Area Plots with Intensity Coloring ~ el nino SST anomalies w/ ggplot2

I see many economy indicator graphs that show emphasis by shading in the curve under the area (while x-axis is time). The shade is stronger at higher values (example).

I did this in R below (ggplot2). This was a little more difficult that I’d expected. The color gradients are good to color each individual points depending on the strength, but not good (AFAIK) to color regions outside the dataset. The functions, geom_area, geom_ribbon geom_area, geom_linerange and the like can color themselves for different pieces, but we can’t specify color gradients within the piece. So I basically created a grid of points with different strengths dependent on the location of the y-axis.

Data is here (NOAA, first three lines commented out + I put a space in front of all minus signs) + similar examples at Learn R and Climate Charts and Graphs.

Plot below:

Simpler and faster to just do points, but was not what I wanted.

R code:

# ---
## NINO 3.4 SST Anomaly
## http://chartsgraphs.wordpress.com/2010/01/28/rclimate-script-nino-3-4-sst-anomaly-trends/
## http://learnr.wordpress.com/2009/05/18/ggplot2-three-variable-time-series-panel-chart/
nino <- read.table("/./NINO 3.4 SST Anomaly.txt", head = T)
nino$Week <- strptime(nino$Week, format = "%d%b%Y")
nino.name <- rep(c("Nino1+2", "Nino3", "Nino34", "Nino4"), each = 2*dim(nino)[1])
nino <- data.frame(Week = nino$Week, stack(nino[,-1]), nino.name)
nino$ind <- gsub(".[0-9]", "", nino$ind)

head(nino, 2)
##         Week values ind nino.name
## 1 1990-01-03   23.4 SST   Nino1+2
## 2 1990-01-10   23.4 SST   Nino1+2

qplot(x=Week, y=values, data = subset(nino, ind == "SSTA" & nino.name == "Nino34"), geom = "point", col = values, fill = values, shape = I(15), solid = TRUE) + scale_colour_gradient2(low="blue", mid="white", high="red")

# -- fill area
nino34 <- subset(nino, ind == "SSTA" & nino.name == "Nino34")

cut <- 100
nino34.grid <- with(nino34, outer(values, seq(min(values), max(values), length=cut), function(X,Y) { (X >= 0 & X > Y & Y > 0) | (X <= 0 & X < Y & Y < 0) }))

nino34.grid <- t(apply(nino34.grid, 1, function(x) { x[x == F] <- NA; x*seq(min(nino34$values), max(nino34$values), length=cut) }))

colnames(nino34.grid) <- round(seq(min(nino34$values), max(nino34$values), length=cut), 2)
rownames(nino34.grid) <- as.character(nino34$Week)

nino34.grid <- melt(nino34.grid)

qplot(as.POSIXct(X1), X2, fill = value, data = nino34.grid, geom = "tile") + scale_fill_gradient2(low="blue", mid="white", high="red")

Filed under: Climate Change, ggplot2, R