/PyramidScheme.jl

Building and using pyramids for large raster data

Primary LanguageJuliaMIT LicenseMIT

PyramidScheme

PyramidScheme.jl is a package to easily and efficiently compute pyramids of a given datacube which might be larger than RAM. It uses DiskArrayEngine.jl as the computational backend. The pyramids can then be used to interactively explore the data.

The long term aim of PyramidScheme.jl is to enable computing based on the layers of the pyramid to enable a more interactive exploration of computations based of pyramided datasets.

Screencast.from.02.07.2024.16.24.48.webm

Usage

To compute the pyramids of a given data cube use the following steps:

using Rasters
using RasterDataSources
using ArchGDAL
using Statistics
using Extents
using PyramidScheme: PyramidScheme as PS
ras = Raster(WorldClim{Elevation},:elev, res="30s", lazy=true)
pyr = PS.Pyramid(ras)

# Now we can plot the data in GLMakie and get a nice interactive plot which uses the pyramids to provide a nice smooth experience by only loading the pixels which can fit into the Makie axis from an appropriate pyramid.

using GLMakie
plot(pyr)

Preparing the Pyramid for zarr data

If you want to prepare a pyramid which is saved inside of a Zarr dataset you can use the buildpyramids function. The buildpyramids function adds the Here we use the same example data as above, but we first save the data in a zarr file.

using YAXArrays
elevpath = getraster(WorldClim{Elevation},:elev, res="30s")
elev = Cube(elevpath)
zarrpath = tempname() * ".zarr"
savecube(elev, zarrpath)
@time PS.buildpyramids(zarrpath)

Example above

To reproduce the zooming example shown at the top of the README you can get the Above Ground Biomass here or try to load it from the cloud, but this might have some serious lag.

using YAXArrays, PyramidScheme
using GLMakie
p2020 = PS.Pyramid("https://s3.bgc-jena.mpg.de:9000/pyramids/ESACCI-BIOMASS-L4-AGB-MERGED-100m-2020-fv4.0.zarr")
replacenan(nanval) =  data -> <=(nanval)(data) ? NaN32 : Float32(data)
p2020nan = replacenan(0).(p2020)
plot(p2020nan, colormap=:speed, colorscale=sqrt)

# to save the base data use YAXArrays to load the underlying data
c = Cube("https://s3.bgc-jena.mpg.de:9000/pyramids/ESACCI-BIOMASS-L4-AGB-MERGED-100m-2020-fv4.0.zarr")
savecube(c, "somepath.zarr")
# Build the pyramid from this data locally
buildpyramids("somepath.zarr")