On the failure of standard emission measure analysis for solar extreme-ultraviolet and ultraviolet irradiance spectra

We perform emission measure analysis of new and accurate UV (λ > 1200 A) and extreme-ultraviolet (EUV) (λ < 1200 A) irradiance (‘‘Sun-as-a-star’’) emission-line spectra of the Sun. Our data consist of (1) daily averaged UV irradiances from the SOLSTICE on the UARS spacecraft and (2) EUV irradiances obtained on the same date from a 1/4 m spectrograph flown on a sounding rocket. Both instruments have a spectral resolution of roughly 1 A. The absolute uncertainties in these data are at most ± 15% (±2 α ), one of the highest photometric accuracies yet achieved. We find large, highly significant and systematic discrepancies in the emission measure analysis of transition region lines which can only be accounted for by a breakdown of one or more standard assumptions. All strong lines above 1000 A, which are from the Li and Na isoelectronic sequences, are too strong by factors of between 2.5 and 7 compared with their counterparts in the EUV region. Previous studies were tantalizingly close to finding these discrepancies, but those data lacked the wavelength coverage and relative photometric precision necessary for definitive conclusions. We argue that either dynamical effects, inaccurate treatments of atomic processes, and/or Lyman continuum absorption are the culprits. However, we favor the former explanation. In any event, this study should have implications for models of the solar transition region, for observing programs with the CDS and SUMER instruments on SOHO, and for analysis of UV spectra for stars across the cool half of the H-R diagram. Finally, the discrepancy is not seen for the ‘‘coronal’’ Li-like ions.