issue_comments

I think also I could make you a contributor or transfer the branch to someone. So that might be an option too.

Sorry guys I don't have time to do anything on this (or any open source stuff) at the moment.

This did used to have some tests, but they seem to have got lost somewhere along the way. I'll add the old tests back in, which would be a start.

This adds the extra code that @pelson suggested but it would be nice to create a little unit test with a mixed Dimension and Auxiliary coordinates. Is there an easy way to create a simple cube in Iris that could test this?

@shoyer I have not forgotten about this PR, I just been working a lot on a pandas PR updating the ci and it's been taking up a lot of my spare time.

I have merged upstream and it seems to be failing but I don't think it's related to this PR, as a clean clone of master branch is also failing https://travis-ci.org/nparley/xarray/jobs/132872124

@shoyer @pelson @rhattersley Is there anything you think could be added to this PR or is this ok for now?

OK this update uses iris.fileformats._pyke_rules.compiled_krb.fc_rules_cf_fc as @pelson mentioned in the comment above. I also added a cell method to the test

OK the latest commit tries to do the cell methods:

``` python

print(cube) sea_water_potential_temperature / (degC) (-- : 148; -- : 180) Auxiliary coordinates: latitude x x longitude x x Scalar coordinates: depth: 4.99994 m, bound=(0.0, 10.0) m time: 1-01-01 12:00:00 Attributes: Conventions: CF-1.5 Cell methods: mean: time_counter print(xarray.DataArray.from_iris(cube)) <xarray.DataArray u'votemper' (dim0: 148, dim1: 180)> array([[ nan, nan, nan, ..., nan, nan, nan], [ nan, nan, nan, ..., nan, nan, nan], [ nan, nan, nan, ..., nan, nan, nan], ..., [ nan, nan, nan, ..., nan, nan, nan], [ nan, nan, nan, ..., nan, nan, nan], [ nan, nan, nan, ..., nan, nan, nan]], dtype=float32) Coordinates: deptht float64 5.0 time_counter object 1-01-01 12:00:00 nav_lat (dim0, dim1) float32 -78.1906 -78.1906 -78.1906 -78.1906 ... nav_lon (dim0, dim1) float32 80.0 82.0 83.9999 85.9999 87.9999 ... * dim0 (dim0) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 ... * dim1 (dim1) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 ... Attributes: units: degC long_name: Temperature standard_name: sea_water_potential_temperature cell_methods: time_counter: mean Conventions: CF-1.5 print(xarray.DataArray.from_iris(cube).to_iris()) sea_water_potential_temperature / (degC) (dim0: 148; dim1: 180) Dimension coordinates: dim0 x - dim1 - x Auxiliary coordinates: latitude x x longitude x x Scalar coordinates: depth: 4.99993801117 m time: 1-01-01 12:00:00 Attributes: Conventions: CF-1.5 Cell methods: mean: time_counter ```

The comment was just to mention that I had based my code on the code from cdms2 from #236 as suggested by @shoyer. I guess this comment would not be needed in the final merge.

Latest commit fixes bug if the Iris cube does not have any dimension coordinates. What do people think about keeping or dropping the automatic dimension coordinates that xarray creates? And if dropping should be done should all coordinates named dim{x} be dropped? Or are we happy with just leaving them there?

For example at the moment:

``` python

print(cube) sea_water_potential_temperature / (degC) (-- : 148; -- : 180) Auxiliary coordinates: latitude x x longitude x x Scalar coordinates: depth: 4.99994 m, bound=(0.0, 10.0) m time: 1-01-01 12:00:00 Attributes: Conventions: CF-1.5 Cell methods: mean: time_counter print(xarray.DataArray.from_iris(cube).to_iris()) sea_water_potential_temperature / (degC) (dim0: 148; dim1: 180) Dimension coordinates: dim0 x - dim1 - x Auxiliary coordinates: latitude x x longitude x x Scalar coordinates: depth: 4.99993801117 m time: 1-01-01 12:00:00 Attributes: Conventions: CF-1.5

```

Ok, bounds can just be something to add to the converter if / when it's supported.

There is a problem when it comes to bounds. Xarray adds the bounds as another DataArray to the DataSet, i.e.:

Data variables: votemper (dim0, dim1) float64 nan nan nan nan nan nan nan nan nan ... deptht_bnds (bnds) float64 0.0 10.0

I can get the name of this variable from the Coord attribute in the DataArray for another variable, i.e. ds['votemper'].coords['deptht'].bounds but can't get the bound array from the DataArray? is that correct?

docker pull nparley/xarray_test is a python 2.7 environment with Iris etc. which mirrors the travis CONDA_ENV=py27-cdat+iris+pynio.

It doesn't do everything, but it's a good first go. Here is an example:

``` python

import xarray import iris nc_file = 'E1_north_america.nc' cube = iris.load_cube(nc_file) darray = xarray.open_dataset(nc_file)['air_temperature'] print(darray) <xarray.DataArray 'air_temperature' (time: 240, latitude: 37, longitude: 49)> [435120 values with dtype=float32] Coordinates: * time (time) datetime64[ns] 1860-06-01 1861-06-01 ... * latitude (latitude) float32 15.0 16.25 17.5 18.75 20.0 ... * longitude (longitude) float32 225.0 226.875 228.75 ... forecast_period (time) timedelta64[ns] 449 days 18:00:00 ... forecast_reference_time datetime64[ns] 1859-09-01T06:00:00 height float64 1.5 Attributes: standard_name: air_temperature units: K Model scenario: E1 ukmo__um_stash_source: m01s03i236 source: Data from Met Office Unified Model 6.05 cell_methods: time: mean (interval: 6 hour) print(xarray.DataArray.from_iris(cube)) <xarray.DataArray u'air_temperature' (time: 240, latitude: 37, longitude: 49)> [435120 values with dtype=float32] Coordinates: * time (time) datetime64[ns] 1860-06-01 1861-06-01 ... * latitude (latitude) float32 15.0 16.25 17.5 18.75 20.0 ... * longitude (longitude) float32 225.0 226.875 228.75 ... forecast_reference_time datetime64[ns] 1859-09-01T06:00:00 height float64 1.5 forecast_period (time) timedelta64[ns] 449 days 18:00:00 ... Attributes: Model scenario: E1 STASH: m01s03i236 Conventions: CF-1.5 source: Data from Met Office Unified Model 6.05 standard_name: air_temperature units: K print(cube) air_temperature / (K) (time: 240; latitude: 37; longitude: 49) Dimension coordinates: time x - - latitude - x - longitude - - x Auxiliary coordinates: forecast_period x - - Scalar coordinates: forecast_reference_time: 1859-09-01 06:00:00 height: 1.5 m Attributes: Conventions: CF-1.5 Model scenario: E1 STASH: m01s03i236 source: Data from Met Office Unified Model 6.05 Cell methods: mean: time (6 hour print(darray.to_iris()) air_temperature / (K) (time: 240; latitude: 37; longitude: 49) Dimension coordinates: time x - - latitude - x - longitude - - x Auxiliary coordinates: forecast_period x - - Scalar coordinates: forecast_reference_time: 1859-09-01 06:00:00 height: 1.5 m Attributes: Model scenario: E1 source: Data from Met Office Unified Model 6.05 ukmo__um_stash_source: m01s03i236 ```

Reference https://github.com/pydata/xarray/issues/621 and https://github.com/pydata/xarray/issues/37. Uses template from https://github.com/pydata/xarray/pull/236.