Journal of Geophysical Research: Atmospheres

RESEARCH ARTICLE Lunar influence on equatorial atmospheric angular momentum

10.1002/2014JD022240
Christian Bizouard1 , Leonid Zotov2 , and Nikolay Sidorenkov3
Key Points: 1 Paris Observatory SYRTE, CNRS/UMR 8630, UPMC, 2 Sternberg Astronomical Institute, Moscow State University, Moscow,
• Equatorial atmospheric angular
Russia, 3 Hydrometeorological Centre/Global Atmospheric Circulation Laboratory of Russia, Moscow, Russia
momentum studied in the
nonrotating frame
• Two main peaks: harmonic at 13.6
days (O1 ) and broadband peak Abstract This study investigates the relationship between the equatorial atmospheric angular
around 7 days momentum oscillation in the nonrotating frame and the quasi-diurnal lunar tidal potential. Between 2
• The 13.6 day cycle is explained by an
equilibrium lunar tide model
and 30 days, the corresponding equatorial component, called Celestial Atmospheric Angular Momentum
(CEAM), is mostly constituted of prograde circular motions, especially of a harmonic at 13.66 days, a sidelobe
at 13.63 days, and of a weekly broadband variation. A simple equilibrium tide model explains the 13.66 day
Correspondence to:
C. Bizouard,
pressure term as a result of the O1 lunar tide. The powerful episodic fluctuations between 5 and 8 days
christian.bizouard@obspm.fr possibly reflect an atmospheric normal mode excited by the tidal waves Q1 (6.86 days) and 𝜎1 (7.095 days).
The lunar tidal influence on the spectral band from 2 to 30 days is confirmed by two specific features, not
Citation:
occurring for seasonal band dominated by the solar thermal effect. First, Northern and Southern
Bizouard, C., L. Zotov, and Hemispheres contribute equally and synchronously to the CEAM wind term. Second, the pressure and wind
N. Sidorenkov (2014), Lunar influence terms are proportional, which follows from angular momentum budget considerations where the
on equatorial atmospheric angular
momentum, J. Geophys. Res. Atmos.,
topographic and friction torques on the solid Earth are much smaller than the one resulting from the
119, doi:10.1002/2014JD022240. equatorial bulge. Such a configuration is expected for the case of tidally induced circulation, where
the surface pressure variation is tesseral and cannot contribute to the topographic torque, and tidal winds
Received 28 JUN 2014 blow only at high altitudes. The likely effects of the lunar-driven atmospheric circulation on Earth’s nutation
Accepted 2 OCT 2014 are estimated and discussed in light of the present-day capabilities of space geodetic techniques.
Accepted article online 8 OCT 2014

1. Introduction
According to investigations going back to the nineteenth century (encapsulated in Chapman and Lindzen
[1970]) and more recent ones like Sidorenkov [2009, 2010a, 2010b], the lunar tide influences the atmospheric
circulation. Its effect has been detected in wind and pressure fields as well as in the globally integrated
atmospheric angular momentum. Given the angular momentum exchange between the solid Earth and the
atmosphere, the atmospheric lunar tide also couples to Earth rotation variations, namely, to length of day
[Li, 2005; Li et al., 2011], possibly to fortnightly polar motion [Bizouard and Seoane, 2010], and to the 13.6 day
nutation [Brzeziński et al., 2002; Sidorenkov, 2009].
Besides prominent quasi-biennial, seasonal, and rapid oscillations, the Equatorial Atmospheric Angular
Momentum (EAAM) exhibits retrograde quasi-diurnal variations of comparable magnitude. Its main peaks
at 24 h (S1 ), 24.07 h (P1 ), and 23.93 h (K1 ) are commonly interpreted as the effect of diurnal solar thermal
heating (24 h) subject to a yearly amplitude modulation [Zharov, 1997; Bizouard et al., 1998]. Squeezed in a
frequency band around 24 h in the Terrestrial Reference Frame (TRF), the corresponding periodicities stretch
from 2 days to several years with respect to the nonrotating reference frame. This interdependence follows
from the fundamental relationship that—with Ω = 2𝜋 1.002738 rad/d being Earth’s mean stellar angular
frequency—any diurnal component of frequency 𝜎 = −Ω + 𝜎 ′ with 𝜎 ′ ≪ Ω is mapped to a long peri-
odic celestial component of frequency 𝜎 ′ . Removal of the diurnal carrier in the TRF-based EAAM therefore
produces Celestial Atmospheric Angular Momentum (CEAM) [Brzeziński, 1994] that represents retrograde
diurnal oscillations as slow periodic variations. Such “demodulated” time series form the basis for a charac-
terization of the effects of the major retrograde diurnal components (S1 , P1 , and K1 ) on Earth’s precession
nutation [Bizouard et al., 1998; Yseboodt et al., 2002; Brzeziński et al., 2002].
Some of the cited studies mention the existence of a minor but sharp 25.82 h peak [Brzeziński et al., 2002]
and evasively attribute it to the O1 lunar tesseral tide, considering the absence of thermal processes at this
frequency. It has been also noticed that in the nonrotating equatorial frame, this equatorial oscillation has
a period of 1∕(24∕25.82 − 1.00273) = 13.6 days and thus excites the well-known 13.6 day nutation. The
present paper attempts to deepen the insight into this signal component by a detailed analysis of CEAM

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 1

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

variations from 2 to 30 days in the nonrotating frame. Specifically, the coherence of the 13.6 day harmonic
with the lunar gravitational tide will be shown.

2. Basic Relationships and Celestial Atmospheric Angular Momentum

Let Hi be the Cartesian components of the Atmospheric Angular Momentum (AAM) vector in the TRF,
C = 8.0370 1037 kg m2 the mean axial moment of inertia, and A = 8.0101 1037 kg m2 the mean equatorial
moment of inertia. At subsecular time scales, Earth rotation variations are commonly investigated by the lin-
ear Liouville equations, where the excitation of any surface fluid layer maps into a nondimensional terrestrial
quantity 𝜒i of the angular momentum of this layer:
H1 H2 H
𝜒1 = ,𝜒 = ,𝜒 = 3 . (1)
(C − A)Ω 2 (C − A)Ω 3 CΩ
The 𝜒i are called Angular Momentum Functions (AMF) and are composed of two terms: (i) the first, produced
by the rotation of the air mass distribution, can be computed from surface pressure data and is usually called
pressure term and (ii) the second, caused by the winds, is proportional to the relative angular momentum of
the atmosphere and might be denoted as wind term [Barnes et al., 1983]. Note that in order to account for
Earth’s nonrigidity in the Liouville equations, the AMF have to be multiplied by appropriate coefficients close
to 1, yielding the so-called Effective Angular Momentum Function (EAMF) [Barnes et al., 1983].
The components 𝜒 ′′ in the true equatorial frame GXYZ—characterized by a fundamental plane perpen-
dicular to the Celestial Intermediate Pole and the Conventional Intermediate Origin [IERS Conventions,
2010]—can be derived from their terrestrial counterparts 𝜒 by applying the matrix transformation

𝜒 ′′ = R3 W𝜒 , (2)

where R3 represents the diurnal rotation and W the polar motion. The major part of this transformation is
caused by the uniform rotation in R3 with the mean angular velocity Ω. We argue that the residual rota-
tion can be considered as a perturbation. At subsecular time scales, polar motion produces oscillations of
the geographic pole below 1 arc sec, and the nonuniform part of the rotation angle in R3 is smaller than
20 arc sec. As the 𝜒i are small quantities of the order of 10−7 rad, the maximum effect of the variable rota-
tion is 20 × 10−7 = 2 10−6 arc sec over 100 years, which can be neglected in light of the AAM accuracy
(> 10−4 arc sec). Thus, for subsecular phenomena, the above transformation can be restricted to an axial
rotation reading

𝜒 ′′ = R3 [𝜃(t)]𝜒 , (3)

where 𝜃(t) is the uniformly varying rotation angle

𝜃(t) = 𝜃(TAI0 ) + Ω(TAI − TAI0 ) , (4)

with TAI0 being a conventionally chosen instant of TAI (Temps Atomique International). Taking TAI0 as 1
January 2000 at UT 1 noon and following IERS Conventions [2010, chapter 5, equation (5.14)], the angle
𝜃(TAI0 ) of this reference date is 𝜃(TAI0 ) = 2𝜋 0.779 057 273 2640 ≈ −79.53◦ .
The rotation in (3) leaves the axial component unchanged (𝜒3′′ = 𝜒3 ), while the complex equatorial quantity
𝜒 = 𝜒1 + i𝜒2 is converted to

𝜒 ′′ = 𝜒ei𝜃(t) . (5)

To estimate the nutation influence produced by a fluid layer excitation, Brzeziński [1994] derived the Liouville
equation in the nonrotating equatorial system GXYZ and introduced the quantity CEAM function defined by

𝜒 ′ = −𝜒ei GST ≈ −𝜒ei 𝜃(t) = −𝜒 ′′ , (6)

where GST signifies the Greenwich Sidereal Time. In order to comply with the recommendations issued by
the International Earth Rotation and Reference System Service (IERS) for parameterizing Earth’s rotation,
GST is replaced by the Earth rotation angle 𝜃 [IERS Conventions, 2010], and hence, the CEAM function is
defined by

𝜒 ′ = −𝜒ei𝜃(t) . (7)

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 2

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

10
CEAM W - overall spectrum
-365 365
1 -182 182
7 121

mas
13.6 91

0.1

0.01
-400 -300 -200 -100 0 100 200 300 400
days
1 1
around 13.6 days around 7 days
0.8 0.8

0.6 13.66 0.6

mas

mas
13.63
0.4 0.4

0.2 0.2

0 0
13.5 13.55 13.6 13.65 13.7 13.75 13.8 5 6 7 8 9 10
days days
1
CEAM P - overall spectrum
-365 365
-182 182
7 121
mas

0.1 13.6 91

0.01
-400 -300 -200 -100 0 100 200 300 400
days
0.2 0.2
around 13.6 days around 7 days

0.15 0.15
13.66
mas

mas

0.1 13.63 0.1

0.05 0.05

0 0
13.5 13.55 13.6 13.65 13.7 13.75 13.8 5 6 7 8 9 10
days days

Figure 1. Amplitude spectrum of the Celestial Angular Momentum, (top) Wind (W) and (bottom) Pressure (P) compo-
nents over the period 1949–2012. Data source: NCEP/NCAR.

3. Analysis of the CEAM

Using 6-hourly EAMF estimates from the reanalysis model of NCEP/NCAR (National Center for Environmental
Prediction/National Center of Atmospheric Research) over the period 1949–2013, as provided by the Global
Geophysical Fluids Center of the IERS [SBA, 2014], we computed the associated CEAM according to (7) for
both pressure and wind terms. Prior to demodulation, we removed the long-term components (periods
larger than 2 days) of the AAM. After application of (7), a low-band-pass filter was used to eliminate residual
diurnal/subdiurnal signal content and to obtain the celestial excitation limited to periods larger than 2 days,
corresponding to the precession-nutation frequency band.
The complex Fourier spectrum of the obtained CEAM is displayed in Figure 1. It is dominated by a prograde
annual oscillation at 365 days (S1 in the TRF) and its subharmonics at 182 (P1 ), 121 (𝜋1 ), and 91 days (1.0083

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 3

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

20
2.1 χ’yp
15 w
χ’y
10
5

mas
0
-5
-10
-15
-20
50000 50020 50040 50060 50080 50100 50120
MJD
15
p
2.1 χ’x
10
χ’xw
5
0
mas

-5
-10
-15
-20
50000 50020 50040 50060 50080 50100 50120
MJD

Figure 2. Equatorial components X and Y of the Celestial Atmospheric Angular Momentum for wind term 𝜒w and NIB
pressure term 𝜒p multiplied by 2.1 (linear regression coefficient from 1949 to 2014) computed from NCEP AAM series
after eliminating long-term periods above 1 month. Extract over 130 days, date in modified Julian day (MJD).

cpd in the TRF). The retrograde part of the spectrum only exhibits power at −365 days (𝜓1 ) and −182 days
(Φ1 ). These seasonal oscillations have been extensively analyzed by various authors, especially with regard
to their observable effects on Earth’s nutation [Zharov, 1997; Bizouard et al., 1998; Yseboodt et al., 2002]. We
rather focus on the rapid band of the CEAM between 2 days and 30 days, of which the spectral zoom clearly
unveils a harmonic at +13.6 days (25.8 h in the TRF) and a broadband peak around +7 days (28 h in the TRF).
This band is isolated in time domain by using an appropriate high-band-pass filter, having an admittance
of 99% at 30 days. An extract over 130 days (from modified Julian day (MJD) 50,000 to MJD 50,130) of the
obtained time series is depicted in Figure 2, showing both wind terms in X and Y components as well as
the full Noninverted Barometer (NIB) pressure terms that are void of any approximative correction for the
oceanic response to air pressure variations. It is remarkable that for X and Y coordinates, wind and pressure
terms are evidently proportional. This finding is not limited to the considered time interval, as correlations
throughout the period 1949–2014 amount to 0.57 for both X components and Y components. Linear regres-
sion over the entire analysis time gives 𝜒w′ ∼ 2.1𝜒p′ . For the case of a hydrostatic response of the oceans
(Inverted Barometer or IB approximation) imposed on the CEAM, the ratio 𝜒w′ ∕𝜒p′ increases up to 5, whereas
the correlation drops to 0.43.
The detected proportionality appears to be a feature of the short-term CEAM from 2 days to 1 month, but it
does not extend to the complementary spectral bands, ranging from 1 month to several years, where cor-
relations between pressure and wind terms drop to 0.1. Another striking feature distinguishing the 2–30
day band from other parts of the spectrum is the fact that contributions of Northern and Southern Hemi-
spheres to the wind terms have synchronous variations with similar phases and amplitudes, as evidenced
by Figure 3. Such a behavior is not observed for the S1 thermal wave for which Southern and Northern
Hemispheres contribute asymmetrically to 𝜒w′ with a phase lag of 6 months.
All these results are confirmed by an analogous analysis of European Centre for Medium-Range Weather
Forecasts (ECMWF) AMF time series as downloaded from the GeoForschungsZentrum Postdam ftp server
ftp://ftp.gfz-postdam.de/pub/home/ig/ops/.
Meaning of the proportionality between pressure and wind term: Let Γ ⃗ be the torque that the atmosphere
⃗
exerts on the solid Earth. It is composed of the bulge torque Γb acting on the equatorial bulge because of
pressure and gravitational forces, and of a local torque Γ ⃗ l caused by pressure on the local topography as
well as the friction drag on Earth’s surface [de Viron et al., 1999; Schindelegger et al., 2013]. In the nonrotating
′
frame we have the following complex quantities: Hw∕p for the equatorial wind/pressure term, Γ′l for the local

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 4

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

10
SH
NH
5

χ’y (mas)
0

-5

-10
50000 50020 50040 50060 50080 50100 50120
MJD
10
SH
NH
5
χ’x (mas)

-5

-10
50000 50020 50040 50060 50080 50100 50120
MJD

Figure 3. Contributions of the Southern Hemisphere (SH) and Northern Hemisphere (NH) to the wind term in
(bottom) 𝜒X′ and (top) 𝜒Y′ for the 2–30 day band. Time series of 130 days, commencing at modified Julian day 50,000
(10 October 1995).

torque, Γ′b for the bulge torque, and Γ′ext for the external gravitational torque on the atmosphere. Following
our derivations in Appendix A (equation (A10)), it can be established in frequency domain that

𝜎 ′ 𝜎 ′ Ĥ w
′ −Γ̂ ′l + Γ̂ ′ext
1− − = , (8)
Ω Ω Ĥ ′ Γ̂ ′b
p

where the sign “ ̂ ” corresponds to the Fourier transform. If the residual torque −Γ̂ ′l + Γ̂ ′ext is much smaller
than the bulge torque Γ′b then

Ω − 𝜎′ ̂ ′
Ĥ w′ ≈ Hp . (9)
𝜎′
For positive angular frequencies 𝜎 ′ of the filtered CEAM, with periods from 2 days to 1 month, we have
1∕30 Ω ≤ 𝜎 ′ ≤ 1∕2 Ω (the retrograde part of the spectrum is much smaller); so according to (9) pressure
and wind terms become almost proportional. This is in contrast with the seasonal band (S1 in the TRF)
where the smallness of the local torque with respect to the bulge torque is not satisfied [Marcus et al., 2004].
Considering for the lunar tidal band a typical magnitude of |𝜒p′ | ∼ 0.2 milliseconds of arc (mas) (see spec-
trum of Figure 1), the bulge torque magnitude, given by |Γ′b | = Ω|Hp′ | = Ω2 (C − A)|𝜒p′ | [Marcus et al., 2004],
amounts to ∼1.5 1018 Nm. According to Bizouard and Lambert [2001], the external torque Γ′ext is mostly com-
posed of a 13.6 day component with an amplitude of ∼1017 Nm, which is at least 10 times smaller than the
equatorial bulge torque Ω|Hp′ |. Hence, as far as the local torque does not exceed the order of magnitude of
the external torque, the above condition holds.

4. Tidally Coherent Fit of the 13.6 Day Term

As emphasized by the spectral zoom of Figure 1, the main peak is at 13.66 days. The most natural hypothesis
for its origin is the diurnal tidal wave O1 determined by the Delaunay argument 2(F + ) in the nonrotating
frame, with F = 𝜔 + l being the sum of the perigee argument 𝜔 and the mean anomaly l, and being the
longitude of the ascending node of the Moon on the ecliptic plane. Note that s = F + is the mean tropic
longitude of the Moon.
As expected from tidal theory, the main peak is accompanied by a sidelobe at 13.63 days having the argu-
ment 2F + . These two components differ by the frequency of the displacement of the ascending
node of the Moon, that is 1∕18.6 cpy (cycles/year). The fact that we observe both of these peaks in CEAM
substantiates the tesseral lunar influence on CEAM, in particular on the wind component.

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 5

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

The celestial oscillations of arguments Φ1 = 2F + 2 = 2s (13.66 days) and Φ2 = 2F + (13.63 days) are
fitted by a least squares method to the model

∑
2
𝜒′ = (mcj + imsj )ei(𝜙j +𝜋∕2) . (10)
j=1

For the period 1949–2013 we obtain

𝜒p′IB [mas] = (0.05 − i0.02) ei(𝜙1 +𝜋∕2) + (0.02 − i0.00) ei(𝜙2 +𝜋∕2)
𝜒p′NIB [mas] = (0.17 − i0.06) ei(𝜙1 +𝜋∕2) + (0.06 − i0.01) ei(𝜙2 +𝜋∕2) (11)
𝜒w′ [mas] = (0.73 − i0.04) e i(𝜙1 +𝜋∕2)
+ (0.23 − i0.01) e i(𝜙2 +𝜋∕2)

The ms terms are small relatively to mc , except for the IB term where about two thirds of the regional contri-
bution of the pressure field—the oceanic one—has been replaced by a time-variable mean value computed
from all pelagic points. Disregarding the IB solution, the harmonic coefficients are therefore almost in phase
with the tidal wave of argument 𝜙i + 𝜋∕2, confirming the proportionality of wind and pressure terms at this
period and supporting their common tidal gravitational cause. The ratio 𝜒w′ ∕𝜒p′NIB = Hw′ ∕Hp′ ∼ 4 for both
tidal frequencies does not match the numerical value of the condition (9), namely, Hw′ ∕Hp′ = 13.6 − 1 ∼ 13.
To explain this difference, it does seem unreasonable to assume that the wind term is underestimated, e.g.,
as a consequence of the 10 mbar vertical boundary of the NCEP/NCAR model that neglects only about 1%
of the total atmospheric mass. On the other hand, the ratio 𝜒w′ ∕𝜒p′IB ≈ 14 much better fits the expected ratio
of 12.6, as if the effective pressure term around the O1 frequency was the one restricted to continents and a
static IB ocean. This is quite peculiar, since an IB response of the oceans is generally observed above 10 days
but not at diurnal periods in the TRF [Salstein and Rosen, 1989].
Moreover, at the O1 frequency the fit is consistent with the external gravitational torque in the true
equatorial frame [Bizouard and Lambert, 2001]. Indeed, this quantity reads

Γ′ext (O1 ) ≈ −0.8 1017 [N m]ei𝜙1 (12)

and is in quadrature of phase with respect to the corresponding harmonics of the wind and IB pressure
terms of expression (11).
Temporal variability: By reference to the high-frequency CEAM model

∑
2
( j )
𝜒′ = mc + i mjs ei(𝜙j (t)+𝜋∕2)
j=1

with 𝜙1 = 2(F + ) (13.66 days, O1 tidal wave in the TRF) and 𝜙2 = 𝜙1 + 2lm (lm is the mean anomaly of
the Moon) representing the weekly signal (6.86 days, 2Q1 tidal wave in the TRF), the parameters mc and
ms were adjusted repeatedly in a sliding window of 6 years. The results are reported in Figure 4 for pressure
and wind terms. The IB pressure term at 13.6 days, and the NIB pressure term at 6.8 days are multiplied by
the corresponding theoretical ratio from (9) for 𝜒 ′w ∕𝜒p′ . For these harmonics, the coherency in the variations
of 𝜒p′ and 𝜒w′ are overwhelming. We notice an 18–20 year modulation for the 13.66 day term, reflecting the
beating produced by the 13.63 day term.

5. The Quasi-Weekly Band

As shown by the Gabor decomposition in Figure 5, the broadband peak between 5 and 8 days is much more
powerful than the thin peaks around 13.6 days, showing episodes with an amplitude of 10 mas. Some stud-
ies like Brzeziński et al. [2002] attribute this weekly signal to the retrograde Rossby-Haurwitz atmospheric
normal mode Ψ11 , having in the TRF the geometry of a spherical harmonic cos(𝜙)ei𝜆 (𝜙 is the latitude, 𝜆 is
the longitude) propagating to the west [Volland, 1996]. On the other hand, in the nonrotating frame, this
resonant mode propagates from the west to the east as the Moon, and with an averaged period of ∼7 days
it could be amplified at planetary scale by the minor lunar tides 2 Q1 (6.86 days) and 𝜎1 (7.09 days) (at least
100 times smaller than O1 ). As shown by the associated sliding window least squares fit plotted in Figure 4,
the linear correlation between variable parts of the wind and NIB pressure term is even more striking than
at 13.6 days (above 0.7). Moreover, at quasi-weekly periods (𝜎 ′ ∕Ω ≈ 1∕7) the ratio 𝜒w′ ∕𝜒p′NIB ≈ 5.8 fits the

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 6

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

3 1.5
2.5 13.6d P - IB 12.6*mc 1 13.6d P - IB 12.6*ms
2 W mc 0.5 W ms

mas
mas
1.5 0
1 -0.5
0.5 -1
0 -1.5
1950 1960 1970 1980 1990 2000 2010 1950 1960 1970 1980 1990 2000 2010

2 1.5
1.5 6.8d P - NIB 5.8*mc 1 6.8d P - NIB 5.8*ms
1 W mc 0.5 W ms
mas

mas
0.5 0
0 -0.5
-0.5 -1
-1 -1.5
1950 1960 1970 1980 1990 2000 2010 1950 1960 1970 1980 1990 2000 2010

Figure 4. Variability of 13.66 day and 6.8 day components of the winds and pressure terms. Six year sliding window fit of
(left column) mc and (right column) ms parameters in the model (mc + ims ) eiΦ , where Φ is the tidal argument. The NIB
or IB pressure terms are multiplied by the theoretical ratio (9) corresponding to the considered period.

condition (9) reading Hw′ ∕Hp′ ≈ 7 − 1 = 6 and is thus valid for the full pressure term in contrast to what
is observed at 13.6 days. Our validation against complementary series from the ECMWF model reproduced
quite well the results associated with NCEP data over the period 1985–2013, except for the wind component
at 13.6 days, which contained a relatively large out-of-phase term ms (0.5 instead of 0.07 mas).

6. Hydrostatic Model of the Tidal O1 Oscillation of the Pressure Term

Why is the phase of the O1 oscillation of the CEAM precisely Φ1 = 2(F + ) + 𝜋∕2? Let us investigate
this question by assuming a simple hydrostatic redistribution of the air masses under the action of the
lunar tidal force. Because of the lunar tidal potential U1 , surface pressure undergoes a variation from P0 to
P0 + P1 , satisfying

∇P ⃗ 1 = −(𝜌0 + 𝜌1 )∇(U
⃗ 0 + ∇P ⃗ 0 + U1 ) (13)

where U0 is the nonperturbed geopotential, 𝜌0 ≈ 1.3 kg/m3 is the nonperturbed uniform air density and 𝜌1
denotes the variation of the air density caused by the tide. Assuming that there is no tidally induced circu-
lation at Earth’s surface (references are given at the end of this section), the air density 𝜌0 is constant at any
place according to the continuity
equation, and 𝜌1 ≈ 0. After neglecting
second-order terms in (13), and account-
ing for hydrostatic equilibrium ∇P ⃗ 0 =
⃗
−𝜌0 ∇U0 in the nonperturbed state, we
obtain by omission of the gradient of 𝜌0

∇P ⃗ 0 U1 ) .
⃗ 1 = −∇(𝜌 (14)

In turn we have

P1 = −𝜌0 U1 . (15)

According to classic tidal theory

[Sidorenkov, 2009; Simon et al., 2013], at
the Earth surface the tesseral daily tide
caused by the Moon is expressed by

U1 ≈ 2GM [1 + m cos(s − p)]

Figure 5. Gabor decomposition of the CEAM—wind term from 1950 (16)
to 2010 in the lunar tidal band 2–20 days. sin 2𝜙 sin 𝜀 sin s cos H ,

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 7

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

where GM = 2.64 m2 /s2 is the Doodson constant for the Moon; s is the mean tropic longitude of the Moon,
s = F + , already introduced; p is the mean tropic longitude of the Moon perigee; m = 0.165 is the
amplitude modulation associated with the elliptical orbit of the Moon; 𝜙 is the latitude; 𝜀 = 23.7◦ is the
obliquity of the ecliptic; and H is the hour angle of the Moon.
It is straightforward to show that

U1 ≈ GM [1 + m cos(s − p)] sin 2𝜙 sin 𝜀(sin(s − H) + sin(s + H)) . (17)

The main terms are those directly proportional to sin(s + H) corresponding to the K1 tide (constant term in
the nonrotating frame), and to sin(s − H), corresponding to the O1 tide, and have the same amplitude. In
fact, the K1 peak in AAM mostly results from annual modulation of the thermal diurnal cycle S1 [Bizouard
et al., 1998], and the tidal contribution represents a small fraction of this spectral component. We focus on
the O1 tesseral potential

U1 (O1 ) ≈ GM sin 2𝜙 sin 𝜀 sin(s − H) . (18)

As s = GST+𝜆−H, where 𝜆 is the longitude of the place under consideration, we have −H = −GST−𝜆+s and
2
U1 (O1 ) = G P1 (cos 𝜙) sin 𝜀 sin(2s − GST − 𝜆)
3 M 2
2
= GM P21 (cos 𝜙) sin 𝜀 (sin(2s − GST) cos 𝜆 − cos(2s − GST) , sin 𝜆) , (19)
3
where P21 (cos 𝜙) = 3 cos 𝜙 sin 𝜙 is the second-order tesseral Legendre polynomial. So according to (15)

P1 = (p21 cos 𝜆 + p̃ 21 sin 𝜆)P21 (cos 𝜙) (20)

with p21 = −𝜌0 32 GM sin 𝜀 sin(2s − GST), p̃ 21 = 𝜌0 32 GM sin 𝜀 cos(2s − GST). However, according to equation (B3)
(Appendix B), the AAM pressure term can be directly derived from the tesseral component of the
surface pressure

4𝜋 r04
𝜒p ≈ − (p + ip̃ 21 ) , (21)
5 (C − A)g 21
where r0 ≈ 6371 km is the mean equatorial radius of the Earth and g ≈ 9.81m/s2 represents the mean
gravity acceleration in the vicinity of Earth’s surface. So the pressure term variation engendered by the
Moon is
4𝜋 r04 2
𝜒p ≈ − 𝜌 G sin 𝜀 (− sin(2s − GST) + i cos(2s − GST))
5 (C − A)g 0 3 M
8𝜋 r04
≈− 𝜌 G i sin 𝜀 ei(2s−GST) , (22)
15 (C − A)g 0 M
and the corresponding CEAM reads

8𝜋 r04
𝜒p′ = −𝜒p eiGST ≈ 𝜌 G sin 𝜀 ei(2s+𝜋∕2) , (23)
15 (C − A)g 0 M
not only accounting for the order of magnitude of the actually observed O1 pressure term in (11) (0.2 mas)
but also reproducing its phase.
Can we also explain the amplitude and the phase of the O1 wind term? If the hydrostatic assumption is true
for surface pressure, tidal winds do not blow at Earth’s surface, but at high altitudes. Tidal lunar variation of
the horizontal wind have been indeed reported in the mesosphere (above 50 km) [Sandford et al., 2006], in
the stratosphere [Nastrom and Belmont, 1976; Sakazaki et al., 2010], in the upper troposphere [Krahenbuhl
et al., 2011], but never in the low troposphere. So tidal winds do not produce any notable friction torque.
On the other hand, a tesseral tidal pressure field exerts a torque on the bulge but cannot contribute to the
topographic torque, which results from spherical harmonics of degree higher than 3 [Dehant et al., 1996]. We
thus conclude that the tidal atmospheric circulation does not contribute significantly to the local torque, in
accordance with condition (9) and that it accounts for the proportionality of 𝜒w′ and 𝜒p′ . As noted earlier, we
obtain the expected ratio for 7 day oscillations, but for 13.6 days the theoretical constraint is satisfied only if
we resort to the IB pressure term and not to the total matter term of the atmospheric angular momentum.

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 8

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

7. Effect on Earth’s Nutation

The action of the lunisolar tesseral forces on the equatorial bulge yields prograde terms of the nutation
below 30 days, with amplitudes of up to 100 mas (13.6 day component). These terms are modeled precisely
by taking into account Earth’s nonrigidity and the ocean tides. However, the celestial pole offsets, i.e., the
corrections to the IERS conventional nutation model, as obtained from the processing of VLBI (Very Long
Baseline Interferometry) observations, have still a standard deviation of about 100 μas in the 5–15 day band,
thereby revealing deficiencies of the model. The latter could result from the atmospheric lunar tide. Consid-
ering the effective CEAM 𝜒e′p = 1.01𝜒 ′p and 𝜒e′w = 1.6𝜒 ′w , the nutation effect can be estimated from the
frequency transfer function of Sasao and Wahr [1981] or Brzeziński [1994, equation (4.9)]:

P(𝜎 ′ ) = Tp (𝜎 ′ )𝜒e′p (𝜎 ′ ) + Tw (𝜎 ′ )𝜒e′w (𝜎 ′ ), (24)

in which
[ ]
1 ap
Tp (𝜎 ) = 𝜎c
′
+ , (25)
𝜎c′ − 𝜎 ′ 𝜎f′ − 𝜎 ′

and a similar expression can be phrased for Tw using the coefficient aw . Here Tp and Tw are transfer functions
of the pressure term and of the wind term, respectively, 𝜎 ′ is the nutation angular frequency, 𝜎c′ = 𝜎c + Ω
and 𝜎f′ ≈ −Ω∕430 (1 − i∕40, 000) are the space-referred angular frequencies of the Chandler wobble and of
the Free Core Nutation (FCN) modes, 𝜎c ≈ Ω∕433 (1 + i∕200) is the Earth-referred Chandler frequency, and
ap = 9.509 × 10−2 , aw = 5.489 × 10−4 are dimensionless coefficients expressing the response of the FCN
mode to pressure and wind excitations [Brzeziński, 1994].
Using the above expression, it can be estimated that for prograde circular excitations of periods spanning
from 2 days to 1 month, Tw (𝜎 ′ ) is at least 2 times larger than Tp (𝜎 ′ ), with Tw presenting a rather constant
value: Tw (Ω∕7) ∼ 2.6 10−3 and Tw (Ω∕13.6) ∼ 2.3 10−3 . Given that the pressure term is about 5 times smaller
than the wind term, its effect can be neglected. With amplitudes in 𝜒w′ reaching 10 mas, broadband oscilla-
tion of the wind term around 7 days can produce a nutation of about 10 × 2.6 10−3 ∼ 0.03 mas; cf. estimates
of Schindelegger et al. [2011]. Until now, these periodicities are poorly covered by space geodetic techniques,
since nutation is routinely monitored by aid of VLBI observations with a mean temporal resolution of about
7 days. However, various attempts to densify nutation determination by jointly processing GNSS (Global
Navigation Satellite System) data seem to confirm this broadband nutation with the estimated order of
magnitude [Bizouard et al., 2010]. The tidal atmospheric effect on the 13.6 day nutation has an average mag-
nitude of 1 × 2.3 10−3 = 0.003 mas. The 13.6 day term fit in celestial pole offset time series depends on the
series produced by the various analysis centers of the International VLBI Service, with amplitude discrepan-
cies ranging from 5 to 15 μas, and phase offsets of up to 180◦ (on the website of the IERS Earth Orientation
Center, http://hpiers.obspm.fr, this fit can be performed with eight available VLBI series). Considering the
smallness of the corresponding atmospheric effect, it is not possible to make any sound conclusion about
its observability in nutation.

8. Conclusion
Below periods of 1 month, the Celestial Equatorial Atmospheric Angular Momentum is mostly composed
of prograde oscillations, dominated by a harmonic at 13.6 days and a broadband peak around 7 days, ini-
tially detected in the terrestrial frame [Brzeziński et al., 2002]. The present study aimed at showing that this
band has a lunar tidal origin, suggested by two specific features that contrast with what is observed for sea-
sonal CEAM variations of thermal origin: (i) Northern and Southern Hemispheres contribute equally and
synchronously to the globally integrated signal and (ii) pressure and wind terms are almost proportional to
each other.
The 13.6 day pressure term clearly results from the main lunar tide, explained by a hydrostatic reaction to
the tesseral O1 tidal forcing. The origin of the broadband peak around 7 days in the nonrotating frame is less
evident. The weekly variation possibly results from the lunar tide, of which the effect is amplified by the Ψ11
atmospheric resonance.
As the surface pressure response to the lunar tesseral potential seems to be hydrostatic, the tidal flow is
only significant at high altitudes, and the topographic and friction torques are negligible compared to the

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 9

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

bulge torque for periods between 2 days and 1 month in the nonrotating celestial frame. This constellation
accounts for the proportionality between pressure and wind terms. While around 7 days (28 h in the TRF)
the observed ratio fits the theoretical ratio for the full (NIB) pressure term, at the tidal period of 13.6 days
(25.8 h in the TRF) the theoretical ratio oddly matches the case of the IB pressure term, as if the oceans and
the atmosphere were hydrostatically coupled.
Our quintessential conclusion is as follows: from 24.8 h to 48 h in the TRF the retrograde equatorial atmo-
spheric angular momentum variations are mostly triggered by the Moon. In order to observe a possible
effect on nutation, improvements of the VLBI/GNSS data processing as well as optimized observation sched-
ules are required. Characterization of the regional atmospheric contributions is likely to yield further insights
into the associated mass transport mechanism.

Appendix A: Relation Between Atmospheric Torques and Angular Momentum

in Nonrotating Frame
⃗ be the torque that the atmosphere exerts on the solid Earth. It is composed of the bulge torque Γ
Let Γ ⃗b
⃗
acting on the equatorial bulge because of the pressure and gravitational forces, and of a local torque Γl
caused by pressure on the topography and friction drag at Earth’s surface. Let Hp be the equatorial pressure
term of AAM and recall that [Marcus et al., 2004]
Γb = −iΩHp , (A1)
⃗ b in the TRF. The angular momentum balance of the
where Γb is the complex equatorial component of Γ
atmosphere states that
⃗
dH ⃗b − Γ
= −Γ ⃗l + Γ ⃗ ext (A2)
dt
⃗ is the total atmospheric angular momentum and Γ
where H ⃗ ext is the lunisolar torque on the atmosphere,
of which the estimation can be found in Bizouard and Lambert [2001]. Let Hw be the equatorial complex
component of the wind term. Written in the TRF, the former equation allows us to establish that
Ḣ p + Ḣ w + iΩ(Hp + Hw ) = −Γb − Γl + Γext , (A3)
⃗l Γ
where Γl and Γext are the components of Γ ⃗ ext in the TRF, respectively. Considering (A1), we can eliminate
Γb and obtain

Ḣ p + Ḣ w + iΩHw = −Γl + Γext . (A4)

This equation is transferred in the true equatorial frame by multiplying both sides by ei𝜃 :
(Ḣ p + Ḣ w )ei𝜃 + iΩHw′ = −Γ′l + Γ′ext , (A5)
where Γ′l and Γ′ext represent the complex equatorial components of the corresponding torques and Hw′ is the
celestial wind term. Equivalently, we have
( )
d Hp ei𝜃 + Hw ei𝜃
̇ p + Hw )ei𝜃 + iΩH′ = −Γ′ + Γ′ .
− i𝜃(H (A6)
w l ext
dt
As 𝜃̇ = Ω, the former equation becomes
Ḣ ′ p + Ḣ ′ w − iΩ(Hp′ + Hw′ ) + iΩHw′ = −Γ′l + Γ′ext (A7)
or

Ḣ ′ p + Ḣ ′ w − iΩHp′ = −Γ′l + Γ′ext . (A8)

Applying the Fourier transform (denoted by the symbol ̂) with the celestial angular frequency 𝜎 ′ ≪ Ω yields
i(𝜎 ′ − Ω)Ĥ p′ + i𝜎 ′ Ĥ w′ = −Γ̂ ′l + Γ̂ ′ext . (A9)
Then, by dividing by Γ̂ ′b
= −iΩĤ p′
deduced from (A1), we obtain an equation similar to equation (11) of
Marcus et al. [2004] expressed in the TRF but including now the external gravitational torque
𝜎 ′ 𝜎 ′ Ĥ w
′ −Γ̂ ′l + Γ̂ ′ext
1− − = . (A10)
Ω Ω Ĥ ′ Γ̂ ′b
p

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 10

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

Appendix B: Pressure Term Reduced to the Tesseral Pressure Field

The pressure term can be expressed by [Bizouard, 2014]

r04 𝜋 2𝜋
𝜒p = − Ps Y21 sin 𝜃 d𝜃 d𝜆 (B1)
3N21 (C − A)g ∫𝜃=0 ∫𝜆=0

where 𝜆 is the longitude, 𝜃 the colatitude, r0 the mean Earth radius, g the mean surface gravity acceleration,
Ps the surface pressure, and Y21 the complex spherical harmonic function P21 (sin 𝜃)ei𝜆 normalized by N21 =
√
(5∕24𝜋). The surface pressure is developed in complex spherical harmonics

∑∑
l
Ps = p′lm Ylm (𝜃, 𝜆) (B2a)
l=0 m=−l

with
plm − ip̃ lm pl0 plm + ip̃ lm
p′lm = for m > 0 , p′l0 = , p′lm = for m < 0 . (B2b)
2Nl|m| Nl0 2Nl|m|

Inserting this development of Ps in (B1) and taking into account orthonormality relations between Y21 and
the other Ylm , we obtain

r04 4𝜋 r04
𝜒p = − p′2,−1 = − (p + ip̃ 21 ) , (B3)
3N21 (C − A)g 5 (C − A)g 21

which is reduced to the tesseral component (2, 1) of the pressure field.

Acknowledgments
We are grateful to the Paris Observa-
References
tory for allocating a 2 month position Barnes, R. T. H., R. Hide, A. A. White, and C. A. Wilson (1983), Atmospheric angular momentum fluctuations, length-of-day changes and
to Leonid Zotov. This greatly helped polar motion, Proc. R. Soc. A, 387, 31–73.
us to carry out this work. We are Bizouard, C. (2014), Le Mouvement du Pôle de L’Heure au Siècle, 284 pp., Presses Académiques Francophones. [Available at
indebted to Michael Schindelegger of www.presses-academiques.com.]
TU Vienna for improving the quality of Bizouard, C., and S. Lambert (2001), Lunisolar torque on the atmosphere and Earth’s rotation, Planet. Space Sci., 50(3), 323–333.
this paper. The atmospheric angular Bizouard, C., and L. Seoane (2010), Atmospheric and oceanic forcing of the rapid polar motion, J. Geodesy, 84, 19–30,
momentum data (NCEP/NCAR doi:10.1007/s00190-009-0341-2.
reanalyses) associated with the results Bizouard, C., A. Brzeziński, and S. Petrov (1998), Diurnal atmospheric forcing and temporal variations of the nutation amplitudes,
shown in this paper were prepared J. Geodesy, 72, 561–577.
by the Special Bureau for Atmosphere Bizouard, C., J. Y. Richard, S. Bolotin, and O. Bolotina (2010), High frequency nutation from GRGS multitechnique combination, paper
within the IERS Global Geophysical presented at 6th Orlov Conference, Kiev, July 2009.
Fluids Center (downloaded from Brzeziński, A. (1994), Polar motion excitation by variations of the effective angular momentum function. II: Extended model, Manuscr.
http://ftp.aer.com/pub/anon_ Geod., 19, 157–171.
collaborations/sba/). Brzeziński, A., C. Bizouard, and S. Petrov (2002), Influence of the atmosphere on Earth rotation: What new can be learnt from the recent
atmospheric angular momentum estimates?, Surv. Geophys., 23, 33–69.
Chapman, S., and R. Lindzen (1970), Atmospheric Tides: Thermal and Gravitational, D. Reidel Publishing Company, Dordrecht, Holland.
Dehant, V., C. Bizouard, J. Hinderer, H. Legros, and M. Leffzt (1996), On atmospheric pressure perturbations on precession and nutations,
Phys. Earth Planet Inter., 96, 25–39.
de Viron, O., C. Bizouard, D. Salstein, and V. Dehant (1999), Atmospheric torque on the Earth and comparison with atmospheric angular
momentum variations, J. Geophys. Res., 4861–4875(B3).
IERS Conventions (2010), IERS Technical Note 36, G. Petit and B. Luzum (ed.), 179 pp., Verlag des Bundesamts fuer Kartographie und
Geodäsie, Frankfurt am Main. [Available at http://tai.bipm.org/iers/conv2010/conv2010.html.]
Krahenbuhl, D. S., M. B. Pace, R. S. Cerveny, and R. C. Balling Jr. (2011), Monthly lunar declination extremes influence on tropospheric
circulation patterns, J. Geophys. Res., 116, D23121, doi:10.1029/2011JD016598.
Li, G. (2005), 27.3-day and 13.6-day atmospheric tide and lunar forcing on atmospheric circulation, Adv. Atmos. Sci., 22(3), 359–374.
Li, G., H. Zong, and Q. Zhang (2011), 27.3-day and average 13.6-day periodic oscillations in the Earth’s rotation rate and atmospheric
pressure fields due to celestial gravitation forcing, Adv. Atmos. Sci., 28(1), 45–58.
Marcus, S., O. de Viron, and J. Dickey (2004), Atmospheric contributions to Earth Nutation: Geodetic constraints and limitations of the
torque approach, J. Atmos. Sci., 61, 352–356.
Nastrom, G. D., and A. D. Belmont (1976), Diurnal stratospheric tide in meridional wind, 30 to 60 km by season, J. Atmos. Sci., 33, 315–320.
Sakazaki, T., M. Fujiwara, and H. Hashiguchi (2010), Diurnal variations of upper tropospheric and lower stratospheric winds over Japan
as revealed with middle and upper atmosphere radar (34.85◦N, 136.10◦E) and five reanalysis data sets, J. Geophys. Res., 115, D24104,
doi:10.1029/2010JD014550.
Salstein, D., and R. Rosen (1989), Regional contributions to the atmospheric excitation of rapid polar motions, J. Geophys. Res., 94(D7),
9971–9978.
Sandford, D. J., H. G. Muller, and N. J. Mitchell (2006), Observations of lunar tides in the mesosphere and lower thermosphere at Arctic
and middle latitudes, Atmos. Chem. Phys., 6(12), 4117–4127.
Sasao, T., and J. Wahr (1981), An excitation mechanism for the free “core nutation”, Geophys. J. R. Astron. Soc., 64, 729–746.

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 11

 Journal of Geophysical Research: Atmospheres 10.1002/2014JD022240

SBA (2014), Website of the IERS Special Bureau for the Atmosphere. [Available at http://geophy.uni.lu/ggfc-atmosphere.html. Accessed
date: October 2014.]
Schindelegger, M., J. Böhm, D. Salstein, and H. Schuh (2011), High-resolution atmospheric angular momentum functions related to Earth
rotation parameters during CONT08, J. Geod., 85(7), 425–433.
Schindelegger, M., D. Salstein, and J. Böhm (2013), Recent estimates of Earth-atmosphere interaction torques and their use in studying
polar motion variability, J. Geophys. Res. Solid Earth, 118, 4586–4598, doi:10.1002/jgrb.50322.
Sidorenkov, N. S. (2009), The Interaction Between Earth’s Rotation and Geophysical Processes, 305 pp., Wiley-VCH, Weinheim, Germany.
Sidorenkov, N. S. (2010a), Luni-solar tides in the Earth’s atmosphere, paper presented at 6th Orlov Conference, Kiev, July 2009.
Sidorenkov, N. S. (2010b), About Inaccurate Estimation of the Role of Tidal Phenomena in Geophysics, vol. 11, special issue: 119–128,
Geophys. Res., Russian Acad. Sc., in Russian.
Simon, B., A. Lematre, and J. Souchay (2013), Oceanic tides, in Lecture Note in Physics, vol. 861, pp. 83–114, Springer,
Berlin Heidelberg.
Volland, H. (1996), Atmosphere and Earth’s rotation, Surv. Geophys., 17(1), 101–144.
Yseboodt, M., O. de Viron, T. M. Chin, and V. Dehant (2002), Atmospheric excitation of the Earth’s nutation: Comparison of different
atmospheric models, J. Geophys. Res., 107(B2), 2036, doi:10.1029/2000JB000042.
Zharov, V. E. (1997), Sol. Syst. Res., 31(6), 501.

BIZOUARD ET AL. ©2014. American Geophysical Union. All Rights Reserved. 12