Overview¶
Within this notebook, we will cover:
How to load a Kerchunk pre-generated reference file into Xarray as if it were a Zarr store.
Prerequisites¶
| Concepts | Importance | Notes |
|---|---|---|
| Kerchunk Basics | Required | Core |
| Xarray Tutorial | Required | Core |
Time to learn: 45 minutes
Opening Reference Dataset with Fsspec and Xarray¶
One way of using our reference dataset is opening it with Xarray. To do this, we will create an fsspec filesystem and pass it to Xarray.
# create an fsspec reference filesystem from the Kerchunk output
import fsspec
import xarray as xr
fs = fsspec.filesystem(
"reference",
fo="references/ARG_combined.json",
remote_protocol="s3",
remote_options={"anon": True},
skip_instance_cache=True,
)
m = fs.get_mapper("")
ds = xr.open_dataset(m, engine="zarr", backend_kwargs={"consolidated": False})---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
Cell In[1], line 13
9 remote_options={"anon": True},
10 skip_instance_cache=True,
11 )
12 m = fs.get_mapper("")
---> 13 ds = xr.open_dataset(m, engine="zarr", backend_kwargs={"consolidated": False})
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/xarray/backends/api.py:619, in open_dataset(filename_or_obj, engine, chunks, cache, decode_cf, mask_and_scale, decode_times, decode_timedelta, use_cftime, concat_characters, decode_coords, drop_variables, create_default_indexes, inline_array, chunked_array_type, from_array_kwargs, backend_kwargs, **kwargs)
607 decoders = _resolve_decoders_kwargs(
608 decode_cf,
609 open_backend_dataset_parameters=backend.open_dataset_parameters,
(...) 615 decode_coords=decode_coords,
616 )
618 overwrite_encoded_chunks = kwargs.pop("overwrite_encoded_chunks", None)
--> 619 backend_ds = backend.open_dataset(
620 filename_or_obj,
621 drop_variables=drop_variables,
622 **decoders,
623 **kwargs,
624 )
625 ds = _dataset_from_backend_dataset(
626 backend_ds,
627 filename_or_obj,
(...) 638 **kwargs,
639 )
640 return ds
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/xarray/backends/zarr.py:1747, in ZarrBackendEntrypoint.open_dataset(self, filename_or_obj, mask_and_scale, decode_times, concat_characters, decode_coords, drop_variables, use_cftime, decode_timedelta, group, mode, consolidated, storage_options, zarr_format, store, engine, use_zarr_fill_value_as_mask, cache_members)
1745 filename_or_obj = _normalize_path(filename_or_obj)
1746 if not store:
-> 1747 store = ZarrStore.open_group(
1748 filename_or_obj,
1749 group=group,
1750 mode=mode,
1751 consolidated=consolidated,
1752 consolidate_on_close=False,
1753 storage_options=storage_options,
1754 use_zarr_fill_value_as_mask=None,
1755 zarr_format=zarr_format,
1756 cache_members=cache_members,
1757 )
1759 store_entrypoint = StoreBackendEntrypoint()
1760 with close_on_error(store):
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/xarray/backends/zarr.py:797, in ZarrStore.open_group(cls, store, mode, group, consolidated, consolidate_on_close, storage_options, append_dim, write_region, safe_chunks, align_chunks, zarr_format, use_zarr_fill_value_as_mask, write_empty, cache_members)
774 @classmethod
775 def open_group(
776 cls,
(...) 790 cache_members: bool = True,
791 ):
792 (
793 zarr_group,
794 consolidate_on_close,
795 close_store_on_close,
796 use_zarr_fill_value_as_mask,
--> 797 ) = _get_open_params(
798 store=store,
799 mode=mode,
800 group=group,
801 consolidated=consolidated,
802 consolidate_on_close=consolidate_on_close,
803 storage_options=storage_options,
804 use_zarr_fill_value_as_mask=use_zarr_fill_value_as_mask,
805 zarr_format=zarr_format,
806 )
808 return cls(
809 zarr_group,
810 mode,
(...) 819 cache_members=cache_members,
820 )
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/xarray/backends/zarr.py:1951, in _get_open_params(store, mode, group, consolidated, consolidate_on_close, storage_options, use_zarr_fill_value_as_mask, zarr_format)
1947 else:
1948 # we have determined that we don't want to use consolidated metadata
1949 # so we set that to False to avoid trying to read it
1950 open_kwargs["use_consolidated"] = False
-> 1951 zarr_group = zarr.open_group(store, **open_kwargs)
1953 close_store_on_close = zarr_group.store is not store
1955 # we use this to determine how to handle fill_value
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/api/synchronous.py:567, in open_group(store, mode, cache_attrs, synchronizer, path, chunk_store, storage_options, zarr_format, meta_array, attributes, use_consolidated)
496 def open_group(
497 store: StoreLike | None = None,
498 *,
(...) 508 use_consolidated: bool | str | None = None,
509 ) -> Group:
510 """Open a group using file-mode-like semantics.
511
512 Parameters
(...) 564 The new group.
565 """
566 return Group(
--> 567 sync(
568 async_api.open_group(
569 store=store,
570 mode=mode,
571 cache_attrs=cache_attrs,
572 synchronizer=synchronizer,
573 path=path,
574 chunk_store=chunk_store,
575 storage_options=storage_options,
576 zarr_format=zarr_format,
577 meta_array=meta_array,
578 attributes=attributes,
579 use_consolidated=use_consolidated,
580 )
581 )
582 )
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/core/sync.py:155, in sync(coro, loop, timeout)
152 return_result = next(iter(finished)).result()
154 if isinstance(return_result, BaseException):
--> 155 raise return_result
156 else:
157 return return_result
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/core/sync.py:116, in _runner(coro)
111 """
112 Await a coroutine and return the result of running it. If awaiting the coroutine raises an
113 exception, the exception will be returned.
114 """
115 try:
--> 116 return await coro
117 except Exception as ex: # noqa: BLE001 -- the caller re-raises the returned exception
118 return ex
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/api/asynchronous.py:894, in open_group(store, mode, cache_attrs, synchronizer, path, chunk_store, storage_options, zarr_format, meta_array, attributes, use_consolidated)
828 """Open a group using file-mode-like semantics.
829
830 Parameters
(...) 882 The new group.
883 """
885 _warn_unimplemented_kwargs(
886 {
887 "cache_attrs": cache_attrs,
(...) 891 }
892 )
--> 894 store_path = await make_store_path(store, mode=mode, storage_options=storage_options, path=path)
895 if attributes is None:
896 attributes = {}
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/storage/_common.py:481, in make_store_path(store_like, path, mode, storage_options)
476 raise ValueError(
477 "'path' was provided but is not used for FSMap store_like objects. Specify the path when creating the FSMap instance instead."
478 )
480 else:
--> 481 store = await make_store(store_like, mode=mode, storage_options=storage_options)
482 return await StorePath.open(store, path=path_normalized, mode=mode)
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/storage/_common.py:413, in make_store(store_like, mode, storage_options)
408 return FsspecStore.from_url(
409 store_like, storage_options=storage_options, read_only=_read_only
410 )
412 elif _has_fsspec and isinstance(store_like, FSMap):
--> 413 return FsspecStore.from_mapper(store_like, read_only=_read_only)
415 else:
416 raise TypeError(f"Unsupported type for store_like: '{type(store_like).__name__}'")
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/storage/_fsspec.py:210, in FsspecStore.from_mapper(cls, fs_map, read_only, allowed_exceptions)
186 @classmethod
187 def from_mapper(
188 cls,
(...) 191 allowed_exceptions: tuple[type[Exception], ...] = ALLOWED_EXCEPTIONS,
192 ) -> FsspecStore:
193 """
194 Create an FsspecStore from an FSMap object.
195
(...) 208 FsspecStore
209 """
--> 210 fs = _make_async(fs_map.fs)
211 return cls(
212 fs=fs,
213 path=fs_map.root,
214 read_only=read_only,
215 allowed_exceptions=allowed_exceptions,
216 )
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/zarr/storage/_fsspec.py:57, in _make_async(fs)
52 return fs
53 if fs.async_impl:
54 # Convert sync instance of an async fs to an async instance. Reuse the original
55 # constructor arguments rather than round-tripping through JSON, since storage
56 # options may hold objects that are not JSON-serializable (e.g. credentials).
---> 57 return type(fs)(*fs.storage_args, **{**fs.storage_options, "asynchronous": True})
59 if fsspec_version < parse_version("2024.12.0"):
60 raise ImportError(
61 f"The filesystem '{fs}' is synchronous, and the required "
62 "AsyncFileSystemWrapper is not available. Upgrade fsspec to version "
63 "2024.12.0 or later to enable this functionality."
64 )
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/fsspec/spec.py:128, in _Cached.__call__(cls, *args, **kwargs)
125 if inst is not None:
126 return inst
--> 128 obj = super().__call__(*args, **kwargs, **strip_tokenize_options)
129 # Setting _fs_token here causes some static linters to complain.
130 obj._fs_token_ = token
File ~/micromamba/envs/kerchunk-cookbook/lib/python3.14/site-packages/fsspec/implementations/reference.py:785, in ReferenceFileSystem.__init__(self, fo, target, ref_storage_args, target_protocol, target_options, remote_protocol, remote_options, fs, template_overrides, simple_templates, max_gap, max_block, cache_size, **kwargs)
783 self.fss[k] = AsyncFileSystemWrapper(f, asynchronous=self.asynchronous)
784 elif self.asynchronous ^ f.asynchronous:
--> 785 raise ValueError(
786 "Reference-FS's target filesystem must have same value "
787 "of asynchronous"
788 )
ValueError: Reference-FS's target filesystem must have same value of asynchronousOpening Reference Dataset with Xarray and the Kerchunk Engine¶
As of writing, the latest version of Kerchunk supports opening an reference dataset with Xarray without specifically creating an fsspec filesystem. This is the same behavior as the example above, just a few less lines of code.
storage_options = {
"remote_protocol": "s3",
"skip_instance_cache": True,
"remote_options": {"anon": True}
} # options passed to fsspec
open_dataset_options = {"chunks": {}} # opens passed to xarray
ds = xr.open_dataset(
"references/ARG_combined.json",
engine="kerchunk",
storage_options=storage_options,
open_dataset_options=open_dataset_options,
)