lair.hrrr.decompress_chunk#

lair.hrrr.decompress_chunk(zarr_id, compressed_data)[source]#

Decompress s3 chunk data.

Parameters:
  • zarr_id (ZarrId) – ZarrId instance

  • compressed_data (bytes) – compressed data

Returns:

np.array – decompressed chunk

See also

retrieve_object

Gets the bytes it takes.

get_value

Does both, and picks the point out of the chunk.

Examples

An analysis chunk is 150 by 150 half-precision numbers; here, a made-up one:

>>> import datetime as dt
>>> import numcodecs
>>> import numpy as np
>>> from lair.hrrr import ZarrId, decompress_chunk
>>> zarr_id = ZarrId(
...     dt.datetime(2024, 1, 15, 12), "sfc", "10m_above_ground", "UGRD", "anl"
... )
>>> raw = np.arange(150 * 150, dtype="<f2").tobytes()
>>> chunk = decompress_chunk(
...     zarr_id, bytes(numcodecs.blosc.compress(raw, b"zstd", 5))
... )
>>> chunk.shape, chunk[0, :3]
((150, 150), array([0., 1., 2.], dtype=float16))