Raster of missing values inside polygon in R

Question

I have a raster rrepresenting values from 0 to 50 for the world. There are large areas with missing values NA values, including:

• All oceans
• Large inland regions such as deserts

The problem is that when I plot r all NAs appear with the same colour regardless of whether they are oceans or deserts. This does not look nice, so I want to display inland NAs in a different colour than oceans.

I was thinking perhaps I could grab a map representing continents (e.g. map("world")), and create a layer of all NAs in r that fall inside continents. Then, I can just overlap this new raster on top of r in a different colour.

How can I create a new raster with all the inland missing values in a raster?

Answer 1

As you haven't provided any datasets to start with I have made some reproducible ones:

Firstly, here is a random raster. Missing values are inserted fo 70% of the cells:

library(raster)
rast <- raster(ncol=100, nrow=100, xmn=-130, xmx=-70, ymn=0, ymx=60)
rast[] <- 1:(100*100)
rast[runif(100*100) >= 0.30] <- NA

Next, I will load the US boundaries from mapdata. I used the package sf to create the polygons, but sadly these are not compatible with rasters yet, so I converted it back to a spatial polygon at the end:

library(sf)
library(sp)
library(mapdata)
states <- sf::st_as_sf(map("state", plot = FALSE, fill = TRUE))
states <-  as(states, 'Spatial')

Comparing these two shapefiles:

plot(states)
plot(rast, add = TRUE)

The easiest way of locating cells over land is to use the mask function. This only leaves cells which overlay the polygon provided. Before doing the mask though, I will set all NAs to a specified value, in this case -9999.

# Replace all NAs with a value
rast[is.na(rast)] <- -9999

overlay <- mask(rast, states)