tilearray.create_array

tilearray.create_array(service_url, bbox, crs, *, service_type=None, chunk_size=None, grid_shape=None, output_format=None, cache_dir=None, compute=False, compute_num_workers=None, dtype=dtype('float32'), tile_decoder=None, on_progress=None, **service_options)

Create an xarray DataArray backed by Dask from a remote service.

When compute=True, tile fetches run on Dask’s threaded scheduler with num_workers=compute_thread_pool_size(fetch_policy) (override via compute_num_workers) so AIMD can use the full fetch concurrency ceiling.

For lazy arrays (compute=False), prefer compute_with_policy() over plain .compute() so Dask worker count matches the fetch ceiling. Fetch policy metadata is stored on DataArray.attrs for that path.

Parameters:

• service_url (str | ServiceConfig)

• bbox (BoundingBox | tuple[float, float, float, float])

• crs (CRS | str | int)

• service_type (ServiceTypeEnum | None)

• chunk_size (tuple[int, int] | None)

• grid_shape (tuple[int, int] | None)

• output_format (Format | None)

• cache_dir (str | Path | None)

• compute (bool)

• compute_num_workers (int | None)

• dtype (str | dtype[Any])

• tile_decoder (Callable[[TileResponse, TileRequest], ndarray[Any, dtype[floating]]] | None)

• on_progress (Callable[[int, int, TileRequest, TileResponse], None] | None)

• service_options (Any)

Return type:

DataArray