See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258107907

Seismic Characterization of the Chelyabinsk Meteor's Terminal Explosion

Article  in  Seismological Research Letters · November 2013

DOI: 10.1785/0220130042

CITATIONS READS
29 810

5 authors, including:

Álvaro González Simone Cesca

Centre de Recerca Matemàtica Helmholtz-Zentrum Potsdam - Deutsches GeoForschungsZentrum GFZ
35 PUBLICATIONS   507 CITATIONS    164 PUBLICATIONS   2,315 CITATIONS   

SEE PROFILE SEE PROFILE

Torsten Dahm
Helmholtz-Zentrum Potsdam - Deutsches GeoForschungsZentrum GFZ
290 PUBLICATIONS   5,039 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

2016 Kaikōura Earthquake View project

Historical Seismology View project

All content following this page was uploaded by Álvaro González on 06 June 2014.

The user has requested enhancement of the downloaded file.

 ○
E

Seismic Characterization of the Chelyabinsk

Meteor’s Terminal Explosion
by Sebastian Heimann, Álvaro González, Rongjiang Wang, Simone
Cesca, and Torsten Dahm

Online Material: Figures of waveform fit, apparent source time explosion (airburst) of the meteor southwest of Chelyabinsk
functions, and video of impact of shock wave at factory. city, and had an equivalent moment magnitude of 3.60. This
implies that this is the second largest meteor explosion ever
seismically recorded, only surpassed by the 1908 Tunguska
INTRODUCTION event (Ben-Menahem, 1975).
Impacts with our planet cause seismic shaking by a variety of
mechanisms. Catastrophic ground motion, even at antipodal DESCRIPTION OF THE GROUND SHAKING
distances, can be generated by the extremely infrequent, hyper-
sonic collisions with large asteroids or comets (Meschede et al., The seismic ground shaking caused by the Chelyabinsk meteor
2011). Fortunately, the atmosphere effectively shields the was exceptionally well registered at planetary scale. It can be
smaller (and far more common) meteoroids, greatly reducing observed in more than 70 digital, broadband seismic recordings
their initial kinetic energy at high altitude, causing them to from stations located at least up to 4000 km away, sampling
slow down, break up, and even vaporize, producing a meteor most azimuths. At further distances, the identification of the
(Ceplecha and Revelle, 2005). In most instances, the ground meteor signal is hampered by the coincidental interference
shaking is triggered by the atmospheric shock wave of a meteor, with wave arrivals from a tectonic earthquake with magnitude
not by the impact of the surviving meteorites (Edwards M w 5.7, originated in Tonga at 03:02:23 UTC. In compari-
et al., 2008). son, when the 1908 Tunguska meteor explosion took place,
A particularly strong shock wave can be generated by seismic monitoring was in its infancy; the resulting surface
explosive fragmentation of the meteoroid in one or several final waves were observed only in four seismic stations up to
airbursts (Ceplecha and Revelle, 2005; Edwards et al., 2008). 5300 km away (Ben-Menahem, 1975).
Such disruptions are triggered when the pressure (ram pres- The ground shaking caused by the Chelyabinsk meteor is
sure) caused by atmospheric drag exceeds the internal strength dominated by Rayleigh waves. Clearly visible both on the
of a meteoroid. They are accompanied by a sudden increase in vertical and radial components of the displacement records,
the meteor luminosity (a flare), because they imply a sharp they travel at up to 3:9 km s−1 . These surface waves can be
increase in the surface area being subject to ablation. excited below a meteor when its atmospheric shock wave hits
On 15 February 2013 at 03:20 UTC, an exceptionally the ground (Edwards et al., 2008). In this case, the shock wave
large meteor in the region of Chelyabinsk, Russia, produced was able to generate broadband Rayleigh waves, with most en-
a powerful shock wave, which caused unprecedented damage ergy released at about 0.04 Hz. This dominant frequency is
to people and property. According to official news reports, glass related to the resonance of the Earth’s crustal structure. Unlike
windows were shattered in over 7300 buildings (some of these tectonic earthquakes or underground explosions, the body
even experienced slight structural damage), and falling debris waves are almost absent, and sharp onsets cannot be identified
hurt more than 1600 people. The meteorite fragments that even at the closest station, ∼220 km away. Almost no seismic
survived the atmospheric entry hit the ground at subsonic ter- energy is detected above 0.1 Hz, or on transversal components,
minal velocity (Schiermeier, 2013), and did not cause any seis- in contrast to tectonic earthquakes with similar seismic moment.
mic shaking detectable at regional distances. However, the
meteor produced the strongest atmospheric infrasound signal SOURCE INVERSION BY FULL WAVEFORM
ever recorded (Stone, 2013) and remarkable ground motion, FITTING
which is the topic of this paper.
Here we describe and model the resulting seismic surface We determined the parameters of the Chelyabinsk seismic
waves, observed at distances of over 4000 km. Our modeling source, modeling it for simplicity as an isotropic atmospheric
indicates that the ground shaking was caused by the terminal airburst. The lack of sharp wave arrivals for this event hampers

doi: 10.1785/0220130042 Seismological Research Letters Volume 84, Number 6 November/December 2013 1021
 Synthetic Rayleigh waves for a range of possible times,
epicentral locations, and explosion altitudes up to 50 km were
KBS
generated to simulate the observed 28 vertical and 6 radial
high-quality displacement seismograms at 28 broadband seis-
LVZ
mic stations up to a distance of ∼4100 km (Fig. 1; and Ⓔ
Fig. S1, available in the electronic supplement to this paper).
We chose to fit the seismograms in the frequency range 0.005–
0.025 Hz (periods from 40 to 200 s), because within it the
seismic signals are not significantly modified by lateral hetero-
ARU
MUGIO geneities of the Earth’s atmosphere and crustal structure. For
the modeling we assume a 5 s boxcar source time function,
200 km which is effectively point-like in time in the frequency range
BRVK analyzed. The source-parameter optimization minimizes the
TIRR
1000 km misfits between the observed and synthetic traces, computed
according to a L1 norm (e.g., Edwards et al., 2008); uncertain-
KIV 4000 km
ABKAR
ties were estimated through bootstrap (e.g., Press et al., 1992, as
implemented by Heimann, 2011).
KKAR
The best-fitting epicenter is found 32 km south-southwest
of Chelyabinsk city, exactly in the terminal, most luminous part
of the meteor trajectory, around which the shock-wave damage
was observed (Table 1 and Fig. 2). The depicted ground path of
▴ Figure 1. Examples of waveform fits at different distances and the meteoroid is based on eyewitness observations and videos,
azimuths around the meteor explosion. Triangles, seismic stations U.S. Government satellite sensor data (Chesley, 2013), and the
used for the analysis. Black traces, observed seismograms. Gray location of the meteorite-strewn field at Lake Chebarkul and
traces, synthetic ones. Star, explosion source. For the complete surroundings. In particular, a major meteorite fragment
set of station codes, seismograms, and synthetic fits, see Ⓔ Fig- splashed in this frozen lake, punching a hole ∼6 m in diameter,
ure S1 in the supplement. exactly at 54.95966° N, 60.32074° E. We determined these co-
ordinates from public pictures of the site and satellite imagery
(by DigitalGlobe, taken 16 February 2013).
location techniques based on them (e.g., its USGS epicenter is The epicentral location is relatively accurate, considering
located more than 30 km north of the meteor ground path), so that it is based only on seismic observations at regional distan-
we performed a full-waveform inversion. Our seismogram sim- ces, without any station in the immediate vicinity of the
ulation code was tailored to consider wave propagation in the meteor. The meteoroid speed of about 18:6 km s−1 (Yeomans,
atmosphere and solid Earth, and the coupling at the interface 2013) places an intrinsic limitation to the location accuracy.
between them (Wang, 2010; applied by Raveloson et al., 2012). The explosion altitude is poorly resolved by seismic data alone
It considers a spherically symmetric, viscoelastic, self-gravitating at large distances, because the Rayleigh-wave excitation mecha-
Earth, with seismic velocities specified by the global model nism is almost independent of explosion height in the observed
AK135Q (Kennett et al., 1995), and atmospheric sound veloc- frequency range. These seismic observations themselves are
ities from a standard model (Committee on the Extension of able to constrain the arrival of the shock wave to the ground,
the Standard Atmosphere, 1976). but not its atmospheric travel time.

Table 1
Source Parameters Calculated for the Meteor Explosion at Unconstrained Origin Time and with Origin Time Equal to the
Moment of Peak Meteor Brightness
Unconstrained Origin Time Fixed Origin Time
Parameter Value 68% Confidence Interval Value 68% Confidence Interval
Longitude 61.22 [60.99, 61.30] Same as unconstrained
[−15 km, +5 km]
Latitude 54.88 [54.65, 55.01] Same as unconstrained
[−25 km, +15 km]
Altitude >0 [0, ∼50] 22.5 km [21.0, 24.0]
Magnitude (M w ) [3.1, 4.0] 3.60 [3.57, 3.62]
Origin time (hh:mm:ss) [03:19:02, 03:21:43] UTC 03:20:33 UTC (fixed)

1022 Seismological Research Letters Volume 84, Number 6 November/December 2013

 M w 3.8.
p The reason is that, for an atmospheric explosion,
M 0 ∝ ρ (assuming constant sound speed) in which M 0 is
the seismic moment and ρ is the air density (which decreases
with altitude).
If the explosion had been underground, we calculate that a
much higher moment magnitude (M w 4.8 for a source at 1 km
depth) would be required to generate the large-amplitude Ray-
leigh waves observed. For comparison, we observed that (in the
same frequency band and at comparable distances) Rayleigh
waves excited by the Chelyabinsk meteor had amplitudes about
three times larger than those produced by the North Korean
nuclear underground explosion (M w 4.5) three days earlier.

REFINEMENTS BASED ON THE AIRBURST

ORIGIN TIME
To calculate the airburst altitude and the precise moment mag-
nitude, we use the timing of the peak meteor brightness as an
independent constraint. The most luminous flare, with appar-
▴ Figure 2. Seismic source, terminal meteor trajectory, and
distribution of shock-wave damage. Star, our optimally fitting
ent brightness larger than the Sun’s (Brown, 2013), took place
epicenter, with contours indicating the confidence regions (for un-
at 03:20:33 UTC (Yeomans, 2013). Using this as the origin
constrained origin time); pentagons, the most damaged popula- time of the airburst, we calculate an airburst altitude of 22:5

tion centers (Russian Ministry of Internal Affairs, 2013): (A) Chelya- 1:5 km (Table 1), in agreement with independent, preliminary
binsk, (B) Kopeysk, (C) Korkino, (D) Etkul’, (E) Yemanzhelinsk, and estimates of 23.3 km of the meteor altitude at peak brightness
(F) Yuzhnouralsk. Lakes are shown in gray (after Lehner and Döll, (Yeomans, 2013).
2004). The arrow depicts the terminal part of the approximate me- The shock wave generated by a meteoroid entry or explo-
teoroid ground path. Past Korkino (black pentagon), the meteor sion is initially hypersonic, and slows to acoustic wave speeds
had its brightest flare (coincident with the seismic explosion after travelling a few tens to hundreds of meters away from the
source), after which it faded abruptly and eventually vanished meteoroid trajectory (Edwards et al, 2008). Our calculations
(thin trace). omit this nonlinear effect, and assume that the shock wave
travels at acoustic speed. Considering that the speed is initially
faster would yield a slightly higher source altitude (e.g.,
On the seismic records, we do not find evidence of a pref- Edwards and Hildebrand, 2004).
erential direction of the shock wave. This anisotropy, if present, In turn, using a more realistic atmospheric profile (e.g.,
should be manifested by azimuthal variations in the arrival Edwards and Hildebrand, 2004; Arrowsmith et al., 2007)
times (e.g., Edwards et al., 2008), amplitude, duration, or dom- for the Chelyabinsk region would tend to yield slightly lower
inant frequencies. Instead, the records are azimuthally homo- source altitudes. This is because the temperature (at least near
geneous in the analyzed frequency range (as shown, for the ground) was lower than assumed by the standard atmos-
example, by the apparent source time functions, Ⓔ Fig. S2; pheric model (at the airport of Chelyabinsk–World Meteoro-
see supplement). Overall, an isotropic point-like airburst repro- logical Organization observatory ID 28645, it was −21:8° C at
duces very well these seismic observations, without requiring a 3:00 UTC), which implies slower sound speeds.
more complex explanation, such as a moving source. This in- Nevertheless, the most important source of altitude uncer-
dicates that, as in some other cases (Edwards et al., 2008), only tainty is the hypersonic vertical component of the meteoroid
the shock wave from the meteor outburst was strong enough to speed (−2:4 km s−1 ; Yeomans, 2013). Because of it, further
excite the measured seismic shaking, at least at the distances constraining the source altitude would require subsecond
from which there is seismic record. In contrast, the atmos- accuracy of the timing of peak brightness. Overall, the uncer-
pheric infrasound signals radiating from a stretch of the tainties imply an error range of a few km in altitude, as dem-
trajectory of a large meteor can be strong enough to be detected onstrated by the modeling results (Table 1).
at long distances and to indicate a moving, hypersonic source Constraining the origin time allows a precise estimate
(e.g., Edwards, 2010). of the moment magnitude of the ground shaking: M w 3.60
The calculated moment magnitude decreases with the (Table 1). This value (or the M w 4.8 for an assumed shallow
modeled explosion altitude. For example, a source with M w 3.3 explosion source) is the second largest ever recorded for a
at 40 km altitude produces the same seismic amplitudes in meteor, only surpassed by the M w ∼ 5:0 of Tunguska (Ben-
the modeled seismograms as another source at 10 km with Menahem, 1975).

Seismological Research Letters Volume 84, Number 6 November/December 2013 1023

VALIDATION OF THE SOURCE LOCATION WITH A POSTSCRIPT
VIDEO RECORDING
After this manuscript was accepted for publication, other pa-
The Chelyabinsk meteor was recorded in dozens of casual or pers related to the Chelyabinsk meteor have been published.
surveillance videos and associated audio. They lack exact Particularly, Tauzin et al. (2013) made a seismic analysis com-
timing, a common problem faced when trying to characterize plementary to ours, and Le Pichon et al. (2013) presented the
meteor trajectories (Gural, 2012), but a few of them can be first results of the infrasound records.
useful for estimating atmospheric travel times of the
shock wave. ACKNOWLEDGMENTS
We use a particularly valuable surveillance video (Ⓔ see
supplement S3), showing the direct shock-wave arrival to We acknowledge all the seismographic institutions and the
Korkino (the town closest to the brightest meteor flare; Fig. 2) Incorporated Research Institutions for Seismology (IRIS) con-
to provide an independent validation of the explosion location. sortium for providing the seismograms, the Internet commu-
It was recorded inside a corrugated carton factory located at nity for contributing with videos and eyewitness reports, and
61.347° E, 54.902° N. We measured the building orientation an anonymous reviewer for valuable comments. The surveil-
in a detailed map, and inspected the video frame by frame. The lance video was kindly provided by YuzhUralKarton, LLC,
windows facing south-southeast (N169°E), toward the meteor thanks to Dan Chernovalov. The figures were elaborated using
trajectory, were smashed by the shock wave 87.5 s after the Generic Mapping Tools (Wessel and Smith, 1998). Á. G. is
apparent peak meteor brightness. supported by a grant of Caja Madrid Foundation (Spain)
The travel time of the shock wave from the explosion and the Spanish Government project FIS2010-19773. S. C.
source to this building (calculated from our simulation code is supported by the German Federal Ministry of Education
with the origin time fixed as before) is 88 s, with a 68% and Research (BMBF) project MINE.
uncertainty range of [82–118] s, due to the uncertainties in
the source location and altitude. Thus, this video recording REFERENCES
supports the preferred estimate of the explosion location.
Arrowsmith, S. J., D. P. Drob, M. A. H. Hedlin, and W. Edwards (2007).
A joint seismic and acoustic study of the Washington State bolide:
Observations and modeling, J. Geophys. Res. 112, D09304, doi:
CONCLUSIONS 10.1029/2006JD008001.
Ben-Menahem, A. (1975). Source parameters of the Siberian explosion of
The Chelyabinsk meteor experienced a major terminal explo- June 30, 1908, from analysis and synthesis of seismic signals at four
sion, manifested as the flare with peak brightness. The resulting stations, Phys. Earth Planet. In. 11, 1–35, doi: 10.1016/0031-9201
atmospheric shock wave caused the damage observed around (75)90072-2.
Brown, P. (2013). A preliminary report on the Chelyabinsk fireball/air-
the explosion source (Fig. 2). Seismic shaking was generated burst, J. Int. Meteor Organ. 41, 22.
by direct coupling of the atmospheric shock wave with the Ceplecha, Z., and D. O. Revelle (2005). Fragmentation model of mete-
ground, and recorded hundreds to thousands of kilometers oroid motion, mass loss, and radiation in the atmosphere, Meteoritics
away (Fig. 1). This terminal explosion took place in the strato- Planet. Sci. 40, 35–54, doi: 10.1111/j.1945-5100.2005.tb00363.x.
Chesley, S. (2013). Chelyabinsk impactor ground track, in D. Yeomans
sphere, at an altitude of ∼23 km. Had it occurred at a lower and P. Chodas, Additional Details on the Large Fireball Event over
altitude, its shock wave could have been even more damaging. Russia on Feb. 15, 2013, National Aeronautics and Space Admin-
It had an equivalent M w 3.60, implying that the Chelyabinsk istration, http://neo.jpl.nasa.gov/news/fireball_130301.html (last ac-
meteor explosion was the second largest ever seismically re- cessed 15 June 2013).
Committee on the Extension of the Standard Atmosphere (1976). U.S.
corded, only surpassed by the Tunguska event. Standard Atmosphere 1976, U.S. Government Printing Office,
Our results highlight that distant seismic recordings can Washington, D.C.
agree remarkably well with local meteor observations, and her- Edwards, W. N. (2010). Meteor generated infrasound: Theory and
ald the potential for characterizing future large meteors with observation, in A. Le Pichon, E. Blanc, and A. Hauchecorne
(Editors), Infrasound Monitoring for Atmospheric Studies, Springer,
scarce direct data. This could be particularly useful in remote Dordrecht, Netherlands, 361–414.
areas or above the oceans, and potentially also for fast location Edwards, W. N., and A. R. Hildebrand (2004). SUPRACENTER:
and early warning. That most of the seismic energy of the Locating fireball terminal bursts in the atmosphere using seismic
meteor was released from a terminal explosion highlights arrivals, Meteoritics Planet. Sci. 39, 1449–1460, doi: 10.1111/
j.1945-5100.2004.tb00121.x.
the danger posed by such phenomena, and we hope that this Edwards, W. N., D. W. Eaton, and P. G. Brown (2008). Seismic
study would help increase public awareness. observations of meteors: Coupling theory and observations, Rev.
We expect that our seismic analysis could be used as a basis Geophys. 46, RG4007, doi: 10.1029/2007RG000253.
of more detailed reconstructions of the Chelyabinsk meteor, Gural, P. S. (2012). A new method of meteor trajectory determination
applied to multiple unsynchronized video cameras, Meteoritics Planet.
which potentially will benefit from other imagery, eyewitness Sci. 47, 1405–1418, doi: 10.1111/j.1945-5100.2012.01402.x.
chronicles, and modeling of signals recorded in infrasound sta- Heimann, S. (2011). A Robust Method to Estimate Kinematic Earth-
tions worldwide. quake Source Parameters, Ph.D. Thesis, University of Hamburg,

1024 Seismological Research Letters Volume 84, Number 6 November/December 2013

 available at http://ediss.sub.uni‑hamburg.de/volltexte/2011/5357/ yabinsk meteor, Geophys. Res. Lett. 40, 3522–3526, doi: 10.1002/
(last accessed 6 September 2013). grl.50683.
Kennett, B. L. N., E. R. Engdahl, and R. Buland (1995). Constraints on Wang, R. (2010). QSSP2010: FORTRAN code for calculating complete
seismic velocities in the Earth from traveltimes, Geophys. J. Int. 122, synthetic seismograms of a spherical Earth using the normal mode
108–124, doi: 10.1111/j.1365-246X.1995.tb03540.x. theory, available at http://www.gfz‑potsdam.de/forschung/ueberblick/
Lehner, B., and P. Döll (2004). Development and validation of a global departments/department‑2/erdbeben‑und‑vulkanphysik/services/
database of lakes, reservoirs and wetlands, J. Hydrol. 296, 1–22, doi: downloads/ (last accessed 6 September 2013).
10.1016/j.jhydrol.2004.03.028. Wessel, P., and W. H. F. Smith (1998). New, improved version of the
Le Pichon, A., L. Ceranna, C. Pilger, P. Mialle, D. Brown, P. Herry, and Generic Mapping Tools released, Eos Trans. AGU 79, 579, doi:
N. Brachet (2013). The 2013 Russian fireball largest ever detected 10.1029/98EO00426.
by CTBTO infrasound sensors, Geophys. Res. Lett. 40, 3732–3737, Yeomans, D. (2013). Fireball and Bolide Reports, National Aeronautics
doi: 10.1002/grl.50619. and Space Administration, http://neo.jpl.nasa.gov/fireballs/ (last
Meschede, M. A., C. L. Myhrvold, and J. Tromp (2011). Antipodal accessed 15 June 2013).
focusing of seismic waves due to large meteorite impacts on Earth,
Geophys. J. Int. 187, 529–537, doi: 10.1111/j.1365-246X.2011.05170.x. Sebastian Heimann
Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery
(1992). Quick-and-Dirty Monte Carlo: The Bootstrap method, Álvaro González
in Numerical Recipes in C: The Art of Scientific Computing, Second Rongjiang Wang
Ed., Cambridge University Press, Cambridge, United Kingdom, Torsten Dahm
994 pp. GFZ German Research Centre for Geosciences
Raveloson, A., R. Wang, R. Kind, L. Ceranna, and X. Yuan (2012). Section on Earthquake Risk and Early Warning
Seismic and acoustic-gravity signals from the source of the 2004 In-
dian Ocean tsunami, Nat. Hazards Earth Syst. Sci. 12, 287–294, Helmholtzstr. 7
doi: 10.5194/nhess-12-287-2012. 14467 Potsdam, Germany
Russian Ministry of Internal Affairs (2013). Public Safety in the sebastian.heimann@gfz‑potsdam.de
Chelyabinsk Region, available at http://mvd.ru/news/item/845541/
(last accessed 6 September 2013) (in Russian). Simone Cesca
Schiermeier, Q. (2013). The death of the Chebarkul meteor, Nature 495,
16, doi: 10.1038/495016a. University of Potsdam
Stone, D. (2013). Siberian meteor spurs dash for data, calls for safeguards, Institute of Earth and Environmental Science
Science 339, 1135, doi: 10.1126/science.339.6124.1135. Karl-Liebknecht-Str. 24-25
Tauzin, B., E. Debayle, C. Quantin, and N. Coltice (2013). Seismoacous- 14476 Potsdam, Germany
tic coupling induced by the breakup of the 15 February 2013 Chel-

Seismological Research Letters Volume 84, Number 6 November/December 2013 1025

View publication stats