[CF-metadata] Salinity units

alison.pamment at stfc.ac.uk alison.pamment at stfc.ac.uk
Wed May 27 11:11:48 MDT 2015

Dear Nan, All,

Certainly this topic has come up several times and we never seem to quite get to a solution that suits everybody.

I don’t know why 1e-3 was originally chosen for use in the standard name table, but even if you go back to version 1 it is in there, which means it was agreed prior to 2006 when the CF website at PCMDI was set up.

The last time the question of salinity units was aired in detail was during the TEOS-10 discussions in 2011. Unfortunately, the mailing list archive seems to be unavailable at the moment, but I can vouch for the fact that the current definitions of the salinity names came from the very detailed discussions that we had at that time. In particular, we added the following wording to the definition of sea_water_practical_salinity: ‘Practical Salinity, S_P, is defined on the Practical Salinity Scale of 1978 (PSS-78) and is calculated from the electrical conductivity of sea water (as well as temperature and pressure). Officially S_P is dimensionless so that, while convenient, and while it is common practice, it is not officially sanctioned to say S_P = 35 psu. Often authors use PSS-78, as in S_P = 35 PSS-78. If salinity was measured using remote sensing techniques and not conductivity, then it is recommended that additional metadata (calibration/validation information) be described in the variable comment attribute.’

Once upon a time (back in 2009) there was a discussion about allowing CF to use ‘psu’ as a unit in its own right, but I think the TEOS-10 discussion made it clear that ‘psu’ is not really a unit at all, so that idea was dropped and we continued to use 1e-3.

We should remember that the canonical unit of ‘1e-3’ doesn’t prevent anyone using ‘1’ in their files if they prefer it, and vice versa. As Jim has already pointed out, UDunits can certainly cope with that. So in one sense, it doesn’t really matter to CF which we choose as the canonical unit as long as we can agree and, most importantly, make the definition really really clear so that consumers of the files know how to interpret the data.

In previous discussions there has never been unanimous agreement about whether it is better to use ‘1e-3’ or ‘1’. My niggling concern about changing the unit after all these years is whether it will lead to misinterpretation of existing data files. Is that going to be a problem?  We have in the past changed the canonical units of standard names, but only to correct outright errors, rather than to change the interpretation of a name. How big a problem is it for the oceanographic community if we don’t change the unit?

If we do decide to go with ‘1’ as the canonical unit, is there a reference, such as TEOS-10, which we can use to support our decision? It would be useful to include it in the definition and hopefully reduce the need to keep revisiting this same question.

Either way, I think we can improve further on the definition to help people better understand the data.

Best wishes,

Alison Pamment                                 Tel: +44 1235 778065
NCAS/Centre for Environmental Data Archival    Email: alison.pamment at stfc.ac.uk<mailto:J.A.Pamment at rl.ac.uk>
STFC Rutherford Appleton Laboratory
R25, 2.22
Harwell Oxford, Didcot, OX11 0QX, U.K.

From: CF-metadata [mailto:cf-metadata-bounces at cgd.ucar.edu] On Behalf Of Nan Galbraith
Sent: 27 May 2015 15:45
To: Rich Pawlowicz
Cc: cf-metadata at cgd.ucar.edu
Subject: Re: [CF-metadata] Salinity units

Hi all -

The '.001' units for P.S. doesn't mean that stored values of practical
salinity differs from A.S. by 'a factor of around a 1000', as far as I
know. If that's the logical inference, then this unit is really a problem,
and maybe we should do something about it.

I wish my CF email archive went back a little further, because there's
nothing (since 2004) that I can find that explains the rationale for
this unit. It certainly *looks* like a compromise between a unit for a
non-dimensional variable and PPT ... When this was originally under
discussion, way back when, I'll bet someone argued that it would eventually
be a big problem.  I'd really love to see that email thread!

Cheers -

On 5/26/15 11:52 AM, Rich Pawlowicz wrote:

I’m not sure what the best answer is either, but I think the “correct” way is
to have people deal with Practical Salinity in some special fashion in
their workflow, because it *is* defined in a weird way that is generally
incompatible with the general idea of ‘quantities with units’) - getting
a salinity definition that is aligned with the way all other quantities
in the world are defined was one of the motivating factors behind TEOS-10!

So, essentially people would have to make their own choice about
what to do with ‘practical salinity’ for whatever they are doing.

I will point out, though, that having two kinds of data that differ
numerically by a factor of around a 1000 is a good way of getting
them to realize that they really are not exactly compatible - you
wouldn’t *want* Practical Salinity and Absolute salinity on the same
plot (“look - salinity increased by 0.16 g/kg everywhere in 2010!”)

But I understand that one might want to make this as painless as

On May 26, 2015, at 8:48 AM, Signell, Richard <rsignell at usgs.gov<mailto:rsignell at usgs.gov>> wrote:

Thanks for this.   Yes, I guess my concern is that folks will do a
catalog search for *salinity* variables, and with a few spot checks,
see that they are have data values in the range of 29-36 or so, and
then go ahead and run a workflow that converts all units using the
units attribute.   And if "practical salinity" has units of "1" and
"absolute salinity" has units of "g/kg" = "0.001", then the data might
not appear on that fixed y-axis plot with [29 36].     But I don't
have a good alternative.   I guess we have to rely that people will
realize from the standard_names that for comparison, you need to
estimate absolute salinity from practical salinity using tools like
GSW toolbox.


On Fri, May 22, 2015 at 3:59 PM, Rich Pawlowicz <rpawlowicz at eos.ubc.ca<mailto:rpawlowicz at eos.ubc.ca>> wrote:

Ummm…I’m not entirely what you are asking, but

a) PSS-78 Practical Salinity is a dimensionless number.  It was defined
such that "the numerical values of practical salinity would be similar to the
numerical values of previous salinity data, when expressed in ‰”, but
it isn’t in fact ppt or anything, and you shouldn’t be multiplying it up or
down by factors of 1000.

b) "Previous salinity data”, (Cox or Knudsen salinity) which
was obtained from titrations, does in fact represent a
mass fraction of something (because you are titrating
with a mass of silver). This was denoted by the ppt ‘unit'.

c) TEOS-10 Absolute Salinity is also a mass fraction (of dissolved solute
on the Reference Composition Salinity Scale). However, nowadays the
SI brochure suggests that different quantities should be distinguished
by their symbols, not their units.  So, there isn’t actually a recommended
unit for Absolute Salinity. You can write

S_A = 35 g/kg = 0.035 kg/kg = 35000 mg/kg

or, again using SI rules and treating the units as a ‘thing’:

S_A/(g/kg) = 35

and any of these are valid - the same way lengths can be in
meters or km or mm or whatever is handy (this is also
true for preformed salinity).

‘ppt’ is discouraged as a unit of mass fraction because (for example) it
could be confused with ‘part per trillion’

Now, the gsw toolbox assumes ‘g/kg’ for its TEOS-10 salinity inputs
and outputs, but YOU don’t have to do that if you don’t want to.

I admit it is a little magic how we can ESTIMATE Absolute Salinity (with
units) from Practical Salinity (without units), but keep in mind that this
is only ONE possible way of estimating Absolute Salinity, and in fact it is
a method that is metrologically somewhat suspect because of the
definition of PSS-78. S_A could also be obtained from density
measurements, for example - and then there is some other
conversion factor involving different units.


On May 22, 2015, at 1:01 PM, Signell, Richard <rsignell at usgs.gov<mailto:rsignell at usgs.gov>> wrote:


For sure dimensionless.  But "1.0", "0.001" or "g/kg"?

The latest version (27) of the CF Standard Name list

sea_water_salinity: "0.001"
sea_water_absolute_salinity: "g kg-1"
sea_water_practical_salinity:    "0.001"
sea_water_preformed_salinity:  "g kg-1"
sea_water_cox_salinity: "0.001"

and units packages, of course, would treat "g kg-1" the same as "0.001".

Yet in the IOC manual on equation of seawater:
it states (PDF page 176, printed page 166) that Practical Salinity
should have units of "1", while "Absolute Salinity" (the argument used
in the toolbox functions) and "Preformed Salinity" (used in numerical
ocean models) should have units "g kg-1".

So it appears that TEOS agrees with CF on units for Absolute Salinity
and Preformed Salinity, but not on Practical Salinity.

And OceanSites (as least here:
is using "sea_water_practical_salinity" with units of "1", so they are
consistent with the TEOS publication, but not the current CF
convention (v27).

On the TEOS site, there is software to calculate Absolute Salinity
from Practical Salinity.   So it would seem that the technically
correct thing to do would be to use the "gsw_SA_from_SP" routine to
convert OceanSites Practical Salinity (in units of "1") to Absolute
Salinity (in units of "g/kg") before comparing with the "Preformed
Salinity" output "g/kg" from ocean models.

I'm pretty confused though, so I'm cc'ing Rich Pawlowicz on this,
hoping for his input.


On Fri, May 22, 2015 at 1:49 PM, Lowry, Roy K. <rkl at bodc.ac.uk><mailto:rkl at bodc.ac.uk> wrote:

Dimensionless. Please????

This is the view of physical oceanographers for whom I have the greatest respect.

Cheers, Roy.
From: Reyna Jenkyns [reyna at uvic.ca<mailto:reyna at uvic.ca>]
Sent: 22 May 2015 18:06
To: cf-metadata at cgd.ucar.edu<mailto:cf-metadata at cgd.ucar.edu>; OceanSITES Data Management Team; Nan Galbraith
Subject: Re: [CF-metadata] Salinity units

I'm interested in this topic since I didn't realize what had been discussed previously, and now I think we must be non-compliant as well.  Is this documented formally in the CF documentation?

Reyna Jenkyns | Data Stewardship Team Lead - Digital Infrastructure
Ocean Networks Canada | T 250 853 3908 | oceannetworks.ca
University of Victoria PO Box 1700 STN CSC 2300 McKenzie Avenue Victoria, BC V8W 2Y2

From: CF-metadata <cf-metadata-bounces at cgd.ucar.edu><mailto:cf-metadata-bounces at cgd.ucar.edu> on behalf of Nan Galbraith <ngalbraith at whoi.edu><mailto:ngalbraith at whoi.edu>
Sent: Friday, May 22, 2015 10:03 AM
To: cf-metadata at cgd.ucar.edu<mailto:cf-metadata at cgd.ucar.edu>; OceanSITES Data Management Team
Subject: Re: [CF-metadata] Salinity units

Hello all -

It's been a long time, but is anyone interested in re-visiting the subject
of units for practical salinity in CF?

I was recently notified that my salinity data was likely to be
overlooked by
some users, because I'd used '1' as the units, not '.001'. Somehow, I'd
forgotten the (long-ago) discussion on the CF list about salinity units.

Some members of  the OceanSITES project are interested in revising our
format spec to encourage the use of '1' as an indication that salinity does
not have units - but, of course, we'd mostly rather remain CF-compliant.

Thanks for any feedback on this.

Cheers - Nan

On 6/17/09 2:48 AM, Lowry, Roy K wrote:

Dear All,

During an exercise with Alison mapping the CF Standard Names to a
units vocabulary in the BODC vocabulary server I noticed that the
units for salinity were '1.00E-03', i.e. parts per thousand. My
understanding in that since the introduction of the Practical
Salinity Scale that salinity is dimensionless with units of '1'.  Is
there agreement for our changing the units in the Standard Name

Cheers, Roy.



* Nan Galbraith        Information Systems Specialist *

* Upper Ocean Processes Group            Mail Stop 29 *

* Woods Hole Oceanographic Institution                *

* Woods Hole, MA 02543                 (508) 289-2444 *


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20150527/63c9ca1d/attachment-0001.html>

More information about the CF-metadata mailing list