Remote I/O#

Utilities for loading OceanTACO data from HuggingFace.

load_hf_dataset#

ocean_taco.dataset.retrieve.load_hf_dataset(url='https://huggingface.co/datasets/nilsleh/OceanTACO/resolve/main/')[source]#

Load OceanTACO catalog from HuggingFace via tacoreader.

Parameters:

url (str) – Base URL of the HuggingFace dataset repository.

Returns:

TacoDataset catalog object.

load_tile_nc#

ocean_taco.dataset.retrieve.load_tile_nc(dataset, date, tile, data_source, cache_dir=None)[source]#

Load one TACO-format tile file, optionally caching to disk.

Parameters:
  • dataset – TacoDataset from load_hf_dataset().

  • date (str) – Date string YYYY-MM-DD.

  • tile (str) – Region tile name, e.g. ‘NORTH_PACIFIC_EAST’.

  • data_source (str) – Source token or filename, e.g. ‘l4_ssh’ or ‘l4_ssh.nc’.

  • cache_dir – If provided, cache to {cache_dir}/{date}/{tile}/{resolved_filename}.

Return type:

Dataset | None

Returns:

xr.Dataset or None if not found.

load_bbox_nc#

ocean_taco.dataset.retrieve.load_bbox_nc(dataset, date, bbox, data_source, cache_dir=None)[source]#

Load and merge tiles overlapping bbox for one requested data source.

Parameters:
  • dataset – TacoDataset from load_hf_dataset().

  • date (str) – Date string YYYY-MM-DD.

  • bbox (tuple) – Tuple (lon_min, lat_min, lon_max, lat_max).

  • data_source (str) – Source token or file name, e.g. ‘l4_ssh’ or ‘l4_ssh.nc’.

  • cache_dir – If provided, cache each tile to {cache_dir}/{date}/{tile}/{resolved_filename}.

Return type:

Dataset | None

Returns:

Merged xr.Dataset for the requested source, or None if no files found.

load_bbox_swot_nc#

ocean_taco.dataset.retrieve.load_bbox_swot_nc(dataset, date, bbox, cache_dir=None)[source]#

Load and merge L3 SWOT tiles overlapping bbox for a given date.

SWOT tiles carry per-tile auxiliary variables (track_ids, track_times) indexed by a track dimension whose size differs between tiles. These cannot be spatially merged, so they are dropped before combining. All spatial data variables (ssha_filtered, etc.) are preserved.

Parameters:
  • dataset – TacoDataset from load_hf_dataset().

  • date (str) – Date string YYYY-MM-DD.

  • bbox (tuple) – Tuple (lon_min, lat_min, lon_max, lat_max).

  • cache_dir – If provided, cache each tile to {cache_dir}/{date}/{tile}/l3_swot.nc.

Return type:

Dataset | None

Returns:

Merged xr.Dataset or None if no files found.