CME propagation in the interplanetary medium

Transcription

1 CME propagation in the interplanetary medium (A review talk) Jens Kleimann Theoretische Physik IV, Ruhr-Universität Bochum, Germany 3rd SOLAIRE Network Meeting November 05, 2009 Puerto de la Cruz, Tenerife

2 A CME s life: It... gets born, [Initiation models: flux emergence, C. Jacobs] evolves, [Coronal development of CMEs, T. Török] and leaves home to see places. [ This talk ]

3 A CME s life: It... gets born, [Initiation models: flux emergence, C. Jacobs] evolves, [Coronal development of CMEs, T. Török] and leaves home to see places. [ This talk ] 1 Motivation 2 Trajectory Arrival times Geo-effectiveness 3 analytical numerical 4

4 Why care about CMEs? Major manifestation of solar activity M kg, W J rate (1...6)/day, 10 % of which hit Earth! CMEs relate to many other fields of Solar physics flares CMEs particle acceleration at shocks global flux removal... Commercial application: space weather safety concerns for astronautics, satellite communication failures, etc.

5 Why care about CMEs? Major manifestation of solar activity M kg, W J rate (1...6)/day, 10 % of which hit Earth! CMEs relate to many other fields of Solar physics flares CMEs particle acceleration at shocks global flux removal... Commercial application: space weather safety concerns for astronautics, satellite communication failures, etc.

6 Why care about CMEs? Major manifestation of solar activity M kg, W J rate (1...6)/day, 10 % of which hit Earth! CMEs relate to many other fields of Solar physics flares CMEs particle acceleration at shocks global flux removal... Commercial application: space weather safety concerns for astronautics, satellite communication failures, etc.

7 Trajectory Arrival times Geo-effectiveness S/C observations from: LASCO on SOHO (white-light coronagraph, 32R FoV, since 1995) ACE, Wind L 1, since 1997) Helios 1/2 0.3 AU, ) SMEI on Coriolis (white-light, all-sky, since 2003, r > 70R ) STEREO A/B (since 2007) anecdotal: ICME detection via H + enhancement by Voyager 2 (@ 58 AU) & Ulysses [Paularena et al. 2001]...

8 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

9 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

10 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

11 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

12 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

13 Trajectory Arrival times Geo-effectiveness Main goals: Predict a CME s trajectory, 2. arrival time (at Earth), 3. geo-effectiveness 1st order assumption: CMEs expand radially. Only "halo" CMEs will hit Earth. [Schwenn 2005]: 10% of events involved non-halo CMEs (missing alarms) 10% of halo CMEs miss Earth (false alarms) Eastward deflection due to Parker spiral? slow fast CMEs go East West (noticeably) (slightly) [Wang et al. 2004]

14 CME tracking Motivation Trajectory Arrival times Geo-effectiveness Problems: STEREO s dual view can be used to reconstruct the 3-D trajectory via stereoscopy. (Well-posed problem for points and curves [Inhester 2006]). CMEs are extended, partially translucend objects tricky to identify common features in images. S/C launched into "deep" solar min. only few events to study. Preliminary results indicate "quasi-radial" paths.

15 CME tracking Motivation Trajectory Arrival times Geo-effectiveness Problems: STEREO s dual view can be used to reconstruct the 3-D trajectory via stereoscopy. (Well-posed problem for points and curves [Inhester 2006]). CMEs are extended, partially translucend objects tricky to identify common features in images. S/C launched into "deep" solar min. only few events to study. Preliminary results indicate "quasi-radial" paths.

16 CME tracking Motivation Trajectory Arrival times Geo-effectiveness Problems: STEREO s dual view can be used to reconstruct the 3-D trajectory via stereoscopy. (Well-posed problem for points and curves [Inhester 2006]). CMEs are extended, partially translucend objects tricky to identify common features in images. S/C launched into "deep" solar min. only few events to study. Preliminary results indicate "quasi-radial" paths.

17 CME tracking Motivation Trajectory Arrival times Geo-effectiveness Problems: STEREO s dual view can be used to reconstruct the 3-D trajectory via stereoscopy. (Well-posed problem for points and curves [Inhester 2006]). CMEs are extended, partially translucend objects tricky to identify common features in images. S/C launched into "deep" solar min. only few events to study. Preliminary results indicate "quasi-radial" paths. [Maloney et al., submitted]

18 Arrival times 1AU) Trajectory Arrival times Geo-effectiveness Required data: 1 initial speed (from coronagraphs, modulo projection) 2 pos.(+) / neg.( ) acceleration a en route due to thermal pressure (+), magnetic forces (±), gravity ( ), aerodynamic drag ( ), "snow plough-effect" ( ) For r < 32R : linear height-time-fit ok. [St.Cyr et al. 2000] (gradual CMEs: a > 0 out to 6 R [Schwenn et al. 2006]) For r 1 AU : empirical models for single CMEs Problem: direct CME data only available near Sun (coronagraphs) and Earth (in-situ spacecrafts)

19 Arrival times 1AU) Trajectory Arrival times Geo-effectiveness Required data: 1 initial speed (from coronagraphs, modulo projection) 2 pos.(+) / neg.( ) acceleration a en route due to thermal pressure (+), magnetic forces (±), gravity ( ), aerodynamic drag ( ), "snow plough-effect" ( ) For r < 32R : linear height-time-fit ok. [St.Cyr et al. 2000] (gradual CMEs: a > 0 out to 6 R [Schwenn et al. 2006]) For r 1 AU : empirical models for single CMEs Problem: direct CME data only available near Sun (coronagraphs) and Earth (in-situ spacecrafts)

20 Arrival times 1AU) Trajectory Arrival times Geo-effectiveness Required data: 1 initial speed (from coronagraphs, modulo projection) 2 pos.(+) / neg.( ) acceleration a en route due to thermal pressure (+), magnetic forces (±), gravity ( ), aerodynamic drag ( ), "snow plough-effect" ( ) For r < 32R : linear height-time-fit ok. [St.Cyr et al. 2000] (gradual CMEs: a > 0 out to 6 R [Schwenn et al. 2006]) For r 1 AU : empirical models for single CMEs Problem: direct CME data only available near Sun (coronagraphs) and Earth (in-situ spacecrafts) Need to bridge [ 0.2, 1.0] AU interval.

21 Trajectory Arrival times Geo-effectiveness Solution #1: Trace type II radio emissions from the CME s upstream shock, occurring at harmonic f = 9 n/m 3 MHz n at source region position but: stand-off distance not known ( 0.25 AU at Earth) Solution #2: Identify CME ICME pairs, use quadrature observations (1 coronagraph + 1 in-situ S/C over limb) to minimize projection effects. Idea: Relate travel time T to initial (v 0 ) vs. final (v e ) speed.

22 Trajectory Arrival times Geo-effectiveness Solution #1: Trace type II radio emissions from the CME s upstream shock, occurring at harmonic f = 9 n/m 3 MHz n at source region position but: stand-off distance not known ( 0.25 AU at Earth) Solution #2: Identify CME ICME pairs, use quadrature observations (1 coronagraph + 1 in-situ S/C over limb) to minimize projection effects. Obs.1 Obs Idea: Relate travel time T to initial (v 0 ) vs. final (v e ) speed.

23 Arrival time statistics (I) Trajectory Arrival times Geo-effectiveness Bruecker et al. [1998]: T avg 80 h (not too bad, esp. at solar min) Lindsay et al. [1999]: linear fit of v e = v e (v 0 ) v approaches v sw Gopalswamy et al. [2001]: linear a = a(v 0 ) fit to (v 0, T ) data kinematic eq. v 0 T + at 2 /2 = R s/c, T = T (v 0 ), T 10 h best match if a = 0 beyond 0.75 AU

24 Arrival time statistics (I) Trajectory Arrival times Geo-effectiveness Bruecker et al. [1998]: T avg 80 h (not too bad, esp. at solar min) Lindsay et al. [1999]: linear fit of v e = v e (v 0 ) v approaches v sw Gopalswamy et al. [2001]: linear a = a(v 0 ) fit to (v 0, T ) data kinematic eq. v 0 T + at 2 /2 = R s/c, T = T (v 0 ), T 10 h best match if a = 0 beyond 0.75 AU

25 Arrival time statistics (I) Trajectory Arrival times Geo-effectiveness Bruecker et al. [1998]: T avg 80 h (not too bad, esp. at solar min) Lindsay et al. [1999]: linear fit of v e = v e (v 0 ) v approaches v sw Gopalswamy et al. [2001]: linear a = a(v 0 ) fit to (v 0, T ) data kinematic eq. v 0 T + at 2 /2 = R s/c, T = T (v 0 ), T 10 h best match if a = 0 beyond 0.75 AU

26 Arrival time statistics (II) Trajectory Arrival times Geo-effectiveness Schwenn et al. [2005]: Correlation v rad speed of lateral expansion (defineable w/o projection effects!) v rad 0.88 v exp viscous drag for decel. to v sw = 0 T ( )] [203 h = vexp ln km/s Cargill [2004]: "aerodynamic" drag a(v) (v v sw ) 2 little difference.

27 Arrival time statistics (II) Trajectory Arrival times Geo-effectiveness Schwenn et al. [2005]: Correlation v rad speed of lateral expansion (defineable w/o projection effects!) v rad 0.88 v exp viscous drag for decel. to v sw = 0 T ( )] [203 h = vexp ln km/s Cargill [2004]: "aerodynamic" drag a(v) (v v sw ) 2 little difference.

28 Arrival time statistics (II) Trajectory Arrival times Geo-effectiveness Schwenn et al. [2005]: Correlation v rad speed of lateral expansion (defineable w/o projection effects!) v rad 0.88 v exp viscous drag for decel. to v sw = 0 T ( )] [203 h = vexp ln km/s Cargill [2004]: "aerodynamic" drag a(v) (v v sw ) 2 little difference.

29 Trajectory Arrival times Geo-effectiveness In summary: 1 On average T 80 h 2 v 0 highly variable, v cme v sw for r R 3 empirical formulas T = T (v 0 ) or T (v exp ), but: 4 large scatter due to oversimplifaction, CMEs and IP medium both too variable/structured for simple fitting laws

30 Trajectory Arrival times Geo-effectiveness In summary: 1 On average T 80 h 2 v 0 highly variable, v cme v sw for r R 3 empirical formulas T = T (v 0 ) or T (v exp ), but: 4 large scatter due to oversimplifaction, CMEs and IP medium both too variable/structured for simple fitting laws "[We propose] that a number of CMEs be dropped from La Torre di Pisa and their drag force be directly measured." Reiner et al. [2003]

31 Trajectory Arrival times Geo-effectiveness Geo-effectiveness := ability to cause severe magnetic storms/ energetic particle flux at Earth. B z < 0 favors interaction with Earth s magnetosphere: { > 0 mainly FL repulsion shielding effect B z,cme < 0 dayside reconnection particle influx (plus magnetosphere compression by p cme ) B cme may stem from 1 original flux rope field and/or 2 draped/compressed IMF ahead of CME Fast CMEs have stronger B (but not B z ). [Lindsay et al. 1999]

32 Trajectory Arrival times Geo-effectiveness Geo-effectiveness := ability to cause severe magnetic storms/ energetic particle flux at Earth. B z < 0 favors interaction with Earth s magnetosphere: { > 0 mainly FL repulsion shielding effect B z,cme < 0 dayside reconnection particle influx (plus magnetosphere compression by p cme ) B cme may stem from 1 original flux rope field and/or 2 draped/compressed IMF ahead of CME Fast CMEs have stronger B (but not B z ). [Lindsay et al. 1999]

33 Trajectory Arrival times Geo-effectiveness Geo-effectiveness := ability to cause severe magnetic storms/ energetic particle flux at Earth. B z < 0 favors interaction with Earth s magnetosphere: { > 0 mainly FL repulsion shielding effect B z,cme < 0 dayside reconnection particle influx (plus magnetosphere compression by p cme ) B cme may stem from 1 original flux rope field and/or 2 draped/compressed IMF ahead of CME Fast CMEs have stronger B (but not B z ). [Lindsay et al. 1999]

34 Trajectory Arrival times Geo-effectiveness Geo-effectiveness := ability to cause severe magnetic storms/ energetic particle flux at Earth. B z < 0 favors interaction with Earth s magnetosphere: { > 0 mainly FL repulsion shielding effect B z,cme < 0 dayside reconnection particle influx (plus magnetosphere compression by p cme ) B cme may stem from 1 original flux rope field and/or 2 draped/compressed IMF ahead of CME Fast CMEs have stronger B (but not B z ). [Lindsay et al. 1999]

35 analytical numerical MagnetoHydroDynamics preferred tool to model underlying physics, esp. with respect to non-linear v B interaction. Analytical works (few in number): 1 The Gibson & Low [1998] flux rope. time-dependent 3-D MHD config; assumes self-similar evolution structure of ρ(r) used to create synthetic white-light images also used as init for simulations

36 analytical numerical MagnetoHydroDynamics preferred tool to model underlying physics, esp. with respect to non-linear v B interaction. Analytical works (few in number): 1 The Gibson & Low [1998] flux rope. time-dependent 3-D MHD config; assumes self-similar evolution structure of ρ(r) used to create synthetic white-light images also used as init for simulations

37 analytical numerical MagnetoHydroDynamics preferred tool to model underlying physics, esp. with respect to non-linear v B interaction. Analytical works (few in number): 1 The Gibson & Low [1998] flux rope. time-dependent 3-D MHD config; assumes self-similar evolution structure of ρ(r) used to create synthetic white-light images also used as init for simulations

38 analytical numerical MagnetoHydroDynamics preferred tool to model underlying physics, esp. with respect to non-linear v B interaction. Analytical works (few in number): 1 The Gibson & Low [1998] flux rope. time-dependent 3-D MHD config; assumes self-similar evolution structure of ρ(r) used to create synthetic white-light images also used as init for simulations

39 analytical numerical MagnetoHydroDynamics preferred tool to model underlying physics, esp. with respect to non-linear v B interaction. Analytical works (few in number): 1 The Gibson & Low [1998] flux rope. time-dependent 3-D MHD config; assumes self-similar evolution structure of ρ(r) used to create synthetic white-light images also used as init for simulations

40 Analytical models (cont d) analytical numerical 2 Flux tube model to probe a CME s internal properties [Wang et al. 09] requires: self-similarity, J B, ϕ = 0 fixing coefficients c by [R, L](t) fit to obs. data gives Γ cme (t) critical values: Γ 4/3 : (f em /f th ) decreases with r Γ 2/3 : no more net acceleration

41 Analytical models (cont d) analytical numerical 2 Flux tube model to probe a CME s internal properties [Wang et al. 09] requires: self-similarity, J B, ϕ = 0 fixing coefficients c by [R, L](t) fit to obs. data gives Γ cme (t) critical values: Γ 4/3 : (f em /f th ) decreases with r Γ 2/3 : no more net acceleration

42 Analytical models (cont d) analytical numerical 2 Flux tube model to probe a CME s internal properties [Wang et al. 09] requires: self-similarity, J B, ϕ = 0 fixing coefficients c by [R, L](t) fit to obs. data gives Γ cme (t) critical values: Γ 4/3 : (f em /f th ) decreases with r Γ 2/3 : no more net acceleration

43 Analytical models (cont d) analytical numerical 2 Flux tube model to probe a CME s internal properties [Wang et al. 09] requires: self-similarity, J B, ϕ = 0 fixing coefficients c by [R, L](t) fit to obs. data gives Γ cme (t) critical values: Γ 4/3 : (f em /f th ) decreases with r Γ 2/3 : no more net acceleration

44 analytical numerical Space weather prediction relies on large-scale numerical MHD. CSEM [Toth 2005] CISM [Odstrcil 2008] Major (technical) challenge: High resolution requirements due to 1 need to track features R across > 200 R 2 Lack of symmetry solar min: B is 2-D, but CME expansion dipolar axis solar max: B is 3-D itself

45 analytical numerical Space weather prediction relies on large-scale numerical MHD. CSEM [Toth 2005] CISM [Odstrcil 2008] Major (technical) challenge: High resolution requirements due to 1 need to track features R across > 200 R 2 Lack of symmetry solar min: B is 2-D, but CME expansion dipolar axis solar max: B is 3-D itself

46 analytical numerical Space weather prediction relies on large-scale numerical MHD. CSEM [Toth 2005] CISM [Odstrcil 2008] Major (technical) challenge: High resolution requirements due to 1 need to track features R across > 200 R 2 Lack of symmetry solar min: B is 2-D, but CME expansion dipolar axis solar max: B is 3-D itself

47 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007]

48 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007]

49 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007]

50 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007]

51 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007] Solution #2: Performance tuning specially tailored grids, esp. spherical with radially varying r = r(r) multi-scale models [e.g. Riley et al. 06] mesh refinement techniques [BATS-R-US, AMRVAC,...]

52 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007] Solution #2: Performance tuning specially tailored grids, esp. spherical with radially varying r = r(r) multi-scale models [e.g. Riley et al. 06] mesh refinement techniques [BATS-R-US, AMRVAC,...]

53 analytical numerical Solution #1: Ignore ϕ dependence anyway. expansion along polar axis: interesting but somewhat unrealistic expansion near ecliptic (implies torus-shaped "CME") 2-D/3-D comparison [Jacobs et al. 2007] Solution #2: Performance tuning specially tailored grids, esp. spherical with radially varying r = r(r) multi-scale models [e.g. Riley et al. 06] mesh refinement techniques [BATS-R-US, AMRVAC,...]

54 analytical numerical Two types of MHD models "principal": idealized settings, few control parameters Goal: assess importance of initial config/ physical effects for resulting development "realistic": init near real situation, uses as much physics as possible Goal: reproduce (remote/ insitu) data from actual events 1 Different initialisation methods ( previous review talks) 2 Different realisations of the background solar wind: uniform [Vandas et al. 1998, 2002] structured [Odstrcil & Pizzo 1999; Manchester et al. 2004] realistic [Hayashi et al. 2006; Shen et al. 2007]

55 analytical numerical Two types of MHD models "principal": idealized settings, few control parameters Goal: assess importance of initial config/ physical effects for resulting development "realistic": init near real situation, uses as much physics as possible Goal: reproduce (remote/ insitu) data from actual events 1 Different initialisation methods ( previous review talks) 2 Different realisations of the background solar wind: uniform [Vandas et al. 1998, 2002] structured [Odstrcil & Pizzo 1999; Manchester et al. 2004] realistic [Hayashi et al. 2006; Shen et al. 2007]

56 analytical numerical Two types of MHD models "principal": idealized settings, few control parameters Goal: assess importance of initial config/ physical effects for resulting development "realistic": init near real situation, uses as much physics as possible Goal: reproduce (remote/ insitu) data from actual events 1 Different initialisation methods ( previous review talks) 2 Different realisations of the background solar wind: uniform [Vandas et al. 1998, 2002] structured [Odstrcil & Pizzo 1999; Manchester et al. 2004] realistic [Hayashi et al. 2006; Shen et al. 2007]

57 analytical numerical 3 Different physics, e.g. treatment of the energy budget: isothermal adiabatic, γ = γ 0 5/3 γ = γ(r) [e.g. Fahr et al. 76, Lugaz et al. 07] complete energy equation with a) ad-hoc heating [e.g. Hartle & Barnes 70, Manchester 04] e.g. Q(r) = q(r) [T 0 γp/ρ] T T 0 "target temp." b) consistent Alfvenic wave heating (and pressure) p w : t ε ± + [(v ± v A )ε ± ] = ε ± 2 v and p w = ε + + ε 2 or t P + [...] =... and p w = 1 2 fh f 0 P(f, r) df

58 analytical numerical 3 Different physics, e.g. treatment of the energy budget: isothermal adiabatic, γ = γ 0 5/3 γ = γ(r) [e.g. Fahr et al. 76, Lugaz et al. 07] complete energy equation with a) ad-hoc heating [e.g. Hartle & Barnes 70, Manchester 04] e.g. Q(r) = q(r) [T 0 γp/ρ] T T 0 "target temp." b) consistent Alfvenic wave heating (and pressure) p w : t ε ± + [(v ± v A )ε ± ] = ε ± 2 v and p w = ε + + ε 2 or t P + [...] =... and p w = 1 2 fh f 0 P(f, r) df

59 analytical numerical 3 Different physics, e.g. treatment of the energy budget: isothermal adiabatic, γ = γ 0 5/3 γ = γ(r) [e.g. Fahr et al. 76, Lugaz et al. 07] complete energy equation with a) ad-hoc heating [e.g. Hartle & Barnes 70, Manchester 04] e.g. Q(r) = q(r) [T 0 γp/ρ] T T 0 "target temp." b) consistent Alfvenic wave heating (and pressure) p w : t ε ± + [(v ± v A )ε ± ] = ε ± 2 v and p w = ε + + ε 2 or t P + [...] =... and p w = 1 2 fh f 0 P(f, r) df

60 analytical numerical 3 Different physics, e.g. treatment of the energy budget: isothermal adiabatic, γ = γ 0 5/3 γ = γ(r) [e.g. Fahr et al. 76, Lugaz et al. 07] complete energy equation with a) ad-hoc heating [e.g. Hartle & Barnes 70, Manchester 04] e.g. Q(r) = q(r) [T 0 γp/ρ] T T 0 "target temp." b) consistent Alfvenic wave heating (and pressure) p w : t ε ± + [(v ± v A )ε ± ] = ε ± 2 v and p w = ε + + ε 2 or t P + [...] =... and p w = 1 2 fh f 0 P(f, r) df

61 analytical numerical Findings from principal models (I) Some indication of nearly self-similar evolution [e.g. Kleimann et al. 09] cf. constancy of cone angle [Schwenn et al. 2005] (relevant for analytical models, etc.)

62 analytical numerical Findings from principal models (I) Some indication of nearly self-similar evolution [e.g. Kleimann et al. 09] cf. constancy of cone angle [Schwenn et al. 2005] (relevant for analytical models, etc.)

63 analytical numerical Findings from principal models (I) Some indication of nearly self-similar evolution [e.g. Kleimann et al. 09] cf. constancy of cone angle [Schwenn et al. 2005] (relevant for analytical models, etc.)

64 analytical numerical Findings from principal models (II) { "Inverse" "Normal" CME evolution depends strongly on 1 background SW (e.g. higher speeds in fast, dilute winds, depends on physics included) [Jacobs et al. 2005] and 2 initial topology: } { faster prom.s give slower } { equatorward. CMEs which deflect poleward. } [Chané et al. 2006], as predicted by Zhang & Low [2004]

65 analytical numerical Findings from principal models (II) { "Inverse" "Normal" CME evolution depends strongly on 1 background SW (e.g. higher speeds in fast, dilute winds, depends on physics included) [Jacobs et al. 2005] and 2 initial topology: } { faster prom.s give slower } { equatorward. CMEs which deflect poleward. } [Chané et al. 2006], as predicted by Zhang & Low [2004]

66 analytical numerical Further challenges: CMEs exhibit diverse structure: Sometimes three parts ("light bulb" [Hundhausen 1988]), but often not. 10(!) different morphological classes acc. to Howard et al. [1985] Interaction/merging: About 2 of 3 CMEs are "complex ejecta" [Burlaga 2002] Incomplete knowledge of IMF structure (accessible only via extrapolation of near-surface fields + in-situ data at single points)

67 analytical numerical Further challenges: CMEs exhibit diverse structure: Sometimes three parts ("light bulb" [Hundhausen 1988]), but often not. 10(!) different morphological classes acc. to Howard et al. [1985] Interaction/merging: About 2 of 3 CMEs are "complex ejecta" [Burlaga 2002] Incomplete knowledge of IMF structure (accessible only via extrapolation of near-surface fields + in-situ data at single points)

68 analytical numerical Further challenges: CMEs exhibit diverse structure: Sometimes three parts ("light bulb" [Hundhausen 1988]), but often not. 10(!) different morphological classes acc. to Howard et al. [1985] Interaction/merging: About 2 of 3 CMEs are "complex ejecta" [Burlaga 2002] Incomplete knowledge of IMF structure (accessible only via extrapolation of near-surface fields + in-situ data at single points)

69 Motivation CMEs show a very diverse phenomenology, therefore purely kinematic models have limited predictive power. Modelling results crucially depend on physical effects included (e.g. wave heating). As numerical models become more sophisticated, they benefit from input due to high-quality S/C observations. Nearly self-similar evolution of single(!) CMEs realistic modelling at small radii is essential!

70 (Thank you!)