How to write to zarr and read from zarr?

[psarka]

I can't find information on how to write and read from zarr 🤔 I was expecting this to work, but it does not :(

import rioxarray
import xarray

with rioxarray.open_rasterio("test.tif") as rds:
    rds.to_zarr("test.zarr", mode="w")

with xarray.open_dataset("test.zarr") as xds:
    xds.rio.to_raster("test2.tif")

It raises rioxarray.exceptions.TooManyDimensions: Only 2D and 3D data arrays supported. What am I missing?

[snowman2]

What does xds look like?

[psarka]

Like this:

<xarray.Dataset>
Dimensions:                        (band: 1, y: 10, x: 10)
Coordinates:
  * band                           (band) int64 1
  * x                              (x) float64 5.0 15.0 25.0 ... 75.0 85.0 95.0
  * y                              (y) float64 -5.0 -15.0 -25.0 ... -85.0 -95.0
Data variables:
    __xarray_dataarray_variable__  (band, y, x) float64 ...
    spatial_ref                    int64 ...

Full example script:

import numpy as np
import rasterio
import rioxarray
import xarray
from rasterio.transform import from_origin

rows = 10
cols = 10
data = np.arange(rows * cols).reshape((rows, cols)).astype(np.int16)
transform = from_origin(0, 0, 10, 10)

profile = {
    "driver": "GTiff",
    "dtype": "int16",
    "count": 1,
    "height": rows,
    "width": cols,
    "crs": "EPSG:4326",
    "transform": transform,
}

with rasterio.open("test.tif", "w", **profile) as dst:
    dst.write(data, 1)

# Open with rioxarray and save to Zarr
with rioxarray.open_rasterio("test.tif") as rds:
    rds.to_zarr("test.zarr", mode="w")

# Open zarr with xarray and save to raster
with xarray.open_dataset("test.zarr") as xds:
    xds.rio.to_raster("test2.tif")