As part of this migration, we can build tooling that converts this Parquet output to CF-trajectory compliant Zarr output (i.e., as a post-processing step) allowing for users to keep their existing scripts analysis scripts if they'd like, and for easier integration into packages like Trajan.
It's important that this tooling is able to handle large datasets without saturating working memory.
As part of this migration, we can build tooling that converts this Parquet output to CF-trajectory compliant Zarr output (i.e., as a post-processing step) allowing for users to keep their existing scripts analysis scripts if they'd like, and for easier integration into packages like Trajan.
It's important that this tooling is able to handle large datasets without saturating working memory.