Solar Irradiance Variability Claus Frhlich PhysikalischMeteorologisches Observatorium World

Solar Irradiance Variability Claus Fröhlich, Physikalisch-Meteorologisches Observatorium, World Radiation Centre Davos Dorf, Switzerland cfrohlich@pmodwrc. ch § Construction of a composite TSI: 1978 to present § Characterization of TSI variability § Development of a proxy model of TSI § Comparison of the model with TSI § Variability of spectral solar irradiance (SSI) § Conclusions 05. 03. 2021 • • • VIRGO Radiometry, an update What is the difference with the ACRIM composite (1997, 2003) Long-term trend? Power spectrum Difference between mode and mean Calibration of PSI and Mg. II index over 3 solar cycles Comparison with mode Bi-variate analysis of TSI versus model VIRGO SPM data set and time series Multivariate analysis of TSI versus RGB of SPM ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 1

Construction of a composite TSI: 1978 to present VIRGO Radiometry: Since version 5. 006 a linear trend for the period after the SOHO vacations has been added to the non-exposure dependent corrections. This trend is now (version 5. 007) allocated to PMO 6 V-B and may be due to an underestimation of its early increase. Construction of the Composite: Besides these small changes due to the new VIRGO data the composite remains the same. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 2

What is the difference with the ACRIM composite The corrections of the original data for the construction of the composite are as follows: • HF: Early degradation, glitch in late 1980, corrections in 1989 -92 • ACRIM-I: different accounting of degradation in 1980 • ACRIM-II: scaling to ACRIM-I via comparison with HF and ERBS, some corrections of glitches due to operations of UARS ACRIM-I and the shifted ACRIM-II are the basis and all the other time series adjusted accordingly. The absolute scale is based on the SARR value of ACRIM-II. The difference to the ACRIM composite is obviously due to the omission of the HF corrections during the gap between the ACRIMs. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 3

Characterization of TSI variability Just by looking at the time series the most obvious result is that all three solar cycles are different ! The mean peak-to-peak amplitude of the solar cycle is 670 ppm and the overall standard deviation amounts to ± 425 ppm. The next question is: What about a long-term trend? If we take the difference between the minima 1986 and 1996 a small trend of – 4. 9 ppm/a is determined. Still another way of looking at variability is the power spectrum. With nearly 25 years of observations we access periods reliably up to 20 years and with the sampling of VIRGO down to the 5 -minute oscillations. From the latter we can also determine the ration between the quiet and active Sun. 05. 03. 2021 Difference: -52 ppm ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 4

Characterization of TSI variability Another way to look at variability is the mode, the peak of the distribution. As the changes are due to two influences of opposite sign, sunspot and faculae, the distribution is asymmetric. For the influence on Earth the mean the relevant, for solar studies, however, the mode. When calculating the mode, some compensation of facular increases and spot decreases are expected. Thus, both the facular and sun-spot influence is underestimated when looking at the deviation of TSI from the mode. The comparison with the model will be discussed later. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 5

Development of a proxy model of TSI We follow here the standard way of a proxy model. It is based on sun-spot darkening modeled by PSI and increases due faculae and network modeled by Mg. II index (Mg core-to-wing ratio). The specifics of this model is that we separate the Mg. II index in a short-term and long-term component. Moreover, we calibrate PSI, Mg. IIST and Mg. IILT for each of the three solar cycle seperately and add to Mg. IILT a linear trend, which accounts for a possible difference between the model and observed TSI. The comparison of the time series shows very good agreement, at least the 81 -day filtered ones. 05. 03. 2021 122. 0 121. 2 117. 5 71. 2 86. 1 79. 6 0. 9244 1. 0580 1. 1564 Long-term trend: -0. 52 ppm/a ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 6

Comparison of the model with TSI The comparison with the results from the mode analysis shows indeed some very interesting features. Note the dependence of the differences between model and mode as a function of time within the cycle. Another way for comparing the proxy model with TSI is bi-variate spectral analysis, which allows to study the frequency dependence of the association between the two time series in terms of coherence, and gain and phase of the filter to translate one into the other. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 7

Variability of spectral solar irradiance The VIRGO investigation contains filter radiometers, called sun photometer (SPM), with 3 channels centered at 402, 500 and 862 nm. Although measurements with a much less exposed spare SPM are available, the correction for degradation cannot be done as for the radiometers. So, after correcting with the back-up data, polynomial are fitted before and after the So. HO vacations. To assess the influence of these fits, TSI is treated with a polynomial with the same degree and the power spectra compared. The time series seem to be reliable up to periods of about 4 -500 days. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 8

Correlation between SSI and TSI To study in more detail how TSI changes are related to SSI changes, or more specifically how the energy is redistributed, we use multi-variate spectral analysis, which is an extension of the bivariate one used to compare model and TSI. 05. 03. 2021 ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 9

Conclusions § A reliable composite TSI is available • • • § A new calibration scheme of the proxy model improves its reliability • § Comparison of the model with TSI reveals interesting features Time series of the 3 -channel SPM on VIRGO are reliable up to periods of 4 -500 d • 05. 03. 2021 Corrections are needed for essentially all contributing radiometers Especially the correction of HF during the gap of the ACRIMs is essential and supported by the independent ERBE data. Its omission yields an unrealistic upward trend By comparing the two minima a difference of 51 ppm is determined with the recent minimum being lower. The uncertainty of this difference is mainly given by the uncertainty of the HF correction of about ± 50 ppm. This means that the trend is most probably not significantly different from zero. Comparison with TSI by means of multivariate spectral analysis shows interesting differences in the contribution of the three colours to TSI variability ISCS Meeting, Tatranská Lomnica, 24 -28. June 2003 10