One variable per history file

[anton-seaice]

In the CMIP data request, and often data folk prefer only have one diagnostic variable per file.

This only makes sense if there are multiple diagnostic timesteps per file (#556)

[dabail10]

We have a post-processing tool that converts the CESM history slice files to single variable timeseries files. Then there is another layer "CMORizing" that changes the attributes and to correspond to the CMIP data request. I'm not sure it is worth just writing a single variable per file. This would just create a lot of files. We had also hoped at some point to allow CESM to write the single variable timeseries files directly. That is, not just say each daily slice in a month. This would be daily for 100 years.

[anton-seaice]

The other argument is that it makes data easier to handle. e.g. its much easier to load 100 years of aice data if there are files with only aice, rather than all the history variables in them.

That is, not just say each daily slice in a month. This would be daily for 100 years.

Yeah, it only makes sense I think if the duration for a file can be long (ish)