Input units for metpy.calc.density() and mixing_ratio_from_relative_humidity()

[figaroleo]

Dear All,

I can not find any such information, nor examples, in the website, sorry.

When passing temperature and pressure in the metpy formulas, in general, and in particular for the following functions:

metpy.calc.density(pressure, temperature, mixing_ratio, molecular_weight_ratio=<Quantity(0.62195691, 'dimensionless')>)
metpy.calc.mixing_ratio_from_relative_humidity(pressure, temperature, relative_humidity)

which are the input units to be used?

I suppose:

temperature: Kelvin
pressure: Pascal

but I can not find any clear statement.

Thanks for your time.

[Z-Richard]

There is no "required" input unit in MetPy because the arguments to these functions (pressure, temperature) are Pint.Quantity, which means that they have units attached to the numerical values such that MetPy could process them correctly. An example:

>>> import metpy.calc as mpcalc
>>> from metpy.units import units
>>> p = 1013 * units.millibar
>>> t = 273 * units.K
>>> q = 0.023 * units.dimensionless
>>> mpcalc.density(p, t, q)
<Quantity(1.27525864, 'kilogram / meter ** 3')>

Note that you can use something like t = 20 * units.celsius too.

[figaroleo]

Thanks!

I searched under the unidata.github.io/MetPy for the term ".magnitude" and I got only some mentions in examples.
Quite frustrating.
This should be clearly stated in the units discussion.

Thank really a lot for your support!

[dopplershift]

@figaroleo Apologies for the frustration. Your experience has identified a clear shortcoming in our Unit Tutorial that we definitely need to address. I've opened #2651 to remind us to address that, hopefully in the next release.

[figaroleo]

Thank you. I think that will be helpful for future users.
I found some questions open around the web, and someone suggests to extract the number from the string, but that is too time-consuming and not really efficient.

[dopplershift]

It's covered in the documentation for Pint, the library we use for unit-handling, but I realize many MetPy users are relying on MetPy's own user guide here to help.

[figaroleo]

Yes, definitely. Also because it is a specific format that is not so "common", and when you install MetPy it installs also dependencies, but pint is one of the several dependencies installed.
E.g., Spyder isn't able to correctly classify and visualise this data type in its variable explorer.