Measurements of spectral irradiance during the solar eclipse of 21 August 2017: reassessment of the effect of solar limb darkening and of changes in total ozone

Measurements of spectral irradiance between 306 and 1020 nm were performed with a GUVis-3511 multi-channel filter radiometer at Smith Rock State Park, Oregon, during the total solar eclipse of 21 August 2017. The radiometer was equipped with a shadowband, allowing to separate the global (sun and sky) and direct components of solar radiation. Data were used to study the wavelength-dependent changes of solar irradiance at Earth's surface. Results were compared with theoretical predictions using three different parameterizations of the solar limb darkening (LD) effect, which describes the change of the solar spectrum from the Sun's center to its limb. Results indicate that the LD parameterization that has been most widely used during the last 15 years underestimates the LD effect, in particular at UV wavelengths. The two alternative parameterizations are based on two independent sets of observations from the McMath-Pierce Solar Telescope. When these parameterizations are used, the observed and theoretical LD effects agree to within 4 % for wavelengths larger than 400 nm and occultation of the solar disk of up to 97.8 %. Maximum deviations for wavelengths between 315 and 340 nm are 7 %. These somewhat larger differences compared to the visible range may be explained with varying aerosol conditions during the period of observations. Aerosol optical depth (AOD) and its wavelength dependence was calculated from measurements of direct irradiance. When corrected for the LD effect, AOD monotonically decreases over the period of the eclipse: from 0.41 to 0.32 at 319 nm and from 0.05 to 0.04 at 1018 nm. These results show that AODs can be accurately calculated during an eclipse if the LD effect is corrected. The total ozone column (TOC) was derived from measurements of global irradiance at 306 and 340 nm. Without correction for the LD effect, the retrieved TOC increases by 20 DU between the 1st and 2nd contact of the eclipse. With LD correction, the TOC remains constant to within natural variability (±2.6 DU or ±0.9 % between 1st and 2nd contact and ±1.0 DU or ±0.3 % between 3rd and 4th contact). In contrast to results of observations from earlier solar eclipses, no fluctuations in TOC were observed that could be attributed to gravity waves, which can be triggered by the supersonic speed of the Moon's shadow across the atmosphere. Furthermore, systematic changes in the ratio of direct and global irradiance that could be attributed to the solar eclipse were not observed. This finding agrees with results of three-dimensional radiative transfer models but contradicts reports from earlier observations, which indicate that the diffuse-to-direct ratio may change by 30 %. Our results advance the understanding of the effects of solar LD on the spectral irradiance at Earth's surface, the variations of ozone during an eclipse, and the partitioning of solar radiation in direct and diffuse components.