Update open_mfdataset() to avoid data vars dim concatenation

[tomvothecoder]

Description

• Closes KeyError in spatial_average when open multiple datasets #136

This PR updates xcdat.open_mfdataset() with a default option for data_vars="minimal to avoid data vars dim concatenation (which breaks a reduction in spatial averaging).

The default xr.open_mfdataset() behavior is to concat dims across data vars. Adding data_vars="minimal" stops this from happening. Notice below that the time dimension is no longer added to dims of lat_bnds and lon_bnds when passing this kwarg.

>>> dir = "/p/user_pub/PCMDIobs/PCMDIobs2/atmos/3hr/pr/TRMM-3B43v-7/gn/v20200707"
>>> file_pattern = "pr_3hr_TRMM-3B43v-7_BE_gn_v20200707_1998*.nc"
>>> files = os.path.join(dir, file_pattern)

>>> ds_xr = xr.open_mfdataset(files)
>>> ds_xr2 = xr.open_mfdataset(files, data_vars="minimal")

>>> print(ds_xr)
<xarray.Dataset>
Dimensions:    (time: 2920, bnds: 2, lat: 400, lon: 1440)
Coordinates:
  * time       (time) datetime64[ns] 1998-01-01 ... 1998-12-31T21:00:00
  * lat        (lat) float64 -49.88 -49.62 -49.38 -49.12 ... 49.38 49.62 49.88
  * lon        (lon) float64 0.125 0.375 0.625 0.875 ... 359.1 359.4 359.6 359.9
Dimensions without coordinates: bnds
Data variables:
    time_bnds  (time, bnds) datetime64[ns] dask.array<chunksize=(248, 2), meta=np.ndarray>
    lat_bnds   (time, lat, bnds) float64 dask.array<chunksize=(248, 400, 2), meta=np.ndarray>
    lon_bnds   (time, lon, bnds) float64 dask.array<chunksize=(248, 1440, 2), meta=np.ndarray>
    pr         (time, lat, lon) float32 dask.array<chunksize=(248, 400, 1440), meta=np.ndarray>

>>> print(ds_xr2)
<xarray.Dataset>
Dimensions:    (time: 2920, bnds: 2, lat: 400, lon: 1440)
Coordinates:
  * time       (time) datetime64[ns] 1998-01-01 ... 1998-12-31T21:00:00
  * lat        (lat) float64 -49.88 -49.62 -49.38 -49.12 ... 49.38 49.62 49.88
  * lon        (lon) float64 0.125 0.375 0.625 0.875 ... 359.1 359.4 359.6 359.9
Dimensions without coordinates: bnds
Data variables:
    time_bnds  (time, bnds) datetime64[ns] dask.array<chunksize=(248, 2), meta=np.ndarray>
    lat_bnds   (lat, bnds) float64 dask.array<chunksize=(400, 2), meta=np.ndarray>
    lon_bnds   (lon, bnds) float64 dask.array<chunksize=(1440, 2), meta=np.ndarray>
    pr         (time, lat, lon) float32 dask.array<chunksize=(248, 400, 1440), meta=np.ndarray>

In #136, the time dimension was being added to lat_bnds and lon_bnds.
As a result, the lat and lon dims were being reduced down, leaving just the time dimension. This would produce the error KeyError: 'Check weights DataArray includes lat dimension.'
https://github.com/XCDAT/xcdat/blob/ca9dee4390cad2ad9856f9959e75981575cc69d7/xcdat/spatial_avg.py#L592-L596

More info on xarray.open_mfdataset().

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules

If applicable:

• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass with my changes (locally and CI/CD build)
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have noted that this is a breaking change for a major release (fix or feature that would cause existing functionality to not work as expected)

[codecov-commenter]

Codecov Report

Merging #143 (adb16c4) into main (68e8c00) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main      #143   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            7         7           
  Lines          345       346    +1     
=========================================
+ Hits           345       346    +1     

Impacted Files	Coverage Δ
xcdat/dataset.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 68e8c00...adb16c4. Read the comment docs.

[tomvothecoder]

Test script

# flake8:noqa F401

import os

import xarray as xr

import xcdat

dir = "/p/user_pub/PCMDIobs/PCMDIobs2/atmos/3hr/pr/TRMM-3B43v-7/gn/v20200707"
file_pattern = "pr_3hr_TRMM-3B43v-7_BE_gn_v20200707_1998*.nc"
files = os.path.join(dir, file_pattern)

#%%
# Test using `xr.open_mfdataset()`
ds_xr1 = xr.open_mfdataset(files)
pr_global1 = ds_xr1.xcdat.spatial_avg(
    axis=["lat", "lon"], data_var="pr"
).pr  # raises error

#%%
ds_xr2 = xr.open_mfdataset(files, data_vars="minimal")
pr_global2 = ds_xr2.xcdat.spatial_avg(axis=["lat", "lon"], data_var="pr").pr  # no error

#%%
# Test using `xcdat.open_mfdataset()` and `data_vars="minimal"`
ds_xcdat = xcdat.open_mfdataset(files)
pr_global3 = ds_xcdat.xcdat.spatial_avg(axis=["lat", "lon"]).pr  # no error

[tomvothecoder]

Hey Steve, here's the related xarray issue with the data_vars="minimal" fix: pydata/xarray#438