odc.geo.cog.save_cog_with_dask

odc.geo.cog.save_cog_with_dask(xx, dst='', *, compression=<odc.geo.types.Unset object>, compressionargs=None, level=None, predictor=<odc.geo.types.Unset object>, blocksize=<odc.geo.types.Unset object>, bigtiff=True, overview_resampling='nearest', tile_batching=4, tile_batching_threshold=20, merge_compressed_overviews=4, aws=None, azure=None, client=None, stats=True, **kw)[source]

Save a Cloud Optimized GeoTIFF to S3, Azure Blob Storage, or file with Dask.

To optimise performance, pre-chunk your data to the desired COG block sizes. Then configure batching and overview merging as needed:

  • tile_batching is how many tiles get grouped into a single compression task. If it’s an integer, that fixed number of tiles are compressed together. If it’s a callable, it is given the total tile count and returns a number of tiles to compress at once. Combining small tile tasks can reduce overhead.

  • tile_batching_threshold is the minimum number of tiles in a given layer before any tile batching occurs. If a layer has fewer tiles than this threshold, batching is skipped because overhead might outweigh the benefits.

  • merge_compressed_overviews merges n topmost overview layers after they have been compressed, reducing overhead from having multiple small overview layers. If you only have a few overviews, you can disable or lower this. By default, it’s set to merge up to four of the highest overview layers.

Parameters:

  • xx (DataArray) – Pixels as xarray.DataArray backed by Dask.

  • dst (str) – S3, Azure URL, or file path.

  • compression (Union[str, Unset]) – Compression to use, default is DEFLATE.

  • level (Union[int, float, None]) – Compression “level”, depends on chosen compression.

  • predictor (Union[int, bool, Unset]) – TIFF predictor setting.

  • compressionargs (Optional[Any]) – Any other compression arguments.

  • overview_resampling (Union[int, str]) – Resampling method used for computing overviews.

  • tile_batching (Union[int, Callable[[int], int]]) – How many tiles to group into each compression task. Can be an integer or a function of total tile count. Default is 4.

  • tile_batching_threshold (int) – Minimum number of tiles in a layer before batching is applied. Default is 20.

  • merge_compressed_overviews (Optional[int]) – Merges that many highest-level overview layers after compression into a single chunk. Default is 4. Set to None or 0 to disable merging.

  • blocksize (Union[Unset, int, list[Union[int, tuple[int, int]]]]) – Configure block sizes for main and overview images.

  • bigtiff (bool) – Generate BigTIFF by default, set to False to disable.

  • aws (Optional[dict[str, Any]]) – Configure AWS write access.

  • azure (Optional[dict[str, Any]]) – Azure credentials/config.

  • client (Optional[Any]) – Dask client.

  • stats (bool | int) – Set to False to disable stats computation. If True, uses a mid-level layer to compute stats.

Return type:

Any

Returns:

Dask delayed.