1. Energy deposition and Infrasonic measurement of Bolides P. Brown Dept of Physics and Astronomy, Western University, London CANADA Work sponsored by: NASA Meteoroid Environment Office (MEO)

2. Impact Frequency 2 Meter-sized Impactors @ 20 km/s (0.08 kT) @ 11 km/s (0.03 kT) @ 30 km/s (0.2 kT) Mass ~2 T One such event globally every ~week – ten days Any one optical site on Earth can “see” a meter-sized impact once every ~two decades

3. Some large (D>5 m) impacts over the last two decades (having speed and peak height) 3

4. Observational data for meter- sized impactors Three Sources: –Ground-based fireball networks (European Network, Prairie Network, MORP) [6 meter-sized events] –Fireball producing meteorites [23 to date] with instrumental flight data [10 produced by >1m diameter] –US Government (USG) sensor data [>50 with speed and energy] http://neo.jpl.nasa.gov/fireball/ 4

5. Impact Statistics Speed  Mean = 18.5 ± 0.7 km/s  Median = 17.9 km/s Entry angle = 46 ◦ ± 3 ◦ Mean height of peak brightness  33 km  >90% lie between 20 – 40 km 5

6. Orbital Characteristics: Meter-sized impactors 6

7. Impact Statistics - II Orbital origins  Source population mainly from ν 6 (inner main-belt) 7% Halley Type Comet (HTC) orbits; similar fraction Jupiter-family comet (JFC) origin No trend in strength with size/energy 7 Jupiter Family Comet (JFC) Outer Main-Belt (OB) 3:1 MMR with Jupiter (P_31) Intermediate Mars Crossers (P_IMC) ν 6 secular resonance (P_Nu6 ) Bottke et al (2002) Source Regions :

8. Meter-class impactors : Physical Characteristics Median energy = 0.4 kT Triggered Progressive Fragmentation Model (TPFM) [ReVelle 2005] used for comparison 8 Fireball Class Frag Pres (Mpa) ΔH frag-peak (km) I0.710-14 II0.214-17 IIIa0.0117-19 IIIb0.00119-24 AS – Almahata Sitta (Ure-Anom) K – Kosice (H5) TL – Tagish Lake (C2 ung) BC– Buzzard Coulee (H4) C - Chelyabinsk (LL5) PF – Park Forest (L5) B – Benesov (LL3.5, H5,Primitive Achondrite) SM – Sutter’s Mill (CM2) M – Mariboo (CM2) Meteorites – individual symbols:

9. Energy Deposition Light curve ~ energy deposition Caveat: τ(m,h,v,comp) Of 5m class events with speeds LC only available for Chelyabinsk (top) and Feb 1, 1994 (bottom) 9 Brown et al (2013) Tagliaferri et al (1995)) W/ster

10. Detailed data (Borovicka and Spurny 1996) including light curve, fragmentation behavior, precise astrometry and spectra  Benesov meteorites recovered 2011 (Spurny et al 2014) – mixture of OC types?  Detailed model comparisons to Benesov observations by Borovicka and Popova (1998) Benesov: V =21 km/s; I max ~ -19.5 mag H peak ~24 km Mass estimates: 3000-4000 kg (Borovicka et al.,1998) (ReVelle&Ceplecha, 2002) D ~ 1.3m E = 0.20 kT Benesov (EN 070591) & Sumava (EN 041274) Sumava: V =27 km/s; I max ~ -21.5 mag H peak ~67 km Mass estimates: 5000 kg (Borovicka & Spurny,1996) D ~ 3m E = 0.4 kT Benesov spectra of final flares Borovicka and Spurny (1996)

11. One model interpretation (Borovicka et al., 2013): Earliest fragmentation at ~45 km altitude at P = 0.7 MPa Earliest fragmentation at ~45 km altitude at P = 0.7 MPa Large scale disruption at 30 – 37 km height with P= 1 – 5 MPa Large scale disruption at 30 – 37 km height with P= 1 – 5 MPa By 29 km object was ~20 boulders of 1-2m sizes based on changes in lightcurve By 29 km object was ~20 boulders of 1-2m sizes based on changes in lightcurve These boulders break again at 26 km under P~10 MPa These boulders break again at 26 km under P~10 MPa Lateral fragment speeds ~400 m/s Lateral fragment speeds ~400 m/s Another (bottom by Popova et al (2013)) Based on a number of plausible simulation realizations to encompass large parameter space of fragmentation behavior (in particular) 11 Chelyabinsk - Fragmentation Borovicka et al (2013) Popova et al (2013)

12. 12 Kosice (H5) Fell Feb 28, 2010 – producing 11 kg of H5 meteorites in 200+ fragments Detailed data (Borovicka et al 2015) including light curve, fragmentation behavior, precise astrometry  Very weak meteoroid – fragmented under < 0.1 MPa  Catastrophic disintegration at P ~ 1 MPa  Infrasound constrained energy from I43 RU @ 1400 km range: Kosice: V =15 km/s; I max ~ -18 mag H peak ~36 km Mass estimates: 3500 kg (Borovicka et al., 2015 D ~ 1.2m E=0.1- 0.2 kT Pressure (Pa) Arrival Azimuth Height (km) 57 39 29 22

13. 13 Extreme meteoroid velocities between ~Mach 30 – 240 produce long, narrow shocks over very short time scales. Characteristics (period and amplitude) of the shock wave are related to meteoroid energy deposition Meteor generated infrasound provides another means of determining meteoroid MASS & KINETIC ENERGY @ M = 30, β = 1.9° @ M = 240, β = 0.2° β RORO

14. Meteors (bolides/airbursts) produce low frequency sound when they detonate in the atmosphere Detectable at infrasound arrays at long distances due to low attenuation of sound and natural sound waveguides in atmosphere METEORS! Gravity Waves Infrasound Audible 14

15. Blast Radius dictates the frequency at which a meteor will produce its sound Small R o : High Frequency / Large R o : Low Frequency cm – m size objects  Infrasound: 0.1 – 10 Hz >10 m size (eg Chelyabinsk): < 0.1 Hz As Frequency ↓ Attenuation ↓ Large, energetic sources produce IS which goes further –Lower frequencies, lots of energy Bolide Infrasound 15 (ReVelle 1974/1976)

16. Ground-level overpressure from cylindrical line source theory Numerical implementation of ReVelle (1974) meteor shock theory and comparison to observations of cm-sized meteoroids by Silber et al (2015) –Main finding – weak-shock to linear transition distance from source is larger than originally assumed Crude rule of thumb – ΔP at the ground scales as ~E 1/2 16

17. Application to Chelyabinsk At Chelyabinsk, nuclear airblast relations (Glasstone and Dolan, 1977) predict 5 – 10 kPa overpressure (0.5-1 MT) Cylindrical weak-shock theory predicts ~ 2-3 kPa Need more instrumental records of ground-level overpressure from large bolides Cylindrical theory using lightcurve 1 MT Nuclear 0.5 MT Nuclear

18. International Monitoring System 45/60 Stations Complete (75%) Stations are arrays composed of 4 – 12 microbarometers Signals found through cross-correlation Arrival direction and steepness directly measureable Local wind noise and stratospheric wind system determines detection efficiency Varies with geography and time of year 18 1 km

19. IMS Bolide events : 2010 – mid 2014 Removed auto detected IS events correlated with mining, rocket launches, volcanoes, repeating sources etc. –Total events examined: 1462 –Total number of potential airbursts : 69 (4% of 1462) –Expected number of meter-sized impacts from Brown et al (2002) : 29/yr (130 vs 69) –Expected number of kiloton (~2 m) class airbursts : 4/yr (18 vs 69) 2014 IMS event also detected by USG sensors

20. IMS bolide detection efficiency – Cued vs. “survey” Based on 2014 cued search with USG Sensors: – IMS identifies ~1/4 of all meter-sized impacts –Approximately 3/4 of all such impacts are detectable infrasonically –Cuing important! IMS raw (survey) detections (~15 /yr)  Implications As a stand alone system, current IMS system identifies minority (<0.5) of all meter-sized impactors Cued impacts from next Gen asteroid surveys (eg. ATLAS) should expect most impacts to be detectable by IMS –Will give an estimate of total energy and geolocation

21. 21 Airburst Energy Estimation : Period – Yield Source energy estimates based on periods/amplitudes calibrated to explosive sources Small events at short ranges usually better estimated with amplitudes (but need to include winds) Larger events show good agreement with ground- truth/USG energies (Ens et al., 2012) particularly by averaging periods across many stations USG

22. Example: Regional Infrasound for the Chatham Island Airburst May 16, 2014 @1242 UT (0.8 kT) Pa USG Measurements: Energy = 0.8 kT Mass ~ 25 T Diameter 2.5 – 3 m V = 16.5 km/s Burst altitude = 44 km Entry Angle = 66 degs Infra Measurements: End Height ~ 33 km Begin Height ~ 69 km Burst Height ~ 47 km Infrasound cross-correlation in 15 sec windows with 80% overlap 1 2 3 4 123 4

23. Bolide Infrasound measurements Pilger et al (2015) Chelyabinsk Infrasound only estimated terminal burst altitude = 20 ± 4 km Energy ~ 50 kt Line source Point source US Government Sensor data (2014): Terminal burst altitude = 19 km Energy ~ 33 kt Silber et al (2011) Oct 8, 2009 - Indonesia

24. Future Research Luminous efficiency – calibrate from different techniques Infrasound models for validation of cylindrical line source overpressure estimates at the ground (particularly amplitude model constraints) –(Silber et al., 2015) applied to cm-sized meteoroids, need to expand to meter-sizes –Search IMS for regional IS airburst detection and apply/modify model –Adapt Whitham weak-shock theory to cylindrical hypersonic sources (eg. Haynes and Millet, 2013) 24

25. General Observations Meter-sized impactors begin fragmentation under 0.1 – 1 MPa ram pressure (Popova et al 2011) –peak luminosity is reached 1-2 scale heights lower Fragmentation is complex Lightcurves are crucial to constraining atmospheric energy deposition in individual cases –Not enough meter-class LCs available to make any generalizations about fragmentation behavior Spectra very helpful, but rare Recovered meteorites provide ground-truth Multi-instrumental observations critical – each measurement technique suffers different systematic biases 25

26. Need for Model validation Models have now incorporated very elaborate physics BUT we have few constraints from observations to guide choices in a very large parameter space (Fragmentation!). 1.Compare various models (particularly fragmentation characteristics) to existing published/detailed large bolide measurements (Chelyabinsk, Benesov, Sumava, Moravka, Kosice) 2.Apply models to USG data for statistical studies 3.Process/extract existing but unpublished precise large fireball data and apply models (EN – eg. EN 171101) 26

27. Modeling Capabilities FM ablation model (theoretical and observational fit to bolide data) [Ceplecha and ReVelle 2005] Triggered Progressive Fragmentation ablation model (TPFM) [ReVelle 2007a] Acoustic Gravity Wave production from bolides [ReVelle 2007b] Numerical Bolide - cylindrical line source weak shock model [Edwards et al., 2007] Seismic hypocenter geolocation of bolide airbursts [Edwards et al. 2004] Infrasound bolide airwave measurement and empirical energy estimation [Ens et al., 2012] Monte Carlo Dark Flight model of meteorite fall ellipse production [Brown et al., 2011] 27

28. 28 The End

29. Backup

30. The Cylindrical Blast Radius 30 RORO V popo dE/dL Meteor is effectively an oriented cylindrical line source. Shock propagation is approximately perpendicular from trajectory Atmospheric pressure = Meteor energy loss/length

31. 31 IMS Network requires 1kt globally thresholds vary primarily with stratospheric seasonal wind pattern. NH Winter –Westerly NH –Easterly SH NH Summer –Easterly NH –Westerly SH (Le Pichon 2009)

32. Source Energy Estimation Source energy estimates for bolides from IS has much interstation variability due to: –Unique characteristics: Line source + quasi-Point source –Wide range of source altitudes detected at different stations –Compounded by numerous phases & extreme distance Two approaches: 1.First principles: Weak shock model (Revelle, 1976) – works at short ranges (<250 km) 2.Empirical Energy/Attenuation Relations –Signal Periods, Amplitudes + USG data. –Now calibrated by multi-instrumental, well-characterized events (Ens et al (2012) –Use existing explosion relations (eg. Clauter and Blandford 1998). 32

33. Schematic: Bolide Entry IS Propagation

34. Weak Shock Theory (ReVelle, 1974) x = R/R 0 Weak shock regime Linear regime  Applicable for direct arrivals – short (<200 km) range  Can be used to estimate overpressure at the ground 34

35. http://neo.jpl.nasa.gov/fireballs/

36. Source Information from Observations Single Station –Alternative data source required! (Optical/radar/eyewitness etc.) –If trajectory information available Travel time/Arrival modelling (met.data required): Source Altitude Forward theoretical modelling  energetics Multiple stations –0 th order: Intersection of observed back-azimuths –1 st improvement: Intersection + travel-time fit  phase ID –2 nd improvement: Intersection + tt + wind correction + source model –Empirical attenuation & period relationships  energetics NOTE: Energetics will often represent meteor at source position! The more observations the better for energetics – large variation! 36

37. Cylindrical Blastwave Theory (ReVelle 1974/1977) Propagation from source to observer goes through 2-3 stages –Nonlinear: very high overpressures, strong attenuation –Weakly Non-linear: high overpressures & attenuation, lengthening period –Linear: low overpressure & attenuation, stable period 37 Blast Radius RoRo Weakly Non-linear Propagation Δp ≤ p o : Increasing Period Non-linear Propagation ~10 R o : Δp >> p o 10 R o Linear Propagation Δp << p o : Period Stable d′ < d Linear Transition

38. 38 Meteor Propagating Northward (red line) Source Altitudes 100 – 70 km Radiant Altitude 7° 30° 50° Radiant Altitude: 30° Radiant Altitude: 50° e.g. Grazers, Genesis, Stardust, Hayabusa Most meteors fit in these categories Steeper is lost to atmosphere via refraction

39. Fireball Infrasound Range Discriminators 39

40. Modeling of Bolides as Infrasonic Sources: Hypersonic Aerodynamics Line source blast wave analogy: Hypersonic flow –Ma >> 1 and dV/dt  0; Very narrow Mach cone Nearly cylindrical source symmetry Line source energy deposition: Nonlinear blast wave relaxation radius: Ro –Ro  Square root of energy deposited per length/pressure Ro  Mach no.  diameter (No fragmentation assumed) –Detectable Ro and source energy, Es, ranges from: ~10 m to > 6 km (Tunguska) ~10 -5 kt to > 10 Mt (Tunguska) –Wave period  Ro/(local thermodynamic sound speed): Near-field weak shock wave valid for distances > ~10  Ro Modified line source effects (fragmentation): Larger Ro at the same size and speed 40

41. Part III: Detailed numerical models IDG group – SOVA 3D are only radiation hydrocode to date IDG group – SOVA 3D are only radiation hydrocode to date Lack of good observations to calibrate the high fidelity in the models for large objects Lack of good observations to calibrate the high fidelity in the models for large objects Some USG lightcurves, but often lacking kinematic information or heights (or both) Some USG lightcurves, but often lacking kinematic information or heights (or both) 41

42. Fragmentation - Assumptions Most models assume that stagnation pressure = material strength is trigger for breakup –ie. Breakup occurs when ρv 2 =strength Need to account for dust and macroscopic fragments – dust important in light production If time between successive fragmentation epochs is short compared to separation timescale (big objects, weak objects etc.) details of individual fragmentation can be ignored (large bodies) treat material as liquid-like object with no material strength (SL9 – like) Standard assumption in many models is interaction of individual fragment shocks produces pressure gradient produces lateral fragment speeds of order This gives fragment separatiuon speeds of a few tens of meters per second for meter-tens of meter sized bolides Passey and Melosh (1980) 42

43. Bolide Fragmentation Modeling Break-up schemes: {A (drag) = A (heat transfer) for the single-body model} –Baldwin and Sheaffer (1971): Frontal area, A (drag)  N^1/3, where N = number of fragments produced during the gross-fragmentation process –Petrov and Stulov (1975), Padavet (1973; 1977; 1978): A (heat transfer) >> A (drag) due to turbulent mixing of air/ablated vapor –Liu (1978): A (heat transfer) >> A (drag) due to meteoroid porosity effects –Grigoryan (1977, 1979): “Pancake” break-up process (no ablation case) Once impactor is heavily fragmented the pressure difference between the front and back of the body compresses the impactor and it if forced to “flow” out the sides expanding in area at a rapid rate (pancaking) –Bess (1979): Break-up: Progressive fragmentation process –Zahnle (1992), Hills and Goda (1993), Chyba et. al. (1993), Bronshten (1994; 1995 {after Grigoryan (1976; 1979) including ablation}, Svetsov (1995), Lyne, Tauber and Fought (1996), Nemtchinov et. al. (1995, 1997), Stulov (1997): Airburst, “Pancake” model development, tests and applications Break-up mechanisms: –Thermal effects: Very inefficient (too long of a time delay is necessary) –Mechanical effects: 1-D stagnation pressure exceeds the bolide’s “strength”

44. Pancake model : Predictions The pancake model makes a number of predictions/assumptions - cf. Grigoryan (1979), Melosh (1981), Zahnle(1992), Chyba et al (1993); Hills and Goda (1993;1998)); Korycansky et al (2002) Bland and Artemieva (2004) –Most of the airburst energy is released as a nearly point source (assumption when used to calculate ground damage) (H&G, 1998) –Lateral fragment speeds are a few tens of m/s (prediction) –Mass surviving to the ground as >100g fragments is ~50% of initial mass (prediction) (B&A, 2006) 44 Collins et al., (2005)

45. Bolide masses Dynamic vs photometric mass Long standing issue as photometric mass 10-100x larger than dynamic masses for bolides (Ceplecha et al 1980) Root cause – fragmentation (Ceplecha and ReVelle 2005) Luminous efficiency depends on velocity, mass (and maybe height and composition) 45

46. Panchromatic Luminous Efficiency: Near the End of the Entry (TPFM) 46

47. Radiation Efficiencies Panchromatic efficiencies calibrated using Lost City –6% at 13 km/s (Ceplecha, 1996) Large uncertainty in extrapolating results –camera network bolides much smaller than satellite events Calibrate satellite energies by –cross-fusion with other sensors – either ground-truthing or infrasound (Brown et al 2002) –Using hydrodynamic models which treat complete radiative aspects of entry (Nemtchinov et al., 1997) 47

48. Filled Circle – Meteorite events where energy is known well from many other techniques Open Circles – Energy determined from infrasound observations alone Brown et al (2002) 48

49. Numerical modeling results from Nemtchinov et al. (1997) Graph shows result from entry model for pure Irons and H-Chondrites Equation (14) is solid line in graph Equation (15) is dashed line in graph Much spread, but average η close to empirical result found by Brown et al (2002) at smaller energies 49

50. Integral Bolometric Efficiency Based on Calibrated Satellite – Sensor Events (Brown et al 2002) 50

51. Masses and Sizes Total energy is determined by taking the observed optical yield (Er) and dividing by efficiency (η) –E t =E r / η With total yield (=energy = kinetic energy of impactor) known, mass is found from E t =1/2mv 2 –Assumes velocity is known Size is found using mass and assuming a spherical shape and bulk density –Bulk density can be determined if meteorites are found 51

52. Video Calibrations – Chelyabinsk Lightcurve Uses indirect scattered light and corrected for autogain Calibrated using meteorite-dropping fireball events and radiant intensity from US Gov Sensors Total deposited energy assuming η = 17% is >471 kT 52 Brown et al (2013))

53. Chelyabinsk : Shock wave – Cylindrical or Spherical? Shock wave causing damage was cylindrical not spherical Ray tracing establishes origin height – arrivals are from various heights, not a single point Secondary, weaker shocks after main arrival are spherical - from discrete fragmentation 53

54. Top-down Modelling Now find the blast radius from photometric measurements and entry model  run weak shock model to obtain the predicted signal amplitude and period

55. 1)Meteoroid intersects Earth and “collides” Typical velocity ~20 km/s (11.7 – 73 km/s) Meteoroid size: 0.1 – 10 m 2) Around 80 – 100 km Meteor becomes luminous 3) Meteoroid produces shock wave Line source sound produced ┴ to trajectory 4) 15 – 40 km: Fragmentation Point source sound produced 5) Direct Acoustic heard, Seismic Detections ? Meteorites ? 6) Ducted sound “heard” at microbarometer array 7) Hydroacoustic in ocean (impact or airwave)

56. 56  Theoretical work has shown range scales with energy in a power law (use USG sensor energy here)  Apply wind correction of form  Apply a multivariate linear least squares regression in log-log space RESULT: Airburst Energy Estimation : Amplitude – Yield  Depends on knowing the wind well  Lots of scatter  Amplitude not very reliable at large ranges