The Greatest GOES Soft X-ray Flares: Saturation and Recalibration over Two Hale Cycles

The solar soft X-ray observations from the GOES satellites now span two full Hale cycles and provide one of the best quantitative records of solar activity, with nearly continuous flare records since 1975. We present a uniform new reduction of the entire time series for 1975 to 2022 at NOAA class C1 level or above, to characterize the occurrence distribution function (ODF) of the flares observed in the 1 – 8 Å spectral band. The analysis includes estimates of the peak fluxes of the 12 flares that saturated in the 1 – 8 Å time series. In contrast to the standard NOAA classifications, these new estimates use the full time resolution of the sampling and have a preflare background level subtracted for all events. Our new estimates include NOAA’s latest calibrations for the GOES-1 through GOES-15 data covering 1975 – 2016. For each of the 12 saturated events we have made new estimates of peak fluxes based on fits to the rise and fall of the flare time profile, and have validated our extrapolation schemes by comparing with artificially truncated but unsaturated X10-class events. In this new estimation, SOL2003-11-04 (the most energetic unambiguously observed event) has a peak flux of \(4.32 \times 10^{-3}\text{ W}/\text{m}^{2}\). This corresponds to X43 on the new scale, or X30 on the old scale. We provide a list in the Appendix for peak fluxes of all 37 events above \(10^{-3}\text{ W}/\text{m}^{2}\), the GOES X10 level, including the 12 saturated events. The full list now gives us a first complete sample from which we obtain an occurrence distribution function (ODF) for peak energy flux \(S\), often represented as a power-law \(dF/dE \propto E^{-\alpha }\), for which we find \(\alpha = 1.973 \pm 0.014\) in the range M1 to X3. The power-law description fails at the high end, requiring a downward break in the ODF above the X10 level. We give a tapered power-law description of the resulting CCDF (complementary cumulative distribution function) and extrapolate it into the domain of “superflares,” i.e., flares with bolometric energies \(>10^{33}\text{ erg}\). Extrapolation of this fit provides estimates of 100-yr and 1000-yr GOES peak fluxes that agree reasonably well with other such estimates using different data sets and methodology, although there is some tension between our 10,000-yr (the Holocene time-scale) estimate and the GOES class obtained for the out-sized 774 AD solar proton event as inferred from cosmogenic nuclide records.