1 SEP Spectral and Compositional Variability at High Energies Tylka et al., ApJ 625, 474-495(2005) Tylka et al, ApJS 164, 536-551 (2006) Tylka and Lee,

1

SEP Spectral and Compositional SEP Spectral and Compositional Variability at High EnergiesVariability at High Energies

Tylka et al., ApJ 625, 474-495(2005)

Tylka et al, ApJS 164, 536-551 (2006)

Tylka and Lee, ApJ 646, 1319-1334 (August 1, 2006)

Tylka et al., poster at this conference

Allan J. TylkaAllan J. TylkaUS Naval Research Laboratory, Washington DCUS Naval Research Laboratory, Washington DC

2A Model for Spectral and Compositional Variability A Model for Spectral and Compositional Variability

at High Energies in Large Gradual SEP Eventsat High Energies in Large Gradual SEP EventsA. J. Tylka & M. A. Lee, A. J. Tylka & M. A. Lee, ApJ, August 2006ApJ, August 2006

Tylka et al. 2005

Event-to-event variability in high-energy SEP spectral & compositional characteristics due to the interplay of two factors:

Evolution in the geometry of the CME-driven shock as it moves outward from the Sun, (quasi-perpendicular v. quasi-parallel)

A compound seed population, typically comprising at least suprathermals from flares and suprathermals from the corona and/or solar-wind.

Bn vs. RS

3

Fe/O vs. EnergyFe/O vs. Energy Spectral VariabilitySpectral Variability 33He/He/44He vs. EnergyHe vs. Energy

Average Fe/O EnhancementAverage Fe/O Enhancement <Q> vs. Energy<Q> vs. Energy

A Model for SEP Variability at High Energies, A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)

The model addresses a number of issues in high-energy SEP phenomenology, The model addresses a number of issues in high-energy SEP phenomenology, with both qualitative and quantitative successes.with both qualitative and quantitative successes.

4

Model parameters chosen to match spectral shape of O and Fe (next slide).

Using nominal composition and charge states, the model reproduces the SEP heavy-ion fractionation effects discovered by Breneman & Stone.

C, N, O, Ne, Mg, Si, S, Ca, Fe

SIS data provided by C.M.S. Cohen

A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)

The Origin of Breneman & Stone (1985) FractionationThe Origin of Breneman & Stone (1985) Fractionation

5

Model Model parameters were parameters were selected to selected to roughly match roughly match spectral shapes of spectral shapes of Fe and O above a Fe and O above a few few MeV/nucleon.MeV/nucleon.

Model Model normalized to normalized to oxygen data at ~8 oxygen data at ~8 MeV/nucleon MeV/nucleon

Normalizations Normalizations of all other of all other species fixed species fixed automatically by automatically by the model.the model.

A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)

The Origin of Breneman & Stone (1985) FractionationThe Origin of Breneman & Stone (1985) Fractionation

6

The Origin of Breneman & Stone (1985) FractionationThe Origin of Breneman & Stone (1985) Fractionation

These calculations are the first quantitative explanation of the Breneman & Stone effect.

A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)A Model for SEP Variability at High Energies, A. J. Tylka & M. A. Lee (ApJ, Aug. 2006)

7Enhanced Fe/O Asymptote at High EnergiesEnhanced Fe/O Asymptote at High Energies

Regardless of the size of the flare-Regardless of the size of the flare-component in the seed population, re-component in the seed population, re-acceleration by a quasi-perp shock give an acceleration by a quasi-perp shock give an asymptotic Fe/O value at high-energies:asymptotic Fe/O value at high-energies:

(Fe/O)(Fe/O)shockshock = (Fe/O)= (Fe/O)flareflare• [Q/A] • [Q/A] Fe, Flare Fe, Flare / [Q/A]/ [Q/A]O, FlareO, Flare

= 8 • (20/56) / (8/16)= 8 • (20/56) / (8/16)

= 5.9= 5.9

using average values for impulsive events:using average values for impulsive events:

(Fe/O)(Fe/O)flareflare = 8x <Q = 8x <QFeFe>=20 <Q>=20 <QOO>=8>=8

Do we see this in the data?Do we see this in the data?

Modeling Modeling (Tylka & Lee (Tylka & Lee 2006)2006)

8Average Fe/O Enhancement at High EnergiesAverage Fe/O Enhancement at High Energies

Based on Based on average characteristics characteristics of flare ions, we’ve made a of flare ions, we’ve made a prediction for the prediction for the average value value of enhanced Fe/O at high-of enhanced Fe/O at high-energies in gradual events.energies in gradual events.

Thus, we can understand this Thus, we can understand this average behavior as the average behavior as the consequence of flare seed consequence of flare seed particles accelerated to high particles accelerated to high energies by shocks. It is a energies by shocks. It is a challenge for other models.challenge for other models.

Numbers on the datapoints tell how many events were used in the average.Numbers on the datapoints tell how many events were used in the average.

30-40 MeV/nuc30-40 MeV/nuc 40-50 MeV/nuc40-50 MeV/nuc 50-75 MeV/nuc50-75 MeV/nuc

9

Very Large SEP Events of Cycle 23Very Large SEP Events of Cycle 23

ACE Fe/O (normalized to corona) at 30-40 MeV/nucIn 1997-2002:In 1997-2002:

13 out of 38 events with 13 out of 38 events with Fe/O Fe/O >> 4x corona 4x corona

Sole Event Selection Criterion: >30 MeV GOES proton fluence > 2 x 105/cm2-sr

• An unbiased survey of high-energy heavy-ions in the Cycle’s largest events

10

Very Large SEP Events of Cycle 23Very Large SEP Events of Cycle 23

ACE Fe/O (normalized to corona) at 30-40 MeV/nucIn 1997-2002:In 1997-2002:

13 out of 38 events with 13 out of 38 events with Fe/O Fe/O >> 4x coronal 4x coronal

In 2003-2005:In 2003-2005:

0 out of 20 events with 0 out of 20 events with Fe/O Fe/O >> 4x coronal 4x coronal

Where has all the Fe gone?

?

Sole Event Selection Criterion: >30 MeV GOES proton fluence > 2 x 105/cm2-sr

• An unbiased survey of high-energy heavy-ions in the Cycle’s largest events

11Flare Sizes and Longitudes in the Very Large Flare Sizes and Longitudes in the Very Large SEP Events of Cycle 23SEP Events of Cycle 23

In both time periods, most of the events are associated with large flares at western longitudes.

It appears that differences in characteristics of the associated flare cannot explain the disappearance of events with enhanced Fe/O.

12

In 1997 – 2002, 17 out of 35 events had CME Speed < 1500 km/s

In 2003 – 2005, 2 out of 17 events had CME Speed < 1500 km/s

Why have the ‘slow’ CMEs almost disappeared from our event sample?*

* 2003-2005 had 5 more very fast (>2500 km/s) CMEs than we saw in 1997-2002. But this fact has no bearing on this question.

CME Speeds in the Very Large SEP Events of Cycle 23CME Speeds in the Very Large SEP Events of Cycle 23

?

13CME Speeds in the Very Large SEP Events of Cycle 23CME Speeds in the Very Large SEP Events of Cycle 23

Although SEP event statistics are limited in 2003-2005, it appears that the probability of getting a very large SEP event with a CME at ~1000-1500 km/s dropped by nearly an order of magnitude late in the Cycle.

14Hypotheses on the Origin of Enhanced Fe/O Hypotheses on the Origin of Enhanced Fe/O at High Energies in Very Large SEP Eventsat High Energies in Very Large SEP Events

1. A direct flare component that dominates at high energies:

• Either from the “impulsive phase” of the flare

• Or from reconnection beneath the CME after launch (Cane et al. 2003; 2006)

• According to Cane et al., this direct flare component is most likely to be observed when the associated CME is comparatively slow.

2. CME-driven shocks accelerate “fresh” suprathermal seed ions that have escaped from the flare that accompanied the CME launch.

• Li & Zank 2005

• Requires the existence of open field lines connecting the flare site to the region upstream of the shock

3. CME-driven shocks accelerate “remnant” suprathermal flare ions, left over in the corona and interplanetary medium from earlier impulsive SEP events.

• Mason et al. 1999; Tylka et al. 2001.

• The flare that accompanied the CME-launch does not contribute to the shock’s seed population because it energized particles mostly on closed loops (Reames 2002).

15

Unless the magnetic topology of CME-source regions or some relevant flare characteristic is fundamentally different late in the Cycle, hypotheses 1 & 2 cannot explain the disappearance of Fe-rich events in 2003-2005.

But hypothesis 3 may be consistent with the disappearance:

• The overall level of flare activity started to decline in 2003;

• The remnant flare suprathermal population (at least near Earth) also declined.

Monthly Flare Count (SGD Online)

Hypotheses on the Origin of Enhanced Fe/O Hypotheses on the Origin of Enhanced Fe/O at High Energies in Very Large SEP Eventsat High Energies in Very Large SEP Events

Wind Suprathermal Fe (Desai et al. 2006)

16Why are slower (< 1500 km/s) CMEs less effective at Why are slower (< 1500 km/s) CMEs less effective at producing very large SEP events late in the Cycle?producing very large SEP events late in the Cycle?

Our event selection requires that the fluence of >30 MeV protons exceed 2 x 105 /cm2-sr

Tylka & Lee (2006) suggested that slower CMEs can produce large numbers of high-energy SEPs only if the shock is quasi-perpendicular near the Sun.

• Otherwise, the SEP spectra are too steep at high energies.

Late in the Cycle, the suprathermal seeds needed by these quasi-perp shocks have diminished.

• These CMEs therefore become less effective in producing large fluences of high energy SEPs.

(In some events – such as 2005 January 20, which may have been quasi-perp near the Sun – the suprathermal seeds probably came from previous gradual SEP events.)

17

Some Caveats …Some Caveats …

An implication of this proposed scenario is that the population of suprathermal protons has also decreased in 2003-2005.

But this has not yet been proven: Desai et al. [2006] showed only that the population of remnant suprathermal flare Fe has apparently diminished.

It is hoped that Wind/3DP and ACE/EPAM can address the Solar Cycle dependence of ~100 keV protons during solar-quiet times. But this will depend upon the long-term stability and background levels in these instruments.

It will also be important to see how the absolute intensity and spectral shape of the suprathermal populations have evolved.

Thus, although this idea appears compelling, there are still open issues requiring further investigation.

18CME Speeds in the Very Large SEP Events of Cycle 23CME Speeds in the Very Large SEP Events of Cycle 23

Note the data points at 500-1000 km/s:

1997-2002: 5 SEP events in 5 years

2003-2005: 1 SEP event in 3 years

Only one “very slow” CME in 2003-2005 (925 km/s, on 2004 Nov 01) produced a SEP event big enough to meet our criterion. This particular event was preceded by three days of high flare activity and large impulsive events. This event has the second largest Fe/O at 30 MeV/nuc in the 2003-2005 event sample, with even larger Fe/O at ~3 MeV/nuc (see table below). This event also has the highest 3He/4He (= 6.1 + 0.5%) among the 64 large SEP events surveyed by Desai et al. 2006b. This exceptional event requires further study.

19

Cane et al. (GRL 30, 8017, 2003) wrote:Cane et al. (GRL 30, 8017, 2003) wrote:

“It has also been suggested that enrichments of 3He (and Fe) in large gradual events could result from shock acceleration of remnant flare suprathermal particles in the interplanetary medium from previous small flares (Mason et al. 1999). However, since it has now been shown that flare particles escape directly in all SEP events (Cane et al. 2002), we see no need to invoke a population from earlier, smaller flares.”

20

Cane et al. (GRL 30, 8017, 2003) wrote:Cane et al. (GRL 30, 8017, 2003) wrote:

“It has also been suggested that enrichments of 3He (and Fe) in large gradual events could result from shock acceleration of remnant flare suprathermal particles in the interplanetary medium from previous small flares (Mason et al. 1999). However, since it has now been shown that flare particles escape directly in all SEP events (Cane et al. 2002), we see no need to invoke a population from earlier, smaller flares.”

Cane et al. 2002: type IIIL radio emissions:

• Implicit assumption here: the acceleration process that makes ~keV electron beams also makes >30 MeV/nuc ions (esp. Fe).

• All the events of 2003-2005 have type IIILs; but there are no Fe enhancements.

• This assumption requires reconsideration.

21SummarySummary

A simple analytical model based on two factors:

• Evolution in the geometry of the CME-driven shock as it moves outward from the Sun, (quasi-perpendicular v. quasi-parallel)• A compound seed population, typically comprising at least suprathermals from flares and suprathermals from the corona and/or solar-wind.

accounts qualitatively and quantitatively for many aspects of high-energy SEP variability. This suggests that these ideas should be pursued more rigorously.

The disappearance of Fe-rich events late in Cycle 23 may serve to clarify various hypotheses in the origins of these events. In particular, it favors shock acceleration of remnant flare suprathermals

22

BackupsBackups

23

24Effects of Averaging over Shock-Normal AngleEffects of Averaging over Shock-Normal AngleTylka & Lee, submitted to ApJTylka & Lee, submitted to ApJ

Power-law Power-law index the same index the same for all species.for all species.

Species “X” Species “X” and and BnBn

dependence dependence reside in reside in rollover, Erollover, E0X0X

SW component SW component has steeper has steeper spectrum by Espectrum by E-1-1

Flare & SW Flare & SW components components have different have different Q/A dependenceQ/A dependence

25How these calculations work…

When we average over =cos Bn, we pull a term ~(Q/A) out of the integral. Neglecting non-leading terms,

For our example, = 1.5

This is close to what we observe for the pure coronal case.

But we still have a positive slope for the flare component!

26How these calculations work…

But in the flare component of the seed particles, the composition already has an inherent Q/A dependence:

(Glenn will explain this!)

Substituting this in, for flare particles processed through a shock, we get:

By parameter variation, we can get various slopes on the Breneman & Stone plot.

27Energy Dependence of Energy Dependence of 33He/He/44He and <QFe>He and <QFe>

• Model calculations roughly reproduce the observed values and energy dependence, except below ~1 MeV/nucleon.

• Our calculations overestimate the flare component of the seed population at low energies. This problem can probably be solved with a more careful treatment of the shock’s actual Bn evolution.

Data from G. Mason, M. Wiedenbeck, and W.F. Dietrich Data from J. Mazur and A. Labrador.

Tylka & Lee 2006

28Spectral and Compositional Variability at High EnergiesSpectral and Compositional Variability at High Energies

The event with suppressed high-energy Fe/O :The event with suppressed high-energy Fe/O :

Exponential rollovers at high energiesExponential rollovers at high energies

Fe softer than O.Fe softer than O.

The event with enhanced high-energy Fe/O :The event with enhanced high-energy Fe/O :

Power-law spectra at high energiesPower-law spectra at high energies

Fe harder than O.Fe harder than O.

This correlation This correlation is a is a generalgeneral characteristic of the SEP data. characteristic of the SEP data.

29Correlations in SEP Characteristics at High EnergiesCorrelations in SEP Characteristics at High Energies

<<QQFeFe>> vs. Fe/O vs. Fe/O

We thus have correlations among measures of We thus have correlations among measures of composition (Fe/O), charge states, spectral composition (Fe/O), charge states, spectral shape, and event size. shape, and event size.

These correlations are powerful constraints for These correlations are powerful constraints for models.models.

Fe/OFe/O vs. Spectral Steepeningvs. Spectral Steepening

30The “Zoo” of Fe/O vs. EnergyThe “Zoo” of Fe/O vs. EnergyData Data (ACE and Wind)(ACE and Wind)Modeling Modeling (Tylka & Lee (Tylka & Lee

2006)2006)

Vary shock geometry and size of flare component Vary shock geometry and size of flare component in seed population: in seed population:

R = (Seed Pop Flare O) / (Seed Pop Coronal O)R = (Seed Pop Flare O) / (Seed Pop Coronal O)

31Spectral CharacteristicsSpectral Characteristics

Tylka et al. [2005] characterized spectra by quantifying the steepening of the oxygen spectrum between Wind/LEMT (at ~3-10 MeV/nuc) and ACE/SIS (at >30 MeV/nuc).

Tylka et al. [2005] found an anti-correlation between high-energy Fe/O and the steepening of the oxygen spectrum.

-- events with strong Fe/O enhancements tended to be power-laws

In 2003-2005, only one event (2005 January 20) showed a power-law spectrum over the combined Wind-ACE energy range of ~3-100 MeV/nuc.

32Very Large SEP Events of 2004-2005Very Large SEP Events of 2004-2005

Event list for 1997-2003 published in Tylka et al., ApJ 625, 474-495 (2005).