... split a layer in tiles?

This document illustrates the use of the tiles function, to rapidly split a layer into multiple layers. This is useful if you want to perform operations on raster that can easily be made parallel, and do not require the full raster.

using SpeciesDistributionToolkit
using CairoMakie

To illustrate the tiling, we will grab the tree cover as given in the EarthEnv dataset, for a small spatial extent.

dataprovider = RasterData(EarthEnv, LandCover)
spatial_extent = (left = 8.412, bottom = 41.325, right = 9.662, top = 43.060)
trees = sum([
    SDMLayer(dataprovider; layer = i, full = true, spatial_extent...) for
    i in 1:4
])

SDM Layer with 31559 UInt8 cells
	Proj string: +proj=longlat +datum=WGS84 +no_defs
	Grid size: (209, 151)

Splitting the data into tiles can be done by calling the tiles function.

This will return a matrix with the same type as the layer given as its first argument. The second argument (the size of the matrix) can be omitted, and will default to (5, 5).

tree = tiles(trees, (2, 2));

This can now be plotted:

Code for the figure

tile_plot = heatmap(
    rand(tree);
    colormap = :Greens,
    figure = (; size = (800, 350)),
    axis = (;
        aspect = DataAspect(),
        xlabel = "Latitude",
        ylabel = "Longitude",
        title = "Relative tree cover",
    ),
)
Colorbar(tile_plot.figure[:, end + 1], tile_plot.plot; height = Relative(0.5))

This construct is very useful when your problem lends itself to naive parallelism.