pvlib-python icon indicating copy to clipboard operation
pvlib-python copied to clipboard

Update default solar constant

Open kdebrab opened this issue 3 years ago • 5 comments

The default solar constant used in irradiance.get_extra_radiation is currently 1366.1 W/m².

Recent work revised the average solar constant to 1361.1 W/m² (see https://www.researchgate.net/publication/324269776_A_reevaluation_of_the_solar_constant_based_on_a_42-year_total_solar_irradiance_time_series_and_a_reconciliation_of_spaceborne_observations).

This new value seems now to be gaining broad acceptance, a.o. in certain standards.

Maybe it's a good time to update also the default value in pvlib to 1361.1 W/m².

kdebrab avatar Oct 05 '22 14:10 kdebrab

~I think this should be an input with a recommended default~ I realize I don't know what I'm talking about.

mikofski avatar Oct 05 '22 18:10 mikofski

I think this should be an input with a recommended default

I've been frustrated by irradiance.disc not exposing the solar constant value as an input, so +1 from me.

Regarding default values: if a model was tuned for a specific value of the solar constant then I'm not sure it makes sense to go against its reference by changing the default.

kandersolar avatar Oct 05 '22 19:10 kandersolar

That function is a convenience wrapper for four extraterrestrial irradiance models, which all differ in how the earth-Sun distance is computed. Two of the models also specify a solar constant: 'asce' uses 1367.7 and 'spencer' uses 1366.1. 'pyephem' refers to the ephem python package, which doesn't contain a solar constant that I have found, and 'nrel' refers to SPA and there is no specified solar constant for either.

So we could make the solar constant an input to this function, but I'd want to specify the values for 'asce' and 'spencer'.

cwhanse avatar Oct 05 '22 21:10 cwhanse

@cwhanse The constant and the method to calculate sun-earth distance are independent of each other.

The constant probably only matters for the clear-sky models, where the best value to use may be the one used at the time the model was developed. Plopping in the new value everywhere would probably create a bit of a bias.

adriesse avatar Oct 13 '22 11:10 adriesse

The constant and the method to calculate sun-earth distance are independent of each other.

I agree, I'm thinking to maintain consistency with the underlying reference for those two extraterrestrial irradiance models which specify a constant.

cwhanse avatar Oct 13 '22 14:10 cwhanse