Predictability and interrelations of spectral indicators for PV performance in multiple latitudes and climates

When PV is installed in the field, the module technologies are rated according to their output energy yield under local operating conditions rather than at standard test conditions (STC), where the spectrum is set to AM1.5G. Care must be taken as this standard is not optimal for all latitudes and the solar spectral distribution variations are one primary influencing factor on PV performance. In addition, obtaining an accurate estimate of the spectral effects on PV performance, as set out in standard procedures, is hampered by the cost of gathering the inputs and the large amount of spectral data required for such a calculation. In this work, based on measured spectral irradiance data from nine sites of different latitudes and climates, we first show a characteristic trend in the spectral distribution over the year concerning the location latitude. The closer a site is to the equator, the more blue-rich the solar spectrum is and the fewer seasonal spectral variations it will contain. Then, we calculate and correlate the most popular metrics (device-independent and device-dependent) used to describe the influence of solar spectra on PV performance. In particular, the monthly irradiance-weighted Spectral Mismatch Factor for different PV technologies and Average Photon Energy show a global linear correlation for data from these nine sites. We use this global linear relationship to propose PV technology-dependent equations that predict annual and monthly spectral gains/losses within a prediction half-interval of up to ± 1.66% by only inserting the monthly or annual irradiance-weighted Average Photon Energy potentially for any site. Reducing the required spectral data sets for performance estimation through our methodology facilitates a more accessible and less costly communication of databases than complete spectral data sets. Finally, using this spectral data, we demonstrate statistically that the Spectral Mismatch Factor and Integrated Useful Fraction Ratio can be replaced by alternative spectral metrics, which require only averaged spectra and, thus, reduce the computational effort to estimate the above indicators.