solve chunking warning

[veenstrajelmer]

import dfm_tools as dfmt
file_nc = r'p:\11209231-001-dwsm-slib2023\02_Modelling\04_Mud\Mud_v2_20162017_rst2\DFM_OUTPUT_DWSM-FM_200m\DWSM-FM_200m_0000_map.nc'
uds = dfmt.open_partitioned_dataset(file_nc)

or:

import xugrid as xu
file_nc = r'p:\11208983-epd-support-tools\toolbox\short_simulation\HK-DFM_scenario\DFM_OUTPUT_HK-FMWAQ\HK-FMWAQ_0000_map.nc'
uds = xu.open_dataset(file_nc, chunks={'time':1})

Raises " UserWarning: The specified chunks separate the stored chunks along dimension "time" starting at index 1. This could degrade performance. Instead, consider rechunking after loading."

This might be because of the default chunks={"time":1} of open_partitioned_dataset. It can be supressed by using auto chunks (or auto chunks for time) but that has negative impact on performance:

uds = dfmt.open_partitioned_dataset(file_nc, chunks="auto")

When plotting a single timestep with uds.mesh2d_s1.isel(time=-1).ugrid.plot(), the performance decreases enormously when using auto chunks with large model outputs like 20-partitions DCSM model. How to proceed?

Additional info, this also happens with plain xarray example:

import xarray as xr
file_nc = r"p:\11209231-003-bes-modellering\hydrodynamica\simulations\sss_017\DFM_OUTPUT_sss\sss_0000_map.nc"
uds = xr.open_dataset(file_nc, chunks={"time":1})

Apply filterwarning zoals in xarray pytest (maybe only if chunkstime=1): https://github.com/pydata/xarray/blob/62a4d0f45a4d275c32387522051bbe493414e090/xarray/tests/test_backends.py#L5082