saturation vapor pressure

[IslandOfDreams]

The Magnus type formula used fopr the calculation of saturation vapor pressure (thermo.py, line 771)

return sat_pressure_0c * np.exp(17.67 * (temperature - 273.15 * units.kelvin)
                                       / (temperature - 29.65 * units.kelvin))

refers to [Bolton1980].

Therein, Bolton explains (p. 1047):
'The following formula was fitted to Wexler's results, extrapolated for T < 0°C, to an accuracy of 0.1% for -30°C <= T <= 35°C: ...'

Since the Earth atmosphere shows a much greater temperature range of roughly -65°C <= T <= 50°C, Bolton's formula regularly gives less accurate results outside his recommended interval.

For a fruitful discussion of this topic see, e. g.,
Alduchov, Eskridge - Improved Magnus Form Approximation of Saturation Pressure.

The following formula as given by [Sonntag, 1990] assures a higher accuracy over a much extended temperature range (Python code):

return (math.exp ( -6096.9385 / T 
                 + 16.635794
                 - 2.711193 * T / 1.0e2
                 + 1.673952 / 1.0e5 * T*T
                 + 2.433502 * math.log (T)
                 ) * 100.0)      # conversion from hPa to Pa
             * units.Pascal.
with T: Temperature in KELVIN in the Temperature Scale 1990.
and     207,25 K <= T <= 334,05 K (i.e. -65.9°C <= T <= 60.9°C)

Reference:
Sonntag D., 1990: Important new values of the physical constants of 1986,
vapor pressure formulations based on ITS-90, and psychrometer formulae.
Z. Meteorol., 70, 340-344.

[dopplershift]

Thanks for the suggestions. We'd definitely like to improve our calculation, see for instance #508. One other challenge is trying to make sure we maintain sufficient run-time performance, which of course requires defining what exactly that is.

Do you have some concrete examples (specifically, input values and references for output) where the current implementation is proving insufficient? That would help us document and justify changes we make.

[wangxing181]

The Magnus type formula used fopr the calculation of saturation vapor pressure (thermo.py, line 771)

return sat_pressure_0c * np.exp(17.67 * (temperature - 273.15 * units.kelvin)
                                       / (temperature - 29.65 * units.kelvin))

refers to [Bolton1980].

Therein, Bolton explains (p. 1047): 'The following formula was fitted to Wexler's results, extrapolated for T < 0°C, to an accuracy of 0.1% for -30°C <= T <= 35°C: ...'

Since the Earth atmosphere shows a much greater temperature range of roughly -65°C <= T <= 50°C, Bolton's formula regularly gives less accurate results outside his recommended interval.

For a fruitful discussion of this topic see, e. g., Alduchov, Eskridge - Improved Magnus Form Approximation of Saturation Pressure.

The following formula as given by [Sonntag, 1990] assures a higher accuracy over a much extended temperature range (Python code):

return (math.exp ( -6096.9385 / T 
                 + 16.635794
                 - 2.711193 * T / 1.0e2
                 + 1.673952 / 1.0e5 * T*T
                 + 2.433502 * math.log (T)
                 ) * 100.0)      # conversion from hPa to Pa
             * units.Pascal.
with T: Temperature in KELVIN in the Temperature Scale 1990.
and     207,25 K <= T <= 334,05 K (i.e. -65.9°C <= T <= 60.9°C)

Reference: Sonntag D., 1990: Important new values of the physical constants of 1986, vapor pressure formulations based on ITS-90, and psychrometer formulae. Z. Meteorol., 70, 340-344.

Hi IslandOfDreams ， can u share the Reference: Sonntag D., 1990: Important new values of the physical constants of 1986, vapor pressure formulations based on ITS-90, and psychrometer formulae. Z. Meteorol., 70, 340-344. or share the psychrometric equation. thanks a lot

[adriesse]

See Sonntag 1994. The work by Sonntag is also quoted in the WMO CIMO guide.

[dopplershift]

#3726 updated MetPy's implementation to use the Ambaum (2020) formulation for saturation vapor pressure (and include saturation over ice). This is a much more physically based formulation and should address the concerns above. Feel free to re-open if you feel this is not the case.