Comparison of methods for estimating the impact of spectrum on PV output

Spectral correction factors are often used to characterize the impact of spectral variations on PV module output. We compare three different methods to predict a spectral correction factor and validate these against outdoor spectroradiometer measurements. It is found that commonly used models lead to a constant error due to assumptions made when determining coefficients describing geometric air mass dependence. We term this error the site spectral offset. This offset directly impacts estimations of the available solar resource and therefore affects the accuracy of energy yield predictions. We determine its value for the temperate coastal climate of Newcastle, Australia to be -2%, +1% and -3% for c-Si, CdTe and CIGS respectively. Errors in the daily resource estimation compared to pyranometer values of up to 15% are found for one module type while absolute variation of up to 60 Wm^-2 in hourly AM1.5 equivalent insolation is seen for all examples studied. Applying a modified version of the CREST estimation method results in a significant reduction in estimation errors over all timescales