GEOPHYSICAL RESEARCH LETTERS, VOL. 38, L17109, doi:10.

1029/2011GL048805, 2011

Bow and stern waves triggered by the Moons shadow boat

J. Y. Liu,1,2,3 Y. Y. Sun,1 Y. Kakinami,4 C. H. Chen,5 C. H. Lin,6 and H. F. Tsai7
Received 6 July 2011; revised 12 August 2011; accepted 15 August 2011; published 14 September 2011.

[1] It has been predicted that the Moons shadow, the

cooling region, sweeping over the Earths atmosphere with a
supersonic speed could trigger bow waves since 1970. The
longest total solar eclipse within next hundred years occurring on 22 July 2009 sweeps over the Eastern Asia region
during the noontime period. An analysis of the HilbertHuang
transform (HHT) is applied to study ionospheric TEC (total
electron content) derived from groundbased GPS receivers
in Taiwan and Japan. We not only find the feature of the
predicted bow wave but also the stern wave on the equator
side of the eclipse path, as well as the stern wake right behind
the Moons shadow boat. The bow and stern waves are
formed by acoustic gravity waves of periods about 3 and/or
5 minutes traveling equatorward with a phase speed of about
100 m/s in the ionosphere. Citation: Liu, J. Y., Y. Y. Sun,
Y. Kakinami, C. H. Chen, C. H. Lin, and H. F. Tsai (2011), Bow
and stern waves triggered by the Moons shadow boat, Geophys.
Res. Lett., 38, L17109, doi:10.1029/2011GL048805.

1. Introduction
[2] Chimonas and Hines [1970] speculate a localized heat
sink traveling through a gravitationally stratified atmosphere,
and suggest this heating region in the lower atmosphere,
troposphere and/or stratosphere [e.g. Brasseur and Solomon,
2005; Wang and Liu, 2010], acting as a continuous source of
traveling waves during a solar eclipse period. They predict by
analogy with similar systems in hydrodynamics [Comstock,
1967] that, if the source velocity exceeds the wave propagation speed, a bow wave formed by acoustic gravity waves
appears around the source region travelling away from the
totality path. Numerous attempts, including model simulations and observations, have been put forward to find the
eclipse generated waves in the ionosphere. Although predictions were made by model simulations in early 1970s
[Beer and May, 1972; Beer et al., 1976], no obvious signatures and/or conclusive evidences of the eclipse generated
bow waves have been observed by means of various ground

1
Institute of Space Science, National Central University, ChungLi,
Taiwan.
2
Center for Space and Remote Sensing Research, ChungLi,
Taiwan.
3
National Space Program Origination, HsinChu, Taiwan.
4
Institute of Seismology and Volcanology, Hokkaido University,
Sapporo, Japan.
5
Department of Geophysics, Graduate School of Science, Kyoto
University, Kyoto, Japan.
6
Department of Earth Science, National Cheng Kung University,
Tainan, Taiwan.
7
Center Weather Bureau, Taipei, Taiwan.

Copyright 2011 by the American Geophysical Union.

00948276/11/2011GL048805

based satellite receivers, ionosondes, and Doppler sounders

[e.g. Davis and da Rosa, 1970; Schodel et al., 1973, Hunter
et al., 1974, Hajkowicz, 1977; Morton and Essex, 1978;
Hanuise et al., 1982; Cheng et al., 1992; Farges et al.,
2003; Jakowski et al., 2008]. The bow wave signatures
of solar eclipses have not yet been observed due to very
limited observation stations being used and the triggered
acoustic gravity waves being too small to be extracted and
detected.
[3] The longest total solar eclipse within next hundred
years occurring on 22 July 2009 sweeps over the Eastern Asia
region and passes through two worlddensest groundbased
GPS networks in Taiwan and Japan (1400s stations in total)
during the noontime period (Figure 1a). The average zonal
speed of the Moons shadow boat computed from the time
and displacement of the totality reported by Espenak and
Anderson [2008] is 951 m/s within 120125E. Since each
receiver could track 910 GPS satellites, about 13000 TEC
points with a time resolution of 30sec are used to simultaneously observe waves induced by the solar eclipse. We
further apply the HilbertHuang transform (HHT) [Huang
et al., 1998] to extract characteristics of the waves associated with the total solar eclipse. Finally the temporal evolution and the spatial distribution of the triggered waves are
reported and discussed.

2. Methodology
[4] HilbertHuang transform (HHT) [Huang et al., 1998]
consists of Empirical Mode Decomposition (EMD) and
Hilbert spectral analysis. In the time domain, EMD is utilized
to decompose the data set into several orthogonal components
which termed Intrinsic Mode Function (IMF), and the normalized Hilbert transform analysis is able to computes the
instantaneous amplitude and frequency from each IMF. The
contour plot of amplitude or power in timefrequency domain
H(w, t) is named as the Hilbert spectrum. The time integration
of the amplitude or power within the specific frequency range
is given by the marginal spectrum, by contrast, the superposition of the amplitude or power within the specific time
interval which can be defined as,
Z
At

H !; t d!

We use H(w, t) to display the wave periods, and A(t) to

demonstrate the movements of the waves on the map.

3. Observation and Interpretation

[5] The eclipse shadow swept through TaiwanJapan
region about 0200 UT (1000 or 1100 LT, local time, LT = UT
+ 8 or 9 hours). Taking measurements of two GPS stations in

L17109

1 of 4

L17109

LIU ET AL.: SOLAR ECLIPSE BOW AND STERN WAVES

Figure 1. The groundbased GPS TEC observation in Taiwan and Japan during the 22 July 2009 total solar eclipse.
(a) The gray curves exhibit the obscurations of the eclipse.
The color dots indicate the location of the groundbased
GPS receivers in various areas. We examine the TEC observations of 196 stations in Taiwan (orange dots) and 1234 stations in Japan (black dots). Data of 196 stations in Taiwan
(orange dots), 34 stations in Okinawa (blue dots), 146 stations
in Kyushu (cyan dots), and 71 stations in Hokkaido (red dots)
are employed to calculate the Hilbert spectrum (see Figure 2).
Possible bow waves are highlighted by the red circles.
The vertical TEC derived from groundbased GPS stations,
(b) gs46 (Taiwan), (c) 0498 (Okinawa), (d) 0729 (Okinawa),
(e) 0688 (Kyushu), in the totality region, and (f) 0015
(Hokkaido), respectively, during 00300340 UT. The gray
red solid lines in Figures 1b, 1c, and 1d show the TEC values
which are derived from the paths of gs46, 0498, and 0729 to
GPS satellite 1, satellite 26, and satellite 18, respectively. The
red segments of the grayred solid lines denote the ionospheric pierce points when the disturbances occurred. The
corresponding satellite tracks project on the ionospheric
height 350 km altitude are shown in Figure 1a. The vertical
dark lines in Figures 1b1g indicate the time of the maximum
of the obscuration.

Taiwan (gs46) (Figure 1b) and Okinawa (0498) (Figure 1c) as

examples, prominent TEC fluctuations/waves with periods
shorter than 10 minutes are recorded during 02100330 UT
and 01000200 UT, respectively. Note that some tiny fluctuations with the similar periods appear at one of the north
Okinawa stations (0729) (Figure 1d) in the totality region
around 0200 UT, while no obvious wave signatures are found
at two stations on the poleward side of the eclipse path
(Figures 1e and 1f).
[6] To extract characteristics of waves associated with the
total solar eclipse, we apply the HHT to calculate the Hilbert
spectra of the TEC in the Taiwan, Okinawa, Kyushu, and
Hokkaido areas one day before to after the 22 July 2009
eclipse (Figure 2). We find, as shown in Figure 1, that the
amplitude with the period of about 3 and/or 5 minutes
prominently enhances in Taiwan, and 23 and/or 5 minutes in
Okinawa during the eclipse day.

L17109

[7] We further examine the temporal and spatial variations

in the waves of period of about 25 minutes. Figure 3 illustrates the distributions of the pronounced wave amplitude
near the totality path during the solar eclipse period of 0055
0240 UT. It can be seen that waves with an amplitude of about
0.10 TEC unit (TECu; 1016 electrons/m2) at 3540N (north
side or poleside of the totality path), 125140E before
the eclipse 00550120 UT (Figures 3a and 3b), and propagate as medium scale traveling ionospheric disturbances in
the northeastsouthwest direction during the eclipse period
01450240 UT (Figures 3c3i). On the other hand, some
waves also preexist at 2432N (southside or equator
side of the totality path), 117133E before the eclipse
00550120 UT (Figures 3a and 3b). However, these pre
existing waves become three equatorward propagating wave
fronts with a pronounced amplitude of about 0.25 TECu
(Figures 3c3h).
[8] To compute the velocity of the wave fronts, we extract
the wave amplitude within 1735N and 120125E during
00550320 UT. Figure 4 displays the amplitude in latitude
versus time on one day before, one day after, and the eclipse
day. It can be seen that no pronounced amplitude can be
observed either on one day before or after the eclipse day
(Figures 4a and 4c). Figure 4b shows that one wave front
closer to the totality path is nearly stationary during 0130
0255 UT, while the other two depart at 0145 UT and 30
minutes later, at 0215 UT, sequentially. The first front speeds
up from 119 m/s to 164 m/s in 20.2N at 0240 UT, and the
second one accelerates from 81 m/s to 122 m/s in 21.5N
at 0250 UT equatorward. The stationary wave front and the
first/second propagating fronts meet the characteristics of
stern wake and bow/stern wave in hydrodynamics, respectively [Comstock, 1967]. The feature of the sudden changes

Figure 2. The Hilbert spectra of the GPS TEC observed

in Taiwan, Okinawa, Kyushu, and Hokkaido during 0000
0600 UT of 2123 July 2010. The frequency higher than
1.67 mHz is corresponding to the period shorter than 10 minutes. The pronounced wave signatures with periods shorter
than 10 minutes appear in (e) Taiwan and (f) Okinawa around
0300 UT and 0200 UT on the eclipse day. By contrast, no
pronounced features can be found in (g) Kyushu and (h) Hokkaido. Furthermore, no short wave signatures can be observed
(ad) one day before and (il) one day after the eclipse day.

2 of 4

L17109

LIU ET AL.: SOLAR ECLIPSE BOW AND STERN WAVES

L17109

Figure 3. The variation of the eclipse generated waves. (ai) Snapshots of the amplitude of the TEC waves (period <10 min)
during the eclipse period. Gray curves and dark circles denote the eclipse path and obscuration contours of 100%, 90% and
80%. The data extracted from the rectangular region (1735N, 120125E) shown in Figure 3i are utilized to construct the
local timelatitude map in Figure 4.
in the velocity of the two propagating fronts also well agrees
with that of experimental simulations of a ship bow wave
[Waniewski et al., 2002].

4. Discussion and Conclusion

[9] It is found that the clear solar eclipse triggered waves
mainly appear on the equator side of the eclipse path
(Figure 3) and propagate equatorward (Figure 4b). A possible
reason might be that the ion movement is influenced both by
the neutral atmosphere movements and ambient geomagnetic
field directions. Their relative importance is determined by
the ratio of the ion neutral collision frequency to the ion gyro
frequency. This ratio is small enough at the F layer height, and
the ion moves mainly in the direction of the magnetic field
irrespectively of the neutral particle movements. Therefore,
the amplitude of the northward waves is smaller than that of
the southward ones [Heki and Ping, 2005; Otsuka et al.,
2006]. Meanwhile, the other reason might be that the poleward blowing winds in the daytime may play an important
role, as the equatorward direction of bow waves follows the
theory of directional filtering effect of AGWs.
[10] A ship moving faster than its spreading ripples forms
the 2dimensional Vshaped waves [Chimonas, 1970]. It is
known that the speed of acoustic gravity waves with periods
of 5 minutes is about 300 m/s in the troposphere and the
stratosphere [Artru et al., 2004; Heki and Ping, 2005]. Let the
pronounced wave observed in this paper be a ripple, the moon
shadow boat speed 954 m/s is much faster than the ripple,
which agrees with the criteria of bow wave formation. Note
that the 300 m/s ripple speed of the wave is greater than the
two propagating fronts of 81164 m/s, which also meets the
criteria of the existence of the two fronts.
[11] Based on the theoretical model of Chimonas [1970],
in the troposphere and stratosphere, the sound speed

Figure 4. The time evolution of the waves with periods

shorter than 10 minutes within 120125E during 0055
0320 UT on one day before to after the eclipse day. (a) One
day before, (b) the eclipse day, and (c) one day after the eclipse
day. A preexists before and B during the moon shadow boat
arriving. C, D, and E appear after the moon shadow boat leaving. Here, C is the stern wake, while E and D are the bow wave
and stern wave, respectively. The resolution in latitude and
time are 0.1 and 1 minute, respectively. The white dots are
the local maxima on the front traces A and B, and the circle
circle and crosscross lines denote the segmented slopes of
these two traces, where the circlecross symbols denote the
bow and/or stern wave speeding up. The gray lines denote
the obscurations within 120125E.

3 of 4

L17109

LIU ET AL.: SOLAR ECLIPSE BOW AND STERN WAVES

300 m/s, buoyancy frequency 102 Hz, the scale height

8 km, and the mean eastward speed 951 m/s of the moon
shadow between 120 and 125E longitude of the current event result in the acoustic cutoff frequency being 2
102 Hz and the bow wave speed being 163 m/s which
well agrees with that observed in this paper. Furthermore, the angle  between the wave fronts and the eastward
moon shadow velocity is 4.9 (=tan1(81 ms1/951 ms1))
to 9.8 (=tan1(164 ms1/951 ms1)).
[12] Figure 3 shows that the two propagating fronts are
generated within the 8090% obscuration area. The average zonal velocity of the totality within 120125E is about
951 m/s. Since the time interval between the bow and stern
waves is 30 minutes (Figure 4b), the ship length should
be approximate 1712 (=951 m/s 30 min) km, where is
corresponding to two times of the distance between 100% and
80% obscuration area. The agreement indicates the moon
shadow boat size is about 80% obscuration.
[13] In conclusion, the ionospheric bow wave, stern wave,
and stern wake are for the first time observed by means
of dense groundbased GPS receivers. The fronts of the bow
and stern waves formed by the pronounced acoustic gravity
waves of the period of about 5 minutes, the amplitude
0.25 TECu, and wavelength 36120 km travel away from the
totality path 81164 m/s during the 22 July 2009 solar eclipse.
[14] Acknowledgments. The Editor thanks two anonymous reviewers
for their assistance in evaluating this paper.

References
Artru, J., T. Farges, and P. Lognonne (2004), Acoustic waves generated
from seismic surface waves: propagation properties determined from
Doppler sounding observations and normalmode modeling, Geophys.
J. Int., 158, 10671077, doi:10.1111/j.1365-246X.2004.02377.x.
Beer, T., and A. N. May (1972), Atmospheric gravity waves to be expected
from the solar eclipse of June 30, Nature, 240, 3032, doi:10.1038/
240030a0.
Beer, T., G. L. Goodwin, and G. L. Hobson (1976), Atmospheric gravity
wave production for the solar eclipse of October 23, 1976, Nature,
264, 420421, doi:10.1038/264420a0.
Brasseur, G. P., and S. Solomon (2005), Aeronomy of the Middle Atmosphere, Chemistry and Physics of the Stratosphere and Mesosphere,
3rd ed., Springer, New York.
Cheng, K., Y.N. Huang, and S.W. Chen (1992), Ionospheric effects of
the solar eclipse of September 23, 1987, around the equatorial anomaly
crest region, J. Geophys. Res., 97(A1), 103111, doi:10.1029/
91JA02409.
Chimonas, G. (1970), Internal gravitywave motions induced in the Earths
atmosphere by a solar eclipse, J. Geophys. Res., 75(28), 55455551,
doi:10.1029/JA075i028p05545.
Chimonas, G., and C. O. Hines (1970), Atmospheric gravity waves induced
by a solar eclipse, J. Geophys. Res., 75(4), 875, doi:10.1029/
JA075i004p00875.

L17109

Comstock, J. P. (1967), Principles of Naval Architecture, Soc. of Nav.

Archit. and Mar. Eng., New York.
Davis, M. J., and A. V. da Rosa (1970), Possible detection of atmospheric
gravity waves generated by the solar eclipse, Nature, 226, 1123,
doi:10.1038/2261123a0.
Espenak, F., and J. Anderson (2008), Total solar eclipse of 2009 July 22,
NASA Tech. Pap. NASA/TP2008214169, NASA Goddard Space Flight
Cent., Greenbelt, Md.
Farges, T., A. Le Pichon, E. Blanc, S. Perez, and B. Alcoverro (2003),
Response of the lower atmosphere and the ionosphere to the eclipse
of August 11, 1999, J. Atmos. Sol. Terr. Phys., 65, 717726,
doi:10.1016/S1364-6826(03)00078-6.
Hajkowicz, L. A. (1977), Periodic fadings in VHF radiosatellite transmissions during the solar eclipse on 23 October 1976, Nature, 266,
147148, doi:10.1038/266147a0.
Hanuise, C., P. Broche, and G. Ogubazghi (1982), HF Doppler observations of gravity waves during the 16 February 1980 solar eclipse,
J. Atmos. Terr. Phys., 44, 963966, doi:10.1016/0021-9169(82)90060-5.
Heki, K., and J. Ping (2005), Directivity and apparent velocity of the
coseismic ionospheric disturbances observed with a dense GPS array,
Earth Planet. Sci. Lett., 236, 845855, doi:10.1016/j.epsl.2005.06.010.
Huang, N. E., Z. Shen, S. R. Long, M. C. Wu, H. H. Shih, Q. Zheng, N.C.
Yen, C. C. Tung, and H. H. Liu (1998), The empirical mode decomposition and the Hilbert spectrum for nonlinear and nonstationary time
series analysis, Proc. R. Soc. London, Ser. A, 454, 903995,
doi:10.1098/rspa.1998.0193.
Hunter, A. N., B. K. Holman, D. G. Fieldgate, and R. Kelleher (1974),
Faraday rotation studies in Africa during the solar eclipse of June 30,
1973, Nature, 250, 205206, doi:10.1038/250205a0.
Jakowski, N., S. M. Stankov, V. Wilken, C. Borries, D. Altadill, J. Chum,
D. Buresova, J. Boska, P. Sauli, F. Hruska, and L. R. Cander (2008),
Ionospheric behavior over Europe during the solar eclipse of 3 October
2005, J. Atmos. Sol. Terr. Phys., 70, 836853, doi:10.1016/j.jastp.2007.
02.016.
Morton, F. W., and E. A. Essex (1978), Total electron content observations
during the 23 October 1976 solar eclipse over south eastern Australia,
J. Atmos. Terr. Phys., 40, 111114, doi:10.1016/0021-9169(78)90115-0.
Otsuka, Y., et al. (2006), GPS detection of total electron content variations
over Indonesia and Thailand following the 26 December 2004 earthquake, Earth Planets Space, 58, 159165.
Schodel, J. P., J. Klostermeyer, and J. Rottger (1973), Atmospheric gravity
wave observations after the solar eclipse of June 30, Nature, 245, 8788,
doi:10.1038/245087a0.
Wang, K.Y., and C.H. Liu (2010), Profiles of temperature responses to
the 22 July 2009 total solar eclipse from FORMOSAT3/COSMIC constellation, Geophys. Res. Lett., 37, L01804, doi:10.1029/2009GL040968.
Waniewski, T. A., C. E. Brennen, and F. Raichlen (2002), Bow wave
dynamics, J. Ship Res., 46, 115.
C. H. Chen, Department of Geophysics, Graduate School of Science,
Kyoto University, Kitashirakawaoiwakecho, Sakyoku, Kyoto 6068502,
Japan.
Y. Kakinami, Institute of Seismology and Volcanology, Hokkaido
University, Kita 10 Nishi 8, Kitaku, Sapporo 0600810, Japan.
C. H. Lin, Department of Earth Science, National Cheng Kung
University, No. 1, University Rd., Tainan 701, Taiwan.
J. Y. Liu and Y. Y. Sun, Institute of Space Science, National Central
University, No. 300, Jungda Rd., ChungLi 32001, Taiwan. (jyliu@jupiter.
ss.ncu.edu.tw)
H. F. Tsai, Center Weather Bureau, 64 Gongyuan Rd., Taipei 110048,
Taiwan.

4 of 4