ZarrDatasets

ZarrDatasets.jl is a julia package to read Zarr datasets based on the native julia implementation Zarr.jl using the CommonDataModel.jl interface.

In the following example, data from Copernicus Marine Service is loaded using ZarrDatasets and a subset is saved as a NetCDF file:

using CommonDataModel: @select
using Dates
using NCDatasets
using STAC
using ZarrDatasets

# get the data set URL from product_id and dataset_id and the STAC catalog
function copernicus_marine_catalog(product_id,dataset_id,
    stac_url = "https://stac.marine.copernicus.eu/metadata/catalog.stac.json",
    asset = "timeChunked")

    cat = STAC.Catalog(stac_url);
    item_canditates = filter(startswith(dataset_id),collect(keys(cat[product_id].items)))
    # use last version per default
    dataset_version_id = sort(item_canditates)[end]
    item = cat[product_id].items[dataset_version_id]
    return href(item.assets[asset])
end

product_id = "MEDSEA_MULTIYEAR_PHY_006_004"
dataset_id = "med-cmcc-ssh-rean-d"

url = copernicus_marine_catalog(product_id,dataset_id)
ds = ZarrDataset(url);

# longitude, latitude and time are the coordinate variables defined in the
# zarr dataset
ds_sub = @select(ds, time == DateTime(2001,1,1)
    && 7 <= longitude <= 11
    && 42.3 <= latitude <= 44.5)

# save selection as a NetCDF file
NCDataset("$(dataset_id)_selection.nc","c") do ds_nc
    write(ds_nc,ds_sub)
end

Using a pre-existing Zarr array or store

It's also simple to wrap an existing Zarr array, or a manually constructed Zarr store, in a ZarrDataset:

zg = zopen("/path/to/zarr")
zd = ZarrDataset(zg)

and you can pass a constructed Zarr.AbstractStore similarly.