Check whether salinity units should be psu or ppt

[jkrasting]

@StephenGriffies agreed to check whether salinity should be expressed as psu or ppt for diagnostic output.

[adcroft]

From the fount of all knowledge (http://en.wikipedia.org/wiki/Salinity), TEOS-10 recommends use of a mass fraction (g/kg). The wiki page has this example:

A sample of seawater from most locations with a chlorinity of 19.37 ppt will have a Knudsen salinity of 35.00 ppt, a PSS-78 practical salinity of about 35.0, and a TEOS-10 absolute salinity of about 35.2 g/kg. The electrical conductivity of this water at a temperature of 15 °C is 42.9 mS/cm.

I think we are using "ppt" unless we are using TEOS-10, in which case we need to adjust all our initial conditions and analyses... :(

[StephenGriffies]

I think we should use ppt, although not yet using TEOS-10. We are in effect ignoring the "correction" term needed for absolute salinity. With ppt, we can then sensibly state that the total salt content of the ocean is the salinity integrated over the ocean. Likewise, we can accept the mass of salt exchanged with sea ice.

So in summary, we should remove "psu" in favor of "ppt".

[StephenGriffies]

The OMDP document for CMIP6 says the salinity units are "1e-3". That is, they should be "ppt". I will start to make the changes in the diagnostic registers for salinity diagnostics that I find.