You are on page 1of 17

Journal of Geodesy (1998) 72: 561±577

Diurnal atmospheric forcing and temporal variations


of the nutation amplitudes
C. Bizouard1, A. BrzezinÂski, S. Petrov2
Space Research Centre, Polish Academy of Sciences, Bartycka 18A, 00±716 Warsaw, Poland
e-mail: bizouard@obspm.fr

Received: 10 September 1997 = Accepted: 5 March 1998

Abstract. A 29-year time-series of four-times-daily at- considerably better agreement with the VLBI nutation
mospheric e€ective angular momentum (EAM) esti- data when using the EAM function without the IB
mates is used to study the atmospheric in¯uence on correction for ocean response, which indicates that this
nutation. The most important atmospheric contribu- correction is not adequate for nearly diurnal variations.
tions are found for the prograde annual (77 las),
retrograde annual (53 las), prograde semiannual
(45 las), and for the constant o€set of the pole Key words. Earth rotation  Nutation  Atmospheric
(dw sin e0 ˆ ÿ86 las, de ˆ 77 las). Among them only angular momentum  Atmospheric tides
the prograde semiannual component is driven mostly by
the wind term of the EAM function, while in all other
cases the pressure term is dominant. These are nonneg-
ligible quantities which should be taken into account in
the new theory of nutation. Comparison with the VLBI
corrections to the IAU 1980 nutation model taking into 1 Introduction
account the ocean tide contribution yields good agree-
ment for the prograde annual and semiannual nutations. The proven causes of the variations in Earth orientation
We also investigated time variability of the atmospheric fall into two categories: (1) the gravitational interaction
contribution to the nutation amplitudes by performing between the Earth and other celestial bodies, primarily
the sliding-window least-squares analysis of both the the Moon and the Sun; (2) exchanges of the angular
atmospheric excitation and VLBI nutation data. Almost momentum between the solid Earth and its environ-
all detected variations of atmospheric origin can be ment, the atmosphere, the hydrosphere, the cryosphere,
attributed to the pressure term, the biggest being the in- and the liquid core. The spatial motion of the Celestial
phase annual prograde component (about 30 las) and Ephemeris Pole (CEP), that is precession and nutation,
the retrograde one (as much as 100±200 las). These is chie¯y the lunisolar e€ect which can be predicted to a
variations, if physical, limit the precision of classical large extent. The atmosphere is the most important
modeling of nutation to the level of 0:1 mas. Compar- driving agent of polar motion at seasonal frequencies
ison with the VLBI data shows signi®cant correlation (Wahr 1983; Barnes et al. 1983), and plays a prominent
for the retrograde annual nutation after 1989, while for role in the excitation of the subseasonal irregular
the prograde annual term there is a high correlation in variations (Eubanks et al. 1988) and of the free Chandler
shape but the size of the atmospherically driven varia- wobble (Furuya et al. 1996). It is also expected that
tions is about three times less than deduced from the diurnal variations in the atmospheric angular momen-
VLBI data. This discrepancy in size can be attributed tum can excite minor but measurable changes in
either to inaccuracy of the theoretical transfer function nutation of the CEP (Sasao and Wahr, 1981; Gross,
or the frequency-dependent ocean response to the unpublished manuscript presented at IUGG XXI
pressure variations. Our comparison also yields a General Assembly 1995; Dehant et al. 1996). These
changes include a contribution to the free core nutation
(FCN) term of variable amplitude between 0.1 and 0.5
milliarcseconds (mas), as well as to the amplitudes of
Correspondence to: C. Bizouard
some important lunisolar constituents of nutation (e.g.,
1
On leave from: Observatoire de Paris, Paris, France the annual prograde nutation corresponding to the S1
2
On leave from: Institute of Applied Astronomy, St.Petersburg, tide). An estimation of the atmospheric in¯uence upon
Russia nutation is a part of the international e€orts aimed at
562

explaining the signi®cant discrepancies between the mantle anelasticity; see Dehant and Defraigne (1997) for
International Astronomical Union (IAU) 1980 theory review. A more ecient way of verifying the estimated
of nutation and the Very Long Baseline Interferometry atmospheric contribution to nutation is to compare the
(VLBI) measurements of the nutation angles, and at time variability of the amplitudes and phases with a
working out a new nonrigid Earth nutation theory similar e€ect derived from the celestial pole o€sets ob-
(McCarthy and Dehant 1995). served by VLBI. Moreover, such a comparison is also an
There are two di€erent methods of modeling the at- important external check of the reliability of the atmo-
mospheric excitation of nutation. The ®rst one, com- spheric angular momentum data at nearly diurnal peri-
monly referred to as the torque approach, consists in the ods.
computation of the atmospheric torque upon the solid Another important problem of interest concerns the
Earth (Bizouard 1995; Dehant et al. 1996; Viron et al. oceanic response to variations of the surface pressure
1998). The second one, the angular momentum ap- which modi®es the atmospheric in¯uence upon Earth
proach, is based on the angular momentum balance rotation. The EAM data set contains two versions of the
between the atmosphere and the solid Earth (Sasao and pressure term: one which is not corrected for the ocean
Wahr 1981; BrzezinÂski 1994; Gross, unpublished response, assuming a rigid ocean, and the other one with
manuscript presented at IUGG XXI General Assembly the so-called inverted barometer (IB) correction assum-
1995; Zharov and Gambis 1996). ing static equilibrium response of the ocean. It is com-
The present paper uses the angular momentum ap- monly believed that the IB correction is adequate for
proach. Earlier attempts applying this method were seasonal and longer-period components of the EAM
unable to reach any de®nitive conclusion due to the function (Wahr 1983), but the barotropic numerical
lack of the appropriate atmospheric angular momen- model indicates that this correction is not reliable at
tum estimates. The situation has improved dramatically periods shorter than 2 days (Ponte et al. 1991); see also
recently. Since the middle of 1992 several meteorolog- Salstein et al. (1993) for further details. By using both
ical centers produce four-times-daily estimates of the versions of the EAM data and comparing results with
atmospheric E€ective Angular Momentum (EAM) the VLBI observations of nutation, one can try to verify
function v de®ned by Barnes et al. (1983) (see also the IB correction at nearly diurnal periods.
Salstein et al. 1993), which can be used to study diurnal We start by recalling in Sect. 2.1 some important
variations. The last important achievement is a 29-year equations relating the atmospheric excitation quantities
homogeneous series of the four-times-daily EAM esti- to nutation derived by BrzezinÂski (1994) on the basis of
mates (Salstein and Rosen 1997) computed on the basis the dynamical theory of Sasao and Wahr (1981). Next,
of results of the U.S. National Center for Environ- we compute from the EAM data the so-called Celestial
mental Prediction (NCEP)/U.S. National Center for E€ective Angular Momentum (CEAM) function and
Atmospheric Research (NCAR) reanalysis project perform its spectral analysis in order to ®nd which
(Kalnay et al. 1996; Salstein 1995), spanning the period nutation terms are in¯uenced by the atmosphere and to
from 1968.0 to 1997.3. Here we use this time-series in make preliminary estimates of the corresponding incre-
order to estimate its power contents at diurnal retro- mental amplitudes (Sect. 2.2). In Sect. 2.3 the
grade frequencies and to estimate the corresponding atmospheric contribution to nutation is estimated by a
nutation. least-squares analysis. In Sect. 2.4 we introduce the
One question addressed here which is extremely im- ocean tide contribution computed by Dehant and De-
portant from the practical point of view is the stability fraigne (1997). The oceanic corrections are then em-
of the atmospheric excitation. If the spectrum contains ployed to compare the atmospheric nutations to the
sharp peaks, implying that the underlying process fol- di€erences between the IAU 1980 nutation theory and
lows a simple sinusoidal model, then the atmospheric the VLBI observations of the nutation angles. Sect. 3 of
e€ect can be included in the future nutation model. If it the paper is devoted to the problem of time variability
is not the case, that is, the spectral peaks express only of the atmospherically driven nutations. We carry out in
pseudo-periodic components with variable amplitudes Sect. 3.1 the sliding-window least-squares analysis of the
and phases, then the atmospheric contributions to nut- CEAM data in order to ®nd to what extent the average
ation can only be estimated a posteriori from observa- amplitudes derived in Sect. 2.3 express the instantaneous
tions, and the size of the variations limits the accuracy of atmospheric excitation. In Sect. 3.2 we make a similar
any nutation model. A simple way of treating this sliding window least-squares analysis of the VLBI nut-
problem is to estimate the ``instantaneous'' phase and ation data and compare results with the time variability
amplitude by narrowing the input data interval, and of the atmospheric contribution. Finally, Sect. 4 gives a
then to check how these quantities vary with time. This brief summary and discussion of the results obtained
is the so-called sliding-window analysis. from this research.
The estimated atmospheric contribution to the nut- Many technical details of our computations are given
ation amplitudes can be compared with the empirical in three appendices, which were mainly motivated by the
corrections to the IAU 1980 model, derived from VLBI fact that there are di€erences in the available literature
observations. Such a comparison, however, cannot be concerning parameterization of nutation, examples be-
conclusive if we do not take into account a variety of ing various conventions applied for representing nutat-
other e€ects such as a nonhydrostatic ¯attening of the ional motion of the pole by complex variables or for
liquid core, in¯uence of the ocean diurnal tide, and labeling prograde and retrograde nutations.
563

2 Computation of the atmospheric contributions BrzezinÂski (1994) introduced the atmospheric celes-
to nutation tial e€ective angular momentum (CEAM) function v0
which is convenient for studying atmospheric and
2.1 Formulation of the problem oceanic excitation of nutation. To ®rst-order approxi-
mation, v0 is the EAM function v expressed in the space-
Nutation is excited by nearly diurnal retrograde (that ®xed frame, which justi®es the terminology. The CEAM
means propagating westward; see Appendix C for a function is de®ned by
strict de®nition of prograde and retrograde motions)
forcings, as seen from the rotating Earth. The main v0 ˆ ÿveiU …1†
contribution is from the lunisolar gravitational torque and similar expressions for the pressure term v0 p and for
exerted on the Earth's equatorial bulge, and this is the the wind term v0 w . Note the convention which will be
only e€ect which is taken into account in the IAU 1980 applied throughout the whole paper: adding the prime
theory of nutation (Wahr 1981; Seidelmann 1982). But means that the quantity is expressed in the space-®xed
nearly diurnal retrograde variations in the distribution frame. In Eq. (1) U denotes the Greenwich sidereal time
of atmospheric mass and/or in the wind pattern change which can be expressed as U ˆ X…t ÿ t0 † ‡ U0 , where X
the atmospheric angular momentum which, in turn, is the sidereal diurnal frequency, t0 denotes the reference
induce nutational motion of the solid Earth through epoch and U0 …t† is a quasi-constant phase referred to t0
the conservation of the total angular momentum of the [see Appendix B for details concerning practical appli-
solid Earth and ¯uid environment system. The atmo- cations of Eq. (1)].
spheric diurnal tides are dominated by the thermal Let P ˆ Dw sin e0 ‡ iDe be the complex combination
tides driven by the diurnal change in solar heating, of the X and Y components of the Celestial Ephemeris
with the biggest term being the Sun-®xed S1 tide Pole (CEP) in the true equatorial plane, where Dw and
(Chapman and Lindzen 1970). These tides are accom- De are the corresponding increments in longitude and
panied by corresponding changes in the equatorial obliquity, and e0 is the mean obliquity at the reference
EAM which we will label in the same manner as the epoch t0 ˆ J 2000:0 (BrzezinÂski and Capitaine 1993). The
tides themselves. The S1 diurnal retrograde EAM atmospheric e€ect on nutation can be expressed by the
component, which has an annual prograde frequency following frequency domain relation (BrzezinÂski 1994,
with respect to the space-®xed frame, undergoes annual Eq. 4.9):
and semiannual modulation which gives rise to the
diurnal retrograde components K1 , P1 , w1 , p1 with p w
P …r0 † ˆ Tp …r0 †v0 …r0 † ‡ Tw …r0 †v0 …r0 † …2†
space-referred frequencies zero, prograde semiannual,
retrograde annual, and prograde terannual, respectively in which
(Eubanks et al. 1986; see also Appendix A for " #
explanation). Unlike the gravitational tides, which can 0 1 ap
Tp …r † ˆ rc 0 ‡ 0 …3†
always be decomposed into spherical harmonics in the rc ÿ r rf ÿ r0
0
discrete frequency domain representation, the thermal
diurnal tides are much more complicated and we and similar expression with Tp ; ap replaced by Tw ; aw ,
should consider them as one band-pass process having respectively. Here Tp ; Tw are transfer functions of the
a continuous spectrum in the diurnal retrograde pressure term and of the wind term, respectively, r0 is
frequency band. the nutation angular frequency, r0c ˆ rc ‡ X and r0f are
The theoretical background of the present study is the space-referred angular frequencies of the Chandler
provided by BrzezinÂski (1994). Let the complex quan- wobble (CW) and of the free core nutation (FCN)
tity v ˆ v1 ‡ iv2 be the equatorial component of the modes, rc is the Earth-referred Chandler frequency, and
atmospheric e€ective angular momentum (EAM) ap ˆ 9:509  10ÿ2 , aw ˆ 5:489  10ÿ4 are dimensionless
function, calculated within a terrestrial frame, which coecients expressing response of the FCN mode to the
consists of: pressure and wind excitation.
Equations (2) and (3) are not sensitive to the choice
1. a term proportional to I13 ‡ iI23 , that is, the complex
of the Chandler frequency which is far from the nutation
combination of the o€-diagonal atmospheric inertia
frequencies. Therefore we adopted in our computations
moments I13 and I23 within the terrestrial frame; this
the observed dissipationless value of rc ˆ 2p=433:3
term can be expressed, assuming hydrostatic equilib-
days. The dependence on the adopted FCN frequency is
rium and thin layer hypotheses, as a function of a
much more important, particularly for the w1 wave (see
surface integral of the surface pressure, and hence is
Fig. 1). The FCN complex frequency r0f ˆ ÿ…2p=Tf0 †
called the pressure term vp ;
…1 ÿ i=2Q0f † is much less constrained from observations
2. a term proportional to the equatorial atmospheric
than is the CW frequency, so we applied Eq. (2) trying
angular momentum with respect to the crust, caused
di€erent values in the range of 415±435 days for the
by the surface atmospheric ¯ow, and hence called the
FCN period Tf0 and 10±100 for the FCN quality factor
wind term vw .
Q0f ˆ Qf =…Tf0 ‡ 2† where Qf is the usually used Earth-
See Barnes et al. (1983) for de®nition of v and referred value (BrzezinÂski 1998).
computational details, and Salstein et al. (1993) for a Finally, let us note that for obvious reasons the the-
report on operational determinations of v. ory presented in this section can also be applied to
564

a Original series S1 ®lter at frequencies 1, 2, 3, and 10 cpy (cycles per year) is


Removed model 0:964, 0:867, 0:725, and 0:028, respectively. This choice
PSD (decibels)

20 K1
P1 π1 of FWHM seems to be a reasonable compromise be-
ψ1 tween two contradictory requirements: (1) amplitudes of
FCN the expected periodical constituents should not be af-
fected by the smoothing, and (2) no power should be left
0 at frequencies higher than the Nyquist frequency 10 cpy
-3 -2 -1 0 1 2 3 in order to avoid aliasing. For the least-squares analyses
performed in the next sections we adopted Dt ˆ 0:0125
40 b S1 P1 year and FWHM ˆ 0:025 year which yields the transfer
PSD (decibels)

function of the Gaussian smoother equal to 0:998, 0:991,


π1 and 0:980 at frequencies 1, 2, and 3 cpy, respectively.
20
ψ1 K1 From the point of view of time-series analysis, the
FCN foregoing procedure is the complex demodulation
0
(Bingham et al. 1967) of the original EAM series at
frequency ÿX. It eliminates all variations with fre-
-3 -2 -1 0 1 2 3 quencies outside the diurnal retrograde band, including
70 c P1 seasonal term and diurnal prograde oscillations. Note,
S1 however, that due to the discrete sampling of the EAM
PSD (decibels)

50 ψ1 function, all periodic terms with frequencies near


K1 π1 …ÿ1 ‡ 4s†X, s ˆ 1; 2; . . ., that means 3, ÿ5, 7, ÿ9, . . .
30 FCN cycles per sidereal day, could alias into the diurnal ret-
rograde band. We do not expect that this undesired ef-
10
fect is signi®cant, on the basis of the following
arguments. First, the estimation of the EAM function
-3 -2 -1 0 1 2 3
Frequency (cycles/year)
involves time averaging (Kalnay et al. 1996), hence it can
be assumed that the ®nal values are not the instanta-
Fig. 1a±c. Maximum entropy power spectra of the original CEAM neous samples but rather the 6-hourly means. Second,
series and after removing the best least-squares ®t of the model though the theoretical models (Chapman and Lindzen
comprising ®rst-order polynomial and sinusoids with periods 1 year, 1970, p. 43) predict 8-hourly local surface pressure
1=2 year, 1=3 year: a pressure term; b pressure term + IB variations, the average kinetic energy spectrum of the
correction; c wind term
free atmosphere shown in Fig. 1.1 of Volland (1988) is
nevertheless rather ¯at between the peaks at 1 cycle per
describe oceanic e€ects on Earth rotation (BrzezinÂski day and about 1 cycle per minute. On the other hand, at
1994). In this case it would be more appropriate to lower frequencies this spectrum shows the same features
adopt slightly di€erent terminology: the matter and as does the EAM spectrum, namely the sharp peaks at 1
motion terms instead of the pressure and wind terms. cycle per year (Wahr 1983) and 1 cycle per day (this
paper), and a broad peak extending approximately from
1 cycle per month to 1 cycle per week (BrzezinÂski 1987;
2.2 Spectral analysis of the equatorial CEAM functions Eubanks et al. 1988).
As a next step we applied the maximum entropy
A ®rst step in our analysis was to calculate the method (MEM) of spectral analysis to the CEAM series.
atmospheric CEAM function v0 by applying the trans- According to BrzezinÂski (1995) this high-resolution
formation given by Eq. (1) to the EAM data (see method is particularly well suited for studying variations
Appendix B for computational details). The nearly in Earth rotation. It yields a realistic estimate of the
diurnal retrograde variations map in this transformation power spectral density (PSD) function, which can be
into low-frequency variations, while all other changes integrated numerically in order to derive amplitudes of
(including seasonal terms and secular variations) appear the detected oscillations. In our approach we treated the
in v0 as high-frequency oscillations which can be easily CEAM functions as well as the corresponding nutation
removed by smoothing. A ®nal step in the initial data as complex quantities v0 ˆ v01 ‡ iv02 and P ˆ dw sin e
processing is to reduce the number of data by making ‡ide, respectively, distinguishing between the prograde
the sample interval Dt longer, in accordance with the (counterclockwise) and retrograde (clockwise) oscilla-
degree of smoothing. tions (see Appendix C). We considered separately the
For the purpose of spectral analysis described in this pressure term vp with and without the inverted barometer
section we adopted Dt ˆ 0:05 year which is suciently (IB) correction for the ocean response bearing in mind
short to express all variations with periods of 1=3 year or that the true solution at diurnal periods which is pres-
longer. We performed simultaneous smoothing and in- ently not known, may be di€erent from these two cases.
terpolation of the CEAM series using the low-pass Spectral estimates presented in Fig. 1 and Table 1
Gaussian ®lter (Feissel and Lewandowski 1984) with full correspond to the CEAM series from 1979:0 to 1997:3.
width at half of maximum (FWHM) equal to 2Dt ˆ 0:1 We noticed that when adding the data prior to 1979, the
year. The theoretical amplitude transfer function of this results become worse in that the average noise back-
565

Table 1a±c. Atmospheric excitation of nutation estimated from the are the values derived from the CEAM series after removal of the
maximum entropy spectral analysis of the CEAM series spanning linear trend and a sum of sinusoids with periods 1 year, 1/2
the period from 1979.0 to 1997.3. Power contribution is expressed year, 1/3 year. Units are mas for the amplitudes. a pressure term
by the unitless ratio of the power P included in a particular spectral vp , sample variance V ˆ 2:11 …0:69† mas2 ; b pressure term vp + IB,
peak to the total power, or sample variance V , of thepinput series. sample variance V ˆ 2:04 …0:11† mas2 ; c wind term vw , sample
Nutation amplitude is the CEAM mean amplitude P multiplied variance V ˆ 512:23 …19:24† mas2
by the modulus of the transfer function T . Numbers in parenthesis

nutation label w1 K1 S1 P1 p1
frequency )1 cpy 0 cpy 1 cpy 2 cpy 3 cpy

a power contribution P=Vp


(%) 3.5 6.9 71.9 7.7 3.8
mean CEAM amplitude P 0.27 0.38 1.23 0.40 0.28
p (0.25) (0.36) (0.83) (0.32) (0.23)
mean nutation amplitude jT j P 0.172 0.035 0.051 0.010 0.005
(0.159) (0.033) (0.034) (0.008) (0.004)
b power contribution P=Vp
(%) 0.5 0.8 82.6 14.1 0.9
mean CEAM amplitude P 0.10 0.13 1.30 0.54 0.14
p (0.10) (0.11) (0.24) (0.07)
mean nutation amplitude jT j P 0.066 0.012 0.053 0.014 0.003
(0.066) (0.010) (0.010) (0.002)
c power contribution P=Vp
(%) 0.6 0.5 41.2 56.8 0.2
mean CEAM amplitude P 1.77 1.60 14.53 17.06 0.94
p (1.10) (1.51) (3.30) (1.30) (0.80)
mean nutation amplitude jT j P 0.011 0.003 0.032 0.038 0.002
(0.007) (0.003) (0.007) (0.003) (0.002)

ground increases and signi®cantly less power can be when the oscillation is strictly harmonic. When the
removed from the series by sinusoidal modeling of the oscillation is pseudo-periodic with variable amplitude
detected spectral lines. A similar e€ect could be observed and phase, the mean amplitude is higher than the
when performing the least-squares analysis in order to amplitude in the least-squares sense.
determine the amplitudes of the atmospheric nutations 3. Additionally, the mean amplitude can also be slightly
(see Sect. 2.3). This degradation of quality of the CEAM overestimated due to the contribution of the mea-
estimates may be due to the fact that the global cover- surement noise included in the data.
age, both spatial and temporal, of the pressure and wind
Note also that p
the
 estimated nutation amplitudes should
data, was rather poor in the sixties and seventies in
be divided by 2 if we like to express them in terms of
comparison to the last two decades. Nevertheless, when
dw sin e and de separately.
studying variability of the atmospheric contribution to
From Table 1 it can be seen that though nearly di-
nutation, reported in Sect. 3, we took into account all
urnal retrograde variations of the wind term are much
the available CEAM data.
larger than those of the pressure term (by more than 1
The main common features of the MEM power
order in amplitude or 2 orders in power), nevertheless
spectra shown in Fig. 1 are the dominating annual
they are strongly attenuated by the low values of the
prograde (S1 ) and semiannual prograde (P1 ) peaks, with
transfer function (aw  ap ) and the corresponding nut-
additional contribution from the terannual prograde
ations are smaller. The atmospheric contribution to the
(p1 ), zero-frequency (K1 ) and annual retrograde (w1 )
annual prograde nutation is of the order of 50 las for
peaks. The only unexpected result is a weak excess of
the pressure term and 30 las for the wind term. In the
power at the semiannual retrograde frequency ÿ2 cpy,
case of the semiannual prograde nutation the wind term
particularly well seen in the wind term, which cannot be
can contribute almost 40 las while the pressure term
explained by annual and semiannual modulation of the
contribution is about three times smaller. The highest
thermal S1 tide (Appendix A). The CEAM amplitudes
values are obtained for the w1 term, up to 170 las, but
computed from the MEM power spectra are shown in
they could be signi®cantly overestimated (cf. Table 3)
Table 1. The corresponding nutation amplitudes given
for the following reason in addition to those presented
in the last rows of a, b, and c of Table should be treated
earlier concerning all peaks. The atmospheric in¯uence
as upper limits of the atmospheric contribution for the
upon nutation shown in Table 1 is computed assuming
following reasons.
that the whole power is concentrated at the central fre-
1. We used the transfer function T …r† given by eq. (2), quency of the peak or that the transfer function does not
with di€erent values of the FCN complex frequency r0f vary within the peak domain. This assumption is satis-
(see Sect. 2.1). Only the highest estimates are shown in ®ed for all but the annual retrograde peak since this
Table 1. In fact, the di€erences are signi®cant only for peak is rather broad and close to the FCN resonance of
the w1 term, up to 20% of the tabulated results. the transfer function (see Fig. 1). This part of the at-
2. Mean amplitude at one particular frequency is the mospheric angular momentum spectrum will be inves-
square root of the power (variance) included in the tigated separately, in the context of the excitation of the
whole peak which equals the true amplitude only FCN free oscillation (BrzezinÂski 1998).
566

An important question concerning the atmospheric and out-of-phase terms of v0j . As explained in Appendix
excitation of nutation is the extent to which this phe- C, hj is a quasi-constant phase of the nutation argument,
nomenon follows a simple sinusoidal pattern which can which can be treated in our applications as constant.
be included in the nutation model. Any signi®cant de- Moreover we added to the model a linear trend cor-
parture from this pattern de®nes accuracy limits for any responding to the zero-frequency peak, obtaining ®nally
future nutation model, at least in the classic form as-
suming constant or secularly changing amplitudes (see X
4
ip op 
v0 ˆ …v0 j ‡ iv0 j †eihj …t† ‡ v0co ‡ v0sl t …6†
Appendix C). Our preliminary attempt to clarify this
jˆ1
problem is to remove from the CEAM series the best
least-squares ®t of the ®rst-order polynomial and the where the summation extends over the prograde annual,
sum of sinusoids with periods 1 year, 1=2 year, 1=3 semiannual, terannual, and the retrograde annual terms,
year, and then repeating the MEM spectral analysis. v0 ip 0 op 0 0
j , v j are real coecients and v co , vsl are complex
Results are shown as dotted lines in Fig. 1 and as am- coecients.
plitude estimates in parenthesis in Table 1. We can see The least-squares analysis was applied separately to
that the model removed most of the power from the the pressure term and to the wind term over two periods:
wind term (96%) and from the pressure term with the IB 1968.0±1997.3 (all the data) and 1979.0±1997.3. There
correction (95%), leaving a relatively smooth spectrum are signi®cant di€erences between the two sets of cor-
with signi®cant excess of power only around the annual responding estimates, up to 1 mas (for the annual term
prograde frequency. In the non-IB case the reduction of of v0 w ). This is due to the irregularity of the atmospheric
total power is signi®cantly lower (about 66%) and none excitation, mentioned already in Sect. 2.2. As seen in
of the peaks is cancelled completely. The remaining ir- Sect. 3, the atmospheric excitation is much more stable
regular contribution to the nutation amplitudes does not after 1979. Therefore the estimates for the period
exceed 7 las for the wind term and 10 las for the pres- 1979.0±1997.3, presented in Table 2, should be treated as
sure term with the IB correction, if we exclude from more meaningful.
considerations the annual retrograde nutation. This ir- The contribution to nutation was then computed
regular contribution is more signi®cant for the pressure simply by multiplying the complex amplitudes
term without the IB correction, about 30 las for the v0 ip ‡ iv0 op from Table 2 by the appropriate term of the
annual prograde nutation and constant o€set/linear transfer function given by Eq. (2); note that the transfer
trend corresponding to the K1 peak, and much more for function is also complex-valued, and therefore this
the annual retrograde nutation due to an ampli®cation transformation changes not only the amplitude of v0j but
by the FCN resonance. also its phase. This method is not directly applicable to
A more detailed analysis of the time variability of the the linear trend of v0 . Instead, we inserted this part of the
atmospherically driven nutations will be given in Sect. model into the time-domain version of Eq. (2), which is
3.1, and comparison with a similar e€ect in the VLBI Eq. (3.15) of BrzezinÂski (1994), and then solved the
observations of nutation will be done in Sect. 3.2. equation analytically. After neglecting the free terms
expressing the FCN oscillation, the solution is also a
®rst-order polynomial
2.3 Atmospheric contribution to nutation  
P ˆ b1 v0co ‡ v0sl t ‡ b2 v0sl …7†
A determination of the amplitude and phase of all
circular terms revealed in the previous section by the in which
spectral analysis, has been performed by a least-squares rc
®t. As our purpose is to estimate the e€ects on the b1 ˆ 0 0 …ar0c ‡ r0f † …8†
rc rf
corresponding circular nutations, the adopted model of
the CEAM function v0 is the same as that for nutation, " #
described in details in Appendix C. For the j-th circular irc …ar0c ‡ r0f †…r0c ‡ r0f †
term we have b2 ˆ 0 0 1 ‡ a ÿ …9†
rc rf r0c r0f
ip op  ip op
v0 j ˆ …v0 j ‡ iv0 j †eihj …t† ˆ i…v0 j ‡ iv0 j †eisj hj …4†
and a ˆ ap or a ˆ aw ; note that b1 is unitless while b2 has
with a dimension of time. For the pressure excitation (a ˆ ap )
we derive b1 ˆ ÿ0:089, b2 ˆ ÿ0:017i years, and for the

hj …t† ˆ sj hj …t† ‡ p=2 ˆ sj …kj t ‡ hj † ‡ p=2 …5† wind term (a ˆ aw ) b1 ˆ 0:002, b2 ˆ ÿ5  10ÿ5 i years.
These are dissipationless values (that means assuming
(cf. Eq. (22) of Appendix C), where t denotes time Q0f ˆ 1), which do not di€er by more than 10% from
referred to the standard epoch J2000.0 and sj equals the values corresponding to the ®nite Q0f .
either 1 or ÿ1 in such a way that sj kj equals exactly the One can notice that b1 equals the transfer function,
angular frequency of nutation, positive for prograde given by Eq. (2), at r0 ˆ 0. In fact, for empirical values
motion and negative for retrograde motion. The real and of v0co , v0sl given in Table 2, the term b2 v0sl is of little
imaginary parts of the complex amplitude, labeled by the importance in Eq. (7), hence a direct extension of the
superscripts ``ip'', ``op'', are amplitudes of the in-phase frequency domain method used for the sinusoidal terms,
567

Table 2a, b. The best least-squares ®t of the model expressing the polynomial. a Circular terms with arguments the same as for nu-
equatorial atmospheric CEAM function v0 spanning the period tations, amplitudes are in mas. b Linear trend referred to J2000.0,
from 1979.0 to 1997.3. The model consists of four circular oscil- units are las for the constant and las/year for the trend coecient
lations corresponding to the nutation terms, and the ®rst-order

a circular term contribution in-phase out-of-phase

‡1=3 year pressure 0:031  0:018 0:231  0:018


pressure + IB ÿ0:025  0:007 0:137  0:007
wind ÿ0:433  0:121 0:023  0:121
‡1=2 year pressure ÿ0:115  0:018 0:206  0:018
pressure + IB 0:548  0:007 ÿ0:190  0:007
wind ÿ2:872  0:121 ÿ18:500  0:122
‡1 year pressure 0:231  0:018 1:156  0:018
pressure + IB 1:033  0:007 0:805  0:007
wind ÿ0:147  0:122 ÿ14:197  0:122
ÿ1 year pressure 0:060  0:018 ÿ0:056  0:018
pressure + IB 0:049  0:007 0:007  0:007
wind 1:001  0:122 1:053  0:122

b linear term contribution real imaginary

constant pressure 0:641  0:044 ÿ0:758  0:042


pressure + IB ÿ0:010  0:018 ÿ0:963  0:017
wind ÿ16:075  0:290 5:011  0:301
secular pressure 0:010  0:003 ÿ0:023  0:003
pressure + IB 0:011  0:001 ÿ0:002  0:001
wind 0:057  0:022 0:120  0:023

that is P ˆ T …r0 ˆ 0†  ‰v0co ‡ v0sl tŠ, gives a reasonably signi®cantly, as illustrated in Fig. 2. The di€erences
good approximation in the present case. between the results given in the present paper and in
The estimated atmospheric contributions to the cir- Dehant et al. (1996) concerning the atmospheric con-
cular nutations, as well as to the constant o€set of the tribution to the annual nutation certainly exceed the
pole and to the linear trend, are displayed in Table 3. As error bounds given by the respective authors. The ex-
could be expected, the most important contribution is to planation of this disagreement is out of the scope of this
the annual nutation, about 75 las for the prograde term. paper and deserves further study.
The retrograde annual term in the CEAM function is The nonzero mean of the CEAM functions v0 causes
about 20 times smaller than the prograde one, but after a constant o€set of the celestial pole of the size almost
a strong ampli®cation by the FCN resonance, the cor- 0:1 mas both for Dw sin e0 and De. Physically, this o€set
responding nutation amplitude reaches the level of 40 to is caused by annual modulation of the atmospheric di-
50 las. There is also a signi®cant semiannual prograde urnal retrograde tide S1 (Appendix A) with a larger
term with the amplitude of about 40 to 45 las driven contribution from the pressure term. The estimated rates
mostly by the wind term. The amplitude of the prograde of the order of 2 las/year for the linear trend in longi-
terannual term, though signi®cant in view of its formal tude and obliquity are small and do not necessarily re-
error, is small when compared to the current precision of ¯ect the true secular e€ect but rather the long-periodic
the nutation determinations. An interesting observation atmospheric perturbation. Again, this estimate di€ers
is that with an exception of the annual retrograde nut- signi®cantly from that computed by Dehant et al.
ation, the atmospheric contributions are dominated by (1996). We do not have any explanation for this di€er-
the out-of-phase terms. If the IB correction is not ap- ence. Note, however, that in the latter estimation only
plied, all phases with respect to the IAU arguments sj hj the pressure torque was taken into account.
are in the range of ÿ5 to ÿ20 , while for the gravita- It should be noted that the original AAM series were
tional lunisolar tides this di€erence is either 90 or ÿ90 not provided with any information on the uncertainties
(see Appendix C). In the case of the annual prograde in the data. In order to estimate the error bars of the
nutation (Fig. 2), which is of particular interest here, the computed amplitudes of the CEAM variations we used
IAU argument sj hj is equivalent to the mean anomally l0 the following procedure. We calculated the sample co-
of the Sun. The phase di€erence of the IAU 1980 annual variances of the smoothed CEAM series at zero lag, c^0 ,
prograde nutation with respect to l0 , hence with respect and at the lag equal to the sampling interval, c^1 . We
to the Sun, is ÿ90 , while for the atmospheric S1 term we applied the standard biased estimate given by
obtained ÿ7 (ÿ34 for IB case) ± see Table 3. Dehant
et al. (1996) computed the e€ect of the diurnal pressure 1 X L
c^k ˆ xl xlÿk …10†
torque on the annual prograde nutation and found the L lˆk‡1
phase di€erence with respect to the gravitational nut-
ation in the range of 123 to 174 (with the most prob- where L is the length of the series and xl ˆ v0 …t0 ‡ lDt†
able value 131 ), that is from 33 to 84 with respect to for l ˆ 1; 2; . . . ; L. If r is the measurement error then the
the Sun. These values di€er from the present results true covariance function at zero lag, c0 , is increased by
568

Table 3a, b. Atmospheric contribution to nutations, constant o€set after Dehant and Defraigne (1997) and the VLBI corrections to the
of the pole, and slopes in longitude and obliquity. Nutation ar- IAU 1980 model. a Contribution to circular terms. b Contribution
gument, denoted in Appendix C by sj hj , means the standard integer to the constant pole o€set at epoch J2000.0 and to the slope in
combination of the Brown's arguments, multiplied additionally by longitude and obliquity. Also given are ocean tide e€ects after
‡1 or ÿ1. Also given for comparison are ocean tide contributions Gross (1993)

a nutation contribution in-phase out-of-phase amplitude phase/ phase/


las las las nut.argument IAU1980 model
+1/3 year pressure )0.5 ‹ 0.3 )4.3 ‹ 0.3 4.3 ‹ 0.4 )7° ‹ 5 )97° ‹ 5
pressure+IB 0.5 ‹ 0.1 )2.5 ‹ 0.1 2.6 ‹ 0.2 17° ‹ 4 )73° ‹ 4
wind )1.0 ‹ 0.3 0.0 ‹ 0.3 1.0 ‹ 0.3 )93° ‹ 17 177° ‹ 17
total )1.5 ‹ 0.6 )4.2 ‹ 0.6 4.5 ‹ 0.8 )20° ‹ 10 )110° ‹ 10
total+IB )0.5 ‹ 0.4 )2.5 ‹ 0.4 2.5 ‹ 0.4 )11° ‹ 10 )101° ‹ 10
VLBI 17 20
+1/2 year pressure 3.0 ‹ 0.5 )5.2 ‹ 0.5 6.0 ‹ 0.6 29° ‹ 6 )61° ‹ 6
pressure+IB )14.0 ‹ 0.2 4.8 ‹ 0.2 14.8 ‹ 0.2 )109° ‹ 1 )199° ‹ 1
wind )6.0 ‹ 0.3 )39.8 ‹ 0.3 40.2 ‹ 0.3 )9° ‹ 1 )99° ‹ 1
total )3.1 ‹ 0.7 )45.0 ‹ 0.7 45.1 ‹ 0.8 )4° ‹ 1 )94° ‹ 1
total+IB )20 ‹ 0.4 )35.0 ‹ 0.4 40.3 ‹ 0.6 )30° ‹ 1 )120° ‹ 1
ocean )570 580
VLBI )606 510
+1 year pressure )9.0 ‹ 1.0 )46.8 ‹ 0.7 47.7 ‹ 0.8 )11° ‹ 1 79° ‹ 1
Pressure+IB )41.6 ‹ 0.3 )32.8 ‹ 0.3 53 ‹ 0.4 )52° ‹ 0.4 38° ‹ 0.4
wind )0.2 ‹ 0.3 )29.2 ‹ 0.3 29.2 ‹ 0.3 0° ‹ 1 90° ‹ 1
total )9.3 ‹ 0.7 )76.0 ‹ 1.0 76.6 ‹ 1.1 )7° ‹ 1 83° ‹ 1
total+IB )41.8 ‹ 0.5 )62.0 ‹ 0.5 74.8 ‹ 0.7 )34° ‹ 1 56° ‹ 1
ocean 30 )30
VLBI 12 )124
)1 year pressure 37.7 ‹ 11.5 )36.5 ‹ 11.5 52.5 ‹ 16.5 46° ‹ 18 )44° ‹ 18
pressure+IB 31.5 ‹ 4.6 3.8 ‹ 4.7 31.7 ‹ 5.2 46° ‹7 224° ‹ 7
wind 6.0 ‹ 0.7 6.2 ‹ 0.7 8.7 ‹ 1.0 97° ‹9 7° ‹ 9
total 43.7 ‹ 12.3 )30.3 ‹ 12.3 53.2 ‹ 17.1 )55° ‹ 18 )35° ‹ 18
total+IB 37.5 ‹ 5.3 10.3 ‹ 5.3 39.0 ‹ 6.5 105° ‹ 10 15° ‹ 10
ocean )210 210
VLBI 1976 )322

b contribution dw sin e0 de w sin e0 e_


las las las/year las/year

pressure )58 ‹ 4 68 ‹ 4 )0.9 ‹ 0.3 2.1 ‹ 0.3


pressure+IB 1‹ 2 86 ‹ 2 )0.9 ‹ 0.1 0.2 ‹ 0.1
wind )29 ‹ 1 9‹ 1 0.1 ‹ 0.0 0.2 ‹ 0.0
total )86 ‹ 5 77 ‹ 4 )0.8 ‹ 0.3 2.3 ‹ 0.3
total+IB )28 ‹ 2 95 ‹ 2 )0.8 ‹ 0.2 0.4 ‹ 0.2
ocean 4698 7518

r2 , yielding c0 ‡ r2 . Bearing in mind that the true on nutation together. Besides, recently the oceanic
autocovariance function at the successive lag, c1 , never dynamical models have been validated by the high-
exceeds c0 , we estimated the a priori formal error for the
p precision TOPEX/Poseidon observations, giving high
least-squares analysis as r^ ˆ c^0 ÿ c^1 . This algorithm credibility to the oceanic nutational corrections.
was applied separately to the real and imaginary parts of The world ocean, along with the atmosphere, is an
v0 . The method described works well if the measurement important agent in the diurnal angular momentum ex-
process is white noise. When the CEAM series were change. Besides the fact that the ocean modi®es the
smoothed prior to the analysis, this assumption is no atmospheric pressure in¯uence upon the solid Earth, it
longer valid and, as a consequence, the error bars also exhibits tidal variations of its surface elevation and
presented in Tables 2 and 3 and in Figs. 3 and 4 may be tidal currents, which both contribute to the oceanic
signi®cantly underestimated with respect to their true angular momentum. Fortunately, it appears that the
values. However, the ratios between the errors for power of the oceanic diurnal tide is concentrated
separate amplitudes are properly conserved. around the K1 wave with period of 1 sidereal day,
whereas that of the atmospheric tide is near the S1 wave
with period of 1 mean solar day. That permits us to a
2.4 Corrections for the ocean tide certain extent to separate the e€ects of these two geo-
physical ¯uids.
Since the atmosphere is coupled with the ocean, it is of The dynamical processes in the ocean are not so ex-
interest to consider the atmospheric and oceanic e€ects tensively studied as those in the atmosphere and the
569

between the IERS 1996 and IAU 1980 precession/


nutation models. Our purpose is not to explain the VLBI
Dehant et al. observations since there exists a number of other
important contributions to nutation. The most up-to-
date summary of those e€ects can be found in Dehant
and Defraigne (1997). Our aim is to assess the role of
-7° Atmospheric prograde the combined atmospheric and oceanic contribution in
annual nutation the nutation budget. From the comparison we con-
-34° With IB correction
clude the following:

+1/3 year: the atmospheric contribution does not ex-


ceed a few las, the oceanic one was not accounted for
in our analysis, but the VLBI correction to the IAU
nutation theory is only about 20 las, which is com-
IAU prograde annual nutation parable to the present accuracy level of the VLBI
observations.
Fig. 2. Direction of the atmospheric prograde annual nutation: our +1/2 year: the oceanic e€ect is certainly dominant over
computation and the results of Dehant et al. (1996) deduced from the that of the atmosphere. It is by two orders of mag-
diurnal pressure torque nitude larger than the atmospheric one in case of the
in-phase term and one order of magnitude in case of
estimates of the oceanic tidal angular momentum the out-of-phase term. The combined atmosphere/
(OTAM) which exist in the literature are based on nu- ocean contribution reduces the observed VLBI cor-
merical models (Seiler 1991; Zhu et al. 1990) and most rection from ÿ606 and ‡510 to ÿ33 and ÿ25 las for
recently on the models validated by altimetric observa- the in-phase and out-of-phase components, respec-
tions (Chao et al. 1996). The diurnal tidal band is usu- tively. However, it should be noted that the prograde
ally represented in the ocean models by the K1 , O1 , and semiannual nutation is signi®cantly a€ected by other
P1 waves. In order to estimate other diurnal terms, Wahr geophysical e€ects, which are not taken into account
and Sasao (1981) derived the ocean admittance function, in the IAU 1980 precession/nutation theory. These
modi®ed later by Zhu et al. (1990) to account for the are the mantle anelasticity, non/hydrostatic mass re-
observed value of the FCN period. Dehant and De- distribution in the mantle and other, totally
fraigne (1997) computed the contributions of the OTAM amounting to ÿ110 las for the in-phase term and
to the largest nutation waves from the results obtained ÿ130 las for the out-of-phase one (Dehant and De-
by Chao et al. (1996) using the afore mentioned admit- fraigne 1997).
tance function. We cite their estimates in Table 3a. +1 year: in this case the atmospheric contribution ex-
Since the strongest gravitational tide is the K1 tide ceeds that of the ocean tide which agrees with the
then the ocean tide should cause a signi®cant constant remark made in Sect. 2.4. Accounting for the com-
pole o€set. Indeed, Gross (1993) found about 8 mas bined atmospheric and oceanic contribution reduces
(Table 3b), but this estimate should be considered as an the observed correction from ‡12 and ÿ124 las to ÿ8
upper bound since it probably contains also a term re- and ÿ18 las for the in-phase and out-of-phase terms,
sponsible for the 18.6 year nutation. However, this respectively. There is only a small, about 10 las,
oceanic contribution to the constant pole o€set could mantle anelasticity correction to the in-phase term
not be compared to observations since it is absorbed in (Dehant and Defraigne 1997).
the de®nition of the CEP, as had been pointed out by ) 1 year: in this case the atmospheric contribution is
Gross (1993). more than one order of magnitude smaller than the
From Table 3 one can judge that generally the at- VLBI correction, whereas the oceanic e€ect is op-
mospheric contribution to nutation is roughly one order posite in sign. Dehant and Defraigne (1997) recently
of magnitude weaker than the e€ect of the ocean tide explained the in-phase component by e€ects of
with the exception of the annual prograde component. mantle dynamics but there still remains an unex-
This con®rms the fact mentioned at the beginning of this plained correction of about ÿ0.3 mas in the out-of-
section indicating that the e€ects of those two geo- phase term. As it follows from that work, the solid
physical ¯uids on nutation could be, to some extent, Earth dynamics alone fails to account for this dis-
separated. crepancy. Thus, the ocean seems to be one of the
reasonable candidates. Note that the power of the w1
ocean is at the necessary level, hence the discrepancy
2.5 Comparison with the VLBI observations can be attributed to the error of phase. In fact, the
w1 OTAM component was estimated indirectly from
We compared in Table 3 the atmospheric contribution other tidal terms using an approximation of the
to the nutation amplitudes, after taking into account the oceanic admittance function, see Sect. 2.4 for details.
ocean tide e€ect, with the empirical corrections to the Thus, this correction may not be so certain as, e.g.,
IAU 1980 nutation model derived from VLBI observa- in the case of the P1 component. Additionally, as
tions. These corrections are computed here as di€erences proposed by Bu€ett (1992, 1993), the electromag-
570

netic core-mantle coupling could contribute to the our computations (not shown in the ®gures) is that the
out-of-phase term at the observable level. wind contributions are very stable in most cases, which
means that almost all variations shown in Fig. 3 come
Summarizing the foregoing description of combined
from the pressure term of the CEAM function.
atmosphere/ocean diurnal forcing we conclude that the
The prograde semiannual oscillation which is driven
unexplained part of the annual prograde term is
mainly by the wind term of v0 , has a very stable phase
dominated by the atmospheric and oceanic e€ects,
over the whole period, and quite stable amplitude with
which reduce the observed discrepancy to 15 las. This
only a small positive jump in 1978 and slightly higher
is not the case of the semiannual prograde nutation
values over 1978±1987 introduced by the pressure term.
where other geophysical e€ects have the same order of
In the case of the prograde annual term there can be
magnitude as the oceanic and atmospheric ones. Al-
observed a signi®cant discrepancy between the param-
though the observed discrepancy is diminished to about
eters estimated prior to and after 1979, with large
30 las by the oceanic/atmospheric corrections, the
variations (about 70 las) in the ®rst part and rather
mantle anelasticity and nonhydrostatic mass redistribu-
smooth behavior (variations less than 30 las) in the
tion corrections totally reach the level of 100 las. The
second part. The third periodical constituent shown in
retrograde annual nutation is by far more complicated
Fig. 3, the retrograde annual term, exhibits much larger
due to its high sensitivity to the solid-Earth geophysical
variations than the ®rst two (note the change of scale
modeling and to the possible inaccuracy of the oceanic
on vertical axes), up to almost 200 las, and has much
correction. Despite the mentioned problems, our esti-
bigger uncertainties. The reason is that the driving term
mation of the atmospheric prograde annual nutation
in the CEAM function is rather weak (see Fig. 1 and
receives an important observational con®rmation.
Tables 1 and 2) but it is multiplied by high values of
the transfer function due to the proximity of the FCN
3 Time variability of the atmospheric nutations resonance.
Finally we note that the constant celestial o€set of the
3.1 Sliding-window method applied to the CEAM function pole caused by the atmosphere becomes more and more
stable over the last 10 to 15 years which, as in the case of
Results of the MEM spectral analysis presented in the other parameters illustrated in Fig. 3, may re¯ect the
Sect. 2.2 suggest that some of the four circular oscilla- improvement in quality of the meteorological data.
tions detected in the CEAM function v0 , prograde and
retrograde annual, prograde semiannual and prograde
terannual, are not stable in that their amplitudes and 3.2 Comparison with the VLBI nutation data
phases can vary in time. This variability a€ects the
corresponding incremental nutations and makes their An important point is that the time variations of the
modeling dicult. The parameters of the circular terms nutation amplitudes, both of the in-phase and out-of-
and the ®rst-order polynomial, estimated in Section 2.3, phase terms, can be attributed solely to the geophysical
are average values over the period from 1979.0 to excitations. Hence, any positive correlation between
1997.3. In this section we investigate how these param- time variations of the atmospherically driven nutations,
eters change in time during the whole period from reported in Sect. 3.1, and of the VLBI observations of
1968.0 to 1997.3 covered by the atmospheric data. We the nutation angles, would be a valuable con®rmation of
®t, by least-squares analysis, the model of Eq. (6) with the reliability of both the CEAM data and our results. It
the linear term reduced to a constant (v0sl ˆ 0), to 3-year may also be possible to check whether the IB correction
sliding window of data. to the ocean response makes this correlation higher or
The estimated time variations of the model parame- lower with respect to the non-IB case. Such a compar-
ters are plotted in Fig. 3. The terannual term is not ative analysis will be made in the present section.
shown there because its amplitude is always below 5 las First, we estimated variability of the observed nut-
and negligible. A ®rst general observation from Fig. 3 is ations by the use of a similar method to that applied in
that the inverted barometer correction introduces sig- Sect. 3.1. The input data is the time-series of the celestial
ni®cant changes to di€erent terms of the data. It shows a pole o€sets dw sin e0 , de with respect to the IAU 1980
systematic shift of phase, ranging from ÿ100 to ÿ20 conventional nutation model, obtained from the VLBI
for the prograde annual term and about ÿ25 for the observations. In order to get more reliable results, we
prograde semiannual term, and is responsible for a analyzed two di€erent series submitted to the IERS
temporary phase jump in the retrograde annual term, (Charlot 1995): one produced by the Goddard Space
opposite in direction to that corresponding to the Flight Center (GSFC) obtained from the IERS Central
CEAM function without the IB correction. Note also a Bureau, courtesy of Dr. M. Feissel, the second from the
systematic di€erence between the constant pole o€sets US Naval Observatory (USNO) copied from the USNO
estimated in the IB and non-IB cases, about ‡50 las in ftp server. Our algorithm consists of two steps.
Dw sin e0 and ‡20 las for De (cf. also average di€erences The ®rst step is a preliminary ``cleaning'' of the series
in Table 3). The IB correction also makes the parameters from the incremental terms of nutation. We applied to
less variable and more smooth; one exception is the the entire input series the procedure developed by Bi-
phase of the prograde annual term with a rapid negative zouard (1995) in order to estimate by the least-squares
jump around 1975. A second general observation from method the in-phase and out-of-phase corrections to the
571

With IB correction

Phase 1/2 year prograde Phase 1 year prograde Phase 1 year retrograde
10 250

0
0
150

-10 -50
50
-20
-100 -50
-30

-40 -150 -150


1970 1980 1990 1970 1980 1990 1970 1980 1990

Amplitude 1/2 year prograde Amplitude 1 year prograde Amplitude 1 year retrograde
100 100 300

80 80
200

60 60
100
40 40
0
20 20

0 0 -100
1970 1980 1990 1970 1980 1990 1970 1980 1990

δψ sin ε, constant offset δε , constant offset


0 100

-50 50

-100 -0

-150 -50
1970 1980 1990 1970 1980 1990

Fig. 3. Variability of the atmospheric circular nutations and of the constant pole o€set, from a sliding-window estimation. Units are las for the
amplitude and the o€set, and degrees for the phase

largest 15 nutation constituents with the following pe- The second step is the sliding-window least-squares
riods: 18.6 years, 182.62 days, 13.66 days, 365.26 days, analysis assuming a model consisting of the annual
9.3 years, 27.55 days, 121.75 days, 13.63 days, 9.13 days, prograde and retrograde, prograde semiannual, pro-
31.81 days, 27.09 days, 177.84 days, 14.77 days, 9.12 grade terannual, and retrograde FCN terms. We
days, 9.56 days. Additionally we added to the model the adopted the window length of 5 years, which is a com-
linear function and the FCN oscillation assuming con- promise between the time resolution of the estimated
stant period and amplitude. Concerning the FCN peri- variations and the need to resolve two complex sinusoids
od, there are di€erences among di€erent empirical with close periods, the annual retrograde, and the FCN.
values (cf. Sect. 2.1). It is commonly believed that when Our results are compared to the atmospheric contribu-
using Tf0 ˆ 431 days instead of the theoretical value tions in Fig. 4. Note the di€erences with respect to the
Tf0 ˆ 460 days in the transfer function, there is the best atmospheric estimates shown in Fig. 3: (1) the window
agreement between the theoretical amplitudes and their length is increased from 3 to 5 years in order to have
values derived from the VLBI observations (e.g., Gwinn agreement with the VLBI data analysis, and (2) atmo-
et al. 1986). On the other hand, the direct estimation spheric contribution is presented as the in-phase and
from the celestial pole o€sets yields lower values, in out-of-phase amplitudes instead of the amplitude and
some cases below 420 days (BrzezinÂski 1998). In the phase with respect to the standard IAU argument. The
present estimation we tried di€erent values between 415 comparison is made for the nutation data after 1984
and 435 days. because earlier estimates are too noisy for our purpose.
572

CEAM contribution CEAM contribution amplified by factor 3


VLBI observations (USNO)
VLBI observations (GSFC)

1 year retrograde out-of-phase 1 year prograde out-of-phase 1 year prograde out-of-phase


50 100

200

100
-50 0

-100 -150 -100


1985 1990 1995 1985 1990 1995 1985 1990 1995
1 year retrograde in-phase 1 year prograde in-phase 1 year prograde in-phase
50 50

150

50
-50 -50

-50

-150 -150 -150


1985 1990 1995 1985 1990 1995 1985 1990 1995

Fig. 4. Time variations of the annual nutation estimated from two series of the VLBI observations and from the atmospheric CEAM functions.
Units are las

Only the annual terms of nutation are shown in Fig. 4 phase amplitude the correlation coecient is 0.75 over
because, according to the considerations of Sect. 3.1, the 1990±1995 with maximum 0.84 at lag ÿ0:5 years. In
atmospheric contribution is negligible in the case of the order to make a visual comparison easier, we applied to
prograde terannual nutation and does not vary signi®- the atmospheric annual prograde contributions the lin-
cantly with time in the case of the prograde semiannual ear transformation with the ®rst-order coecient equal
nutation. All the correlation coecients given in the to 3, and plotted results in the right diagrams. The
following are computed for the USNO time-series of the proximity of the atmospheric and VLBI curves over the
VLBI estimates. period 1990 to 1995 becomes almost perfect, and also
From Fig. 4 it can be seen that there are similarities rather good over some earlier time-intervals.
between the time variations of the observed nutation While the foregoing comparison seems to be very
amplitudes and those derived from the atmospheric ex- encouraging, one has to treat with caution the estimated
citation. Particularly striking is the case of the annual values of the correlation coecient, as well as the fact
retrograde in-phase term from 1989 to 1995 and the out- that lagging the amplitude estimates by 0:5 year can
of-phase term from 1990 to 1995. The correlation coef- increase the correlation, which contradicts the physics.
®cient between the atmospheric and VLBI curves is, in Note that the curves representing time variations of the
the ®rst case, 0.85 with maximum 0.88 at lag +0.5 years, nutation amplitudes are sampled once in 0:5 year, which
and in the second case, 0.71 with maximum 0.90 at lag gives only 14 values from 1989 to 1995, and this number
+0.5 years. In the case of the prograde annual terms is further reduced by lagging the series. Shortening the
there are similarities in shape but the atmospheric con- sampling step does not change the estimated numbers
tribution seems to be rather smooth in comparison to signi®cantly, and has no mathematical justi®cation be-
the observed variations. Surprisingly, we obtained for cause here the ``instantaneous'' annual amplitude is an
the out-of-phase amplitudes from 1989 to 1995 a cor- average value over the 5-year time-interval.
relation coecient which is even higher than for the One can ask if the transfer function expressed by
annual retrograde nutation, namely 0.97. For the in- Eq. (2) is not underestimated at the annual prograde
573

frequency. The only way to accomplish the linear semiannual and, to a lesser extent, prograde terannual.
transformation described is to multiply by 3 the pressure In addition, there is a nonnegligible atmospheric con-
response coecient ap in Eq. (3). Note, however, that tribution to the constant o€set of the pole. The
after this modi®cation also the variations of the retro- atmosphere is probably also the important excitation
grade annual amplitudes will be ampli®ed about 3 times source of the observed free core nutation, but this
and the agreement in the left-hand side diagrams be- problem will be a subject of the future study.
come worse, particularly for the in-phase term. Hence, Based on the spectral estimates we modeled obser-
there must be some frequency-dependent e€ect which is vations of the equatorial atmospheric angular momen-
not taken into account in our transfer function. One tum as a sum of four complex sinusoids with periods
possible candidate is the dynamical oceanic response for ÿ1 year, ‡1 year, ‡1=2 year, ‡1=3 year plus a ®rst-
the atmospheric pressure variations. order polynomial expressing zero-frequency peak in the
We did the same comparative analysis (not shown in spectrum. Coecients of the model were determined by
Fig. 4) using the pressure term of the CEAM function the standard least-squares algorithm and multiplied by
with the IB correction. All the correlations with the the theoretical transfer function in order to estimate the
VLBI data became lower, in some cases even negative. corresponding incremental nutations. Our results,
The only exception is the out-of-phase prograde annual shown in Tables 2, 3, and Fig. 2, do not correspond to
nutation where the correlation remains the same as for the entire data set, but only to its last 18 years. We
the non-IB case. We conclude that the IB correction is noticed that the atmospheric excitation data are much
not adequate for nearly diurnal variations. We should more stable since 1979, probably due to the substantial
use in this case either the original pressure term, as- noise reduction. Therefore after rejecting the earlier
suming the rigid ocean response, or look for a more data we expect to obtain more meaningful results. We
re®ned model expressing the dynamical ocean response found the following atmospheric contributions to the
for variable atmospheric loading. It is also worth noting nutation amplitudes: 77 las for prograde annual, 53 las
that the ocean does not react only to the atmospheric for retrograde annual, 45 las for prograde semiannual
pressure variations but also to the diurnal wind forcing. and only about 5 las for prograde terannual. In the ®rst
Even if the wind itself is, in general, less important than three cases these are small but observable e€ects which
the pressure, the wind can induce changes in sea level should be taken into account in the new theory of
that are relatively more important (J. Wahr, private nutation. We also determined a constant o€set of the
communication). For completeness, we should also celestial pole dw sin e0 ˆ ÿ86 las, de ˆ 77 las caused by
mention that both the pressure and the winds would the atmosphere. Only the prograde semiannual com-
cause ocean currents which could contribute to the ex- ponent of the model is driven mostly by the wind term
citation, though taking into account the equations of of the atmospheric angular momentum, while in all
Sect. 2.1 we do not expect that this e€ect is signi®cant. other cases the dominating contribution is from the
There is one more interesting observation concerning pressure term. Contrary to the gravitational nutation,
the oceanic e€ects. As has been stated in Sect. 3.1, the in all cases with the exception of the retrograde annual
prograde semiannual CEAM oscillation which is driven nutation, the out-of-phase term is much larger than the
mainly by the wind term of v0 , has a very stable phase in-phase one.
over the whole period, and quite stable amplitude, but An important problem from the point of view of
the corresponding nutation amplitude computed from modeling the atmospheric in¯uence on nutation is the
the VLBI pole o€sets (not shown in the ®gures) exhibits stability of the incremental amplitudes. We investigated
a rather strong variability, up to 0.1 mas. Therefore one this problem by performing the sliding-window least-
can conclude that the only agent which is most proba- squares analysis. A general observation from the anal-
bly responsible for such variations is the P1 ocean tide, ysis is that the wind contributions are stable in time.
and the variability of corresponding OTAM component That means almost all detected amplitude variations
can be as large as 20% of its amplitude. The foregoing shown in Fig. 3 are generated by the pressure term. A
reasoning may be considered indirect evidence of the striking feature is an anomalous behavior of the most
fact that the ocean response to the tesseral tidal po- important atmospheric contribution to nutation, the
tential is not strictly harmonic but rather is a band-pass prograde annual term, prior to 1979. The largest vari-
process. ability is found in the case of the retrograde annual
wave, up to almost 200 las, which can be explained by
the proximity of the FCN frequency of resonance: small
4 Summary and conclusions variations of the atmospheric excitation are ampli®ed
after multiplication by high values of the transfer func-
We used a 29-year time-series of the four-times-daily tion. If the estimated atmospherically driven variations
atmospheric angular momentum estimates obtained of amplitude of the retrograde annual nutation are real,
recently in a frame of the common NCEP/NCAR then we can conclude that the precision limit for the
reanalysis project, to study the atmospheric contribution classical models of nutation, assuming constant or sec-
to nutation. From the spectral analysis of the series (Fig. ularly changing in-phase and out-of-phase amplitudes, is
1 and Table 1) we found that the only nutation of the order of 0.1 mas.
constituents which are a€ected at an observable level The next step of the research was to introduce cor-
are the annual prograde and retrograde, prograde rections for the ocean tide in order to compare the at-
574

mospheric estimates with the VLBI determinations of magni®cation could only be accomplished in our algo-
the nutation angles. This comparison was an important rithm by increasing coecient ap of the transfer function
veri®cation of our results for the following reasons. The given by Eq. (2) by a factor of three, but this, in turn,
®rst one is that the atmospheric excitation series is would make the agreement of the annual retrograde
provided without information on the uncertainty in the nutation worse. We concluded that there is probably a
data. We know already much about the EAM accuracy frequency-dependent e€ect which was not taken into
at longer periods including seasonal variations, but the account, either in the atmospheric data reduction or in
data reliability at nearly diurnal periods is not known. the transfer function Eq. (2), for instance the dynamical
By use of an empirical procedure we assigned some response of the ocean to the atmospheric pressure vari-
uncertainties to the EAM data, but, as discussed in ations. Our comparison indicates that considerably
Sect. 2.3, the formal errors of our estimates of the at- better agreement with the VLBI data is obtained using
mospheric in¯uence on nutation can be signi®cantly the pressure term which is not corrected for the ocean
lower than their true values. The second reason is that response than after applying the IB correction. This is
the nutation transfer function derived by BrzezinÂski one more argument supporting the thesis that the IB
(1994) on the basis of the dynamical theory of Sasao and correction is not reliable at nearly diurnal frequencies,
Wahr (1981), may not be fully adequate to the problem. and we have to look for a more re®ned model expressing
In particular, it contains two coecients ap , aw ex- the dynamical response to the diurnal atmospheric
pressing sensitivity of the FCN mode to the pressure and pressure variations. Moreover, there can also be changes
wind excitations [see Eqs. (2) and (3)], which are purely in sea level that are driven by the diurnal wind forcing,
theoretical quantities and never received observational which are not negligible in the excitation balance of
con®rmation. Finally, as seen from Tables 1, 2, 3 and nutation. Another observation concerning the oceanic
Figs. 1, 3, there are signi®cant di€erences between re- e€ects is that the variability of the semiannual prograde
sults obtained from the pressure terms with and without nutation amplitude computed from the VLBI pole o€-
the IB correction. It appears that our results favor the sets indicates that the oceanic response to the tidal po-
non-IB oceanic response at the nearly diurnal frequen- tential is not strictly harmonic but is rather a band-pass
cies, unlike the case of the longer time scales, where the process.
IB-correction is proven to be valid.
The atmospheric contribution to nutation is generally Acknowledgements. This research has been supported by the Pol-
an order of magnitude weaker than the e€ect of the ish National Committee for Scienti®c Research (KBN) under
ocean tide with the exception of the annual prograde grants No. 9 S605 022 07 and No. 9 T12E 019 12. One of us (CB)
was ®nanced by a fellowship in the framework of bilateral French-
component (Table 3). In the latter case the in-phase Polish agreement for the period of his work at the SRC. We thank
atmospheric contribution is ÿ9 and ÿ76 las versus the Dr. David Salstein and Mr. Peter Nelson of the IERS Sub-bureau
oceanic corrections of ‡30 and ÿ30 las for the in-phase for Atmospheric Angular Momentum for providing the atmo-
and out-of-phase components, respectively. Accounting spheric angular momentum data. We are also indebted to Dr.
for the combined atmosphere ‡ ocean diurnal forcing VeÂronique Dehant, Dr. Dennis McCarthy, and Prof. John Wahr
explains 85% of the VLBI corrections to the IAU 1980 for their careful reviews which helped us to improve the paper.
amplitudes of the annual prograde component. The re-
maining discrepancy is 15 las, approximately the current
precision for the nutation amplitudes estimated from Appendix A
VLBI observations. In case of the prograde semiannual
and retrograde annual components, it is not possible to Seasonal modulation of the thermal atmospheric tide S1
separate the atmospheric/oceanic e€ects from the solid-
Earth geophysical contributions. At these frequencies The S1 diurnal tide of the atmospheric EAM function v
mantle dynamics plays an important role as shown by can be expressed in the terrestrial frame as
Dehant and Defraigne (1997). Thus, the comparison of
the atmospheric/oceanic nutations to the observations is v…t† ˆ AeÿiX…1ÿ1=366:26†t …11†
not conclusive.
where A is the complex amplitude expressing both the
After accepting the assumption that all variability of
amplitude and the phase of oscillation, X denotes
the observed nutation amplitudes, both of the in-phase
diurnal sidereal frequency and t stands for time.
and out-of-phase terms, can be attributed to the geo-
Equation (11) can be transformed to the celestial frame
physical forcing, we could learn much from the com-
by multiplying by the sidereal rotation factor eiXt which
parison of results obtained from the sliding-window
yields
least-squares analysis of the atmospheric excitation and
VLBI data. We found (Fig. 4) a satisfactory correlation v0 …t† ˆ AeiX=366:26t …12†
between the variability of both the in-phase and out-of-
phase amplitudes of the annual retrograde nutation after a prograde annual oscillation in the atmospheric CEAM
1989. Concerning the annual prograde nutation, there is function v0 . Intuitively speaking, the S1 tide is ®xed to
a good correlation in shape but the agreement in the size the Sun.
of variation was obtained ®rst after multiplying the at- Let us assume now that the amplitude A is not con-
mospheric contribution by the factor of 3. Such a stant but undergoes an elliptical annual modulation:
575

A ˆ A0 ‡ A‡ eiX=366:26t ‡ Aÿ eÿiX=366:26t …13† 2000, taking on values 0:5, 1:5, 2:5; . . . ; ‰MJDŠ
denotes the integer part of MJD, the coecients
After substituting into Eq. (11) we derive are d0 ˆ 6h 41m 50s :54841, d1 ˆ 8640184s :812866, d2 ˆ
0s :093104, d3 ˆ ÿ6s :2  10ÿ15 , and
v…t† ˆ Ao eÿiX…1ÿ1=366:26†t
‡ A‡ eÿiX…1ÿ2=366:26†t ‡ Aÿ eÿiXt …14† r ˆ 1:002737909350795
2
‡ 5:9006  10ÿ11 Tu0 ÿ 5:9  10ÿ15 Tu0 …20†
that is a sum of prograde annual, prograde semiannual
and zero-frequency terms, as seen from space. In a is the ratio of universal to sidereal time. As can be easily
conventional notation these are the S1 , P1 , and K1 tidal checked, an alternative but numerically equivalent form
diurnal waves in the terrestrial frame, respectively. of Eq. (19) can be obtained by putting r ˆ 1 and
In a similar way, assuming semiannual modulation of Tu0 ˆ …MJD ÿ 51544:5†=36525.
the amplitude: In practical computations we replaced the obliquity
eA by its mean value at J2000.0 e0 ˆ 8438100 :448 and
A ˆ A0 ‡ A‡ ei2X=366:26t ‡ Aÿ eÿi2X=366:26t …15† represented Dw only by the principal nutation term,
we derive which gives an accuracy better than 10ÿ5 with respect
to the full expression. Also, by neglecting the second-
v…t† ˆ A0 eÿiX…1ÿ1=366:26†t and third- order terms in the expansion given by Eq.
(19) and the ®rst and second-order terms in Eq. (20),
‡ A‡ eÿiX…1ÿ3=366:26†t ‡ Aÿ eÿiX…1‡1=366:26†t …16† we introduce an error which does not exceed 10ÿ6 in 30
This is a sum of prograde annual, prograde terannual years.
and retrograde annual terms, as seen from space,
corresponding to the S1 , p1 , and w1 tidal diurnal waves Appendix C
in the terrestrial frame, respectively.
Representation of nutations

Appendix B The nutational motion of the CEP axis is classically


described in terms of two angles: nutation in longitude
Computation of the CEAM function v0 Dw, that is displacement in the ecliptic plane of date, and
nutation in obliquity De, that is variation of inclination
The CEAM function is de®ned by with respect to the ecliptic plane of date. These two
angles are represented in the IAU 1980 conventional
v0 ˆ v01 ‡ iv02 ˆ ÿeiU …v1 ‡ iv2 † …17† model by a sum of 106 sinusoidal terms
0p
and similar expressions for the pressure term v and for X
106 X
106
the wind term v0 w (BrzezinÂski 1994; BrzezinÂski and Dw ˆ Dwj sin hj ; De ˆ Dej cos hj …21†
Capitaine 1993). In order to evaluate v0 , the Greenwich jˆ1 jˆ1
sidereal time U has to be calculated for each date of the
four-times-daily EAM series. Each record of the EAM (Wahr 1981; Seidelmann 1982; McCarthy 1996, Chap.
series contains the date in the form: calendar year, 5), where the amplitudes Dwj and Dej are constants with
month and hour which is 0, 6, 12, and 18 hours GMT. a small time-dependent contribution modeled as a ®rst-
First, this civil date was expressed in Modi®ed Julian order polynomial of time, the arguments hj are integer
Days (MJD), and then we applied the standard proce- combinations of the ®ve fundamental arguments
dure of computing U, described, e.g., in Chapter 5 of the (Brown's arguments) describing mutual celestial motion
IERS Conventions (McCarthy 1996): of the Moon and the Sun, each of them being expressed
as a fourth-degree polynomial of time. The nutation
U ˆ k  GMST ‡ Dw sin eA …18† arguments can be expressed as
where eA denotes the obliquity, Dw is nutation in hj …t† ˆ kj t ‡ hj …t† …22†
longitude, k ˆ 2p=86400 is the conversion factor from
the sidereal rotation angle in time seconds to radians, where hj …t† does not di€er from a constant by more than
and GMST is the Greenwich mean sidereal time given by 30 in 30 years. Hence, to a good approximation hj can be
the expression considered as linear functions of time and Eq. (21) as a
standard harmonic expansion.
GMST ˆ r… MJD ÿ ‰MJDŠ†86400 In more re®ned models taking into account inelas-
2
‡ d0 ‡ d1 Tu0 ‡ d2 Tu0 ‡ d3 Tu0
3
…19† ticity or atmospheric and oceanic e€ects, it is necessary
to add to Eq. (21) the so-called out-of-phase terms. The
in which Tu0 ˆ du0 =36525, and du0 ˆ ‰MJDŠ ÿ 51544:5 is the IERS 1996 model of precession/nutation (McCarthy
number of days elapsed since 12 hours UT1, 1 January 1996, Chap. 5), e.g., takes the following form
576

263 
X  and ``ÿ'' in aj correspond only to the sign of hj in the
Dw ˆ Dwip op
j sin hj ‡ Dwj cos hj …23† complex sinusoid, but do not de®ne whether the nutat-
jˆ1 ion is prograde or retrograde. The reason is an ambi-
263 
X  guity in adoption of hj which can always be replaced by
De ˆ Deip op
j cos hj ‡ Dej sin hj …24† ÿhj with simultaneous appropriate change of sign of the
jˆ1 nutation amplitude; see discussion preceding Eq. (30).
From the comparison of Eq. (27) with Eqs. (23) and
where the amplitudes Dwip ip
j and Dej of the in-phase (24), and using Eq. (25) we derive
terms, corresponding to Dwj and Dej in Eq. (21), are ÿ
linear functions of time, and Dwop op ip ip 
j , Dej are the ampli- a‡ 1
j ˆ2 ÿDwj sin e0 ‡ Dej
tudes of the out-of-phase terms, which are assumed to be ÿ op 
constant. This expression for the lunisolar e€ect is ‡ i12 ÿDwop j sin e0 ÿ Dej …28†
ÿ 
further completed by adding a similar sum expressing aÿ 1 ip ip
j ˆ2 Dwj sin e0 ‡ Dej
the planetary in¯uence which is much less important ÿ op 
than the lunisolar one and will not be considered here. ‡ i12 ÿDwop j sin e0 ‡ Dej …29†
The actual precession/nutation is a sum of the model
and the residuals dw, de which are routinely determined Equations (28) and (29) relate amplitudes a‡ ÿ
j , aj of the
by the VLBI technique. complex nutation representation of Eq. (27) to the
A convenient way of representing the nutational amplitudes of the classical expansions given by Eq. (21)
motion of the CEP, which is a two-dimensional process, and Eqs. (23), (24) for Dw, De. By changing the phase by
p
is to express it by the complex variable 2 with respect to hj in Eq. (27) we see that the real
parts of a‡ ÿ
j , aj are only due to the in-phase terms, while
P ˆ Dw sin e0 ‡ iDe ˆ X ‡ iY …25† the imaginary parts correspond to the out-of-phase
terms. Hence, when applying the IAU 1980 model of
that is a combination of the X and Y components of the nutation expressed by Eq. (21), a‡ ÿ
j and aj are real for all
CEP in the mean equatorial frame which is right-handed j, while for the IERS 1996 model the imaginary parts of
in the sense that its Y -axis is directed 90 E with respect a‡ ÿ
j and aj become di€erent from zero but are still much
to the X -axis, and e0 ˆ 8438100 :448 is the mean obliquity smaller than the real parts. In the case of the atmo-
at the reference epoch J2000.0 [kinematical details spheric nutations the out-of-phase contribution can be
concerning Eq. (25) are given by BrzezinÂski and Capit- signi®cant or even dominant.
aine (1993)]. Let us recall here the well-known fact that Eqs. (21)
Any single harmonic term in the complex plane XY and (23), (24) are ambiguous representations of nutation
can be expressed by the complex sinusoid in the sense that any hj can be replaced by ÿhj with a
simultaneous change of Dwip op ip
j , Dej for ÿDwj , ÿDej ,
op
z…t† ˆ z0 eirt …26† respectively. As an example, one can compare the an-
in which r denotes frequency, z0 ˆ reia represents the nual term in the IAU 1980 theory of nutation in the
initial value at t ˆ 0, r ˆ jzj is the radius and a expresses original formulation (Seidelmann 1982) and in that
the initial phase. Equation (26) describes circular presented recently in the IERS Conventions (McCarthy
motion, prograde (counterclockwise) for r > 0 and 1996). Fortunately, Eq. (27) is not sensitive to this
retrograde (clockwise) for r < 0. We also often speak change of the sign convention because, according to
about prograde and retrograde frequencies in sense the Eqs. (28) and (29), it causes an interchange of a‡ ÿ
j with aj
already described above. Each nutation represented by which results only in changing the order of the two
Eq. (25) is an elliptical motion [assuming hj …t† ˆ const: complex sinusoidal terms in Eq. (27).
in Eq. (25), which is almost the case] which can be Let us consider now a single circular term of nutation
expressed as the following sum of two circular motions 

with opposite frequencies Pj …t† ˆ iaj eisj hj ˆ iaj eisj …kj t‡hj † …30†

Pj …t† ˆ a‡ i‰hj …t†‡p2Š


‡ aÿ ÿi‰hj …t†ÿp2Š where sj ˆ 1 or sj ˆ ÿ1. Its angular frequency is sj kj
j e j e
  and, in the case of the IAU 1980 model (aj real), its
ˆ i a‡j e
ihj …t†
‡ aÿj e
ÿihj …t†
…27† phase di€erence with respect to the argument sj hj is
either 90 , for aj > 0, or ÿ90 , for aj < 0. For the
(Bizouard 1995), where a‡ ÿ
j , aj are the amplitudes
models taking into account the out-of-phase terms (e.g.,
(complex, in general). Any constant change of phase the IERS 1996 model of precession/nutation) the
with respect to hj is allowed in Eq. (27), provided it is deviations of phase from 90 are very small. Physically
compensated by an appropriate change of the coe- this phase di€erence re¯ects the well-known fact that
cients a p
j . The reason for choosing 2 will be seen from
nutational displacement is always shifted in phase by 90
Eqs. (28) and (29) below. from the driving tidal torque.
When kj > 0 in hj [see Eq. (22)] then the ®rst term of The atmospherically driven circular nutation is ex-
Eq. (27) represents the prograde nutation and the second pressed as
term represents the retrograde nutation, while for kj < 0
the situation is reversed. Note that the superscripts ``‡'' Pja …t† ˆ Aj ei…sj hj ‡uj † …31†
577

where the amplitude Aj is a real and positive number. Its Furuya M, Hamano Y, Naito I (1996) Quasi-periodic wind signal
phase di€erence with respect to the argument sj hj is uj as a possible excitation of the Chandler wobble. J Geophys Res
101: 25 537±25 546
and the phase di€erence with respect to the IAU model
Gross RS (1993) The e€ect of ocean tides on the Earth's rotation as
is either uj ‡ 90 or uj ÿ 90 . predicted by the results of an ocean tide model. Geophys Res
Lett 20: 293±296
Gwinn CR, Herring TA, Shapiro II (1986) Geodesy by radio in-
References terferometry: studies of the forced nutations of the Earth 2.
Interpretation. J Geophys Res 91: 4755±4765
Barnes RTH, Hide R, White AA, Wilson CA (1983) Atmospheric Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin
angular momentum ¯uctuations, length-of-day changes and L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Chelliah M,
polar motion. Proc R Soc London A 387: 31±73 Ebisuzaki W, Higgins W, Janowiak J, Mo K C, Ropelewski C,
Bingham C, Godfrey MD, Tukey JW (1967) Modern techniques of Wang J, Leetmaa A, Reynolds R, Jenne R, Joseph D (1996) The
power spectrum estimation. IEEE Trans Audio Electroacoust NMC/NCAR 40-year reanalysis project. Bull Am Meteoro Soc
15: 56±66; reprinted In: Childers DG (ed) (1978) Modern 77: 437±471
spectrum analysis. IEEE Press, New York McCarthy DD (ed) (1996) IERS Conventions. IERS Tech Note 21,
Bizouard C (1995) ModeÂlisation astromeÂtrique et geÂophysique de Observatoire de Paris
la rotation de la Terre. PhD thesis, Observatoire de Paris McCarthy DD, Dehant V (1995) The IAU nutation theory and
BrzezinÂski A (1987) Statistical investigations on atmospheric an- perspectives of its change, In: Appenzeller I (ed) Highlights of
gular momentum functions and on their e€ects on polar motion. astronomy. Kluwer Academic Dordrecht, pp 247±248
Manuscr Geod 12: 268±281 Ponte RM, Salstein DA, Rosen RD (1991) Sea level response to
BrzezinÂski A (1994) Polar motion excitation by variations of the pressure forcing in a barotropic numerical model. J Phys
e€ective angular momentum function, II: extended model. Oceanogr 21: 1043±1057
Manuscr Geod 19: 157±171 Salstein DA (1995) Atmospheric Angular Momentum (AAM) and
BrzezinÂski A (1995) On the interpretation of maximum entropy Earth rotation, In: Capitaine N, Kolaczek B, DeÂbarbat S (eds)
power spectrum and cross-power spectrum in earth rotation Proc journeÂes systeÁmes de reÂfeÂrence spatio-temporels 1995.
investigations. Manuscr Geod 20: 248±264 Space Research Centre, Warsaw, pp 139±146
BrzezinÂski A (1998) Estimation of the free core nutation parame- Salstein DA, Rosen D (1997) Global momentum and energy sig-
ters from the observed celestial motion of the pole, In prepa- nals from reanalysis systems. In: Lamb PJ (ed) Proc 7th Conf
ration climate variations, February 1997, American Meterological
BrzezinÂski A, Capitaine N (1993) The use of the precise observa- Society, Boston USA, pp 344±348
tions of the Celestial Ephemeris Pole in the analysis of geo- Salstein DA, Kann DM, Miller AJ, Rosen RD (1993) The sub-
physical excitation of earth rotation. J Geophys Res 98: 6667± bureau for atmospheric angular momentum of the International
6675 Earth Rotation Service: a meteorological data center with
Bu€ett BA (1992) Constraints on magnetic energy and mantle geodetic applications, Bull Am Meteorol Soc 10: 67±80
conductivity from the forced nutations of Earth. J Geophys Res Sasao T, Wahr JM (1981) An excitation mechanism for the free
97: 19 581±19 597 `core nutation', Geophys J R Astron Soc. 64, 729±746.
Bu€ett BA (1993) In¯uence of a toroidal magnetic ®eld on the Seidelmann PK (1982) 1980 IAU theory of nutation: the ®nal re-
nutations of Earth. J Geophys Res 98: 2105±2117 port of the IAU working group on nutation Celest Mech 27: 79±
Chao BF, Ray RD, Gipson JM, Ma C (1996) Diurnal/semidiurnal 106
polar motion excited by oceanic tidal angular momentum. J Seiler U (1991) Periodic changes of the angular momentum budget
Geophys Res 101: 20 151±20 163 due to the tides of the world ocean. J Geophys Res 96: 10287±
Chapman S, Lindzen RS (1970) Atmospheric Tides: Thermal and 10300
Gravitational. D Reidel, Dordrecht Viron O de, Bizouard C, Dehant V, Salstein D (1998) E€ect of the
Charlot P (ed) (1995) Earth orientation, reference frames and at- atmosphere on the Earth rotation by the torque approach with
mospheric excitation functions submitted for the 1994 IERS numerical application to the Earth's nutation. In preparation
Annual Report IERS Tech Note 19, Observatoire de Paris Volland H (1988) Atmospheric tidal and planetary waves. Kluwer
Dehant V, Defraigne P (1997) New transfer functions for nutations Academic, Dordrecht
of a non-rigid Earth. J Geophys Res 102: 27,659±27,688 Wahr JM (1981) The forced nutations of an elliptical, rotating,
Dehant V, Bizouard C, Hinderer J, Legros H, Gre€-Le€tz M, elastic and oceanless earth. Geophys J R Astr Soc 64: 705±727
(1996). On atmospheric pressure perturbations on precession Wahr JM (1983) The e€ects of the atmosphere and oceans on the
and nutations, Phys Earth and Planet Inter 96: 25±39 Earth's wobble and on the seasonal variations in the length of
Eubanks TM, Steppe JA, Sovers OJ (1986) An analysis intercom- day ± II. Results. Geophys J R Astron Soc 74: 451±487
parison of VLBI nutation estimates. In: Mueller II (ed) Proc Int Wahr JM, Sasao T (1981) A diurnal resonance in the ocean tide
Conf Earth rotation and the terrestrial reference frame, Co- and in the Earth's load response due to the resonant free ``core
lumbus, Ohio. IAU Publ vol. 1, pp 326±340 nutation''. Geophys J R Astr Soc 64: 747±765
Eubanks TM, Steppe JA, Dickey JO, Rosen RD, Salstein DA Zharov VE, Gambis D (1996) Atmospheric tides and rotation of
(1988) Causes of rapid motions of the Earth's pole. Nature 334: the Earth. J Geod 70: 321±326
115±119 Zhu SY, Groten E, Reigber C (1990) Various aspects of numerical
Feissel M, Lewandowski W (1984) A comparative analysis of determination of nutation constants II. An improved nutation
Vondrak and Gaussian smoothing techniques. Bull Geod 58: series for the deformable Earth. Astron J, 99: 1024±1044
464±474

You might also like