alternate behavior for range= argument in terra::plot() for SpatRasters to not mask extreme values

[mikoontz]

Hello!

I am trying to plot a SpatRaster with the color palette mapped to a specific range in the data. I can use the range= argument to set the bounds for the data to be mapped, such that the full range of the color palette is used. The default behavior in this case is to mask out values that are beyond the specified range. Is it possible to instead color those points using the extreme value of the color palette?

set.seed(1)
r = terra::rast(nrow = 5, ncol = 5)
terra::values(r) = sample(x = 1:25, size = 25)
terra::plot(r, col = viridis::inferno(100, direction = -1), range = c(5, 20))

If this is possible (or a reasonable feature to request), is it possible to also adjust the legend to suggest that the extreme colors represent values <= (and/or >=) the minimum (and/or maximum) specified in the range?

This would essentially be mimicking matplotlib's extend argument for creating a colorbar (https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.colorbar.html). That implementation uses triangles at the end of the colorbar to suggest the extreme colors of the palette represent all values smaller and/or larger than the range.

teunbrand hacked ggplot2 to do this here. (Note this implementation may not work in recent versions of ggplot2, but the author built a more generalized approach in their package here). And a real life Python example is here (but which uses extra set_under and set_over methods to force the extreme colors to be something different than the extreme colors in the supplied plotting palette.)

I'm certainly open to other ways of visualizing the >= or <= than triangles at the end of the colorbar. Even a label on the extreme values to literally say <= and/or >= next to the minimum value and/or maximum value of the range might suffice?

A real example of this might look like this:

With the barebones code to recreate it looking like this (with range specified, but which masks out values outside that range and with a squared off colorbar).

nightfire = read.csv("https://static-content.springer.com/esm/art%3A10.1038%2Fs41586-021-04325-1/MediaObjects/41586_2021_4325_MOESM2_ESM.csv")
r = terra::rast(nightfire, crs = sf::st_crs(4326)$wkt)
terra::plot(r, range = c(0, 60), col = viridis::inferno(100, direction = -1))

> packageVersion("terra")
[1] ‘1.7.71’
> terra::gdal(lib="all")
    gdal     proj     geos 
 "3.8.2"  "9.3.1" "3.11.2" 

[rhijmans]

Thank you for that suggestion. I have added the argument fill_range=FALSE. Setting it to TRUE may give you what you are after.

library(terra)
set.seed(1)
r = terra::rast(nrow = 5, ncol = 5, vals=sample(1:25, 25))
terra::plot(r, range=c(5, 20), fill_range=TRUE)

nightfire = read.csv("https://static-content.springer.com/esm/art%3A10.1038%2Fs41586-021-04325-1/MediaObjects/41586_2021_4325_MOESM2_ESM.csv")
x = terra::rast(nightfire, crs = "epsg:4326")
terra::plot(x, range = c(0, 60), col=viridis::inferno(100, direction=-1), fill_range=TRUE)

To only limit one extreme, you can now also do

terra::plot(x, range = c(NA, 60), col=viridis::inferno(100, direction=-1), fill_range=TRUE)

[mikoontz]

@rhijmans !! This is very slick! Thank you so much for implementing this suggestion. It gets me exactly what I was after.