Occurrence rate of extreme magnetic storms

Statistical analysis of occurrence rate of magnetic storms induced by different types of interplanetary drivers is made on the basis of OMNI data for period 1976-2000. Using our catalog of large scale types of solar wind streams we study storms induced by interplanetary coronal mass ejections (ICME) (separately magnetic clouds (MC) and Ejecta) and both types of compressed regions: corotating interaction regions (CIR) and Sheaths. For these types of drivers we calculate integral probabilities of storms with minimum Dst < -50, -70, -100, -150, and -200 nT. The highest probability in this interval of Dst is observed for MC, probabilities for other drivers are 3-10 times lower than for MC. Extrapolation of obtained results to extreme storms shows that such a magnetic storm as Carrington storm in 1859 with Dst = -1760 nT is observed on the Earth with frequency 1 event during ~500 year.

💡 Research Summary

The paper presents a statistical investigation of the occurrence rates of geomagnetic storms driven by different interplanetary structures, using the OMNI solar‑wind database for the period 1976–2000 and a catalog of large‑scale solar‑wind types compiled by the authors. Four driver categories are considered: magnetic clouds (MC), ejecta (the non‑cloud component of interplanetary coronal mass ejections, ICME), corotating interaction regions (CIR), and sheath regions that precede ICMEs. For each driver type the authors count the total number of events (Nj) and the number of events that produce a storm with a minimum Dst index below a set of thresholds (‑50, ‑70, ‑100, ‑150, and ‑200 nT). The integral probability Pj(Dst0)=Kj/Nj is then calculated for each threshold.

The results show that MCs have the highest probability of generating storms at all thresholds, and that the probability declines most slowly with increasing storm intensity for MCs. For example, the probability for MCs drops by roughly a factor of ten when the threshold moves from ‑50 nT to ‑200 nT, whereas the probabilities for CIRs, sheaths, and ejecta are 3–4 times lower at the ‑50 nT level and 5–15 times lower at the ‑200 nT level. This confirms earlier findings that MCs are the most geoeffective solar‑wind structures.

Because the observational record contains very few events with Dst ≤ ‑500 nT, the authors extrapolate the probability curves to more extreme values. They fit the logarithm of the probability versus the logarithm of |Dst| using two approaches: a second‑order polynomial (log‑log) and a power‑law with a fixed exponent of –2.5. The log‑log plot is not linear, and the quadratic fit declines too steeply, so the authors adopt the power‑law as the more realistic representation.

Using the fitted power‑law, they compute the average waiting time (or recurrence interval) for storms of a given intensity: Tj =