html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/issues/374#issuecomment-82063927,https://api.github.com/repos/pydata/xarray/issues/374,82063927,MDEyOklzc3VlQ29tbWVudDgyMDYzOTI3,1217238,2015-03-17T03:01:57Z,2015-03-17T03:01:57Z,MEMBER,"By default we use the biggest units that divide all the given times equally. That's not necessarily SI, but it does result in more interpretable time values....
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,62242132
https://github.com/pydata/xarray/issues/374#issuecomment-82035634,https://api.github.com/repos/pydata/xarray/issues/374,82035634,MDEyOklzc3VlQ29tbWVudDgyMDM1NjM0,167164,2015-03-17T02:09:15Z,2015-03-17T02:09:30Z,NONE,"blegh... just noticed the different dates. Never mind :P 

Thanks for the help again
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,62242132
https://github.com/pydata/xarray/issues/374#issuecomment-82034427,https://api.github.com/repos/pydata/xarray/issues/374,82034427,MDEyOklzc3VlQ29tbWVudDgyMDM0NDI3,167164,2015-03-17T02:04:17Z,2015-03-17T02:04:17Z,NONE,"Ah, cool, that looks good, however, I think there might be a bug somewhere. Here's what I was getting originally:

```
In [62]: new_data['time'].encoding

Out[62]: {}

In [60]: new_data.to_netcdf('data/tumba_site_mean_2_year.nc')
```

resulted in:

```
$ ncdump ../projects/synthetic_forcings/data/tumba_site_mean_2_year.nc|grep time --context=2
netcdf tumba_site_mean_2_year {
dimensions:
        time = 35088 ;
        y = 1 ;
        x = 1 ;
variables:
        ...
        float time(time) ;
                time:calendar = ""proleptic_gregorian"" ;
                time:units = ""minutes since 2000-01-01 00:30:00"" ;
...

 time = 0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, 330, 360, 390, 420, 
    450, 480, 510, 540, 570, 600, 630, 660, 690, 720, 750, 780, 810, 840, 
```

And if I copy the encoding from the original file, as it was loaded:

```
In [65]: new_data['time'].encoding = data['time'].encoding

In [66]: new_data['time'].encoding

Out[66]: {'dtype': dtype('>f8'), 'units': 'seconds since 2002-01-01 00:30:00'}

In [67]: new_data.to_netcdf('data/tumba_site_mean_2_year.nc')
```

results in

```
$ ncdump ../projects/synthetic_forcings/data/tumba_site_mean_2_year.nc|grep time --context=1
dimensions:
        time = 35088 ;
        y = 1 ;
--
variables:
        ...
        double time(time) ;
                time:calendar = ""proleptic_gregorian"" ;
                time:units = ""seconds since 2002-01-01T00:30:00"" ;

...
 time = -63158400, -63156600, -63154800, -63153000, -63151200, -63149400, 
    -63147600, -63145800, -63144000, -63142200, -63140400, -63138600, 
```

Now the units are right, but the values are way off. I can't see anything obvious missing from the encoding, compared to the xray docs, but I'm not sure how it works.

Also, since seconds are the base SI unit for time, I think it would be sensible to use seconds by default, if no encoding is given.
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,62242132
https://github.com/pydata/xarray/issues/374#issuecomment-82012686,https://api.github.com/repos/pydata/xarray/issues/374,82012686,MDEyOklzc3VlQ29tbWVudDgyMDEyNjg2,1217238,2015-03-17T01:01:52Z,2015-03-17T01:01:52Z,MEMBER,"This is actually already implemented but poorly documented. See the section on DataArray encoding: http://xray.readthedocs.org/en/stable/io.html#netcdf

Basically, you just need to add a units encoding to the time variable with the desired time units. 
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,62242132