Copernicus functions and tools fail with TypeError: open_dataset_from_arco_series() got an unexpected keyword argument 'opening_dask_chunks'.

[jjrob]

The full exception message looks similar to this:

RuntimeError: Failed to open Copernicus Marine Service dataset "https://s3.waw3-1.cloudferro.com/mdl-arco-time-026/arco/GLOBAL_MULTIYEAR_PHY_001_030/cmems_mod_glo_phy_my_0.083deg_static_202311--ext--mdt/static.zarr". Please check your internet connectivity and that your username and password is correct. The following error, reported by the copernicusmarine.download_functions.download_arco_series.open_dataset_from_arco_series() function, may indicate the problem: TypeError: open_dataset_from_arco_series() got an unexpected keyword argument 'opening_dask_chunks'.

The problem is that copernicusmarine 2.2.0 once again changed the interface of open_dataset_from_arco_series(). This time, the opening_dask_chunks parameter was renamed to optimum_dask_chunking. It also takes a different type of value, but now supports a default so we don't need to provide it at all. In MGET's CMEMS.py, CMEMSARCOArray._Open() should be updated to check whether copernicusmarine 2.2.0 is installed and call open_dataset_from_arco_series() accordingly.

[jjrob]

Actually, optimum_dask_chunking does require a value, but we can pass None. In copernicusmarine's own code, they call their own function get_optimum_dask_chunking() to compute a complex data structure that configures how zarr downloads should be chunked. We don't need to do this because under normal circumstances we already split our calls to copernicusmarine into chunks.