cdutil.ANNUALCYCLE.climatology() failed in cdutil 3.1.0

[chengzhuzhang]

As discussed in an email with @doutriaux1 and @dnadeau4 the same function works in cdutil 2.10, while failed in cdutil 3.1.0

Below is the code:

import cdms2
import cdutil

filename = 'test_data_amip_r1i1p1_mo_regrid_3x3_sgp.nc'
fin = cdms2.open(filename)
var = fin('tas')
cdutil.setTimeBoundsMonthly(var)
var_season = cdutil.ANNUALCYCLE.climatology(var).squeeze()

Below is the error:

Traceback (most recent call last):
  File "cdutil_test.py", line 8, in <module>
    var_season = cdutil.ANNUALCYCLE.climatology(var).squeeze()
  File "/Users/zhang40/anaconda2/envs/arm_diags_env_0207_1/lib/python2.7/site-packages/cdms2/avariable.py", line 1706, in squeeze
    return(MV.squeeze(self))
  File "/Users/zhang40/anaconda2/envs/arm_diags_env_0207_1/lib/python2.7/site-packages/cdms2/MV2.py", line 352, in squeeze
    maresult, axes=axes, attributes=attributes, grid=grid, id=id)
  File "/Users/zhang40/anaconda2/envs/arm_diags_env_0207_1/lib/python2.7/site-packages/cdms2/tvariable.py", line 203, in __init__
    self.initDomain(axes, copyaxes=copyaxes)
  File "/Users/zhang40/anaconda2/envs/arm_diags_env_0207_1/lib/python2.7/site-packages/cdms2/tvariable.py", line 357, in initDomain
    raise CDMSError("Wrong number of axes to initialize domain.")
cdms2.error.CDMSError: Wrong number of axes to initialize domain.

[chengzhuzhang]

test_data_amip_r1i1p1_mo_regrid_3x3_sgp.nc.zip

Here is the data used for testing

[lee1043]

@chengzhuzhang it looks like the error is coming from using squeeze in-correctly. Please try replacing last line of your code as below:
var_season = cdutil.ANNUALCYCLE.climatology(var)(squeeze=1)

[lee1043]

var.shape --> (324, 1, 1): monthly time series on one grid (or area-averaged)
ann = cdutil.ANNUALCYCLE.climatology(var)
ann.shape --> (12, 1, 1): annual cycle of var
var_season = cdutil.ANNUALCYCLE.climatology(var)(squeeze=1)
var_season.shape --> (12,): annual cycle of var, lat/lon dimensions removed

[chengzhuzhang]

Thank you @lee1043. I talked to @dnadeau4 , it seems like using asma() also works. My question is, since the same line worked for an older version cdat2.10. Is it because cdat3.10 dropped support for squeeze() ?

[doutriaux1]

@chengzhuzhang you're right it might needs a cleanup at the cdms level. @dnadeau4 what do you think?

[lee1043]

@chengzhuzhang in my understanding asma returns numpy array instead of transient variable; so you lose coordinate information. I think it might be still useful to keep var type as transient variable.

>>> type(var)
<class 'cdms2.tvariable.TransientVariable'>
>>> var_season = cdutil.ANNUALCYCLE.climatology(var).asma()
>>> var_season.shape
(12, 1, 1)
>>> type(var_season)
<class 'numpy.ma.core.MaskedArray'>

[lee1043]

I agree usage of squeeze seems not very clear in the current documents, for example:
https://cdms.readthedocs.io/en/latest/search.html?q=squeeze&check_keywords=yes&area=default
https://cdms.readthedocs.io/en/latest/generated/MV2.squeeze.html?highlight=squeeze

[durack1]

@doutriaux1 this could be a candidate for migration into the https://github.com/CDAT/cdutil issue list

[github-actions]

Marking issue as stale, since there has been no activity in 30 days.

Unless the issue is updated or the 'stale' tag is removed, this issue will be closed in 7 days.

[durack1]

@jasonb5 I think this is still a valid bug? no