Jiggens/SAPPHIRE:

SAPPHIRE is a model of solar energetic particle event fluence and peak flux. It generates these from virtual events, each with an event peak flux and event fluence and wait time (quiet time between events).

It covers confidence levels from 50-99.9%, missions from 3 months to 55 years, and worst week to worst 5 minutes. 5-300 MeV/nuc for H+, He++, 0.1-1000 MeV/nuc for Z up to 92. Also produces 1-in-n year fluences/peak fluxes, at requested confidence levels and mission durations. Helium model is separate from protons, and then heavier ions are done via abundance ratio.

P. Jiggens et al. JSWSC, 8(A31), doi: 10.1051/swsc/2018010, 2018. P. Jiggens et al. IEEE Trans. Nuc. Sci., 65(2), doi: 10.1109/TNS.2017.2786581, 2018. See https://spitfire.estec.esa.int/trac/SAPPHIRE/

Often lower than JPL and ESP, due to data treatment, allowing use of “more sensible” (i.e., higher) confidence levels.

Available in SPENVIS now and OMERE later in 2019.

Limitations:

• Does not produce virtual time series for detailed scenario analysis
• Uses average heavy ion abundances (varies within and between events)
• RDS v2 energies may change (highest energies only)

Going forward:

• Growing uptake
• Add higher energies for human spaceflight
• Address lower energy peak (<1 MeV)
• Merge with VESPER (virtual time series)
• Away from 1 AU

Questions:

• How do spectra extend to 1 GeV? Extrapolate using a Band fit, checked against neutron monitor data.
• Is the Poisson model of event rate valid? It’s an approximation that seems to be working, but there are some questions that remain, especially on short and long timescales.

Aminalragia-Giamini/VESPER:

VESPER is a probabilistic model that produces virtual time series of solar proton differential fluxes. It must account for wait time, duration, peak flux, fluence, and spectral coherence. Adapted the SAPPHIRE virtual time lines to generate durations and wait times. Model is heavily empirical.

Core of the model is the LF₂ variable, which is the integral of log(E²flux)dlog(E). Sum of LF2 over the duration of events is strongly correlated with the duration of events. Generate random cases of SLF2 given event duration. Scale real events using SLF2 – stretch in time and scale in flux.

Generated many virtual events with VESPER to compare to SEPEM for self consistency. Agreement is good, with only a few offsets.

VESPER is closer to SAPPHIRE than it is to ESP or JPL model, but there are some pending differences even between VESPER and SAPPHIRE yet to be understood.

VESPER has been coupled to MSM (Magnetospheric Shielding Model) to account for dynamic variation of solar particle access to satellite orbit over the course of events. Example shown for GTO-like orbit, with expected suppression of lower energies, and bigger effect for fluences than for peak fluxes.

Caveats/future work:

• Investigation filling/sampling of duration-SLF2 space.
• Sensitivity to event list
• Peak-flux to fluence relationship could be exploited
• Merging with SAPPHIRE

Questions:

• In comparison charts, how was JPL peak flux extracted (JPL does not do this natively)? What’s shown is the JPL fitting methodology applied to peak fluxes rather than fluences. It is not peak fluxes from the JPL model itself.
• When will heavy ions be added? Plan to start He in year 2 of ecIRENE, and rest of heavy ions in following year.

Aran/Time Series SEP:

(via Skype)

Concerned with SEP evolution away from 1 AU. Simple 1/R² assumption may overestimate the SEP fluence. Also does not account for in situ acceleration, e.g., by interplanetary shocks.

Physics-based models (most) can only model individual events, not overlapping series of events. Also, models often don’t model particle intensity after shock crossing. So, we have to split the observed events. What appears to be 1 or 2 events at Earth, may come from 4 different sources at the Sun. The higher energies are often important for revealing this multi-source structure. Dividing the events allows us to more appropriately address radial evolution through the heliosphere.

Generating a virtual event by rescaling an original event composed of multiple sources could cause problems. We may need to break up the events before doing the rescaling. We may also need to remove the backgrounds from the highest energy channels before creating virtual events.

Questions: None.

Bruno/PAMELA SEP:

Nominally two classes of energetic (>100 MeV, GLE) events. Impulsive (flares) and gradual (CMEs). Reality is probably a mixture.

PAMELA adds higher energies, plus pitch angle and species resolution.

Fit the event fluence with a  power law and a broken power-law: A (E/E_s)^gexp(-E/E₀) and simple power law. Broken power law fits better from 80-1000 MeV (26 examples shown). True of GLE and non-GLE events. Consistent with DSA theory. (DSA: Diffusive shock acceleration). Roll-over energy is correlated with event fluence, and starting frequency of Type II Radio Bursts & lower shock formation heights.

2012/0127 event spectral shape shows a break around 10 MeV as well as the broken power law at higher energy.

Well connected events tend to have higher fluence. Poorly connected high fluence events have to be long duration (>6 days). Well connected events have sharp increase and quick decay, whereas poorly-connected events have slower buildup, broader peak, and gradual decay.

GLE vs sub-GLE (sub-GLE seen only by South Pole neutron monitor): no difference in spectral shape.

Pitch angles: low energy population < 1 GV rigidity locally mirroring (scattered), and a high energy distribution > 1.5 GV beaming along field line.

Applied Sandberg calibration method to GOES HEPAD channel energies.

Questions:

• Is there a correlation between the broken power law fit parameters? Yes, the power-law spectral index is positively correlated with the roll-over energy (there is a backup slide showing this)
• Most neutron monitor fits are done in rigidity, not energy (as done here)? A power-law extending to infinity is unphysical, and the exponential cutoff (broken power-law) is more realistic.
• Are you suggesting the NM response function shouldn’t use the power-law, but use broken power-law (cutoff power-law)? Yes.
• Is PAMELA in LEO? Yes.
• Observed spectrum is corrected for geomagnetic cutoff? Yes, by only using observations outside rigidity cutoff.
• PAMELA cannot observe event continuously. How is this accounted for? (See “Data analysis” slide). Use a rigidity-dependent duty cycle, and interpolate over times when cannot see event, plus supplement with calibrated GOES data.

Lei/MSM:

MSM is the Magnetospheric Shielding Model, a tool for computing geomagnetic shielding cutoffs and Earth shadowing. Uses Geant4-based MAGNETOCOSMICS/PLANETOCOSMICS particle tracing code. Based on extended T89: TS89, Boberg’s 1995 extension, plus IGRF 2010 (1955 to 2015). Every 5 years, 5x5 lat/lon grid, at 450 km, every 3 hrs in UT, all values of Kp 0 to 9. Each map provides V(k), Stormer constant.

Average transmission function computed from vertical cutoff and Earth shadowing effect for each rigidity.

Caveats: not good for very disturbed conditions (T89). Day of year (season) not included explicitly, but is included via L interpolation.

Outlook:

• Update w/ latest IGRF
• Finer grid
• Improve map statistics
• Re-implement interpolation algorithms

Questions:

• We know T89 is not the greatest. Maybe do some case comparisons against more recent Tsyganenko models. There are some efforts, but difficult to engage collaborators because it’s not seen as interesting enough.
• It would be nice to have updated Kp-based external field model that’s better than T89. Error analysis so far is just comparing map resolution, not comparing to particle observation.
• We need to validate these cutoffs against in situ particle observations: Van Allen, PAMELA, PROBA-V/EPT (all have narrow field of view, narrow energy channels)

Discussion:

What is the release status of PAMELA data? (A. Bruno not in room to answer.)

A real-time monitor of cutoffs (probably based on LEO data) is needed. A proposal is in to NASA to do so, funding decision not made yet. ESA SSA program may be interested in such a project of its own.

How hard is it to make use of MSM? Right now, it’s a little convoluted, at least in ESPREM.

There is a need to update the CREME96 peaks & worst cases & GCRs.

SEPEM has a radial scaling system that’s roughly 1/R for fluence and closer to 1/R² for peak flux.

Proposing 50% worst case spectrum in a month, 95% worst case spectrum in a month, 95% worst case spectrum over mission. Big question is what time average to use? Also, behind how much shielding? This is a big challenge: there is no good single answer. We know the current approach of using a seemingly arbitrary number of worst days or worst weeks, or scale factors doesn’t work. But, we’re having a hard time defining what should be provided in a specification document, short of going back to the models and doing an onerous calculation with the full time series model.

An education note: many engineers mistakenly think that the solar particles are unaffected by shielding, but that’s only sorta true of GCR.