nc_time_axis

[Simon Tett]

Hi,
  I am reading a pp file containing one dataset. The pp file  was produced by some post processing of UM data in IDL. I can plot the pp field using qplt.pcolormesh but when I try and plot it with contourf I get a ValueError:The values must be numbers or instances of "nc_time_axis.CalendarDateTime". I am running on windows x64 using version 1.13 of iris.

 See below for the traceback. I had a look at nc_time_axis but can't see how to convert the time-cords from floats to whatever nc_time_axis wants. So a couple of questions:

1) Is this a bug?

2) How can I convert the time to whatever nc_time_axis wants?

Simon
----------------------------

The time coords are:
a.coord('time')
Out[100]:
DimCoord(array([ 19770.,  23370.,  26970.,  30570.,  34170.,  37770.,  41370.,
        44970.,  48570.,  52170.,  55770.,  59370.,  62970.], dtype=float32), bounds=array([[ 17970.,  21570.],
       [ 21570.,  25170.],
       [ 25170.,  28770.],
       [ 28770.,  32370.],
       [ 32370.,  35970.],
       [ 35970.,  39570.],
       [ 39570.,  43170.],
       [ 43170.,  46770.],
       [ 46770.,  50370.],
       [ 50370.,  53970.],
       [ 53970.,  57570.],
       [ 57570.,  61170.],
       [ 61170.,  64770.]], dtype=float32), standard_name='time', units=Unit('days since 0000-01-01 00:00:00', calendar='360_day'))



a=iris.load_cube(os.path.join(onePercent,'ot.pp'))
qplt.contourf(a)
Traceback (most recent call last):
  File "C:\Anaconda2\lib\site-packages\IPython\core\interactiveshell.py", line 2882, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-98-cb31dbf8f8c3>", line 1, in <module>
    qplt.contourf(a)
  File "C:\Anaconda2\lib\site-packages\iris\quickplot.py", line 187, in contourf
    result = iplt.contourf(cube, *args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\iris\plot.py", line 803, in contourf
    result = _draw_2d_from_points('contourf', None, cube, *args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\iris\plot.py", line 395, in _draw_2d_from_points
    result = draw_method(u, v, data, *args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\pyplot.py", line 2875, in contourf
    ret = ax.contourf(*args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\__init__.py", line 1892, in inner
    return func(ax, *args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\axes\_axes.py", line 5829, in contourf
    contours = mcontour.QuadContourSet(self, *args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\contour.py", line 864, in __init__
    self._process_args(*args, **kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\contour.py", line 1429, in _process_args
    x, y, z = self._contour_args(args, kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\contour.py", line 1508, in _contour_args
    x, y, z = self._check_xyz(args[:3], kwargs)
  File "C:\Anaconda2\lib\site-packages\matplotlib\contour.py", line 1534, in _check_xyz
    x = self.ax.convert_xunits(x)
  File "C:\Anaconda2\lib\site-packages\matplotlib\artist.py", line 199, in convert_xunits
    return ax.xaxis.convert_units(x)
  File "C:\Anaconda2\lib\site-packages\matplotlib\axis.py", line 1472, in convert_units
    ret = self.converter.convert(x, self.units, self)
  File "C:\Anaconda2\lib\site-packages\nc_time_axis\__init__.py", line 261, in convert
    raise ValueError('The values must be numbers or instances of '
ValueError: The values must be numbers or instances of "nc_time_axis.CalendarDateTime".

[RuthC]

Hi Simon,

I don't know what the problem is with your time coordinate, but since your data appear to be decadal means, I would be tempted to bypass the problem by swapping the time coordinate for a year coordinate:

import iris.coord_categorisation as cat

cat.add_year(a, 'time')
a.remove_coord('time')

Ruth