1. Solar Extreme Events 2005, Armenia September 2005 Extreme Ground Level Enhancements Observed by Spaceship Earth John W. Bieber University of Delaware, Bartol Research Institute and Department of Physics and Astronomy Supported by NSF grant ATM-0000315 Visit our Website: http://neutronm.bartol.udel.edu/

2. OBSERVATION OF COSMIC RAYS WITH GROUND-BASED DETECTORS Ground-based detectors measure byproducts of the interaction of primary cosmic rays (predominantly protons and helium nuclei) with Earth’s atmosphere Two common types: –Neutron Monitor Typical energy of primary: ~1 GeV for solar cosmic rays, ~10 GeV for Galactic cosmic rays –Muon Detector / Hodoscope Typical energy of primary: ~50 GeV for Galactic cosmic rays (surface muon detector)

3. NEUTRON MONITORS Older type “BP28” – proportional counter filled with BF 3 : n + 10 B → α + 7 Li Modern type – counter filled with 3 He: n + 3 He → p + 3 H Both types some- times called “NM64” Neutron Monitor in Nain, Labrador Construction completed November 2000

4. Spaceship Earth Spaceship Earth is a network of neutron monitors strategically deployed to provide precise, real- time, 3-dimensional measurements of the cosmic ray angular distribution: 11 Neutron Monitors on 4 continents Multi-national participation: –Bartol Research Institute, University of Delaware (U.S.A.) –IZMIRAN (Russia) –Polar Geophysical Inst. (Russia) –Inst. Solar-Terrestrial Physics (Russia) –Inst. Cosmophysical Research and Aeronomy (Russia) –Inst. Cosmophysical Research and Radio Wave Propagation (Russia) –Australian Antarctic Dvivision –Aurora College (Canada)

5. Why are all the stations at high latitude? Reason 1: Uniform energy response Plot shows neutron monitor response to a simulated (rigidity) -5 solar particle spectrum Below a geomagnetic cutoff of about 0.6 GV, atmospheric absorption determines the cutoff – all stations have a uniform energy response in this regime

6. Why are all the stations at high latitude? Reason 2: Excellent directional sensitivity Trajectories are shown for vertically incident primaries corresponding to the 10-, 20-, … 90-percentile rigidities of a typical solar spectrum

7. Why are all the stations at high latitude? Reason 3: Focusing of obliquely incident primaries Because of adiabatic focusing in the converging polar magnetic field, primaries with widely divergent angles of incidence have similar asymptotic directions (most easily seen by following the time- reversed trajectory)

8. SPACESHIP EARTH VIEWING DIRECTIONS Optimized for solar cosmic rays 9 stations view equatorial plane at 40-degree intervals Thule and McMurdo provide crucial 3-dimensional perspective Circles denote station geographical locations. Average viewing directions (squares) and range (lines) are separated from station geographical locations because particles are deflected by Earth's magnetic field. STATION CODES IN: Inuvik, Canada FS: Fort Smith, Canada PE: Peawanuck, Canada NA: Nain, Canada MA: Mawson, Antarctica AP: Apatity, Russia NO: Norilsk, Russia TB: Tixie Bay, Russia CS: Cape Schmidt, Russia TH: Thule, Greenland MC: McMurdo, Antarctica

9. Ground Level Enhancements (GLE): Some Recent Highlights 2001 April 15 (Easter) –2 nd largest event of this cycle –Diffusive event 2005 January 20 (Inauguration Day) –Largest GLE in almost half a century –Extreme anisotropy –Possible wave excitation and nonlinear transport 2003 October 28 (“Halloween”) –Direct solar neutrons detected –Strange propagation characteristics 2000 July 14 (Bastille) –Particles reflected from a magnetic bottleneck downstream of Earth 1989 October 22 Revisited –Strange features of this event may be explained by weak scattering within a closed magnetic loop (A GLE in a CME)

10. THE EASTER GLE The 2001 April 15 (Easter) event was the 2 nd largest of the present cycle, with some stations seeing an increase of more than 2X over the Galactic background Plot shows individual time profiles of Spaceship Earth stations at 1-minute time resolution

11. Station Viewing Directions at Start of Easter GLE Each station is sensitive to a range of directions, because different energies experience different amounts of deflection in Earth’s magnetic field. The red squares show the viewing directions of the median rigidity for each station, and the line encompasses the central 50% of energies detected. All stations have excellent directional sensitivity. “O” and “X” denote the Sunward and anti- Sunward Parker spiral directions, respectively. Station codes: AP Apatity, CS Cape Schmidt, FS Fort Smith, IN Inuvik, MA Mawson, MC McMurdo, NA Nain, NO Norilsk, PE Peawanuck, TI Tixie Bay, TH Thule.

12. Results of Fitting to a First-Order Anisotropy Network data were fitted to a first-order anisotropy, f(θ,φ) = n { 1 + ξ X sinθ cosφ + ξ Y sinθ sinφ + ξ Z cosθ }, where f(θ,φ) is the intensity measured by a station viewing in the direction defined by θ (colatitude) and φ (longitude), n is the particle density, and (ξ X,ξ Y,ξ Z ) are the three components of the anisotropy vector.

13. Modeling Interplanetary Transport of the Easter GLE Numerical solutions of the Boltzmann equation, including scattering and focusing in a realistic Parker field, were least squares fitted to both the density and “weighted anisotropy” (density times ordinary anisotropy) derived from Spaceship Earth observations. Modeling yields the injection function at the Sun, shown in the top panel, as well as the scattering mean free path, determined to be λ ║ = 0.34 AU in this event Our derived injection onset of 13:42 UT ± 1 min is 2.6 min earlier than that derived from the “inverse velocity” method (Tylka et al., Proc. 28 th ICRC (Tsukuba), p 3305, 2003). Figure and results are from Bieber et al., Astrophys. J. (Lett.), 601, L103-L106, 2004.

14. Particle Acceleration Timing

15. THE RECORD-SETTING JANUARY 20, 2005 GROUND LEVEL ENHANCEMENT (GLE) McMurdo increase was 30X over 6 min – largest increase at sea level since famous 1956 event –Space weather forecasts must strive for the earliest possible warning of major SEP events South Pole increase was 56X, largest ever recorded by a neutron monitor! This distinction is partly owing to South Pole’s unique location at high latitude and high altitude (2820 m) Event was enormously anisotropic: Neutron rate increase at other high- latitude stations was an order of magnitude smaller – “only” 3X or so –In assessing SEP radiation exposure to pilots and air crews, it is essential to account for the event anisotropy

16. GLE AND GIANT GLE A HISTORICAL PERSPECTIVE Neutron Monitor Era (normalized to sea level) –Largest event ever: February 23, 1956 @ 47X –2 nd largest: January 20, 2005 @ 30X –3 rd largest: September 29, 1989 @ 4X B.N.M. (Before Neutron Monitors) –GLE were first reported by Forbush 1 based upon ionization chamber measurements –The ionization chamber events were giant GLE – By Duggal’s reckoning 2 they would correspond to neutron monitor increases (sea level, high latitude) as follows: Feb 28, 1942 @ 7X Mar 7, 1942 @ 8.5X Jul 25, 1946 @ 12X Nov 19, 1949 @ 21X In planning for human missions into deep space, it is important to recognize that SEP event magnitudes vary over an extremely large range 1.Forbush, S. E., Three unusual cosmic-ray increases possibly due to charged particles from the Sun, Phys. Rev., 70, 771-772, 1946. 2.Duggal, S. P. Relativistic solar cosmic rays, Rev. Geophys. Space Phys., 17, 1021-1058, 1979.

17. ENERGY SPECTRUM: POLAR BARE METHOD South Pole station has both a standard neutron monitor (NM64) and a monitor lacking the usual lead shielding (Bare). The Polar Bare responds to lower particle energy on average. Comparison of the Bare to NM64 ratio provides information on the particle spectrum. This event displays a beautiful dispersive onset (lower panel), as the faster particles arrive first. Later, the rigidity spectrum softens to ~P – 5 (where P is rigidity), which is fairly typical for GLE. The dip around 06:55 UT may be related to the change in propagation conditions indicated by our transport model (see below).

18. STATION VIEWING DIRECTIONS AT GLE ONSET STATION CODES FS: Fort Smith, Canada TH: Thule, Greenland MC: McMurdo, Antarctica NA: Nain, Canada SP: South Pole, Antarctica BA: Barentsburg, Norway MA: Mawson, Antarctica AP: Apatity, Russia NO: Norilsk, Russia TB: Tixie Bay, Russia CS: Cape Schmidt, Russia IN: Inuvik, Canada Squares show the asymptotic viewing direction of a median energy (1.4 GeV) solar cosmic ray. Lines encompass the central 80% of detector energy response, extend- ing from the direction of a 0.5 GeV particle to that of a 4.6 GeV particle. Directions of the nominal inward (“O”) and outward (“X”) Parker spiral are also shown.

19. DENSITY AND ANISOTROPY EXTRACTED FROM NETWORK DATA Data of the 12 individual stations were fitted to the function f(μ) = a 0 + a 1 e bμ where μ is cosine of pitch angle, and a 0, a 1, and b are free parameters. The density and anisotropy of solar cosmic rays were then computed from the fit function. “Weighted anisotropy” is simply density times ordinary anisotropy. (Density and weighted anisotropy may also be viewed as the zeroth- and first-order coefficients of a Legendre expansion of f(μ)).

20. TRANSPORT MODELING Model based on numerical solutions of Boltzmann equation Free parameters are scattering mean free path λ, turbulence “q” parameter (spectrum slope), and injection time profile at the Sun. Could not obtain a satisfactory fit with a single mean free path – extreme anisotropy and fast decay of initial peak demand a long mean free path (~ 1 AU), but this is incompatible with slower decay and smaller anisotropy of the second peak. We solved this dilemma by superposing two separate solutions of the Boltzmann equation, each with its own value of λ and q, and its own injection function. Optimal fits appear at left. Earlier injection (dotted) has λ = 0.9 ± 0.1 AU and q ≈ 0.5 –Later injection (dashed) has λ = 0.6 ± 0.1 AU and q ≈ 1.5 –Heavy dark line shows superposition of the two injections. In upper two panels the model is compared with data (histogram) Interestingly, the superposed injection function displays a typical monotonic rise and monotonic fall. The double peaked structure of the data results from the changed transport parameters, not from a double peaked injection. Injection function is quite similar to 50 MHz radio burst (red). (Black curve is 500 MHz.) Injection onset was 06:41 UT at Sun

21. TRANSPORT MODELING: DID THE EARLY PARTICLES EXCITE WAVES THAT HELPED TO SCATTER THE LATER PARTICLES? It is customary to treat relativistic solar cosmic rays as “test particles” that do not significantly alter the medium through which they travel. However … … the test particle approximation may not be applicable in this extreme event: –First, the turbulent wave intensity in the relevant wavenumber range was quite weak at event onset, as evidenced by the long initial mean free path. –Second, this event was one of the largest ever, as noted above. At GLE maximum, we estimate the energy density of relativistic solar cosmic rays was possibly an order of magnitude larger than the energy density of magnetic turbulence in the relevant wavenumber range. This estimate is preliminary and subject to confirmation by more detailed analysis, but it appears there was sufficient energy for nonlinear wave-particle interactions to occur in this event.

22. RELATIVISTIC SOLAR PARTICLES OBSERVED BY NEUTRON MONITORS: 2003 OCTOBER 28 (a) Solar neutron increase at Tsumeb, Namibia (b) Solar protons at several high-latitude monitors (c) Omnidirectional intensity with model fit (d) Derived injection profile at Sun For details see: Bieber et al., Geophys. Res. Lett. 32, L03S02, doi:10.1029/2004GL02 1492, 2005.

23. A MYSTERY OF OCTOBER 28: WHY DID THE FIRST PARTICLES ARRIVE FROM ANTI-SUNWARD ? The earliest onset was seen at Cape Schmidt (CS), viewing almost directly anti- Sunward, and at Norilsk (NO), viewing almost anti-Sunward along the nominal Parker spiral.

24. Magnetic Mirrors in the Solar Wind GLE particles are sometimes injected into a normal Parker background field, as in (a) – e.g., Easter GLE But frequently they are injected into a background that is highly disturbed as a result of earlier events from the same active region. In (b), particles mirror from a magnetic bottleneck downstream of Earth. We believe this configuration existed in the Bastille Day 2000 event. Sometimes particles are injected into a closed magnetic loop, as in (c). We believe this configuration explains the highly unusual October 22, 1989 event. Figure is from Ruffolo et al., Astrophys. J., submitted, 2005.

25. BASTILLE EVENT: A MAGNETIC “BOTTLENECK” BEYOND EARTH Best fit for a Parker field; λ ║ = 0.15 AU Best fit for a Magnetic Bottleneck; λ ║ = 0.27 AU; Bottleneck is 0.3 AU beyond Earth and is 85% reflecting

26. ACE AND WIND MAGNETIC FIELD DATA SUPPORT THE BOTTLENECK HYPOTHESIS At GLE onset, the disturbance preceding the event would have been ~0.3 AU beyond Earth, and the field enhancement is the correct size to provide 85% reflection, in excellent agreement with our modeling. GLE Onset

27. OCTOBER 22, 1989: A STRANGE COSMIC RAY “SPIKE” From 17:55 – 18:10 UT, McMurdo observed a huge solar particle increase to ~150% of background, while South Pole observed a ~30% increase (corrected to sea level) – 7 other stations saw no increase during this cosmic ray “spike.” After 18:30 the event appears diffusive, with moderate anisotropies and slow decay.

28. BIDIRECTIONAL COSMIC RAY STREAMING: FURTHER EVIDENCE FOR STRONG MIRRORING ON OCT 22, 1989

29. OUR MODELING SUPPORTS THE LOOP HYPOTHESIS FOR THE OCT 22 GLE Left panels show density, right panels show weighted anisotropy Points are data; lines show our best fit model for each of 4 scenarios (top to bottom): 1.Injection onto a normal Archimedean spiral 2.Archimedean spiral with a magnetic bottleneck beyond Earth 3.Closed loop with injection onto one leg of loop 4.Best model: Closed loop with injection onto both legs of loop Figure is from Ruffolo et al., Astrophys. J., submitted, 2005.

30. With Spaceship Earth you can even determine “q” QLT Result: Magnetic power spectrum slope “q” governs scattering rate Ф – If P(k) ~ k -q, then Ф(μ) ~ | μ| q-1 (1- μ 2 ) where μ is cosine of pitch angle Ф(μ) ∂f/∂μ must be finite at 90 o pitch angle for particles to scatter through 90 o – therefore the behavior of f at 90 o is diagnostic of q During the Bastille Day event, the pitch angle distribution displayed a sharp change at 90 o (left panels), indicative of q ~ 5/3 During the Easter event, the distribution was smooth through 90 o (right panels), consistent with q ~ 1 Figures from Bieber et al., Astrophys. J., 567, 622, 2002 and Bieber et al., Astrophys. J. (Lett.), 601, L103-L106, 2004.

31. SPACE WEATHER APPLICATIONS Numerous stations provide realtime displays of raw count rate data Univ Delaware / Bartol site (left) provides higher level data, including realtime displays of cosmic ray –Density –Flow direction –Pitch angle distribution http://neutronm.bartol.udel.edu/spaceweather/

32. SPACESHIP EARTH STATIONS ARE WELL SITUATED TO ALERT / MONITOR RADIATION HAZARD ON POLAR AIRLINE ROUTES Line shows Chicago-Beijing great circle route. Squares are Spaceship Earth stations.

33. Neutron Monitors Can Provide the Earliest Alert of a Solar Energetic Particle Event In the January 20, 2005 GLE, the earliest neutron monitor onset preceded the earliest Proton Alert issued by the Space Environment Center by 14 minutes.

34. In this study, a GLE Alert is issued when 3 stations of Spaceship Earth (plus South Pole) record a 3% increase With 3 stations, false alarm rate is near zero GLE Alert precedes SEC Proton Alert by ~ 10-30 min

35. SUMMARY Ground-based cosmic ray detectors remain the state-of- the-art instrumentation for studying high-energy (>1 GeV) cosmic rays High-latitude sites offer unique advantages for studying the angular distribution of solar cosmic rays Spaceship Earth is an 11-station network of high-latitude neutron monitors. Analysis of Spaceship Earth data permits determination of –Injection onset to ~1 minute precision –Injection time profile –Interplanetary scattering mean free path –Scattering “q” parameter Mirroring from large-scale interplanetary structures (e.g., loops, bottlenecks) is an important factor in modeling GLE transport

36. LESSONS FOR SPACE WEATHER FROM JANUARY 20, 2005 GLE In planning for human missions into deep space, it is important to recognize the extremely large range of SEP event magnitudes. –In neutron monitor era, largest event was this one (56X at South Pole). (But 1956 event was larger if normalized to sea level.) –Second largest was 1956 event (47X at Leeds) –Then a big gap to third largest in 1989 (4X at numerous stations) Space weather forecasts must strive for the earliest possible warning of major SEP events – 6 minutes from onset to peak in this event! Particle anisotropy can cause variations of as much as 10 to 1 in radiation exposure in polar regions of Earth.