In response to Issue https://github.com/google-research/arco-era5/issues/28, where it was identified that the chunk size in the model-level Zarr data was excessively large, I have made significant improvements. The existing chunk size was in the order of gigabytes. To address this, I introduced a modification in the target_chunks method of the XarrayZarrRecipe.
By incorporating the hybrid: 1 parameter in the target_chunks method, I was able to split each chunk per level.
In response to Issue https://github.com/google-research/arco-era5/issues/28, where it was identified that the chunk size in the model-level Zarr data was excessively large, I have made significant improvements. The existing chunk size was in the order of gigabytes. To address this, I introduced a modification in the
target_chunksmethod of theXarrayZarrRecipe.By incorporating the
hybrid: 1parameter in thetarget_chunksmethod, I was able to split each chunk per level.