http://orcid.org/0000-0001-9235-3257 Elizabeth A. Walter‑Shea
Date of this Version
Walter-Shea, E.A., K.G. Hubbard, M. A. Mesarch and G. Roebke. (2018) Improving the calibration of silicon photodiode pyranometers. Meteorology and Atmospheric Physics.
Published in Meteorology and Atmospheric Physics (2018)
Reliable measurements of global irradiance are essential for research and practical applications. Silicon photodiode pyranometers (SiPs) offer low-cost sensors to measure direct and diffuse irradiance despite their non-uniform spectral response over the 300–1000 nm spectral range. In this study, non-adjusted linear and adjusted calibrations were applied at different times of the year to determine sources of estimated errors in global irradiance due to the two calibration approaches, calibration time, and sensor age. 16 SiPs, along with two standards, measured incident global irradiance over a 5-year period under a range of sky conditions. Sensors performed best in the months in which they were calibrated when using the linear calibration approach. With the solar zenith angle adjusted calibration approach, certain calibration months provide a defendable validation for the following 12 months [ranging an average of 13.5–17.4 W m−2 standard error (SE)], while other calibration months do not provide consistent results and sometimes result in very poor validation (31.1–242.7 W m−2 SE). Older sensors (greater than 6 years) in general become more sensitive to solar zenith angle and their response drifts over time, while newer SiPs performed better than older sensors. Calibrations which accounted for solar zenith angle effects improved global irradiance estimates for older SiPs. For the Lincoln NE location, the appropriate calibration is in spring or late summer, regardless of calibration approach. These results indicate that solar zenith angle correction is not needed for largely diffuse components under cloudy conditions, so that in the future, a “smart” calibration may be possible, where diffuse radiation fractions are known.