Structure and Dynamics of Inner Magnetosphere and Their Effects on Radiation Belt Electrons APL Chia-Lin Huang Boston University, MA, USA CISM Seminar, March 24th,
Download ReportTranscript Structure and Dynamics of Inner Magnetosphere and Their Effects on Radiation Belt Electrons APL Chia-Lin Huang Boston University, MA, USA CISM Seminar, March 24th,
Structure and Dynamics of Inner Magnetosphere and Their Effects on Radiation Belt Electrons APL Chia-Lin Huang Boston University, MA, USA CISM Seminar, March 24th, 2008 Special thanks: Harlan Spence, Mary Hudson, John Lyon, Jeff Hughes, Howard Singer, Scot Elkington, and many more Goals of my Research To understand the physics describing the structure and dynamics of field configurations in the inner magnetosphere To assess the performance of global magnetospheric models under various conditions To quantify the response of global magnetic and electric fields to solar wind variations, and ultimately their effects on radial transport of radiation belt electrons. 2 Motivation: Radiation Belts Discovery of Van Allen radiation belts – Explorer 1, 1958 Trapped protons & electrons, spatial distribution (2-7 RE), energy (~MeV) outer belt slot region inner belt J. Goldstein 3 Dynamical Radiation Belt Electrons Why study radiation belt electrons? Because they are physically interesting Radiation damage to spacecraft and human activity in space Goal: describe and predict how radiation belts evolves in time at a given point in space Green [2002] 4 Solar Wind and Magnetosphere Ring Current Average picture of solar wind and magnetosphere (magnetic field, regions, inner mag. plasmas) Variations of Psw, IMF Bz causes magnetospheric dynamics 5 Magnetic Storms Most intense solar wind-magnetosphere coupling IMF Bz southward, strong electric field in the tail Formation of ring current and its effect to field configurations Dst measures ring current development Storm sudden commencement (SSC), main phase, and recovery phase Duration: days 6 Magnetospheric Pulsations Ultra-low-frequency (ULF) MHD waves Frequency and time scale: 2-7 mHz, 1-10 minutes Field fluctuation magnitude First observed in 19th century Waves standing along the magnetic field lines connect to ionospheres [Dungey, 1954] Morphology and generation mechanisms are not fully understood 7 Global Magnetospheric Models Provide global B and E fields needed for radiation belt study Data-based: Tsyganenko models Parameterized, quansi-static state of average magnetic field configurations Physics-based: Global MHD code Self-consistent, time dependent, realistic magnetosphere Global MHD simulation Importance and applications, validation of the global models Empirical model Tsyganenko model LFM MHD code 8 Charged Particle Motion in Magnetosphere Gyro, bounce and drift motions Gyro ~millisecond, bounce ~ 0.1-1 second, drift ~1-10 minutes Adiabatic invariants and L-shell W B J p|| ds BdS To change particle energy, must violate one or more invariants Sudden changes of field configurations Small but periodic variation of field configurations 9 Why is it so Hard? What Would Help? Proposed physical processes Acceleration: large- and small-scale recirculations, heating by Whistler waves, radial diffusion by ULF waves, cusp source, substorm injection, sudden impulse of solar wind pressure and etc. Loss: pitch angle diffusion, Coulomb collision, and Magnetopause shadowing. Transport Difficulties to differentiate the mechanisms: Lack of Measurements Lack of an accurate magnetic and electric field model Converting particle flux to distribution function is tricky Need better understanding of wave-particle interactions Computational resource 10 The Rest of the Talk Magnetospheric field dynamics: data & models Large-scale: Magnetic storms Small-scale: ULF wave fields Effects of field dynamics on radiation belt electrons Create wave field simulations Quantify electron radial transport in the wave fields 11 Lyon-Fedder-Mobbary Code Lyon et al. [2004] Uses the ideal MHD equations to model the interaction between the solar wind, magnetosphere, and ionosphere Simulation domain and grid 2D electrostatic ionosphere Solar wind inputs LFM grid in equatorial plane Field configurations and wave field validations by comparing w/ GOES data 12 Data/Model Case Study 24-26 September 1998 major storm event (Dst minimum -213 nT) LFM inputs: solar wind and IMF data Geosynchronous orbit Compare LFM and GOES B-field at GEO orbit Sep98 event: solar wind data and Dst 13 Statistical Data/Model Comparisons 9 magnetic storms; 2month non-storm interval Field residual B = BMHD – BGOES LFM field lines are consistently understretched, especially during storm-time, on the nightside Predict reasonable nonstorm time field Improvements of LFM Increase grid resolution Add ring current 14 Statistical comparison of Tsyganenko models and GOES data 52 major magnetic storm from 1996 to 2004 TS05 has the best performance in all local time and storm levels Field residual B = BGOES – BTmodel Under-estimate T96 T02 TS05 Perfect prediction Over-estimate 15 Consequence of field model errors Inaccurate B-field model could alter the results of related studies Example: radial profiles of phase space density of radiation belt electrons Discrepancies between Tsyganenko models using same inputs Model field lines traced from GOES-8’s position (left) Pitch angles at GOES-8’s position and at magnetic equator (right) ~15% error between T96 and TS05 16 ULF Waves in Magnetosphere NASA Wave sources: shear flow, variation in the solar wind pressure, IMF Bz, and instability etc. Previous studies: integrated wave power, wave occurrence Next, calculate wave power as function of frequency using GOES data; wave field prediction of LFM and T model. 17 Power Spectral Density (PSD) Calculate PSD using 3hour GOES B-field data Procedures: 1. Take out sudden field change 2. De-trend w/ polynomial fit 3. De-spike w/ 3 standard deviations 4. High pass filter (0.5 mHz) 5. FFT to obtain PSD [nT2/Hz] 18 GOES B-field PSDs in FAC Noon 9 years of GOES data (G-8, G-9 and G-10 satellites) Field-aligned coordinates Separate into 3-hour intervals (8 local time sectors) Dawn Dusk Calculate PSDs Median PSD in each frequency bin Compressional Azimuthal Midnight Radial 19 PSDB [nT2/Hz] Sorting GOES Bb PSD by SW Vx 20 Bz southward PSDB [nT2/Hz] Bz northward Sorting GOES Bb PSD by IMF Bz 21 ULF Waves in LFM code Direct comparisons of ULF waves during Feb-Apr 1996 in field-aligned coord. GOES data Bb compressional Bn radial PSDB [nT2/Hz] LFM output Bφ azimuthal Local Time Much better than expected! 22 Dst and Kp effects on ULF wave power High Dst interval Dst ≤ -40 nT Low Dst interval Dst > -40 nT ULF wave power has higher dependence on Kp than Dst Even though LFM does not reproduce perfect ring current, it predicts reasonable field perturbations High Kp interval Kp ≥ 4 Low Kp interval Kp < 4 23 TS05 model LFM code GOES data ULF wave prediction of Tsyganenko model Underestimates the wave power at geosynchronous orbit Field fluctuations are results of an external driver Lack of the internal physical processes 24 Summary of Model Performance Model Tsyganenko model LFM MHD code Storm config. Non-storm O O X O ULF wave field X O Use LFM’s wave fields during non-storm time to study ULF wave effects on radiation belt electrons Such conditions exist during high speed solar wind streams intervals. 25 ULF Wave Effects on RB Electrons Strong correlation between ULF wave power and radiation belt electron flux [Rostoker et al., 1998] Drift resonant theory [Hudson et al., 1999 and Elkington et al., 1999] ULF waves can effectively accelerate relativistic electrons Elkington et al. [2003] Rostoker et al. [1998] Quantitative description of wave-particle interaction 26 Particle Diffusion in Magnetosphere Diffusion theory: time evolution of a distribution of particles whose trajectories are disturbed by innumerable small, random changes. Pitch angle diffusion (loss): violate 1st or 2nd invariant Radial diffusion (transport and acceleration): violate 3rd invariant f 1 2 D L f LL 2 t L L L (Radial diffusion equation) , where DLL L2 2 day 1 (Radial diffusion coefficient) 27 Radial Diffusion Coefficient, DLL Large deviations in previous studies Possible shortcomings Experimental (solid) and theoretical (dashed) DLL values Walt [1994] Over simplified theoretical assumptions Lack of accurate magnetic field model and wave field map Insufficient measurement M. Walt’s suggestion: follow RB particles in realistic magnetospheric configurations 28 When Does LFM Predict Waves Well? GOES and LFM PSDs sorted by solar wind Vx bins LFM does better during moderate activities Create ULF wave activities by driving the LFM code with synthetic solar wind pressure input X O O O 29 Solar Wind Pressure Variation Histograms of solar wind dynamic pressure from 9 years of Wind data for Vx = 400, 500, and 600 km/s bins Make time-series pressure variations proportional to solar wind Vx 30 Synthetic Solar Wind Pressure (Vx) LFM inputs: Constant Vx; variation in number density. Northward IMF Bz (+2 nT), to isolate pressure driven waves. Idealized LFM Vx simulations using high time and spatial resolutions 31 Idealized Vx Simulations LFM Vx runs GOES data Vx = 400 Vx = 500 Vx=600 GOES statistical study (9 years data) as function of Vx (“mostly” northward IMF) Drive LFM to produce “real” ULF waves with solar wind dynamic pressure variations as function of Vx (“purely” northward IMF) 32 Eφ Wave Power Spatial Distributions 6 mHz Wave power PSDE ( f ) df [(m V / m) 2 ] 0.5 mHz Wave power increases as Vx (Pd variations) increases Wave amplitude is higher at larger radial distance (wave source) 33 Radiation Belt Simulations Test particle code [Elkington et al., 2004] Satisfy 1st adiabatic invariant Guiding center approximation 90o pitch angle electron Push particles using LFM magnetic and electric fields Simulate particles in LFM Vx = 400 and 600 km/s runs Particle initial conditions Fixed μ = 1800 MeV/G Radial: 4 to 8 RE 1o azimuthal direction ~15000 particles /run 34 Rate of Electron Radial Transport (DLL) Convert particle location to L* [Roederer, 1970] L* 2 k0 RE Calculate our radial diffusion coefficient, DLL(Vx) DLL L2 2 DLL increases with Vx DLL increases with L 35 Compare DLL Values I The major differences between previous studies and this work B ~10 nT Amplitude of wave field IMF Bz Magnetic field model Particle energy Calculating method Theoretical assumption Differences make it impossible for a fair comparison B ~2 nT B ~1 nT Highlight: Selesnick et al. [1997] 36 Compare DLL Values II DLL ~ dB2 [Schulz and Lanzerotti, 1974] After scaling for wave power Compare to Selesnick et al. [1997] again Match well with Vx=600 km/s interval (L-dependent) Average Vx of Selesnick et al. [2007] and IMF Bz effect This suggests that radial diffusion is well-simulated, can differentiate from other physical processes DLL(Vx, Bz, Pdyn, Kp etc.) 37 Summary TS05 best predicts GEO magnetic fields in all conditions LFM has good predictions of quiet time fields, but not for storm time ULF wave structures and amplitudes at GEO sorted by selected parameters ULF wave field predictions: LFM is very good, but not TS05 Radial diffusion coefficient derived from MHD/Particle code 38 Conclusions and Achievements Most comprehensive, independent study of state-of-the-art empirical magnetic field models Most quantitative investigation of global MHD simulations in the inner magnetosphere Most comprehensive observational ULF wave fields at geosynchronous orbit dedicated to outer zone electron study First exploration on ULF wave field performance of global magnetospheric models First DLL calculation by following relativistic electrons in realistic, self-consistent field configurations and wave fields of an MHD code 39