Diffraction effects in solar radiometry are nothing new, they have been discussed at several occasions, mostly in conjunction with establishing absolute scales. When it comes to extremely precise measurements, as the ISO 9060 AA-class calls for, it is worth to look at diffraction a little closer.
Let us recall the origin of the diffraction effect: Electromagnetic radiation propagates as a wave, and this means that it does not exactly follow the path one would assume from pure geometry. Diffraction effects occur due to the wave nature of the radiation. Depending on the aperture geometry, diffraction effects can enhance the signal in the detector, or radiation can be lost due to diffraction, which leads to a lower signal. A more detailed description and how it can be derived can be found in Advances in Solar Radiometry, or in several publications from Eric Shirley.
As we know, diffraction effects are wavelength dependent, and the spectrum at our measurement site might differ from the standard spectrum G-173. Fortunately the revised ISO 9060 gives us 17 other spectra we should use for our study (Figure 1). Now what does it mean for the sensitivity change?
Figure 2 shows the diffraction effects calculated for a PMO6 radiometer, for the 18 spectra given in the ISO 9060:2018. We can see that the deviation from the reference spectrum G173 to the other spectra are as large as 500 ppm. The standard defines the clear sky direct normal irradiance spectral error as the error that occurs as a consequence of a change in the spectrum. Its acceptance interval is 100 ppm for the ISO 9060 AA-class. So diffraction alone can cause an error that is five times larger than permitted by the class definition.
What are the consequences from this? First of all it shows that the AA-class is really something new. It calls for a completely new level of accuracy. All known reference instruments, the so called cavities or absolute radiometers such as the PMO6, the HF and others are definitely not AA class, they all suffer from the diffraction effects. Secondly, there are sophisticated methods necessary to overcome the diffraction effects and to match the clear sky direct normal irradiance spectral error criteria.
The diffraction effect for a HF radiometer is about 20% lower, this is because of the larger apertures, used in the HF radiometer. However, the geometry of the HF apertures does not match the geometry as defined in the ISO 9060:2018, and the error is still four times larger than permitted.