Solar Spectral Irradiance Variations in 240-1600 nm During the Recent Solar Cycles 21-23

Regular solar spectral irradiance (SSI) observations from space that simultaneously cover the UV, visible (vis), and the near-IR (NIR) spectral region began with SCIAMACHY aboard ENVISAT in August 2002. Up to now, these direct observations cover less than a decade. In order for these SSI measurements to be useful in assessing the role of the Sun in climate change, records covering more than an eleven-year solar cycle are required. By using our recently developed empirical SCIA proxy model, we reconstruct daily SSI values over several decades by using solar proxies scaled to short-term SCIAMACHY solar irradiance observations to describe decadal irradiance changes. These calculations are compared to existing solar data: the UV data from SUSIM/UARS, from the DeLand & Cebula satellite composite, and the SIP model (S2K+VUV2002); and UV-vis-IR data from the NRLSSI and SATIRE models, and SIM/SORCE measurements. The mean SSI of the latter models show good agreement (less than 5%) in the vis regions over three decades while larger disagreements (10 -aEuro parts per thousand 20%) are found in the UV and IR regions. Between minima and maxima of Solar Cycles 21, 22, and 23, the inferred SSI variability from the SCIA proxy is intermediate between SATIRE and NRLSSI in the UV. While the DeLand & Cebula composite provide the highest variability between solar minimum and maximum, the SIP/Solar2000 and NRLSSI models show minimum variability, which may be due to the use of a single proxy in the modeling of the irradiances. In the vis-IR spectral region, the SCIA proxy model reports lower values in the changes from solar maximum to minimum, which may be attributed to overestimations of the sunspot proxy used in modeling the SCIAMACHY irradiances. The fairly short timeseries of SIM/SORCE shows a steeper decreasing (increasing) trend in the UV (vis) than the other data during the descending phase of Solar Cycle 23. Though considered to be only provisional, the opposite trend seen in the visible SIM data challenges the validity of proxy-based linear extrapolation commonly used in reconstructing past irradiances.