Reference solar spectra: Their evolution, standardization issues, and comparison to recent measurements

The International Organization for Standardization (ISO) is preparing a standard that will certify the process of developing "reference solar spectra". The information needed for this certification is critically reviewed here by processing a test case, using the latest of these reference spectra. It is shown that a few issues remain to be clarified in the current draft standard. In particular, it is not clear what methodology one should use to properly "validate" or assess the performance of a reference composite spectrum primarily based on measured irradiance data. A historical perspective on the evolution of such reference solar spectra is presented. Whereas considerable improvements (particularly noticeable in the UV) took place until about 1978, the six more recent spectra (proposed since 1985) show comparable features and similar magnitudes from the UV-C to the near infrared, when smoothed to the same relaxed resolution. Excellent agreement is found between the latest composite reference spectrum and an experimental irradiance dataset limited to the UV (295-355 nm). Good agreement is also found by comparison with various single-day or average SOLSTICE-UARS spectra over a larger UV region (120-420 nm), even though solar activity interference in this validation attempt appears obvious. Conversely, large differences are noted when comparing the same reference spectrum (as well as other older reference spectra) to spectral data from the SORCE instruments, from 120 to 1600 nm. Whereas these instruments have been designed for extremely high accuracy, the current datasets (versions 3 and 4) must be considered preliminary due to various unresolved problems. Nevertheless, the precision of these instruments seems to be sufficiently high over most of the spectrum to detect short-term changes of 0.1% or less in solar irradiance due to daily changes in activity. Considerably larger daily variations (up to approximate to 20%) in the UV-C spectrum are observed during the intense activity period of October-November 2003. Furthermore, using the latest version of SORCE data, a comparison between periods of moderate activity and periods of elevated sunspot activity (with low total solar irradiance) suggests that the most part of this loss in total irradiance can be explained by small and smooth spectral changes in the 400-1600 nm waveband, thus confirming that sunspots are dark over all or most of the spectrum. (c) 2005 COSPAR. Published by Elsevier Ltd. All rights reserved.