Radiation Belt Relativistic Electron Loss Processes

John SampleUC Berkeley-Space Sciences Lab

Spring AGU 2006With Thanks to Robyn Millan

The Static Radiation Belts

• Even the quiet, ‘static’ radiation belts are constrained by losses.– Slot formation by Wave-

Particle scattering of electrons into the atmosphere

– outer boundary (inside the last closed field line) determined by:

~current sheet scattering

~magnetopause encounters

The Highly Variable Radiation Belts

• Fluxes can change by orders of magnitude on short timescales, especially around magnetic storms

• However, the wide variability in radiation belt response to storm conditions (or lack thereof) demonstrates a competition between acceleration and loss processes

Reeves et al. ‘03

Tools of the Trade

• The end of the field line– LEO satellites in high inclination orbits– Bremsstrahlung observing Balloons– Sounding Rockets

• Mid field line– GEO Satellites – Other Equatorial Orbits– Generally looking at Changes in flux:

convolves acceleration, loss, and adiabatic changes

– Loss cone usually too narrow to resolve

• Theory and Modeling– With good models of field and distributions

we can connect disparate observations, follow evolution of PSD, and make predictions based on input conditions

– still require validation and refinement

Onsager et alBlake et al

Days, Dec. 1999

Frequently, we need multi-point observations, and must rely on models to connect the dots

A Zoo of Loss Processes

• At least three broad categories of flux decrease– pitch angle scattering-atmospheric precipitation– magnetopause encounters– adiabatic decreases

• Classification Schemes– Local Time (e.g. Duskside dropouts, Dawnside microbursts)– Energy Dependence (keV microbursts vs. MeV microbursts)– Time profile (microbursts vs. bands, Duskside REP)– Storm-time Dependence (Main phase vs. Recovery– Even observation method (large historical record waiting to be mined)– Different Physics (e.g. Current sheet scattering vs. gyroresonance, when does QL theory

break down)

• Still to Determine the t, L,MLT, E, and activity dependent relative importance of each

• Should be included in any comprehensive model • Losses can constrain acceleration rates required to give measured fluxes

The Adiabatic Response

• Kim and Chan ‘97 looked at fully adiabatic response of equatorially mirroring electrons for Nov. 93 storm

• changes must occur slower than the drift frequency– Ring current reduces B in

inner magnetosphere

– electrons move outward to conserve third adiabatic invariant

– conservation of the first adiabatic inv. reduces perp. momentum

2211 dsBdsB Used symmetric and asymmetric field modelsFlux decrease occurs because of energy dependence and spatial gradientFound that some real loss was required, suggested magnetopause encounters

(e.g. Dessler + Karplus ’61 McIlwain ‘66)

Precipitation into the atmosphere• Equatorial pitch angles smaller

than αL will encounter the atmosphere before mirroring

• Wave Particle Interactions– Gyroresonance requires

-k||v|| = ±n/– Chorus (~100Hz- few kHz)– EMIC (~Hz, below H+,He+,O+)– Hiss (~100Hz- few kHz)

• Current sheet scattering– when electron gyro-radius is

comparable to radius of curvature of the field line.

– should be most important near midnight

– should also affect protons

• Bounce Loss Cone vs. Drift Loss Cone

after Selesnick 03

Size of Loss cone is a function of longitude

Equatorial pitch angles smaller than this will strike the atmosphere

Degrees East

Available Waves

• Large number of wave modes available for gyroresonance

• Hiss vs. Chorus separation occasionally ambiguous

• Waves have local time, activity dependence– many times we do not have wave

measurements at the same time as particles

Summers et al. ‘98

Meredith et al. ‘04

Plasmaspheric HissHiss magnetic field amplitude:

Electron minimum resonant energy ~1 MeV only inside L~3

(Meredith et al., 2004)

Microburst Precipitation

~100s

X-ray Microbursts

Anderson and Milton, 1964• Early balloon observations of ~100 keV microbursts attributed to scattering by VLF whistler-mode waves (Parks, 1978, Rosenberg et al., 1990)

• Relativistic electron microbursts: first reported by Imhof ‘92, studied extensively with SAMPEX, large geometric factor HILT which measures >1MeV electrons allows for time resolution up to 20ms.

(Lorentzen et al., 2001)

Sampex

MLT / L DistributionMicroburst precipitation

• Occur between L=4-6

• 0200-1000 MLT

• Outside plasmapause

Whistler Chorus

• Also occurs on dayside

• Outside plasmapauseMeredith et al. ‘02

Microbursts

(Courtesy T. P. O’Brien)

MeV microbursts and Whistler Chorus

• local time circumstantial evidence led Lorentzen ’01 to look for conjunction of POLAR (Wave data) with SAMPEX observations of MeV microbursts. – no 1-1 correspondence was

found, but Chorus risers having duration similar to single microbursts were seen at similar times,L-Shells, MLT

• Demonstrated that resonance with MeV electrons would have to occur off equatorially or at a higher harmonic

Distribution of EMIC waves

(from Erlandson and Ukhorskiy, 2001)(from Meredith et al., 2003)

• Peak near dusk meridian ~colocated with bulge in plasmasphere

EMIC Resonance Condition

• In order for EMIC waves to be resonant with electrons, the electrons must overtake the wave at high doppler shift

• this enforces a minimum

v|| minimum Eres

– small pitch angles will resonate at a smaller total E

• Resonance condition depends strongly on wave velocity– Dispersion relation requires

high density or low magnetic field to resonate with ~1-2MeV electrons

• will also resonate with ~few keV protons

Imhof et al ’86 found ~30% of pre-midnight selective high energy events studied using S81 and other satellites were related to or embedded in regions of proton precipitation

Meredith et al. 03

Combination of waves can be important

• Albert 2003 showed lifetimes with hiss alone ~7x longer than with hiss and EMIC

• EMIC still needs high ratio of fpe to fce in order to resonate with ~MeV electrons

L=4, 100pT Hiss, 1pT EMIC(weighted by drift time in relevant region)

Energy Selective Precipitation on the Dusk Side

• We do see strong scattering losses on the duskside– Thorne and Andreoli ‘80

• 3 events in 14 months of s3-3 data (1% of all precipitation events observed)

– Imhof ’86 • 41 events using S-(72,78,81)-1

data. Some very intense.• Filtered on narrow L structures• e-foldings >500keV• Concluded such events were

rare, but poor MLT coverage may have missed peak.

– Foat ’96, Millan ’02• Observed Bremsstrahlung X-ray

bursts from similarly high energy electron precipitation on the dusk side.

Energy Selective Precipitation on the Dusk Side

• We do see strong scattering losses on the duskside– Thorne and Andreoli ‘80

• 3 events in 14 months of s3-3 data (1% of all precipitation events observed)

– Imhof ’86 • 41 events using S-(72,78,81)-1

data. Some very intense.• Filtered on narrow L-structures• e-foldings >500keV• Concluded such events were

rare, but poor MLT coverage may have missed peak.

– Foat ’96, Millan ’02• Observed Bremsstrahlung X-ray

bursts from similarly high energy electron precipitation on the dusk side.

NOON

MIDNIGHT

Loss out the Magnetopause

• Adiabatic drift outward– When Dst drops, particles move outward to conserve their 3rd invariant

• Magnetopause compression– Sharp increases in solar wind dynamic pressure

• Li et al ‘97 used to explain losses into L=4.6 during November ‘93 storm

• Usually thought to be most important at higher L but…

• Outward radial diffusion towards a time dependent boundary has recently been found to be capable of rapidly propagating flux drops from MP encounters inward

• In some cases magnetopause losses should be recognizable in proton fluxes at similar energies

• Very difficult to measure directly

>2MeV flux dropouts

• Sudden drops in high energy flux at GEO identified by Onsager et al. ’02

• Events begin with tailward stretching of nightside magnetopause, an asymmetric process that can occur before Dst drops

• Fluxes aren’t equilibrated in local time within one electron drift.

Looking for a mechanism• Green et al. ’04 Identified 52 similar

events, performed a superposed epoch analysis using GEO,HEO,LEO data

– events first observed at dusk, similar LT progression as in Onsager

– converting to PSD eliminates adiabatic effects

• accounts for local time progression of events

– however fluxes remain low well after field returns to normal so real loss occurred

– Magnetopause encounters?• losses went in to L ~4,

• no similar dropout in proton flux

– Precipitation?

Looking for a mechanism• Green et al. ’04 Identified 52 similar

events, performed a superposed epoch analysis using GEO,HEO,LEO data

– Adiabatic? No

– Magnetopause encounter? Unlikely

– Precipitation?• Do see an increase of Bounce Loss

around time of event onset, however data coverage is sparse

• Millan et al. ’06– One of the events identified by Green et

al. study occurred during the MAXIS balloon campaign of January 2000.

– MAXIS observed an extended, intense X-ray burst with sufficient precipitation to account for the flux drop at GEO

Quantification of Loss Rates

• Lorentzen ’01 Demonstrated that microburst loss from a single storm could flush the radiation belts of MeV electrons

• O’Brien ’04 Principal losses occur during main phase, losses continue through recovery, but with less cumulative effect.

• Millan ’02 Losses from Duskside REP discussed earlier were shown to be capable of emptying outer belt in a few days

• Current work by Shprits et al. has shown outward radial diffusion and MP encounters capable of similar high loss rates

• Ukorhisky et al. used test particles in TS05 to show strong MP losses during Sep. 7, ’02 storm

• Selesnick ’06 By using a model to compare drift loss to trapped population using SAMPEX able to calculate daily energy dependent atmospheric loss rates.

Summary• Wide variety of loss processes act on relativistic

electrons in the radiation belts• Each process (subprocess!) has been shown to be

dominant in some cases• To actually specifically identify any given event

with a particular process usually requires joining multiple satellite data sets with appropriately fit models

• It is possible to quantify a loss rate without understanding the details of the loss process

• Much work to be done