Discrete scale when using tidyterra::geom_spatraster

[schonhose]

Hi,

I just came across your package yesterday and it help me to drastically limit the code I used to plot a spatRaster from the terra package. However, I am looking to plot a raster using a discrete scale, whereas the original has continuous values.

Now I am not sure how to proceed with this, as geom_spatraster does not support breaks. Is this something that can be easily added or is there any other way I can plot the results and then use scale_fill_manual to set the colors I want to use?

[dieghernan]

Hi @schonhose:

On ggplot2 system, you can create a discrete scale on continuos mapped values with the scale_*_binned() functions https://ggplot2.tidyverse.org/reference/scale_colour_continuous.html. Some times this is abbreviated with a _b suffix, like in scale_fill_viridis_b() https://ggplot2.tidyverse.org/reference/scale_viridis.html.

These functions are passed to binned_scale https://ggplot2.tidyverse.org/reference/binned_scale.html that supports a breaks argument, so that may be what you are looking for.

Notice that the scales provided with tidyterra have three versions: *_c for continous values, *_b for binned values and *_d for discrete values.

So, in short, I think you can use *_b/_binned functions for what you want. The other approach is to create a discrete variable and use the scale_fill_manual functions.

Find here a reprex showing several examples:

library(tidyterra)
library(terra)
#> terra 1.5.34
library(ggplot2)

f_volcano <- system.file("extdata/volcano2.tif", package = "tidyterra")
volcano2 <- rast(f_volcano)

volcano2
#> class       : SpatRaster 
#> dimensions  : 174, 122, 1  (nrow, ncol, nlyr)
#> resolution  : 5, 5  (x, y)
#> extent      : 1756969, 1757579, 5917003, 5917873  (xmin, xmax, ymin, ymax)
#> coord. ref. : NZGD2000 / New Zealand Transverse Mercator 2000 (EPSG:2193) 
#> source      : volcano2.tif 
#> name        : elevation 
#> min value   :  76.26222 
#> max value   :  195.5542

# Use scales binned
# With scales from tidyterra package

ggplot() +
  geom_spatraster(data = volcano2) +
  scale_fill_wiki_b()

# Control breaks, and use a scale from ggplot2

ggplot() +
  geom_spatraster(data = volcano2) +
  scale_fill_binned(breaks = c(90, 100, 130, 140, 180))

# Or create a categorial variable

volcano2 <- volcano2 %>%
  mutate(cat = cut(elevation, breaks = c(70, 90, 100, 130, 140, 180, 200),
                   include.lowest=TRUE))


ggplot() +
  geom_spatraster(data = volcano2, aes(fill = cat)) +
  scale_fill_manual(values=hcl.colors(6, "Inferno"))

^{Created on 2022-07-19 by the reprex package (v2.0.1)}

[dieghernan]

Just to complete my previous answer, note that all those approachs plots the raster after it have been categorised (e.g. continuos values are internally discretized). You have another option that is to discretize only the legend, this is achieved with guide_colorsteps():

library(tidyterra)
#> terra 1.5.34
library(ggplot2)

f_volcano <- system.file("extdata/volcano2.tif", package = "tidyterra")
volcano2 <- rast(f_volcano)

# This modify the plot and the guide
ggplot() +
  geom_spatraster(data = volcano2) +
  scale_fill_viridis_b()

# Use guide on _c to modify only the guide
ggplot() +
  geom_spatraster(data = volcano2) +
  scale_fill_viridis_c(guide = guide_colorsteps())

^{Created on 2022-07-19 by the reprex package (v2.0.1)}

[schonhose]

Thank you so much for your response!

I found another way by reclassifying the raster using raster <- terra::classify(raster, c(70, 90, 100, 130, 140, 180, 200)) after which I can use scale_fill_manual.