Spectral irradiance correction of photovoltaic energy yield predictions in six high-latitude locations with measured spectra

Abstract

As solar power is expected to supply over 16% of the world’s electricity mix by 2030, accurate energy yield prediction and forecasting is essential. The incident solar spectrum varies continuously, causing energy prediction error as sunlight deviates from the reference spectrum. This work quantifies spectral error in bifacial photovoltaic energy yield predictions, demonstrating instantaneous spectral impacts of −45% to +32% and annual impacts from +0.7% to +2.7% on energy yield. This work presents solar spectral impact analysis for seven North American locations (39.7–69.1°N). We apply measured spectra to bifacial fixed-tilt silicon modules using a 2D view-factor model. Ground-reflected and diffuse irradiance cause the largest spectral errors while direct irradiance is best-matched to reference conditions. This study indicates that spectral correction methods should be applied for bifacial systems or in locations with high diffuse fractions above 35%, representing over 75% of Earth’s landmass. These effects indicate current PV system designs underestimate diffuse and ground-reflected irradiance contributions.