Philippe Miron

Any help on this? I'm seeing this message even though the created mbtile files are fine.

Is there any info on _when_ it will be included? This issue is 10 years old.

Thanks for the suggestion, I will take a look at Awkward Array. I have not considered pandas, the data set is not that large ~17 GB, but I believe xarray...

Hi @rabernat, We are still testing out file format and actually putting together a Notebook for the EarthCube annual meaning benchmarking typical workflow tasks with xarray, pandas, and awkward array....

See [here](https://data.coaps.fsu.edu/pub/pmiron/array.nc). I'm also working with @jpivarski on a parquet representation that would be handled by [awkward array](https://awkward-array.readthedocs.io/en/latest/).

``` Dimensions: (traj: 5, obs: 10345) Coordinates: ID (traj) int64 ... longitude (obs) float32 ... latitude (obs) float32 ... time (obs) datetime64[ns] ... ids (obs) int64 ... Dimensions without coordinates:...

Hi @dhruvbalwada, The variables are indeed each stored into a 1-d ragged array of size `['obs']`. The separation can be easily done using the `rowsize['traj']` variable that contains the number...

To perform operations per trajectory, you can chunk it like this: ``` # align chunks of the dask array with the trajectories chunk_settings = {'obs': tuple(rowsize.tolist())} ds = xr.open_dataset(path_gdp, chunks=chunk_settings)...

I'm not 100% certain of what is the issue. I managed to create the XarrayZarrRecipe: ``` recipe = XarrayZarrRecipe(pattern, target_chunks={'obs': tuple(obs_per_file)}, process_chunk=preprocess_gdp, xarray_open_kwargs={'decode_times':False}) ``` ``` recipe XarrayZarrRecipe(file_pattern=, inputs_per_chunk=1, target_chunks={'obs': (417,...

> I will try to spend some time playing around with it. I've modified the preprocess function to take a `xarray.Dataset` as a parameter so we can pass it directly...