Chapter 1 · Lesson 3 of 6 · about 14 minutes
The solar cycle and solar events
The environment changes on two time scales. The slow one is the roughly eleven-year solar cycle, which modulates the galactic cosmic-ray flux and sets how likely solar events are. The fast one is the events themselves: irregular, unpredictable in timing, and spread over orders of magnitude in size. Requirements have to cover both, with different tools: a climatological model for the cycle and a statistical model for the events.
What you need to know
The cycle. The sunspot number is the oldest index of solar activity and still the usual way to say where a mission falls in the cycle. Cycle 23 peaked in 2000 with an annual mean of 174, fell to 4.2 in 2008, rose to 113 in 2014 (cycle 24, the weakest in a century), fell to 3.6 in 2019, and cycle 25 has so far peaked at 155 in 2024. Cycles vary in length, amplitude and shape, and the magnetic polarity reverses each cycle, so the full magnetic cycle is about twenty-two years.
What the cycle does to each population. The solar wind and its magnetic field are stronger at maximum and sweep away more of the low-energy galactic cosmic rays, so the GCR flux is highest at solar minimum and lowest at maximum. In the 2008 review's integral LET figure the solar-minimum curve lies about 2.7 times above the solar-maximum curve at LET 1 MeV·cm²/mg (read from a digitization of the figure, so approximate). Large solar particle events are more frequent during the active years, conventionally taken as the roughly seven years from two and a half years before maximum to four and a half years after, though large events have occurred close to minimum. The trapped populations respond too. The upper atmosphere expands at maximum and removes inner-zone protons at low altitude, so AP-8 MIN gives the higher proton flux in LEO. Outer-zone electrons respond to the recurrent high-speed streams of the declining phase.
The events. A solar particle event delivers its fluence over hours to days. The largest events of the space age (August 1972, October 1989) each delivered of order 10¹⁰ protons above 10 MeV per cm² outside the magnetosphere. Because event sizes span orders of magnitude, the engineering question is statistical: what fluence will not be exceeded, at a stated confidence, over the mission's active years. The ESP model answers that for protons and PSYCHIC extends it to heavy ions. Their outputs are confidence levels on a cumulative fluence, not forecasts of any event.
Worked example: how big is a bad year, and how big is one event
The question. A GEO mission launches into the active years of a cycle. What solar-proton fluence should the dose and displacement-damage analyses use for one active year, and what should the single-event and transient analyses use for one event?
Inputs. ESP-PSYCHIC in SPENVIS 4.6 at 95 % confidence: the total fluence for one active year, and the worst single event, both at GEO (the Earth shadow passes 0.994 of free space). For comparison, the August 1972 event as parameterized by King (1974), J(>E) = 7.9 × 10⁹ exp((30 − E)/26.5) protons per cm² for E in MeV from 10 to 200.
| Integral fluence above E, protons per cm² | > 10 MeV | > 100 MeV |
|---|---|---|
| ESP total, one active year, 95 % | 3.6 × 10¹⁰ | 7.4 × 10⁸ |
| PSYCHIC worst single event, 95 % | 2.5 × 10¹⁰ | 6.6 × 10⁸ |
| August 1972 event (King 1974) | 1.7 × 10¹⁰ | 5.6 × 10⁸ |
| Worst event as a share of the year | 70% | 89% |
| Worst event relative to August 1972 | 1.5 | 1.2 |
Steps.
- Take the one-year total at 95 % for the cumulative effects. Dose and displacement damage add up, so the quantity to use is the cumulative fluence over the active years at the project's confidence level. For a longer mission rerun the model with the actual number of active years rather than multiplying the one-year figure, because the model's cumulative distribution is not linear in time.
- Take the worst event at 95 % for the effects that depend on an event. The upset rate during the event, a transient on a power bus and the availability of a function during the event all depend on what one event delivers and how fast it delivers it, not on the yearly sum. The 95 % worst event accounts for 70% of the year's total above 10 MeV and 89% above 100 MeV. At this confidence level the year is essentially one large event.
- Check the size against history. The 95 % worst event is 1.5 times the August 1972 event above 10 MeV and 1.2 times above 100 MeV. The historical event is a useful sanity check and a familiar reference, but it has no confidence level, so the statistical models have replaced it as a requirement basis.
- Add the heavy ions for the SEE analysis. The same PSYCHIC run gives, per event above 10 MeV/n, 3.3 × 10⁸ helium, 9.4 × 10⁶ oxygen and 7.7 × 10⁵ iron nuclei per cm². Above 100 MeV/n, roughly the energy an iron nucleus needs to pass 100 mil of aluminum, the event delivers 3,007 iron nuclei per cm², about 0.09 of a year of galactic iron above the same energy in free space (3.5 × 10⁴ per cm² per year, from the ISS value and the orbit's iron transmission of 0.24), compressed into a few days. A part with a low LET threshold therefore generally sees its heavy-ion event rate rise by an order of magnitude or more for the duration of a large event.
Result. Use 3.6 × 10¹⁰ protons per cm² above 10 MeV for the year (and the full spectrum behind it) in the TID and DDD analyses, and 2.5 × 10¹⁰ per cm² above 10 MeV, with its heavy-ion spectrum, in the SEE and transient analyses as the design event. State the confidence level with every number.
Going deeper
What the confidence level means
The ESP model fits the distribution of event fluences from the space-age record, treats event occurrence in the active years as a Poisson process, and derives the probability that the cumulative fluence over a stated number of active years will not exceed a given value. "95 % confidence for one active year" means that in the model 95 % of such years would have a total below the quoted fluence. It is not the probability that a part survives, and the 95 % values at two energies are not the spectrum of any single year: they are pointwise percentiles. For missions longer than one active year rerun the model with the full duration; the percentiles do not scale linearly.
Scenario models and probability
CREME96 includes three solar-event scenarios built on the October 1989 event: the worst week, the worst day and the peak five minutes. They are flux time profiles, which the cumulative models do not give, and they remain the standard inputs for peak-rate and availability questions. They have no probability of occurrence attached, and stacking several worst weeks to represent a long mission is a convention rather than a statistical statement. Jiggens et al. (2014) compare the two approaches and show that one worst week corresponds to roughly 50 to 65 % confidence over a mission, with the equivalence changing with LET and shielding.
Modulation in one parameter
The force-field approximation describes the solar modulation of galactic cosmic rays with a single potential φ, typically 400 to 500 MV at minimum and 1000 to 1200 MV at maximum, applied as an energy loss to the local interstellar spectrum. It is adequate for engineering spectra above a few hundred MeV per nucleon, which is where most of the SEE-relevant flux is. The ISO 15390 model, which CREME96 adopted, uses its own semi-empirical modulation description tied to the sunspot number, with the same physical content. None of this is complete heliospheric physics. Drift effects give the two magnetic polarities different modulation at the same sunspot number, which is one reason the GCR maximum of 2009 to 2010 and that of 2019 to 2020 differed.
The common mistake
The common mistake is to treat "solar maximum" as the worst case for everything. It is generally the worst case for solar events and the best case for galactic cosmic rays and for inner-zone protons in LEO. A mission that slips from the active years into the declining phase gains on solar protons and loses on cosmic rays and, at low altitude, on trapped protons. The analysis should state which phase was assumed for each population and why.
Takeaway
Treat the cycle with a climatological model and the events with a statistical one: a cumulative fluence at a stated confidence for dose and displacement damage, and one design event with its heavy ions for everything that happens during an event.
Check yourself
A requirement reads "solar proton fluence: 1 × 10¹⁰ cm⁻² above 10 MeV". What is missing?
The confidence level, the number of active years it covers, the model and version, whether it is free space or at the orbit (and if so, which geomagnetic condition), and the full spectrum behind the single integral number, since dose and displacement damage depend on the shape. Without those the number cannot be reproduced, compared with another mission, or defended at review.
A launch slips by two years, from the active years of the cycle into the declining phase, and the project manager calls it good news for radiation. For which populations is it, and for which is it not?
It is generally good news for solar events, which are more frequent in the active years, and bad news for galactic cosmic rays, whose flux rises toward solar minimum as the solar wind weakens (about 2.7 times between the solar-maximum and solar-minimum curves at LET 1 MeV·cm²/mg in the 2008 review). At low altitude it is also bad news for inner-zone protons, since the atmosphere that expands at maximum and removes them contracts again, which is why AP-8 MIN gives the higher proton flux in LEO, and the outer-zone electrons respond to the high-speed streams of the declining phase. The analysis should state which phase was assumed for each population and why, and the slip reopens the sheet (lesson 6).
A reviewer reads "95 % confidence" on the solar-proton fluence as a 95 % probability that the parts will survive, and asks for a higher confidence level to be safe. What does the number actually state?
That in the ESP model 95 % of such active years would have a cumulative fluence below the quoted value. It is a percentile of the environment distribution, derived from the space-age record of event fluences with event occurrence treated as a Poisson process. It is not a survival probability and it is not a margin (lesson 6): part survival depends on the design margin and on the part response, which are handled separately. The 95 % values at different energies are also pointwise percentiles rather than the spectrum of any single year, and the percentiles do not scale linearly with mission length.
For dose the worked example uses the 95 % one-year total, 3.6 × 10¹⁰ protons per cm² above 10 MeV, and for the single-event and transient analyses the 95 % worst event, 2.5 × 10¹⁰. Since the event is 70 % of the year, a colleague suggests using the larger yearly total for everything and calling it conservative. Why does the lesson separate them?
Event effects depend on what one event delivers and how fast it delivers it, not on a yearly sum, so the design event has to be a fluence with a duration and a heavy-ion spectrum attached: per event the PSYCHIC run gives 7.7 × 10⁵ iron nuclei per cm² above 10 MeV/n, and above 100 MeV/n 3,007 per cm², roughly 0.09 of a year of galactic iron in free space compressed into a few days. A part with a low LET threshold therefore generally sees its heavy-ion rate rise by an order of magnitude or more for the duration of the event, which a yearly total spread over a year does not show. For peak-rate and availability questions the CREME96 worst-week, worst-day and peak-five-minute profiles supply the time structure that the cumulative models do not give.
A heritage specification uses the August 1972 event, 1.7 × 10¹⁰ protons per cm² above 10 MeV, as its solar-proton design case, and a reviewer wants it replaced. What is the argument on each side?
In its favor, the 1972 event is a familiar reference and a useful sanity check: the 95 % worst event from PSYCHIC is 1.5 times it above 10 MeV and 1.2 times above 100 MeV, so the two are close. Against it, the historical event carries no confidence level and no statement of how many events it represents over how many active years, so it cannot be compared with another mission or tied to a probability, and the statistical models have replaced it as a requirement basis. The usual resolution is to keep the 1972 figures as a cross-check column and to write the requirement from ESP for the cumulative case and PSYCHIC for the design event.
Further reading
- M. A. Xapsos, G. P. Summers, J. L. Barth, E. G. Stassinopoulos and E. A. Burke, "Probability model for cumulative solar proton event fluences," IEEE Trans. Nucl. Sci., vol. 47, no. 3, pp. 486-490, 2000.
- M. A. Xapsos, C. Stauffer, T. Jordan, J. L. Barth and R. A. Mewaldt, "Model for cumulative solar heavy ion energy and linear energy transfer spectra," IEEE Trans. Nucl. Sci., vol. 54, no. 6, pp. 1985-1989, 2007.
- A. J. Tylka et al., "CREME96: a revision of the Cosmic Ray Effects on Micro-Electronics code," IEEE Trans. Nucl. Sci., vol. 44, no. 6, pp. 2150-2160, 1997.
- P. Jiggens et al., "Long-term destructive SEE risk and calculations using multiple worst-case events versus modelling," IEEE Trans. Nucl. Sci., vol. 61, no. 4, pp. 1695-1702, 2014. doi:10.1109/TNS.2014.2302994.
- M. A. Xapsos et al., "Inclusion of radiation environment variability in total dose hardness assurance methodology," IEEE Trans. Nucl. Sci., vol. 64, no. 1, pp. 325-331, 2017. doi:10.1109/TNS.2016.2607021.
- J. H. King, "Solar proton fluences for 1977-1983 space missions," J. Spacecraft and Rockets, vol. 11, no. 6, pp. 401-408, 1974.
- S. Bourdarie and M. Xapsos, "The near-Earth space radiation environment," IEEE Trans. Nucl. Sci., vol. 55, no. 4, pp. 1810-1832, 2008. NTRS 20080039622.