
title(main=expression(paste("Temperature ","Anomaly ","(",degree,"C)")), 
      adj=0, cex.main=0.8)
title(main="(Difference from 1980-2015 annual mean)", adj=0, cex.main=0.6, 
      line=0, font.main=1)

# add horizontal lines
for (j in -3:3) {
	lines(1:12, rep(j, 12), lty=3)
}

# add yearly temperature lines
for (j in 1:nrow(d)) {
	lines(1:12, as.numeric(d[j,])+sa, col=colours[j])
}

This produces the following image:

This is close enough!

So how do we create the animation?  The old skool way of doing this is to create every image in R and create an animated GIF with ImageMagick.  So we'll do that

The slightly adjusted code to create every image is here (creates 137 PNGs in working directory!):

#
# Assumes you have run the code above to create d and sa
#

doplot <- function(i) {

	x   <- d[i,] + sa
	col <- colours[i]

	mmin <- -3
	mmax <- 3
	par(mar=c(2.5,2.5,2,0))
	plot(1:12, x, type="l", ylim=c(mmin,mmax), 
             yaxt="n", xaxt="n", bty="n", col=col)
	axis(side=1, at=1:12, labels=c("Jan","Feb","Mar",
                                       "Apr","May","Jun",
                                       "Jul","Aug","Sep",
                                       "Oct","Nov","Dec"), 
             lwd=0, lwd.ticks=1, cex.axis=1.5)

	axis(side=2, at=-3:3, labels=-3:3, lwd=0, 
             lwd.ticks=1, las=2, cex.axis=1.5)

	title(main=expression(paste("Temperature ","Anomaly ","(",degree,"C)")), 
              adj=0, cex.main=1.5)
	title(main="(Difference from 1980-2015 annual mean)", 
              adj=0, cex.main=1, line=-0.5, font.main=1)

	# add horizontal lines
	for (j in -3:3) {
		lines(1:12, rep(j, 12), lty=3)
	}

	# add other years
	for (k in 1:i) {
		lines(1:12, d[k,]+sa, col=colours[k])
	}

	# add year label
	text(7, 0.8, rownames(d)[i], col=colours[i], font=2, cex=2)

}

# create plots/PNGs
for (j in 1:nrow(d)) {
	
	name <- paste("plot",sprintf("%03.f", j),".png", sep="")
	png(name, width=800, height=600)
	doplot(j)
	dev.off()
}

Then in ImageMagick, we use convert to create the animated GIF:

convert *.png -delay 3 -loop 1 globaltemp.gif
convert globaltemp.gif \( +clone -set delay 500 \) +swap +delete globaltempwpause.gif

Final result here:

Done