Tweet
I've been slowly working through this Perez/Ineichen model published in 2002,
It held up well in this comparison of models:
Python code here:
That model gets around AOD influence by using the climateological monthly Linke turbidity data available here.
I've been cautious about the making assumptions about how constant the inputs into Bird will be throughout the year, but haven't tried yet to see mow many of them can be populated from publicly available local weather sources. At some point I intend to track down this paper in which the Bird model has supposedly been improved by adapting it to local conditions:
Mostly because of it's availability, but also because it seems to match what I measure, and because it's easy to manipulate the inputs, I use the Bird model in spreadsheet format from NREL. As I mentioned previous, I did the Perez model from 1999 or so and found it a bit optimistic vs. readings by a few % or so after I got the Davis. So, I went back to the Bird model/NREL spreadsheet & fart around with the inputs for ozone/precip. H2O, etc. w/ ref. as listed in the March 2012 Sandia report you reference. I might mess around w/ the Ineichen model if I ever get off this fouling measurement kick, but for now, given the uncertainties of the atmosphere and the instrument, the Bird model and it's ease of use seems fit for purpose as long as I remember most everything is only about +/- 5% or so on a good day.
BTW, found your weatherunderground site. Is your irradiance sensor level ? Any guess at albedo ? Reason for questions: Your GHI around solar noon is pretty consistently about 5.5 - 6.5% higher than mine. Mine pretty much agrees w/ Boulder Knolls on the Weatherunderground. Yours seems to agree w/ a site close to you. I might get another sensor and have both calibrated.
Comment