[desrotour17] Atmospheric tides in Earth-like planets

A&A 603, A107 (2017)
DOI: 10.1051/0004-6361/201628252 Astronomy

© ESO 2017 &
Astrophysics
Atmospheric tides in Earth-like planets

P. Auclair-Desrotour1, 2 , J. Laskar1 , and S. Mathis2, 3
1
IMCCE, Observatoire de Paris, CNRS UMR 8028, PSL, 77 Avenue Denfert-Rochereau, 75014 Paris, France
e-mail: [email protected]
2
Laboratoire AIM Paris-Saclay, CEA/DSM – CNRS – Université Paris Diderot, IRFU/SAp Centre de Saclay, 91191 Gif-sur-Yvette,
France
e-mail: [email protected]
3
LESIA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universités, UPMC Univ. Paris 06, Univ. Paris Diderot,
Sorbonne Paris Cité, 5 place Jules Janssen, 92195 Meudon, France
Received 4 February 2016 / Accepted 8 November 2016
ABSTRACT
Context. Atmospheric tides can strongly affect the rotational dynamics of planets. In the family of Earth-like planets, which includes
Venus, this physical mechanism coupled with solid tides makes the angular velocity evolve over long timescales and determines the
equilibrium configurations of their spin.
Aims. Unlike the solid core, the atmosphere of a planet is subject to both tidal gravitational potential and insolation flux coming from
the star. The complex response of the gas is intrinsically linked to its physical properties. This dependence has to be characterized and
quantified for application to the wide variety of extrasolar planetary systems.
Methods. We develop a theoretical global model where radiative losses, which are predominant in slowly rotating atmospheres,
are taken into account. We analytically compute the perturbation of pressure, density, temperature, and velocity field caused by a
thermogravitational tidal perturbation. From these quantities, we deduce the expressions of atmospheric Love numbers and tidal
torque exerted on the fluid shell by the star. The equations are written for the general case of a thick envelope and the simplified one
of a thin isothermal atmosphere.
Results. The dynamics of atmospheric tides depends on the frequency regime of the tidal perturbation: the thermal regime near
synchronization and the dynamical regime characterizing fast-rotating planets. Gravitational and thermal perturbations imply different
responses of the fluid, i.e. gravitational tides and thermal tides, which are clearly identified. The dependence of the torque on the tidal
frequency is quantified using the analytic expressions of the model for Earth-like and Venus-like exoplanets and is in good agreement
with the results given by global climate models (GCM) simulations. Introducing dissipative processes such as radiation regularizes
the tidal response of the atmosphere, otherwise it is singular at synchronization.
Conclusions. We demonstrate the important role played by the physical and dynamical properties of a super-Earth atmosphere (e.g.
Coriolis, stratification, basic pressure, density, temperature, radiative emission) in its response to a tidal perturbation. We point out
the key parameters defining tidal regimes (e.g. inertia, Brunt-Väisälä, radiative frequencies, tidal frequency) and characterize the
behaviour of the fluid shell in the dissipative regime, which cannot be studied without considering the radiative losses.
Key words. hydrodynamics – waves – planets and satellites: atmospheres – planets and satellites: dynamical evolution and stability
1. Introduction as well as solid tides, and the energy dissipated is even

more dependent on the tidal frequency than the solid tides
Since the pioneering work of Kelvin (1863) on the gravi- are because of a highly resonant behaviour (Ogilvie & Lin
tationally forced elongation of a solid spherical shell, tides 2004; Ogilvie 2014; Auclair Desrotour et al. 2015). As a con-
have been recognized as a phenomenon of major impor- sequence, in both cases the phase shifted elongation of
tance in celestial mechanics. They are actually one of the the shell induces a tidal torque which modifies the rota-
key mechanisms involved in the evolution of planetary sys- tional dynamics of the body. This torque drives the evolu-
tems over long timescales. Because they affect both stars tion of the spin of planets and determines its possible states
and planets, tides introduce a complex dynamical coupling of equilibrium (Gold & Soter 1969; Dobrovolskis & Ingersoll
between all the bodies that compose a system. This cou- 1980; Correia & Laskar 2001; Correia et al. 2003, 2014;
pling results from the mutual interactions of bodies and is Correia & Laskar 2003; Arras & Socrates 2010).
tightly bound to their internal physics. Solid tides have long The increased focus on exoplanets over the past two decades
been the subject of studies (see the reference works by Love and the rapidly increasing number of discovered planetary sys-
1911; Goldreich & Soter 1966). The corresponding tidal dis- tems (Mayor & Queloz 1995; Perryman 2011) have brought to
sipation, which is caused by viscous friction, can have a light a huge diversity of existing orbital structures that can be
strong impact on the orbital elements of a planet (e.g. Mignard very different from our own solar system (Fabrycky et al. 2012).
1979, 1980; Hut 1980, 1981; Neron de Surgy & Laskar 1997; Therefore, tidal effects now need to be studied in a general con-
Henning et al. 2009; Remus et al. 2012). This evolution de- text to characterize these new systems, understand their orbital
pends smoothly on the tidal frequency (e.g. Efroimsky & Lainey and rotational dynamics, and constrain the physics of exoplanets
2007; Greenberg 2009). Fluid tides cause internal dissipation with the information provided by observational measurements.
Article published by EDP Sciences A107, page 1 of 44
A&A 603, A107 (2017)
In this work, we focus on Earth-like exoplanets, which are at synchronization. In these models, the amplitude of perturbed
interesting for their complex internal structure and possible hab- quantities like pressure, density, and temperature diverges at
itability. Indeed, a planet such as the Earth or Venus is composed this point. The case of Venus was studied by Correia & Laskar
of several solid and fluid layers with different physical proper- (2001), who proposed an empirical bi-parameter model for the
ties, which implies that the tidal response of a layer cannot be tidal torque describing two possible regimes: the regime of an
simply correlated to its size. For instance, the Earth’s oceans, in atmosphere where dissipative processes can be ignored, far from
spite of their very small depth compared to the Earth’s radius, synchronization, and its supposed regime at the vicinity of syn-
are responsible for most of the tidal dissipation of the planet chronization, where the perturbation is expected to annihilate it-
(Webb 1980; Egbert & Ray 2000; Ray et al. 2001). The atmo- self. This behaviour was recently retrieved with global climate
spheric envelope of a terrestrial exoplanet sometimes represents models (GCM) simulations by Leconte et al. (2015), who took
a non-negligible fraction of its mass (about 20% in the case of into account dissipative mechanisms, thus showing that these
small Neptune-like bodies). It is also a layer of great complexity mechanisms drive the response of planetary atmospheres in the
because it is submitted to both tidal gravitational potentials due neighbourhood of synchronization.
to other bodies and thermal forcing of the host star. Therefore, it Our objective in this work is thus to propose a new theo-
is necessary to develop theoretical models of atmospheric tides retical global model controlled by a small number of physical
with the smallest possible number of parameters. These models parameters which could be used to characterize the response of
will be useful tools to explore the domain of parameters, and any exoplanet atmosphere to a general tidal perturbation. We aim
to quantify the tidal torque exerted on the spin axis of a planet to answer the following questions:
and quantities used in the orbital evolution of planetary systems.
– How does the amplitude of pressure, density, temperature,
One of these quantities is the second-order Love number (k2 ) in-
and velocity oscillations depend on the parameters of the sys-
troduced by Love (1911), which measures the quadrupolar hy-
tem (i.e. rotation, stratification, radiative losses)?
drostatic elongation. The adiabatic Love number k2 has been
– What are the possible regimes for atmospheric tides?
estimated in the solar system (e.g. Konopliv & Yoder 1996;
– How do Love numbers and tidal torques vary with tidal
Lainey et al. 2007; Williams et al. 2014).
frequency?
According to Wilkes (1949), the first theoretical global mod-
els of atmospheric tides were developed at the end of the eigh- First, in Sect. 2, we derive the general theory for thick atmo-
teenth century by Laplace, who published them in Mécanique spheres, which are characterized by a depth comparable to the
céleste (Laplace 1798). Laplace was interested in the case of a radius of the planet. In this framework, we expand the perturbed
homogeneous atmosphere characterized by a constant pressure quantities in Fourier series in time and longitude, and in se-
scale height and affected by tidal gravitational perturbations. He ries of Hough functions (Hough 1898) in latitude. This allows
thus founded the classical theory of atmospheric tides. At the us to compute the equations giving the vertical structure of the
end of the nineteenth century, Lord Kelvin pointed out the un- atmospheric response to the tidal perturbation. Then, from the
expected importance of the Earth’s semidiurnal pressure oscilla- hydrodynamics we derive the analytic expressions of the vari-
tions in spite of the predominating diurnal perturbation (Kelvin ations of pressure, density, temperature, velocity field, and dis-
1882; Hagan et al. 2003; Covey et al. 2009, see also the mea- placement for any mode. At the end of the section, the density
sures of the daily variation of temperature and pressure, Fig. 13). oscillations obtained are used to compute the atmospheric tidal
This question motivated many further studies (e.g. Lamb 1911, gravitational potential, the Love numbers, and the tidal torque
1932; Taylor 1929, 1936; Pekeris 1937) which contributed to exerted on the spin axis of the planet. In Sect. 3, we apply the
enriching the classical theory of tides. In the 1960s, theoretical equations from Sect. 2 to the case of a thin isothermal stably
models were generalized in order to study other modes, such stratified atmosphere. In this simplified case, the terms associ-
as diurnal oscillations (see Haurwitz 1965; Kato 1966; Lindzen ated with the curvature of the planet disappear and most of the
1966, 1967a,b, 1968), but still remained focused on the case parameters depending on the altitude, such as the pressure scale
of the Earth. They revealed the impact of thermal tides due to height of the atmosphere, become constant. In Sect. 4, we treat
solar insolation by showing that these tides drive the Earth’s the case of slowly rotating convective atmospheres, such as near
diurnal and semi-diurnal waves. Indeed, in the Earth atmo- the ground in Venus (Marov et al. 1973; Seiff et al. 1980). In the
sphere, the contribution of the tidal gravitational potential, which next section, we derive the terms of thermal forcing from the
causes the elongation of solid layers, is negligible compared to Beer-Lambert law (Bouguer 1729; Klett et al. 1760; Beer 1852)
the effect of the thermal forcing. A very detailed review of the and the theory of temperature oscillations at the planetary sur-
general theory of thermal tides and of its history is given by face. In Sect. 6, the model is applied to the Earth and Venus. The
Chapman & Lindzen (1970), hereafter CL70 (see also Wilkes corresponding torque and atmospheric Love numbers are com-
1949; Siebert 1961). puted from analytical equations. We end the study with a discus-
Leaning on this early work, here we generalize the exist- sion and by giving our conclusions and prospects. To facilitate
ing theory to Earth-like planets and exoplanets. We introduce the reading, several technical issues have been deferred to the
the mechanisms of radiative losses and thermal diffusion, which Appendix where an index of notations is also provided.
are usually not taken into account in geophysical models of at-
mospheric tides owing to their negligible effects on the Earth’s 2. Dynamics of a thick atmosphere forced thermally
tidal response (radiation was, however, introduced with a Newto- and gravitationally
nian cooling by Lindzen & McKenzie 1967; Dickinson & Geller
1968, which inspired the present work). These mechanisms can- The reference book CL70 has set the basis of analytical ap-
not be ignored because they play a crucial role in the case proaches for atmospheric tides. This pioneering work focuses on
of planets like Venus, which are very close to spin-orbit syn- fast rotating telluric planets covered by a thin atmosphere, such
chronization. Models based on perfect fluid hydrodynamics, as the Earth.
such as the one detailed by CL70, fail to describe the atmo- The goal of the present section is to generalise this work.
spheric tidal response of such planets because they are singular We establish the equations that govern the dynamics of tides in
A107, page 2 of 44
P. Auclair-Desrotour et al.: Atmospheric tides in Earth-like planets
stand for the Earth and Venus, respectively, which are taken as
examples to illustrate the theory. All functions and fields dis-
played in the figures are computed with the algebraic manipula-
tor TRIP (Gastineau & Laskar 2014) and plotted with gnuplot.
2.1. Forced dynamics equations

The atmosphere is affected by the tidal gravitational potential U
and the heat power per mass unit J. In this approach, the equa-
tions of dynamics are linearized around the equilibrium state.
The small variations in this state induced by the tidal perturba-
tion are the velocity field V “ Vθ eθ ` Vϕ eϕ ` Vr er ; the corre-
sponding displacement field ξ “ ξθ eθ ` ξϕ eϕ ` ξr er defined as
Bξ
V“ , (3)
Bt
and the pressure (δp), density (δρ), and temperature (δT ) fluctu-
ations where
Fig. 1. Spherical coordinate system associated with the equatorial refer-
ence frame of the planet, and the geometrical parameters of the system. p “ p0 ` δp, ρ “ ρ0 ` δρ, T “ T 0 ` δT. (4)
This represents six unknown quantities to compute (Vr , Vθ , Vϕ ,
a thick fluid shell in the whole range of tidal frequencies, from δp, δρ, δT ), and we therefore solve a six-equation system. We
synchronization to fast rotation. We consider a spherical telluric first introduce the Navier-Stokes equation. We assume the Cowl-
planet of radius R covered by a stratified atmospheric layer of ing approximation in which the self-gravitational potential fluc-
typical thickness Hatm . The atmosphere rotates uniformly with tuations induced by tides are neglected (Cowling 1941). It is well
the rocky part at the angular velocity Ω (the spin vector of the adapted to waves characterized by a high vertical wavenumber as
planet being denoted Ω). Therefore, the dynamics are written in are found in the vicinity of synchronization, as detailed farther.
the natural equatorial reference frame rotating with the planet, In the equatorial reference frame RE;T , the linearized Navier-
RE;T : tO, XE , Y E , Z E u, where O is the centre of the planet, XE Stokes equation is thus written
and Y E define the equatorial plane and Z E “ Ω{ |Ω|. We use the
BV 1 g
spherical basis per , eθ , eϕ q and coordinates pr, θ, ϕq, where r is the ` 2Ω ^ V “ ´ ∇p ´ δρ er ´ ∇U, (5)
radial coordinate, θ the colatitude, and ϕ the longitude. As usual, Bt ρ0 ρ0
t stands for the time. which, projected on er , eθ , and eϕ , gives
The atmosphere is assumed to be a perfect gas homoge-
1 B δp
ˆ ˙
neous in composition, of molar mass M, and stratified radially. BVθ
Its pressure, density, and temperature are denoted p, ρ, and T ´ 2Ω cos θVϕ “ ´ Ù , (6)
Bt r Bθ ρ0
respectively. For the sake of simplicity, all dissipative mecha-
nisms (e.g. viscous friction and heat diffusion) are ignored ex-
δp
ˆ ˙
cept for the radiative losses of the gas, which play an impor- BVϕ 1 B
` 2Ω cos θVθ ` 2Ω sin θVr “ ´ ` U , (7)
tant role in the vicinity of synchronization. Tides are considered Bt r sin θ Bϕ ρ0
to be a small perturbation around a global static equilibrium
state. The basic pressure p0 , density ρ0 , and temperature T 0 , are
BVr 1 Bδp δρ BU
supposed to vary only with the radial coordinate. The tidal re- ´ 2Ω sin θVϕ “ ´ ´g ´ ¨ (8)
sponse is characterized by a combination of inertial waves, of Bt ρ0 Br ρ0 Br
typical frequency 2Ω, due to Coriolis acceleration, and gravity Equations (6)–(8) are coupled by the Coriolis terms (character-
waves which are restored by the stable stratification. The typi- ized by the factor 2Ω) on the left-hand side. To simplify the
cal frequency of these waves, the Brunt-Väisälä frequency (e.g. equations, we assume the traditional approximation, i.e. to ne-
Lighthill 1978; Gerkema & Zimmerman 2008), is given by glect the terms 2Ω sin θVr and 2Ω sin θVϕ in Eqs. (7) and (8),
„
1 dp0 1 dρ0
 respectively. This hypothesis, commonly used in literature deal-
N “g
2
´ , (1) ing with planetary atmospheres and star hydrodynamics (e.g.
Γ1 p0 dr ρ0 dr Eckart 1960; Mathis et al. 2008), can be applied to stratified flu-
where Γ1 “ pB ln p0 {B ln ρ0 qS is the adiabatic exponent (the in- ids where the tidal flow satisfies two conditions: (i) the buoyancy
dex S being the specific macroscopic entropy) and g the gravity. (given by gδρ{ρ0 er in Eq. (5)) is strong compared to the Coriolis
We assume that the fluid follows the perfect gas law, term (2Ω ^ V) and (ii) the tidal frequency (introduced below)
is greater than the inertia frequency (2Ω) and smaller than the
p “ R s ρT, (2) Brunt-Väisälä frequency. These conditions are satisfied by the
Earth’s atmosphere and fast rotating planets. The quantitative
where R s “ RGP {M is the specific gas constant of the atmo- precision that can be expected with the traditional approximation
sphere, RGP “ 8.3144621 J mol´1 K´1 (Mohr et al. 2012) be- may be lower in the vicinity of synchronization where the tidal
ing the molar gas constant. In the following, the notations <, frequency tends to zero; moreover, we note that it leads to issues
=, and ˚ are used to designate the real part, imaginary part, and with momentum and energy conservation in the case of deep at-
conjugate of a complex number. The subscript symbols C and B mospheres (for a discussion on the limitations of the traditional
A107, page 3 of 44
A&A 603, A107 (2017)
approximation, see Gerkema & Shrira 2005; White et al. 2005; Like the basic fields p0 , ρ0 , and T 0 , the radiative frequency
Mathis 2009). varies with r and defines the transition between the dynamical
Hence, we obtain: regime, where the radiative losses can be ignored, and the radia-
tive regime, where they predominate in the heat transport equa-
1 B δp
ˆ ˙
BVθ tion. Assuming that the radiative emission of the gas is propor-
´ 2Ω cos θVϕ “ ´ Ù , (9)
Bt r Bθ ρ0 tional to the local molar concentration C0 “ ρ0 {M, it is possible
to express Jrad and σ0 as functions of the physical parameters of
the fluid (Appendix D),
δp
ˆ ˙
BVϕ 1 B
` 2Ω cos θVθ “ ´ Ù , (10)
Bt r sin θ Bϕ ρ0 8a
Jrad “ S T 03 δT (16)
M
BVr Bδp BU
ρ0 “´ ´ gδρ ´ ρ0 ¨ (11) and
Bt Br Br
8κa S 3
In strongly stratified fluids, the left-hand side of Eq. (11) is usu- σ0 prq “ T , (17)
ally ignored because it is very small compared to the terms in δp RGP 0
and δρ in typical wave regimes. The radial acceleration will only
play a role in regimes where the tidal frequency is close to or the parameter a being an effective molar emissivity coeffi-
exceeds the Brunt-Väisälä frequency, which corresponds to fast cient of the gas and S “ 5.670373 ˆ 10´8 W m´2 K´4 the
rotators. Stefan-Boltzmann constant (Mohr et al. 2012). The substitution
The second equation of our system is the conservation of of Eq. (15) in Eq. (14) yields
mass,
1 Bδp 1 Bδρ N 2 ρ0 δT
´ ` Vr “ κ J ´ σ0 ¨ (18)
Bρ
` ∇. pρVq “ 0, (12) Γ1 P0 Bt ρ0 δt g p0 T0
Bt
Finally, the system is closed by the perfect gas law
which, in spherical coordinates, writes
Bδρ 1 B ` 2 ρ0
„
B BVϕ
 δp δT δρ
r ρ0 Vr `
˘
psin θVθ q ` “ 0. (13) “ ` ¨ (19)
Bt
` 2
r Br r sin θ Bθ Bϕ p0 T0 ρ0
The thermal forcing (J) appears on the right-hand side Substituting Eq. (19) in Eq. (18), we eliminate the unknown δT ,
of the linearized heat transport equation (see CL70, and obtain
Gerkema & Zimmerman 2008) given by
κρ0
ˆ ˙ ˆ ˙
1 Bδp N 2 Bξr 1 Bδρ
1 Bδp 1 Bδρ N 2 ρ0 `Γ1 σ0 δp ` “ J` `σ0 δρ .
´ ` Vr “ κ rJ ´ Jrad s , (14) Γ1 p0 Bt g Bt p0 ρ0 Bt
Γ1 p0 Bt ρ0 δt g p0
(20)
where κ “ pΓ1 ´ 1q {Γ1 and Jrad is the power per mass unit ra-
diated by the atmosphere, supposed to behave as a grey body. Because of the rotating motion of the perturber in the equatorial
We consider that Jrad 9 δT . This hypothesis is known as “New- frame (RE;T ), a tidal perturbation is supposed to be periodic in
tonian cooling” and was used by Lindzen & McKenzie (1967) time (t) and longitude (ϕ). So, any perturbed quantity f of our
to introduce radiation analytically in the classical theory of at- model can be expanded in Fourier series of t and ϕ
mospheric tides (see also Dickinson & Geller 1968). Physically, ÿ
it corresponds to the case of an optically thin atmosphere in f “ f m,σ pθ, rq eipσt`mϕq , (21)
which the flux emitted by a layer propagates upwards or down- m,σ
wards without being absorbed by the other layers. In optically
thick atmospheres, such as on Venus (Lacis 1975), this physical the parameter σ being the tidal frequency of a Fourier compo-
condition is not verified. Indeed, because of a stronger absorp- nent and m its longitudinal degree1 . We also introduce the spin
tion, the power emitted by a layer is almost totally transmitted to parameter
the neighbourhood. Therefore, this significant thermal coupling
within the atmospheric shell should ideally be taken into account 2Ω
ν pσq “ , (22)
in a rigorous way, which would lead to great mathematical diffi- σ
culties (e.g. complex radiative transfers, Laplacian operators) in
our analytical approach. However, recent numerical simulations which defines the possible regimes of tidal gravito-inertial
of thermal tides in optically thick atmospheres by Leconte et al. waves:
(2015) show behaviour of the flow that is in good agreement
with a model that uses radiative cooling. Therefore, in this work – |ν| ď 1 corresponds to super-inertial waves, for which the
we assume Newtonian cooling as a first model of the action tidal frequency is greater than the inertia frequency;
of radiation on atmospheric tides. Newtonian cooling brings a – |ν| ą 1 corresponds to sub-inertial waves, for which the tidal
new characteristic frequency, denoted σ0 , which we call radia- frequency is lower than the inertia frequency.
tive frequency and which depends on the thermal capacity of the 1
atmosphere. The radiative power per unit mass is thus written In a binary star-planet system where the planet orbits circularly in
the equatorial plane defined by XE and Y E at the orbital frequency norb ,
p0 σ0 the semidiurnal tide corresponds to m “ 2, σ “ 2 pΩ ´ norb q, and
Jrad “ δT. (15) ν “ Ω{ pΩ ´ norb q.
κρ0 T 0
A107, page 4 of 44
The parameters m and ν determine the horizontal operators separated coordinates,

ÿ
J m,σ “ Jnm,σ prq Θm,ν
n pθq ,
„ 
1 B
Lm,ν
θ “ ` mν cot θ , (23)
p1 ´ ν2 cos2 θq Bθ n
ÿ
„  U m,σ
“ Unm,σ prq Θm,ν
n pθq ,
1 B m
Lm,ν
ϕ “ ν cos θ ` ¨ (24) n
p1 ´ ν2 cos2 θq Bθ sin θ ÿ
ξrm,σ “ ξr;n
m,σ
prq Θm,ν
n pθq ,
n
Here the operator Lm,ν θ is associated with Vθm,σ and ξθm,σ , ÿ
while Lϕ is associated with Vϕm,σ and ξϕm,σ (Lee & Saio 1997;
m,ν ξθm,σ “ ξθ;n
m,σ
θ rΘn pθqs ,
prq Lm,ν m,ν
Mathis et al. 2008). Hence, the traditional approximation allows n

ÿ
us to write the horizontal component of the velocity field only as ξϕm,σ “ ξϕ;n
m,σ
ϕ rΘn pθqs ,
prq Lm,ν m,ν
a function of the variations in the pressure and tidal gravitational n
potential ÿ
Vrm,σ “ m,σ
Vr;n prq Θm,ν
n pθq ,
n
i m,ν m,σ
Vθm,σ “ L py ` U m,σ q ,
ÿ
(25) Vθm,σ θ rΘn pθqs ,
m,σ
σr θ “ Vθ;n prq Lm,ν m,ν
n
1 ÿ
pym,σ ` U m,σ q , Vϕm,σ ϕ rΘn pθqs ,
m,σ
Vϕm,σ “ ´ Lm,ν (26) “ Vϕ;n prq Lm,ν m,ν
σr ϕ n
ÿ
where y “ δp {ρ0 . Since V
m,σ m,σ m,σ
“ Bt ξm,σ , we also get the δp m,σ
“ δpm,σ
n prq Θn pθq ,
m,ν
corresponding horizontal component ξm,σ

H “ ξθ eθ
m,σ
` ξϕm,σ eϕ of ÿ
n
the displacement vector from Eqs. (25) and (26) δρm,σ “ δρm,σ
n prq Θn pθq ,
m,ν
n
1 ÿ
ξθm,σ “ 2 Lm,ν pym,σ ` U m,σ q , (27) δT m,σ
“ δT nm,σ prq Θm,ν
n pθq
σ r θ n
ÿ
ξϕm,σ
i
“ 2 Lm,ν pym,σ ` U m,σ q . (28) y m,σ
“ ym,σ
n prq Θn pθq ,
m,ν
(34)
σ r ϕ n
The substitution of Eqs. (25) and (26) in Eq. (13) yields where the functions Θm,ν
n are the eigenvectors of the Laplace op-
erator, the corresponding eigenvalues being denoted Λm,ν
n (with
1 B ` 2 m,σ ˘ ρ0 ν “ 2Ω{σ). Laplace’s tidal equation, expressed as
δρm,σ “ ´ r ρ0 ξr ´ 2 2 Lm,ν pym,σ ` U m,σ q , (29)
2
r Br σr
n q “ ´Λn Θn ,
Lm,ν pΘm,ν m,ν m,ν
(35)
where L is the Laplace operator, parametrized by ν and m and
m,ν
describes the horizontal structure of the perturbation. The solu-
is expressed by
tions of Eq. (35), tΘm,νn unPZ are called Hough functions (Hough
1898)2 . They form a complete set of continuous functions on
sin θ
ˆ ˙
m,ν 1 B B the interval θ P r0, πs, parametrized by m and ν. They also ver-
L “
sin θ Bθ 1 ´ ν2 cos2 θ Bθ ify the same boundary conditions as the associated Legendre
(30) polynomials (denoted Pm l ), which are solutions of Eq. (35) for
1 ` ν2 cos2 θ
ˆ ˙
1 m2 ν “ 0 with Λm,ν n pν “ 0q “ l pl ` 1q. In Figs. 2 and 3, the func-
´ mν `
1 ´ ν2 cos2 θ 1 ´ ν2 cos2 θ sin2 θ tions Θm,ν
n , L θ Θ
m,ν m,ν
n , and ϕ Θn are plotted for the Earth and
Lm,ν m,ν
Venus in the case of the solar semidiurnal tide defined by m “ 2,

with ν “ 2Ω{σ (Eq. 22). When ν “ 0 (i.e. Ω “ 0), Lm,ν is re- σ “ 2 pΩ ´ norb q, and ν “ Ω{ pΩ ´ norb q, where norb stands for
duced to the horizontal Laplacian. Since the horizontal velocity the mean motion of the planet. In the case of the Earth, the set
field is not coupled with the other variables, only three unknowns is mainly composed of gravity modes. Since the spin parameter
remain. So, the new system is written is slightly greater than 1, Rossby modes exist but are trapped at
the poles. In the case of Venus, Hough functions are composed
Bδpm,σ BU m,σ of gravity modes only because |ν| ă 1. These functions are very
σ2 ρ0 ξrm,σ “ ` gδρm,σ ` ρ0 , (31) similar to the associated Legendre polynomials P2l , with l ě 2.
Br Br The system composed of Eqs. (31) to (33) can then be re-
1 B ` 2 m,σ ˘ ρ0 duced to the pair of first-order differential equations
δρm,σ “´ 2 r ρ0 ξr ´ 2 2 Lm,ν pym,σ ` U m,σ q , (32)
r Br σr $ d `
& dr r ξr;n “ A1 r ξr;n ` B1 yn ` C1 ,
2 m,σ
˘ 2 m,σ m,σ
δpm,σ 2
κρ0 m,σ δρ m,σ ’
N ’
piσ ` Γ1 σ0 q ` iσ ξrm,σ “ J `piσ ` σ0 q ¨
Γ1 p0 g p0 ρ0 m,σ
(36)
% dyn “ A r2 ξm,σ ` B ym,σ ` C ,
’
’
(33) dr
2 r;n 2 n 2
Under the traditional approximation, solutions of Eqs. (31)´(33) 2

The properties of Hough functions and the method used to compute
can be obtained under the form of series of functions of them in this work are detailed in Appendix A.
A107, page 5 of 44
A&A 603, A107 (2017)
2 30
n=0 n= 6
1.5 n=1 n= 5
20
n=2 n= 4
1
n=3 10 n= 3
0.5 n=4 n= 2
2,1.003
2,1.003
0 n=5 0 n= 1
n
n
−0.5
−10
−1
−20
−1.5
−2 −30
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
2
n=0 n= 6
1.5 n=1 20 n= 5
n=2 n= 4
1
n=3 10 n= 3
2,1.003
0.5 n=4
2,1.003
n= 2
0 n=5 0 n= 1
n
−0.5
L
−10
−1
−1.5 −20
−2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
2
n=0 n= 6
1.5 n=1 20 n= 5
n=2 n= 4
1
n=3 10 n= 3
2,1.003
2,1.003
0.5 n=4 n= 2
0 n=5 0 n= 1
n
−0.5
L
−10
−1
−1.5 −20
−2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
Fig. 2. Normalized Hough functions Θm,ν
n , θ Θn ,
Lm,ν m,ν
and ϕ Θn
Lm,ν m,ν
corresponding to the Earth’s semidiurnal tide (m “ 2 and ν “ 1.0030). We note
that ν “ 2Ω{σ. Left: first gravity modes; 0 ď n ď 5. Right: first Rossby modes: ´6 ď n ď ´1. In this case, |ν| is very close to 1 and the Rossby
modes are trapped at the poles. The Earth’s semidiurnal tide is mainly described by gravity modes. For more details about the behaviour of Hough
functions, see Fig. A.2.
with the coefficients and

1 dρ0 iσ N 2
ˆ ˙
, 1 iσ
A1 “ ´
ρ0 dr
´
iσ ` σ0 g A2 “ σ2 ´ N2 ,
r2 iσ ` σ0
Λm,ν
n iσ ` Γ1 σ0 r2 iσ ` Γ1 σ0 g 1 dρ0
B1 “ ´ , (37) B2 “ ´ ´ , (38)
σ2 iσ ` σ0 c2s iσ ` σ0 c s2 ρ0 dr
κr2 ρ0 Λm,ν κgρ0 dUnm,σ
C1 “ Jnm,σ ` n2 Unm,σ , C2 “ Jnm,σ ´ ,
p0 piσ ` σ0 q σ p0 piσ ` σ0 q dr
A107, page 6 of 44
2.0
n=0 with the set of coefficients
1.5 n=1 1 dB1
$
n=2
’
’ A “ Á1 ´ B2 ´ ,
1.0
’
’
’ B1 dr
n=3
’
’
’ ˆ ˙
0.5
& 1 dB1 dA1
2,0.4804
n=4
B “ A1 B2 ` ´ ´ B1 A2 , (41)
0.0 n=5 ’
’ B1 dr dr
’
n
’
’ ˆ ˙
’ dC1 1 dB1
−0.5
` B2 .
’
%C “
’ ` B1C2 ´ C1
dr B1 dr
−1.0
−1.5
Equation (40) gives us the vertical structure of tidal waves gener-
ated in the fluid shell by both gravitational and thermal forcings.
−2.0 We have:
0 0.2 0.4 0.6 0.8 1
/ 2 dρ0 iσ N 2 iσ ` Γ1 σ0 gρ0 1 dB1

A“ ` ` ´ ¨ (42)
2.0 ρ0 dr iσ ` σ0 g iσ ` σ0 Γ1 P0 B1 dr
n=0
1.5 n=1 The last term, pdB1 {drq {B1 , is associated with curvature and can
1.0
n=2 be ignored in the context of the “shallow atmosphere” approxi-
n=3 mation, as discussed in the next section. It is expressed as (see
2,0.4804
0.5 n=4
Eq. (50))
0.0 n=5
1 dB1 1 dε s;n
n
“´ ¨ (43)
−0.5 B1 dr 1 ´ ε s;n dr
L
−1.0 We now transform Eq. (40) in a Schrödinger-like equation by

introducing the radial function
−1.5
żr
−2.0 F prq “ A puq du, (44)
0 0.2 0.4 0.6 0.8 1
R
/
3.0 and the variable change
n=0
1
2.0
n=1 r2 ξr;n
m,σ
“ e´ 2 F prq Ψm,σ
n , (45)
n=2
n=3 where the volume Ψm,σn is the wave function of tidal waves as-
1.0
sociated with the mode of degree n. We note that fixing σ0 “
2,0.4804
n=4
0 allows us to retrieve the usual variable change, r2 ξr;n
m,σ
“
0.0 n=5
´1{2 m,σ
ρ0 Ψn (Press 1981; Zahn et al. 1997; Mathis et al. 2008).
n
With the general variable change, Eq. (40) becomes

L
−1.0
d2 Ψm,σ
„ ˆ ˙
n 1 dA A2 1
−2.0 ` B ´ ` Ψm,σ
n “ Ce 2 F prq , (46)
dr2 2 dr 2
−3.0
0 0.2 0.4 0.6 0.8 1 which gives us the vertical profile of the perturbation3 . The form
of this equation is very common in the literature because it de-
/ scribes the radial propagation of waves in a spherical shell. The
Fig. 3. First gravity modes (0 ď n ď 5) of normalized Hough func- typical wavenumber of these waves, denoted k̂n , is given by
tions Θm,ν
n , Lθ Θn , and Lϕ Θn
m,ν m,ν m,ν m,ν
corresponding to the semidiurnal
tide of Venus-like exoplanets (m “ 2 and ν “ 0.4804). We note that
ˆ ˙
1 dA A2
ν “ 2Ω{σ. For more details about the behaviour of Hough functions, k̂n2 “ B ´ ` ¨ (47)
2 dr 2
see Figs. A.2.
We introduce the corresponding normalized length scale of vari-
where we introduce the sound velocity ations of the mode of degree n,
d
Γ1 p0 2π
cs “ ¨ (39) LV;n “ ˇ ˇ¨ (48)
ρ0 Hatm ˇk̂n ˇ
The system of Eq. (36) can itself be reduced to a single equation This parameter is the typical height over which spatial variations
in r2 ξr;n
m,σ
alone of the mode can be observed.
3
d2 ` 2 m,σ ˘ d ` 2 m,σ ˘ To solve the vertical structure equation numerically in any case, we
2
r ξr;n ` A r ξr;n ` Br2 ξr;n
m,σ
“ C, (40) use the method described in Appendix B.
dr dr
A107, page 7 of 44
A&A 603, A107 (2017)
iσ ` Γ1 σ0 ρ0 σ2 ´ 1 F
dΨm,σ
" „
2.2. Polarization relations n
δρm,σ
n prq “ e 2
iσ ` σ0 c s p1 ´ ε s;n q
2 Λm,ν
n dr
The perturbed quantities are readily deduced from Ψ. Before go-
Γ1 ´ 1 m,σ
*
ing further, let us introduce the Lamb frequency (the cut-off fre-
quency of acoustic waves) associated with the mode (ν, m, n), ` Bn Ψn s ´
m,σ
J ´ Un m,σ
, (58)
iσ ` Γ1 σ0 n
cs
σ ´ 1 F dΨm,σ
1 " 2 „
σ s;n prq “ pΛm,ν
n q
2 , (49) iσT 0 pΓ1 ´ 1q n
r δT nm,σ prq “ 2 e 2
c s piσ ` σ0 q p1 ´ ε s;n q Λm,ν
n dr
and the ratio
Γ1 σ2
ˆ ˙ *
1
` Cn Ψ n s `
m,σ
1´ 2 m,σ
Jn ´ Un m,σ
.(59)
σ s;n
˙2
iσ ` Γ1 σ0 σ
ˆ
iσ
ε s,n prq “ ¨ (50)
iσ ` σ0 σ s;n The coefficients in these expressions are given by
g Γ1 σ0
„ 2 ˆ ˙ 
The parameter |ε s,n | measures the relevance of the anelastic ap- 1 iσ N
An prq “ ´ 2 1í p1 ` Kcurv q ,
proximation, which consists in neglecting all the terms that cor- 2 iσ ` σ0 g cs σ
respond to an acoustic perturbation in Eq. (36). When |ε s,n | ! 1
(i.e. |σ| ! |σ s;n |), this hypothesis can be assumed. However, par- „ ˆ ˙
ticular caution should be used when ignoring some terms. For 1 iσ N2 2
Bn prq “ ´1
instance, since the first gravity mode of the Earth’s semidiurnal 2 iσ ` σ0 g ε s;n
tide is characterized by |σ| « |σ s;0 | « 10´4 s´1 , the anelastic
g Γ1 σ0
ˆ ˙ 
approximation cannot be done. In order to be able to study cases
of the same kind, the polarization relations that follow will be ´ 2 1í p1 ` Kcurv q ,
cs σ
given in their most general form. Hence, by substituting the se-
ries of Eq. (34) in Eqs. (25)–(28), and (31)–(33), we obtain the
radial profiles ξr;n , ξθ;n , ξϕ;n , δpm,σ
n , δρn ,
m,σ m,σ m,σ m,σ m,σ m,σ m,σ
« ff
, Vr;n , Vθ;n , Vϕ;n N2 pΓ1 ` 1q iσ ` 2Γ1 σ0 σ2s;n
δT n , and yn as functions of Ψn and its first derivative. We
m,σ m,σ m,σ Cn prq “ ´ 2
g pΓ1 ´ 1q 2 piσ ` σ0 q σ
obtain for the Lagrangian displacement
g Γ1 σ0
ˆ ˙
1 1 iσ
ξr;n
m,σ
prq “ 2 e´ 2 F Ψm,σ
n , (51) ´ 1í p1 ` Kcurv q , (60)
r 2c s iσ ` σ0
2 σ
σ ´ 1 F dΨn
" 2 „ m,σ 
1 where we have introduced the curvature term
ξθ;n
m,σ
prq “ 2 e 2 ` An Ψm,σ
σ r p1 ´ ε s;n q Λm,ν dr n
σ2 c2s
ˆ ˙
n
d r2
Γ1 ´ 1 σ2 m,σ
* Kcurv “ , (61)
´ J ´ ε s;n Unm,σ , (52) gΛn p1 ´ ε s;n q dr c2s
iσ ` σ0 σ2s;n n
which will be ignored in the thin atmosphere approxi-
σ ´ 1 F dΨm,σ
" 2 „ 
i n mation. These relations and Eq. (46) define a linear op-
ξϕ;n prq “ 2
m,σ
e 2 ` An Ψn
m,σ
σ r p1 ´ ε s;n q Λm,ν
n dr erator, denoted Υσ , which determines the vertical struc-
Γ1 ´ 1 σ2 m,σ
* ture of the atmospheric tidal response. Let be Ynm,σ “
´ J ´ ε s;n Unm,σ , (53) n , ξ n , V n , δpn , δρn , δT n , yn q the output vector of
pΨm,σ m,σ m,σ m,σ m,σ m,σ m,σ
iσ ` σ0 σ2s;n n the model and Xn “ pUn , Jn q the input vector. Eqs. (51)
m,σ m,σ m,σ
to (59) can be reduced to

for the velocity
Ynm,σ “ Υm,σ
n Xn .
m,σ
(62)
iσ 1
m,σ
Vr;n prq “ 2 e´ 2 F Ψm,σ
n , (54) For any quantity f , the component of Υm,σ describing the vertical
r n
i
" 2
σ ´ 1 F dΨn
„ m,σ  profile of f is denoted Υm,σ m,σ
f ;n (Υ p;n for the pressure, Υm,σ
ρ;n for the
m,σ
Vθ;n prq “ e 2 ` An Ψm,σ density, etc.).
σr p1 ´ ε s;n q Λm,ν n dr n
Polarization relations can be naturally divided into two dis-
Γ1 ´ 1 σ2 m,σ
*
tinct contributions. The first contribution, which is called the
´ J ´ ε s;n Unm,σ , (55)
iσ ` σ0 σ2s;n n “horizontal part” in this work, is a linear combination of the forc-
ings Unm,σ and Jnm,σ . It corresponds to the component of the re-
σ ´ 1 F dΨm,σ
" 2 „ 
1 n
m,σ
Vϕ;n prq “ ´ e 2 ` An Ψm,σ sponse that does not depend on the tidal vertical displacement.
σr p1 ´ ε s;n q Λn dr n
The length scale of this horizontal component is the typical
Γ1 ´ 1 σ2 m,σ thickness of the atmosphere (Hatm ). The other part, expressed
*
´ J ´ ε s;n Unm,σ , (56) as a function of Ψm,σ and called “vertical part”, thus results from
iσ ` σ0 σ2s;n n n
the tidal fluid motion along the vertical direction. In the follow-
and for scalar quantities ing, we use the subscripts H and V to designate the horizontal and
vertical components.
ρ0 σ ´ 1 F dΨm,σ
" 2 „  As can be noted, polarization relations point out the neces-
n
δpm,σ
n prq “ e 2 ` An Ψn
m,σ
sity of taking into account dissipative processes to study tidal
1 ´ ε s;n Λm,ν
n dr regimes close to synchronization. Indeed, without dissipation
(57)
Γ1 ´ 1 σ2 m,σ
*
(σ0 “ 0), synchronization (σ “ 0) is characterized by a sin-
´ J ´ Unm,σ ,
iσ ` σ0 σ2s;n n gularity. The terms associated with the horizontal component,
A107, page 8 of 44
which are directly proportional to Unm,σ and Jnm,σ , tend to infin- Love numbers are defined as the ratio between the gravitational
ity at σ Ñ 0. Terms in Ψm,σ n and dΨm,σn {dr behave similarly, potential due to the tidal response of the atmosphere and the forc-
as demonstrated in the next section. These terms are highly os- ing potential at the upper boundary of the layer. Let us denote
cillating along the vertical direction over the wholeˇ ˇ depth of Ratm “ R ` Hatm this upper boundary. Then, at the upper bound-
the atmosphere in this frequency range because ˇk̂n ˇ Ñ `8 ary, the atmospheric tidal potential given by Eq. (66) is simply
when σ Ñ 0. Hence, radiation regularizes the behaviour around expressed:
σ “ 0 and damps the vertical oscillations of waves associated ż Ratm
with the vertical component. Diffusion acts in the same way by 4πG
Ulm,σ pRatm q “ ´ rl`2 δρm,σ
l
prq dr. (68)
flattening the oscillations of smallest length scales (LV;n ) (Press p2l ` 1q Rl`1
atm R
1981; Zahn et al. 1997; Mathis et al. 2008). We return to this
point when we solve analytically the case of the isothermal at- Considering Eq. (72) and using the linearity property of Υm,σ
n ,
mosphere with constant profiles for the forcings (Sect. 3). the density component δρm,σl can be expanded
The whole spectrum of possible tidal regimes is represented ÿ ÿ
in Fig. 4. The radiative regime (blue) characterizes a flow where δρm,σ
l “ l,n,k δρn,k
cm,σ m,σ
(69)
thermal losses due to the linear Newtonian cooling (see the heat nPZ kě|m|
transport equation, Eq. 18) predominate. In the dynamic regime
with δρm,σ “ Υm,σ
ρ;n U k , Jk
` m,σ m,σ ˘
(orange), dissipation can be neglected because the system is gov- n,k and the change-of-basis
erned by the Coriolis acceleration. Finally, the mixed-acoustic coefficients,
regime (red) corresponds to high tidal frequencies comparable
l,n,k “ xPl , Θn yxPk , Θn y,
cm,σ m m,σ m m,σ
(70)
to the Lamb frequency, given by Eq. (49). This regime marks the
limit of the traditional approximation. which allows us to write the atmospheric tidal potential
ż Ratm
2.3. Tidal potential, Love numbers, and tidal torque 4πG ÿ ÿ
Ulm,σ pRatm q “ ´ cm,σ
l,n,k rl`2 δρm,σ
n,k
prq dr.
p2l ` 1q Rl`1
atm nPZ kě|m| R
The new mass distribution resulting from tidal waves generates
a tidal gravitational potential U, which is a linear perturbation (71)
of the spherical gravitational potential of the planet. This atmo-
spheric potential presents disymmetries with respect to the direc- Let us treat the case of a simplified planet-star system, where
tion of the perturber. The tidal torque thus induced will affect the the planet P circularly orbits around its host star, of mass MS .
rotationnal dynamics of the planet over secular timescales and The semi-major axis, obliquity, and mean motion of the planet
determine the possible equilibrium states of the spin, as demon- are denoted a, ε, and norb . In this case, the modes pσ, mq of the
strated by Correia & Laskar (2001). This torque is deduced from gravitational and thermal forcings applied on the atmosphere can
the Poisson’s equation expressed in the co-rotating frame be written (Appendix C)
ÿ
Ulm,σ prq Pml pcos θq e ,
$ m,σ ipσt`mϕq
∆U “ 4πG δρ, (63) ’
’
&
U “
lě|m|
the notation G designating the gravitational constant. Like δρ, ÿ (72)
l pcos θq e
Jlm,σ prq Pm .
m,σ ipσt`mϕq
%J “
’
the potential is expanded in series of functions of separated
’
lě|m|
variables
ÿ Hence the complex Love numbers can be written generically
l pcos θq e
Ulm,σ prq Pm ,
ipσt`mϕq
U“ (64)
Ulm,σ ˇ
ˇ
m,σ,l
m,σ
kl “ m,σ ˇˇ ¨ (73)
where the Pm Ul
l (with l P N such as l ě |m|) are the normal-
r“Ratm
ized associated Legendre polynomials. Decomposing δρm,σ on In celestial dynamics, the second-order Love number (l “ 2) is
this basis, we get the equation describing the vertical structure commonly used to quantify the tidal response of a body. There-
of Ulm,σ , fore, we illustrate the expression given by Eq. (73) by comput-
ing this coefficient for m “ 2. For the sake of simplicity, we
dUlm,σ
ˆ ˙
d set the obliquity ε “ 0 (the spin of the planet is supposed to be
r 2
´ l pl ` 1q Ulm,σ “ 4πG r2 δρm,σ
l . (65)
dr dr aligned with its orbital angular momentum). The tidal frequency
is σSD “ 2 pΩ ´ norb q and the series of Eq. (C.14) are reduced
For the upper boundary condition, the tidal potential is required to the terms characterized by pl, m, j, p, qq “ p2, 2, 2, 0, 0q4
to remain bounded at r Ñ 8. At r “ 0, we impose the same
condition. Therefore, the solution of Eq. (65) is c c
1 3 G MS 2 1 3 J2 prq
U2,2,2,0,0 “ ´ r and J2,2,2,0,0 “ ¨
4πG ” l m,σ ı 2 2 a3 2 2 a2
Ulm,σ prq “ ´ r Fl prq ` r´pl`1q Hlm,σ prq , (66) (74)
2l ` 1
where Flm,σ and Hlm,σ are the functions The tidal potential becomes
4πG ÿ ÿ 2,σSD Ratm 4 2,σSD
$ ż `8 ˇ ż
U22,σSD ˇ c2,n,k r δρn,k prq dr. (75)
ˇ
’
F m,σ
prq “ u´l`1 δρm,σ puq du, “´ 3
5Ratm nPZ kě2
’ Ratm
& l
’ l R
r
żr (67)
4
’ The parameters j, p, and q are the usual indexes of the Kaula’s ex-
% Hlm,σ prq “ ul`2 δρm,σ
’
’
l
puq du. pansion of the tidal gravitational potential; see Appendix C.
R
A107, page 9 of 44
A&A 603, A107 (2017)
Fig. 4. Frequency spectrum of the atmospheric tidal response. The identified regimes are given as functions of the tidal frequency (σ).
If the star can be considered a point-mass perturber (R ! a), l “ 2 can be neglected. Denoting U2m,σ and J2m,σ the quantities
then U22,σSD rapidly decays with k. So, the terms of orders k ą 2 Ulm,σ and Jlm,σ for l “ 2 (which should not be confused with the
are only generated by the thermal forcing. We also assume that projections Unm,σ and Jnm,σ of the forcings on the set of Hough
the horizontal pattern of J is well represented by the function P22 , functions), it follows
which allows us to ignore other terms. Finally, we consider the
case |ν| „ 1 (|norb | ! |Ω|, see Eq. (22)), where c2,σ SD
ż Ratm
2,0,2 « 1 and
ÿ ˇ ˇˇ ˇ ´ ¯
n,2 ˇ sin ∆ϕ2,n,2 dr.
ˇ
T m,σ “ ´πm cm,σ r2 ˇU2m,σ ˇ ˇδρm,σ m,σ
ˇ
2,σSD 2,σSD 2,n,2
0 ă c2,n,2 ! 1 for n ą 0 (Θ0 « P22 ). In this simplified nPZ R
framework, the second-order Love number can be approximated (81)
in magnitude order by
If the forcing U2σ,m is a positive real function, then the previous
G MS 2 J2 prq
3 ż Ratm „ 
expression becomes
4π a
k22,σSD „ r4 Υ2,σ
ρ;2
SD
r ,´ 2 dr.
5 MS R5atm R a3 a ÿ ż Ratm ! )
(76) T m,σ “ ´πm cm,σ
2,n,2 r 2 m,σ
U 2 = δρm,σ
n,2 dr. (82)
nPZ R
To conclude this section, we compute the tidal torque exerted on Substituting the polarization relation of density in Eq. (82), we
the atmosphere with respect to the spin axis. The monochromatic obtain the torque as a sum of two contributions
torque associated with the frequency σ writes (Zahn 1966)
" ż * T m,σ “ THm,σ ` TVm,σ , (83)
1 BU m,σ
T m,σ
“ ´< pδρ q dV ,
m,σ ˚
(77)
2 V Bϕ where
Λm,ν
"ż Ratm
where the notation V designates the˘ volume ` ofm,σthe
˘ atmospheric n ` m,σ ˘2
ÿ
THm,σ “ πm cm,σ
2,n,2 = Hn U2 dr
shell. Denoting ∆ϕm,σ δρ σ2
` m,σ
l “ arg l ´ arg U l the phase lag nPZ R
between the forcing and the response, we obtain
Λm,ν Γ1 ´ 1
ż Ratm *
ÿ ż Ratm ˇ ` Hn n
U m,σ m,σ
J , (84)
r2 ˇUlm,σ ˇ ˇδρm,σ ˇ sin ∆ϕm,σ dr,
m,σ
ˇˇ ˇ ` ˘
T “ ´πm l l (78) R σ2 iσ ` Γ1 σ0 2 2
lě|m| R
dΨm,σ
"ż Ratm „ 
and the total tidal torque, ÿ
´ F2 n
TVm,σ “ ´πm cm,σ
2,n,2 = Hn e ` Bn Ψm,σ
n
ÿ nPZ R dr
T “ T m,σm,s . (79)
pm,sqPZ2
Û2m,σ dr
(
(85)
m,σ
As previously done for the Love numbers, T is expanded us-
with Ψm,σ “ Ψm,σ U2 , J2
` m,σ m,σ ˘
ing Hough modes n n and
ż Ratm iσ ` Γ1 σ0 ρ0 r2 σ2
ÿ ˇ ˇˇ ˇ ´ ¯ Hn “ ¨ (86)
∆ϕ iσ ` σ0 c2s Λm,ν
n p1 ´ ε s;n q
ˇ
T m,σ “ ´πm cm,σ r2 ˇUlm,σ ˇ ˇδρm,σ sin m,σ
l,n,k dr,
ˇ
l,n,k n,k ˇ
l,n,k R
Assuming |ε s;n | ! 1 and noticing that nPZ cm,σ
ř
(80) 2,n,2 “ 1, we estab-
m,σ
lish that TH does not depend on the eigenvalues of Laplace’s
with l ě |m|, n P Z, and k ě |m|. Indeed, we note here tidal equation and can be written
that δρσ,m
n,k is the component of density variations represented
ρ0 r2 Γ1 ´ 1
"ż Ratm *
by Pm l caused by the component of the excitation represented TH “ πm=
m,σ m,σ m,σ
U J dr
by Pm and projected on Θn . Hence, the parameter ∆ϕm,σ c2 iσ ` Γ1 σ0 2 2
´k l,n,k “ "Rż Ratm s 2 (87)
ρ0 r iσ ` Γ1 σ0 ` m,σ ˘2
¯ *
arg δρm,σ ´ arg Ulm,σ is the phase difference between the
` ˘
`πm= U2 dr .
n,k
R c2s iσ ` σ0
component of degree pn, kq of the response and the component
of degree l of the excitation. When the tidal gravitational po- The torque induced by this component is expanded in series
tential is quadrupolar (R ! a), the terms of orders higher than of Hough functions and characterized by the solutions Ψm,σ
n of
A107, page 10 of 44
the vertical structure equation. The typical scales of the vertical which is used in the following. Assuming that p0 , ρ0 , and T 0
component are given by the vertical wavenumbers (see Eq. (47)) only depends on x, we deduce their expressions from Eqs. (2)
and can be very small compared to the scale of the horizontal and (89)
ones. For example, in the vicinity of synchronization, k̂n 9 1{σ
p0 pxq gH pxq
if the radiative damping is ignored (CL70), which means that the p0 pxq “ p0 p0q e´x , ρ0 pxq “ , T 0 pxq “ ¨ (94)
characteristic scales of variations rapidly decay when σ Ñ 0. Fi- gH pxq Rs
nally, we note that the series of modes in Eq. (81) can often be
reduced to one dominating term. For instance, in the case of the In the fluid shell, the gravitational acceleration does not vary
very much
Earth’s semidiurnal tide, Θσ,2 0 „ P22 and the expression above
(Eq. (81)) can be simplified ˇ g pxq ´ g p0q ˇ
ˇ ˇ
ˇ ˇ „ 2 H x, (95)
ż Ratm
ˇ ˇˇ ˇ ´ ¯ ˇ g p0q ˇ R
r2 ˇU2σ,2 ˇ ˇδρσ,2 ∆ϕ σ,2
m,σ
ˇ
T sin 2,0,2 dr. (88)
ˇ
„ ´2π 0,2
which means that g can be considered as a constant. Assuming
ˇ
R
that H also does not vary with the altitude, we obtain a constant
3. Tides in a thin stably stratified isothermal profile of the temperature and exponentially decaying profiles of
the pressure and density:
atmosphere
p0 p0q ´x gH
We establish in this section the analytical equations that describe p0 pxq “ p0 p0q e´x , ρ0 pxq “ e , T0 “ ¨ (96)
the tidal perturbation in the case where Hatm ! R. This theoret- gH Rs
ical approach has been exhaustively formalized for the Earth in
In this case, which corresponds to an isothermal atmosphere,
“atmospheric tides” (CL70) and corresponds to the case of thin
95% of the mass of the gas is contained within the interval
atmospheres. The formalism developed in the previous section
x P r0, 3s. Therefore, we can write Hatm « 3H. For the Earth,
allows us to go beyond this pioneer work, and more particularly
H „ 8 km (CL70) and Hatm „ 24 km.
to study regimes near synchronization thanks to the inclusion of
radiative losses.
3.2. Wave equation and polarization relations
3.1. Equilibrium state The constant H hypothesis is very useful in order to simplify
the expressions of Sect. 2. Indeed, with this approximation, the
For a thin atmosphere (r « R), the equilibrium structure can be
typical frequencies of the system no longer vary with the radial
determined analytically. We introduce the altitude z, such that
coordinate. The Brunt-Väisälä frequency simply writes
r “ R ` z, which is here a more appropriate coordinate than r.
The fluid is supposed to be at hydrostatic equilibrium, which c
κg
leads to N“ , (97)
H
1 Bp
“ ´g. (89) and the acoustic frequency associated with the mode pν, m, nq,
ρ Bz
1 cs
The local acceleration in RE;T , denoted aRE;T , is not equal to σ s;n “ pΛm,ν
n q
2 , (98)
g because of the rotating motion. The total acceleration can be R
written with c s “
a
Γ1 gH. Moreover, the horizontal structure of tidal
aRE;T “ ´g pzq er ` ac , (90) waves does not change because Laplace’s tidal equation is not
modified. So, the vertical structure equation and the radial polar-
where ac is the centrifugal acceleration: ization relations only are affected by the constant H hypothesis.
The coefficients defined by Eq. (41) become
ac “ r sin θ Ω2 psin θ er ` cos θ eθ q . (91)
1
Here, Coriolis acceleration does not intervene because the global A“ ´ ,
H
circulation of the atmosphere is ignored. The centrifugal acceler-
Λm,ν
„ 
ation (Eq. (90)) can be neglected with respect to the gravity if the iσ N 2
B “ n2 ` ε s;n ´ 1 ,
arate of the planet satisfies the condition Ω ! Ωc , where
rotation R iσ ` σ0 σ2
Ωc “ g{R is the critical Keplerian angular velocity. In the case (99)
Λm,ν κ hn dJnm,σ
" „ 
of an Earth-like planet, with g „ 10 m s´2 and R „ 6.0ˆ103 km, C“ n m,σ m,σ
` Kn Jn ´ iσUn
Ωc „ 20 rotations per day. From Eqs. (2) and (89), a typical Hσ2 iσ ` σ0 H dx
dU m,σ
*
depth can be identified,
` ε s;n n ,
p0 Rs T0 dx
H“ “ , (92)
gρ0 g where
which is the local characteristic pressure height scale. The pa- hn
rameter H represents the vertical scale of variations of basic Kn “ 1 ´ ¨ (100)
H
fields. It is of the same order of magnitude as Hatm and allows
us to introduce the reduced altitude The parameter hn , which writes
R2 σ2
żz
dz
x“ , (93) hn “ , (101)
0 H pzq gΛm,ν
n
A107, page 11 of 44
A&A 603, A107 (2017)
is usually called the “equivalent depth” in the literature (e.g. is satisfied, the forcing in the right-hand member is dominated
Taylor 1936; Chapman & Lindzen 1970) and represents a typ- by thermal tides and the contribution of the gravitational tidal
ical length scale associated with the mode of degree n. We note potential can be ignored in Eq. (107), which is rewritten
that the curvature term in Eq. (42) vanishes in the shallow atmo-
sphere approximation because the coefficient B1 (Eq. (37)) is a d2 Ψm,σ 1
„
iσ hc

κHΛm,ν ´x
„
hn dJnm,σ
n n e 2
constant in this case. This allows us to simplify the vertical struc- ` ´ 1 Ψ m,σ
“
ture equation (Eq. (46)). Unlike the vertical wavenumber in the dx2 4 iσ ` σ0 hn n
iσ ` σ0 H dx
thick shell (Eq. (47)), the vertical wavenumber k̂n does not vary `Kn Jnm,σ s .
with x in this case. It is expressed (109)
„ ˆ ˙ 
1 iσ N 2 The polarization relations of the thin atmospheric shell are de-
2
k̂n “ ςn ` ε s;n ´ 1 ´ 1 , (102)
4 iσ ` σ0 σ2 duced from Eqs. (51) to (59)
with the scale ratio ςn “ 4H 2 Λm,ν 2

n {R . It can also be written
1 x m,σ
ξr;n
m,σ
“ e 2 Ψn , (110)
„  R2
1 iσ 4κH
2
´ 1 ` ςn pε s;n ´ 1q . σ2 dΨm,σ
" „ 
k̂n “ (103) 1
4 iσ ` σ0 hn
x n
ξθ;n
m,σ
“ 2 m,ν e
2 ` An Ψm,σ
σ R p1 ´ ε s;n q HΛn dx n
The term ςn pε s;n ´ 1q comes from the radial acceleration in the

Navier-Stokes equation (Eq. (11)) and is negligible for usual val-
ues of Λm,ν ´ Dn Jnm,σ ´ ε s;n Unm,σ u , (111)
n (|ε s;n | ! 1 and ςn 9 H {R ! 1). Therefore, by con-
2 2
sidering the case where σ0 ! |σ|, we recover the wavenumber i

"
σ2 x
„ m,σ
dΨn

obtained by CL70 for the Earth’s semidiurnal tide. In the general ξϕ;n “ 2 e 2 ` A Ψm,σ
n n
case, the expression of Eq. (103) can be approximated by σ R p1 ´ ε s;n q HΛm,ν n dx
„ 
1 iσ 4κH ´ Dn Jnm,σ ´ ε s;n Unm,σ u ,
2
k̂n « ´1 , (104) (112)
4 iσ ` σ0 hn
iσ x m,σ
which gives for the real and imaginary parts,(introducing the
m,σ
Vr;n “ e 2 Ψn , (113)
R2
characteristic depth hc “ 4κH and fixing < k̂n ą 0,
σ2 dΨm,σ
" „ 
i x n
m,σ
Vθ;n “ m,ν e
2 ` An Ψm,σ
$ « ff1{2 σR p1 ´ ε s;n q HΛn dx n
’ 1 σ2 hc ?
´1` ∆ ,
’ (
’ < k̂n “ ?
2 2 σ2 ` σ20 hn
’
’
&
´ Dn Jnm,σ ´ ε s;n Unm,σ u , (114)
« ff´1{2
σ0 σ hc σ2 hc ?
’
’ 1 "
σ2
„
dΨm,σ

´1` ∆ , 1
’ (
% = k̂n “ ? 2 x
’ n
m,σ
Vϕ;n m,ν e ` An Ψnm,σ
’
2 2 σ ` σ20 hn σ2 ` σ20 hn “´ 2
σR p1 ´ ε s;n q HΛn dx
(105)
with ´ Dn Jnm,σ ´ ε s;n Unm,σ u , (115)
ρ0 σ2 dΨm,σ
˜ ¸2 " „ 
σ2 hc σ20 σ2 h2c δpm,σ “ e
x
2
n
` An Ψm,σ ´ Dn Jnm,σ
∆“ ´ 1 ` ¨ (106) n
1 ´ ε s;n HΛm,ν dx n
σ2 ` σ0 hn
2 2
σ2 ` σ20 hn
2 n
` ˘
The behaviour of the vertical wavenumber defines the possible ´ Unm,σ u , (116)
regimes of the perturbation. These regimes are represented by
iσ ` Γ1 σ0 ρ0 σ2 dΨm,σ
" „
the map in Fig. 5, which shows the real and imaginary part of the δρm,σ “
x
e2 n
vertical wavenumber as colour functions of the tidal frequency n
iσ ` σ0 c2s p1 ´ ε s;n q HΛm,ν
n dx
and critical frequency, σc;n , given by σ2c;n “ 4κHgΛm,ν 2
n {r (the
frequency σc;n is proportional to the Lamb frequencyˇ defined
(ˇ by Γ1 ´ 1 m,σ
*
Eq. (98)). Propagative modes are characterized by ˇ= k̂n ˇ !
ˇ ` Bn Ψm,σ
n s ´ J ´ Un m,σ
, (117)
(ˇ iσ ` Γ1 σ0 n
1 (right panel, dark regions), evanescent modes by ˇ= k̂n ˇ "
1 (yellow regions). The transition between the radiative regime iσT 0 pΓ1 ´ 1q
"
σ2 x
„ m,σ
dΨn
and the dynamic regime corresponds to |σ| « σ0 . δT nm,σ “ 2 e 2
In the case of the thin isothermal atmosphere, the equation c s piσ ` σ0 q p1 ´ ε s;n q HΛm,ν
n dx
giving the vertical profiles (Eq. (46)) writes
Γ1 σ2
ˆ ˙ *
1
d2
Ψm,σ
n ` Cn Ψm,σ
n s ` 1´ 2 Jnm,σ ´ Unm,σ , (118)
` k̂n2 Ψm,σ
n “ H 2 e´x{2C pxq . (107) iσ σ s;n
dx2
where C is given by Eq. (99). Moreover, if the condition with
Γ1 ´ 1 σ2
ˇ m,σ ˇ ˇ ˇ
ˇ Un ˇ ˇ H ´ hn ˇ
ˇ!ˇ (108) Dn “ , (119)
iσ ` σ0 σ2s;n
ˇ ˇ
ˇ J m,σ ˇ ˇ σH ˇ
n
A107, page 12 of 44
` (˘
Fig. 5. Real and imaginary part of the vertical wavenumber as functions of the tidal (σ) and critical (σc;n ) frequencies. Left: sign pσq f < k̂n ,
with k̂n defined in Eq. (104), as a function of f pσ{σ0 q (horizontal axis) and f σc;n {σ20 (vertical axis), where f is the function defined by
` 2 ˘
` (˘
f pXq “ sign pXq log p1 ` |X|q. The colour of the map, denoted c, is defined by sign pσq f < k̂n . Hence, the purple region stands for slowly
` (˘
oscillating waves. The luminous (dark) regions correspond to strongly oscillating waves propagating downwards (upwards). Right: log = k̂n
as a function of the same parameters. The luminous (dark) regions correspond to strongly (weakly) evanescent waves. The positions of the planets
(Earth C , Venus B ) on the map are determined by the gravity mode of degree 0 in the case of the semidiurnal tide (ΛC;0 “ 11.1596, ΛB;0 “ 7.1485).
The tidal frequencies of the planets are given by σC {σ0 « 194 and σB {σ0 “ 0.43, with σ0 “ 7.5 ˆ 10´7 s´1 .
and where An , Bn , and Cn are the constant coefficients of The parameter K P C is an integration constant which is fixed by
Eq. (60)5 , the upper boundary condition.
The upper border should be treated very carefully, as has
σ0
ˆ ˙
iσ 1
$
been discussed by Green (1965). The fluid envelopes of stars
’ A n “ κ ´ ` i ,
iσ ` σ0 and giant gaseous planets, which are considered as bounded fluid
’
’
’
’ 2 2σ
’
’ shells, are usually treated with a free-surface condition applied
κ σ0
’ ˆ ˙
& iσ 1 at the upper limit x “ xatm (e.g. Unno et al. 1989), i.e.
Bn “ ´ ì , (120)
’ iσ ` σ0 ε s;n 2 2σ
’ 1 dp0 m,σ
δpm,σ ξ
’
’ ` “ 0. (123)
% Cn “ 1 ´ σ s;n ¨
n
2
H dx r;n
’
’
’
’
2 Γ1 σ2 Hence,
dΨm,σ
n
3.3. Boundary conditions ` PΨm,σ
n “ Q, (124)
dx
Solving Eq. (107) requires that we choose two boundary condi- with the complex coefficients
tions. Following CL70, we fix ξr “ 0 at the ground. This cor-
σ2
$
responds to a smooth rigid wall and is equivalent to ξr;n m,σ
“ 0
’ P “ An ´ p1 ´ ε s;n q s;n ,
’
’
at x “ 0. In the case of Earth-like exoplanets where the telluric ’
& Γ1 σ2
surface is well defined, this condition is relevant. Thus, given the (125)
form of Eq. (107), Ψm,σ κR2 HΛm,ν
„ 
n can be written ’ xatm
n
´ m,σ m,σ
.
’
% Q “ e 2 g piσ ` σ q Jn ` σ2 Un
’
’
0 x“xatm
Ψm,σ
n “ A pxq eik̂n x ` B pxq eík̂n x , (121)
The atmosphere then behaves as a wave-guide of typical thick-
where A and B are the complex functions ness xatm (Appendix E). In this case, atmospheric tides are analo-
gous to ocean tides (see Webb 1980), the amplitude of the pertur-
H 2 x ´pik̂n `1{2qu
$ ż
bation being highly frequency-resonant. However, this condition
’
’
’ A pxq “ K ´ i e C puq du, appears to be inappropriate for the thin isothermal atmosphere
& 2k̂n 0
(122) because the fluid is not homogeneous and there is no discon-
H 2 x pik̂n ´1{2qu
ż
’
% B pxq “ ´K ` i
’ tinuous interface in this case. Setting the free-surface condition
’ e C puq du.
2k̂n 0 would give birth to unrealistic resonances depending on xatm (see
Fig. E.1 in Appendix F). Therefore, we have to choose a condi-
5
These coefficients are the same as those given by Eq. (60) to an tion consistent with the exponentially decaying density and pres-
H factor. sure. Following Shen & Zhang (1990), we consider that there is
A107, page 13 of 44
A&A 603, A107 (2017)
no material escape when x Ñ `8, i.e. Vr | xÑ8 “ 0. This means i

ξϕ;n
m,σ
“ tEn e x FAn pxq ´ Dn Jnm,σ ´ εs;n Unm,σ u ,
that the amplitude of oscillations decrease with the altitude and, σ2 R p1 ´ ε s;n q
consequently, that we have to eliminate the diverging term in (133)
Eq. (121)6 . Λ m,ν
H ” 1
ı
n
m,σ
“ i 2 En 1 ´ epik̂n ` 2 q x ,
(
Let us assume = k̂n ą 0. As shown by CL70 (Eq. (88)), Vr;n (134)
Rσ
the above condition requires that i
m,σ
Vθ;n “ tEn e x FAn pxq ´ Dn Jnm,σ ´ εs;n Unm,σ u ,
dΨm,σ
n
σR p1 ´ ε s;n q
´ ik̂n Ψm,σ
n “0 (126) (135)
dx
1
at the upper boundary (x “ xatm ). To make this condition rel-
m,σ
Vϕ;n “´ tEn e x FAn pxq ´ Dn Jnm,σ ´ εs;n Unm,σ u ,
σR p1 ´ ε s;n q
evant, xatm must be chosen so that xatm {xcrit " 1, where xcrit is
the typical damping depth computed from the case where C is a (136)
constant, and given by ρ 0 p0q
δpm,σ En FAn pxq ´ pDn Jnm,σ ` Unm,σ q e´x , (137)
(
“
n
1 ´ εs;n
1
xcrit “ , (127) iσ ` Γ1 σ0 ρ0 p0q
δρm,σ tEn FBn pxq
(
= k̂n ` 1{2 “
n
iσ ` σ0 c2s p1 ´ εs;n q
we usually fix xatm “ 10 in computations. The integration con-
Γ1 ´ 1 m,σ
ˆ ˙ *
stant of Eq. (122) is then obtained as ´ J ` Un m,σ
e ´x
, (138)
iσ ` Γ1 σ0 n
H 2 xatm pik̂n ´1{2qu
ż
K“i e C puq du. (128) iσ pΓ1 ´ 1q T 0
δT nm,σ “ 2 tEn e x FCn pxq
2k̂n 0 c s piσ ` σ0 q p1 ´ εs;n q
Γ1 σ2
ˆ ˙ *
1
` 1´ 2 Jnm,σ ´ Unm,σ , (139)
3.4. Analytical expressions of the semidiurnal tidal torque, iσ σs;n
lag angle, and amplitude of the pressure bulge
with
If Unm,σ and Jnm,σ are assumed to be constant (with Un , Jnm,σ P C),
κ pKn Jnm,σ ´ iσUnm,σ q
C does not vary with x and the expression of Eq. (121) can be En “ ˘ ¨ (140)
piσ ` σ0 q k̂n2 ` 41
`
simplified significantly. Indeed, in this case it can be written
H 2C ” ´ x ı From the analytical solution and the polarization relation of den-

Ψm,σ
n pxq “ e 2 ´ eik̂n x . (129) sity given by Eq. (138), we then compute the second-order Love
k̂n2 ` 14
number and the tidal torque. Integrating δρ over the whole thick-
( ness of the atmosphere, we obtain
We note that we use the convention = k̂n ą 0, which implies
Ψm,σ
n Ñ 0 at infinity. Substituting this simplified solution in ż xatm
iσ ` Γ1 σ0 ρ0 p0q
"
pΓ1 ´ 1q Jnm,σ
polarization relations (Eqs. (110) to (118)), we compute three- δρm,σ pxq dx “ ´
dimensional variations of the velocity field, displacement, pres- 0
n
iσ ` σ0 c2s p1 ´ ε s;n q iσ ` Γ1 σ0
κ Bn ` 12 rKn Jnm,σ ´ iσUnm,σ s
` ˘
sure, density, and temperature caused by the semidiurnal tide as
explicit functions of the physical parameters of the atmosphere ` ˘2
piσ ` σ0 q ik̂n ´ 12
`
and tidal frequency. Let us introduce the function FP of the al-
titude, parametrized by P (which can be An , Bn , or Cn ), and ` Un u ,
m,σ
defined by (141)
where the expressions of Bn and Kn are given by Eqs. (120)

ˆ ˙
1 ´x ` 1
FP pxq “ P ´ e ´ P ` ik̂n epik̂n ´ 2 q x .
˘
(130) and (100), respectively. The semidiurnal tide is supposed to
2
be represented by spatial forcings of the form J px, θ, ϕq “
Then, polarization relations write J2 P22 pcos θq e2iϕ and U px, θ, ϕq “ U2 P22 pcos θq e2iϕ with the
tidal frequency σ “ 2 pΩ ´ norb q. The second-order Love num-
Λm,ν
n H
” ı
pik̂n ` 12 q x , ber, deduced from Eq. (75), is thus expressed
ξr;n
m,σ
“ E n 1 ´ e (131)
R2 σ2
4π G HRatm ÿ 2,σ xatm 2,σ
ż
1 k22,σ “ ´ c2,n,2 δρn,2 pxq dx. (142)
ξθ;n
m,σ
“ 2 tEn e x FAn pxq ´ Dn Jnm,σ ´ εs;n Unm,σ u , 5 U2 nPZ
σ R p1 ´ ε s;n q 0
(132) Assuming |ε s;n | ! 1, we identify the different contributions. It

6
In the case of the semidiurnal tide treated by CL70 (no dissipative
follows
processes), this condition cannot be applied because k̂n P R. To address
this particular situation, CL70 proposes applying a radiation condition k22,σ “ k2;V;therm
2,σ 2,σ
` k2;V;grav 2,σ
` k2;H;therm 2,σ
` k2;H;grav , (143)
at the upper boundary, i.e. considering that the energy is only propagat-
ing upwards and eliminating the terms of the solution corresponding to where the subscripts therm and grav indicate the origin of a con-
an energy propagating downwards. However, in the present work, when tribution (thermally or gravitationally forced) and H and V its
σ0 ą 0, = k̂n ‰ 0, so we do not have to consider this case.
(
horizontal and vertical components. The terms of this expansion
A107, page 14 of 44
are expressed as functions of the parameters of the atmosphere The tidal torque exerted on the stably stratified isothermal at-
and tidal frequency mosphere is thus very small compared to the horizontal thermal
component,
2,σ κ
piσ ` Γ1 σ0 q Bn ` 21 Kn J2
` ˘
ÿ
˘2 U , (144)
2,σ
k2;V;therm “ ´W c2,n,2
piσ ` σ0 q2 ik̂n ´ 21
`
2 R2 ρ0 p0q σ
nPZ TH2,σ “ ´2πκ U 2 J2 2 , (156)
g σ ` σ20
κ piσ ` Γ1 σ0 q Bn ` 12
` ˘
ÿ
˘2 ,
2,σ 2,σ
k2;V;grav “ W c2,n,2 iσ (145) where we have assumed that the thermal forcing is in phase
piσ ` σ0 q2 ik̂n ´ 21
`
nPZ
with the perturber (J2 P R). This behaviour had been identi-
Γ1 ´ 1 J2 fied before, for instance by Arras & Socrates (2010) who stud-
2,σ
k2;H;therm “ ´W , (146) ied thermal tides in hot Jupiters. According to Arras & Socrates
iσ ` σ0 U2
(2010), at the vicinity of synchronization, the vertical displace-
iσ ` Γ1 σ0 ment of fluid due to the restoring effect of the Archimedean
2,σ
k2;H;grav “ ´W , (147) force (N 2 Vr term in the heat transport equation, Eq. (14)) ex-
iσ ` σ0
actly compensates the local density variations generated by the
the parameter Dn being a dimensionless constant given by thermal forcing, which annihilates the quadrupolar tidal torque.
We discuss this effect in Sect. 6 when considering the isother-
4π G HRatm ρ0 p0q mal and stably stratified atmospheres of the Earth and a Venus-
W“´ ¨ (148)
5 c2s like planet (see Fig. 14). For a stably stratified structure, the
2,σ 2,σ
tidal torque exerted on the atmosphere is weak and cannot
Contrary to gravitational Love numbers (k2;V;grav and k2;H;grav ) balance the solid torque, which leads the planet’s spin to the
2,σ synchronization configuration. We note here that the impor-
which are intrinsic to the planet, thermal Love numbers (k2;V;therm
2,σ tance of stable stratification has been pointed out for the case
and k2;H;therm ) are proportional to the forcings ratio J2 {U2 and
of Jupiter-like planets by Ioannou & Lindzen (1993b) (see also
thus depend on the properties of the whole star-planet system, Ioannou & Lindzen 1993a, 1994), who showed that the atmo-
particularly the semi-major axis and the stellar luminosity (e.g. spheric tidal response was very sensitive to variations in the
Correia et al. 2008). Brunt-Väisälä frequency (N).
In the same way as we obtained Love numbers, the torque
exerted on the atmosphere may be computed by substituting
Eq. (141) in Eq. (80). This torque writes 3.5. Comparison with CL70
"ż xatm *
ÿ The vertical wavenumber is a key parameter of the tidal pertur-
2,σ
T “ ´2πR HU2 2 2,σ
c2,n,2 = δρn,2 pxq dx .
2,σ
(149) bation. Hence, to highlight the interest of taking into account
nPZ 0
dissipative mechanisms such as radiation, we consider the rela-
Finally, assuming |ε s;n | ! 1 and introducing the factor tive difference between the k̂n given by Eq. (103) and the value
established via the classical theory of tides without dissipation
2πR2 Hρ0 p0q (e.g. Wilkes 1949, CL70), denoted k̂CL;n in reference to CL70
H “´ (150) and given by
c2s
allows us to write T 2,σ under the form

„ 
1 hc
2
k̂CL;n “ ´1 . (157)
4 hn
2,σ
T 2,σ “ TV;therm 2,σ
` TV;grav 2,σ
` TH;therm 2,σ
` TH;grav , (151)
with This difference, written

# +
κ piσ ` Γ1 σ0 q Bn ` 2
` 1
˘ ˇ ˇ
ÿ ˇ k̂ ´ k̂ ˇ
˘2 J2 ,
2,σ
TV;therm “ ´HU2 c2,σ
2,n,2 Kn = η “ 2ˇ
ˇ n CL;n ˇ
ˇ, (158)
piσ ` σ0 q2 ik̂n ´ 21
`
nPZ ˇ k̂n ` k̂CL;n ˇ
(152)
is plotted in Fig. 6 as a function of the reduced tidal fre-
quency σ{σ0 , and of the height ratio hc {hn . The supplementary
# +
κ piσ ` Γ1 σ0 q Bn ` 2
` 1
˘
ÿ
˘2 U 2 ,
2,σ
TV;grav “ HU2 c2,σ
2,n,2 = iσ term of Newtonian cooling in the heat transport equation induces
piσ ` σ0 q2 ik̂n ´ 12
`
nPZ an additional regime with respect to CL70. At low frequencies,
(153) around the radiation frequency, the vertical profiles are damped,
while they are highly oscillating in the case where σ0 “ 0. This
Γ1 ´ 1
" *
2,σ
TH;therm “ ´HU2 = J2 , (154) thermal regime corresponds to the middle white-yellow region
iσ ` σ0 of the map. The two wavenumbers become similar for |σ| " σ0 .
This is the regime of fast rotating planets studied by CL70. Be-
iσ ` Γ1 σ0
" *
2,σ
TH;grav “ ´HU2 = U2 . (155) yond the threshold corresponding to |σ| „ N, the traditional
iσ ` σ0 approximation assumed in Sect. 2 is no longer valid and the
coupling between the three components of the Navier-Stokes
Let us consider the case of the pure thermal tide ˇ(Uˇ2 « 0) at
equation is considered. Finally, the area located at hc {hn « 1
the vicinity of synchronization. If σ Ñ 0, then ˇk̂n ˇ Ñ `8, represents the discontinuous transition between propagative and
|Bn | Ñ `8, and Bn {k̂n2 9 piσ ` σ0 q { piσ ` Γ1 σ0 q. As a con- evanescent waves in CL70, this transition being regular in our
2,σ 2,σ 2,σ 2,σ
sequence, k2;V;therm « ´k2;H;therm and TV;therm « ´TH;therm . model.
A107, page 15 of 44
A&A 603, A107 (2017)
Therefore, the case N 2 “ 0 corresponds to the adiabatic temper-

ature gradient, given by
κg
H pzq “ H p0q ´ κz or T 0 pzq “ T 0 p0q ´ z. (160)
Rs
For this structure, the reduced altitude is expressed by
ˆ ˙
1 H p0q
x “ ln , (161)
κ H p0q ´ κz
and we thus obtain the distributions of the basic pressure height,
temperature, pressure, and density, respectively
gH p0q ´κx
H pxq “ H p0q e´κx , T 0 pxq “ e ,
Rs (162)
p0 pxq “ p0 p0q e ´x
, ρ0 pxq “ ρ0 p0q e pκ´1qx
,
with H p0q “ p0 p0q { rgρ0 p0qs.
4.2. Tidal response

Fig. 6. Relative difference between the complex wavenumber obtained
in this work (Eq. (103)) on the one hand, and the real one given by CL70 For the sake of simplicity, we neglect the contribution of the
for Earth atmospheric tides (k̂CL;n ) on the other hand. The difference η, gravitational forcing in the Navier-Stokes equation and examine
given by Eq. (158), is plotted in logarithmic scale as a function of the the case of pure thermal tides. When σ Ñ 0, the tidal response
normalized tidal frequency σ{σ0 and of the ratio hc {hn . The colour of tends to its hydrostatic component, the equilibrium tide (Zahn
the map, denoted c, is defined by the expression c “ log pηq. The hori- 1966). We thus assume the hydrostatic approximation. This al-
zontal and vertical axis represent f pσ{σ0 q and f phc {hn q, where f is the lows us to reduce the radial projection of the Navier-Stokes equa-
function defined by f pXq “ sign pXq log p1 ` |X|q with X P R. The po- tion (Eq. (8)) to
sitions of the planets are determined by the gravity mode of degree 0 in
the case of the semidiurnal tide. Dark (luminous) regions correspond to 1 Bp
the dynamic (dissipative) regime. The vertical wavenumbers of Venus δρ “ ´ ¨ (163)
are strongly affected by dissipative processes, while those of the Earth gH Bx
are not. Now, by expanding the perturbed quantities in Fourier series and
substituting Eq. (163) in the heat transport equation (Eq. (20)),
we get the simplified vertical structure equation of pressure
4. Tides in a slowly rotating convective atmosphere dδp κρ0 J 1 iσ ` Γ1 σ0
` τp δp “ , where τp pσq “
In this section, we treat the asymptotic case of a slowly rotating dx iσ ` Γ1 σ0 Γ1 iσ ` σ0
convective atmosphere for which the equations of tidal hydro- (164)
dynamics can be simplified drastically. This situation typically
corresponds to Venus-like planets. Indeed, the rotation period of represents the complex damping rate of the perturbation. Follow-
Venus, 243 days, is of the same order of magnitude as its or- ing Dobrovolskis & Ingersoll (1980) and Shen & Zhang (1990),
bital period. It follows that |ν| ă 1 for the semidiurnal tide of we consider that thermal tides are generated by an increase in
Venus. As a consequence, the Coriolis acceleration can be ne- heat at the ground and apply the profile of thermal forcing
glected as the first step. The atmospheric layer located below
J pxq “ J0 τJ e´τJ x , (165)
60 km is characterized by a strongly negative temperature gra-
dient (Seiff et al. 1980). In this layer, the temperature decreases where 1{τJ ! 1 is the thickness of the heated region (see
from 750 K to 250 K, which make it subject to convective in- Eq. (213) for a physical expression of this parameter) and J0
stability (Baker et al. 2000). Therefore, the Brunt-Väisälä fre- the mean absorbed power per unit mass. At the upper boundary
quency of Venus is far lower than the value given by the isother- xatm " 1, we apply the free-surface condition δp “ 0 and get the
mal approximation. As this frequency represents the “stiffness” pressure profile
of the Archimedean force which restores gravity waves, these
waves cannot propagate in the unstable region above the surface. κρ0 p0q J0 ´τJ x
δp pxq “ ´ e , (166)
iσ ` σ0
4.1. Equilibrium distributions of pressure, density, from which we deduce the vertical profiles of the Lagrangian
and temperature horizontal displacements
κJ0
We assume Ω “ 0 and N “ 0. According to Eq. (1), the Brunt- ξθ pxq “ ´ e´τJ x , (167)
Väisälä frequency can be expressed as Rσ2 piσ ` σ0 q
g κJ0
ˆ ˙
dH
2
N “ κ` ¨ (159) ξϕ pxq “ ím e´τJ x , (168)
H dz Rσ2 piσ ` σ0 q
A107, page 16 of 44
of the horizontal velocities

1.00
κJ0
Vθ pxq “ í e´τJ x , (169)
Rσ piσ ` σ0 q
0.50
T / max(T)
κJ0
Vϕ pxq “ m e´τJ x , (170) 0.00
Rσ piσ ` σ0 q
and of the density and temperature −0.50
κρ0 p0q τJ J0
δρ pxq “ ´ e´τJ x , (171) Conv
gH p0q piσ ` σ0 q −1.00
CL01
−4 −2 0 2 4
κτJ J0 / 0
δT pxq “ ´ e´τJ x . (172)
Rs piσ ` σ0 q Fig. 7. Tidal torques exerted on the atmosphere in the case of the
slowly rotating convective atmosphere (red line with the label “Conv”,
4.3. Second-order Love number and tidal torque Eq. (174)) and in the model by Correia & Laskar (2001) (green dashed
line with the label “CL01”, Eq. (175)) as functions of the tidal fre-
The previous results enable us to compute the tidal Love num- quency. Both functions are normalized by their respective maximum
bers and torque associated with the atmospheric tidal response. value, reached at σ “ σ0 ; the tidal frequency is normalized by the
Owing to the hydrostatic approximation, these quantities are di- Newtonian cooling frequency (σ0 ). The tidal forcing is quadrupolar,
rectly proportional to δp px “ 0q. We consider that the semid- with U and J given by Sect. 4.3. The parameters a and b of Eq. (175)
iurnal thermal tide can be reduced to its quadrupolar compo- are adjusted so that the peaks of the two functions are superposed.
nent, given by U px, θ, ϕq “ U2 P22 pcos θq ei2ϕ and J px, θ, ϕq “
J2 τJ e´τJ x P22 pcos θq ei2ϕ . Hence, using the expressions Eq. (76) functions of the tidal frequency, T pσq looks like T 2,σ pσq from
for the second-order Love number and Eq. (82) for the tidal Eq. (174) with amplitude maxima located at
torque, we obtain c
2
4π G RHκρ0 p0q J2 σ0 „ ˘ ¨ (176)
2,σ
k2;conv “ (173) 3b
5 iσ ` σ0 U2 This strengthens the statement of Correia & Laskar (2001), who
and expected that “further studies of atmospheres of extra solar syn-
chronous planets r...s may provide an accurate solution for the
2,σ κρ0 p0q σ case σ « 0”. In Fig. 14, the expression of Eq. (175) is plot-
Tconv “ ´2πR2 U2 J2 2 ¨ (174)
g σ ` σ20 ted for the Earth and Venus in addition to the analytical results
of the model presented here with parameters a and b adjusted
numerically.
4.4. Comparison with previous models
The tidal torque plotted in Fig. 7 is identical to the hori- 5. Physical description of heat source terms
zontal component of the torque exerted on the stably strati-
fied isothermal atmosphere given by Eq. (156) because con- On the left-hand side of Eq. (46) and in polarization relations, the
vection eliminates the component resulting from the fluid perturbation is driven by Unm,σ and Jnm,σ . The tidal gravitational
vertical displacement. In agreement with the early result of potential has been expanded in spherical harmonics for many
Ingersoll & Dobrovolskis (1978) (Eq. (4)), it is of the same years and there is no need to come back to it7 . Nevertheless, it is
form as the result given by the Maxwell model (Correia et al. necessary to establish physical expressions of the heat power per
2014), where we identify the Maxwell time T0 “ σ0´1 . It also unit mass to compute a solution. The different contributions must
corresponds to the torque computed by Leconte et al. (2015) be clearly identified. We detail them in the case of the optically
for Venus-like planets with numerical simulations using GCM, thin atmosphere. The most obvious heat source, but not neces-
which suggests that the slow rotation and convective instabil- sarily the one with the largest amplitude, is the absorption of the
ity approximations are appropriate for planets of this kind. This light flux coming from the star. The non-absorbed part of this
torque is stronger than that computed in the case of the sta- flux reaches the ground and causes temperature oscillations at
bly stratified atmosphere. Consequently, it can lead to non- x “ 0. This implies two contributions from the ground: a radia-
synchronized rotation states of equilibrium by counterbalanc- tive emission in infrared and diffusive heating due to turbulence
ing the solid tidal torque. Finally, it must be compared to the within the surface boundary layer.
model introduced by Correia & Laskar (2001) in early theoreti-
cal studies of the tidal torque exerted on Venus atmosphere. In 5.1. Insolation
this model, the tidal torque is given by
The heating fluxes coming from the ground are all oriented ra-
a” 2
ı
dially and proportional to the local insolation flux at x “ 0.
T pσq “ 1 ´ e´bσ , (175)
σ 7
For instance, see the Kaula expansion of the gravitational
where a and b are two empirical positive real parameters. As tidal potential (Kaula 1964) detailed in Appendix C and
can be observed in Fig 7 where (174) and (175) are plotted as Mathis & Le Poncin-Lafitte (2009).
A107, page 17 of 44
A&A 603, A107 (2017)
Therefore, they can be easily written as a product of separable

functions: F pθq G pxq. This is not the case of the insolation flux, x x
which goes through the spherical shell along the star-planet di-
rection (see Fig. 10). Particularly, the atmosphere is partly en-
lightened by stellar rays on the dark side. We note IS , the bolo-
metric flux transmitted to the atmosphere. If we assume that the
I(x+dx) I(x+dx)
star radiates like a black body, this flux is given by the integral
of the spectral radiance IS;λ over a wide range of wavelengths
ż dx dx
IS “ IS;λ pλq dλ, (177)
λ
I(x) I(x)
where IS;λ is given by Plancks’s law (Planck 1901):
R2S 2hc2 πλ´5
IS;λ pλq “ , (178) 0 0
d2 e kBhcTS λ ´ 1
the parameter c being the speed of light in vacuum, h the Planck
constant, kB the Boltzmann constant, d the star-planet distance,
RS the radius of the star, and T S its surface temperature. The flux Fig. 8. Left: absorption of the flux emitted by the star along the star-
is partly reflected by the atmosphere with the albedo Aatm;λ for planet direction. Right: absorption of the flux emitted by the ground.
the wavelength λ, which gives the effective heating flux
I0;λ pλq “ p1 ´ Aatm;λ q IS;λ . (179) Near the planet-star axis, they are simply expressed as
$ ż
To describe the absorption of I0 by the atmosphere, we use the ’
’
’ I inc
“
´x
I0;λ e´τλ e dλ,
Beer-Lambert law (Bouguer 1729; Klett et al. 1760; Beer 1852). & λ
Light rays propagate along a straight line defined by the linear ż (186)
’ 1 ´x
spatial coordinate l. So, the absorption of the incident spectral % J inc “ ελ I0;λ e´τλ e dλ,
’
’
density Iλinc by a gas G of molar concentration CG , illustrated by M λ
Fig. 8, can be written where τλ is the dimensionless optical depth for the radiance of
dIλinc wavelength λ,
“ ´εG;λCG plq dl, (180)
Iλinc ρ0 p0q H
τλ “ ελ . (187)
M
the parameter εG;λ being the molar extinction density coeffi-
cient of the gas, which is supposed to depend on the light wave- The parameter 1{τλ corresponds to an absorption depth of the
length (λ) alone. This dependence should be taken into account flux in a homogeneous fluid normalized by the pressure height
because εG;λ can vary over several orders of magnitude, typi- scale (H). If τλ ! 1, the heating power of a wavelength λ is al-
cally, εG;λ 9 λ´4 in the regime of Rayleigh diffusion (molecules most totally transmitted to the ground. On the contrary, if τλ " 1,
are small compared to the wavelength). From Eq. (180), we get the flux does not reach the surface because it is entirely absorbed
ş by the atmosphere (see Fig. 9). In this case, the atmosphere is not
Iλinc “ I0;λ e´εG;λ l C G dl , (181) optically thin, and it cannot be assumed that its radiative losses
are proportional to δT as was done in Sect. 2, and thermal dif-
and the incident flux
ż fusion has to be considered. Therefore, the case treated by the
I inc “ Iλinc dλ. (182) present work corresponds to τλ ! 1.
λ The proportion of flux density transmitted to the ground αλ
is deduced from Eq. (186) taken at x “ 0,
Then, performing the derivation of Eq. (181), the heat power per
unit volume is obtained, αλ “ e´τλ . (188)
ż ş
J “ CG εG;λ I0;λ e´εG;λ l CG dl dλ,
inc
(183) Like the atmosphere, the ground has an albedo denoted Agr;λ ,
λ which implies that the upcoming flux is partly reflected. There-
and the heat power per unit mass comes straightforwardly fore, we can write the absorption equation
CG
ż dIλref
` τλ e´x Iλref “ 0,
ş
J inc “ εG;λ I0;λ e´εG;λ l CG dl dλ. (184) (189)
ρ0 λ dx
We retrieve the formulation of the thermal forcing proposed by and compute the reflected surface power
CL70. Considering that the gas is homogeneous in composition
Iλref “ Agr;λ αλ Iλinc pR, ψq cos ψeτλ e ,
´x
leads us to CG “ C0 “ ρ0 {M. Therefore, the incident power flux (190)

and heat power per unit mass become the coordinate ψ being the angle between the star-planet direc-
$ ż ş tion and the position vector of the current point. The heat power
inc
’
’
’ I “ I0;λ e´ελ l C0 dl dλ, per unit mass and wavelength provided by the reflected flux is
λ
&
(185) derived from the previous equation
ż
1 ελ
’ ş
inc
ελ I0;λ e´ελ l C0 dl dλ. Jλref “ Agr;λ αλ Iλinc pR, ψq cos ψeτλ e .
´x
’
%J “
’
(191)
M λ M
A107, page 18 of 44
6 6
= 1.0E 02 = 1.0E 02
= 1.0E 01 = 1.0E 01
5 = 1.0E+00 5 = 1.0E+00
= 1.0E+01 = 1.0E+01
4 = 1.0E+02 4 = 1.0E+02
x
3
x
2 2
1 1
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Iinc / max(Iinc) Irad / max(Irad)

6 6
= 1.0E 02 = 1.0E 02
= 1.0E 01 = 1.0E 01
5 = 1.0E+00 5 = 1.0E+00
= 1.0E+01 = 1.0E+01
4 = 1.0E+02 4 = 1.0E+02
3
x
2 x 2
1 1
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
inc inc rad rad
j / max(j ) j / max(j )
6 6
= 1.0E 02 = 1.0E 02
= 1.0E 01 = 1.0E 01
5 = 1.0E+00 5 = 1.0E+00
= 1.0E+01 = 1.0E+01
4 = 1.0E+02 4 = 1.0E+02
3 3
x
x
2 2
1 1
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Jinc / max(Jinc) Jrad / max(Jrad)
Fig. 9. Vertical profiles of radiative fluxes (I), power per unit volume (J ), and power per unit mass (J) for various values of the dimensionless
optical depth τλ (Eq. (187)). As usual, the reduced altitude, x “ z{H, is on the vertical axis. Left: incident flux, defined by Eq. (186), at ψ “ 0.
If τλ ! 1 the ground receives almost 100% of the total power. On the contrary, if τλ " 1, all the power is absorbed by the atmosphere. Right:
radiative flux emitted by the ground, defined by Eq. (206). For small τλ , the flux goes through the atmosphere without being absorbed. For high
values of the parameter, it is absorbed by the lowest layers.
So, the contribution of the reflected flux is finally obtained considered to be a black body of temperature T gr , emitting the
$ ż power flux Igr;BB given by the Stefan-Boltzmann law,
Agr;λ αλ Iλinc pR, ψq cos ψeτλ e dλ
´x
ref
’ I “
Igr;BB “ gr S T gr
4
,
’
’
& λ
(193)
(192)
ż
ελ inc the parameter gr being the emissivity of the ground and S “
I pR, ψq cos ψeτλ e dλ.
’ ´x
% J ref “ Agr;λ αλ
’
M λ 5.670373 ˆ 10´8 W m´2 K´4 the Stefan-Boltzmann constant
’
λ
(Mohr et al. 2012). Moreover, the atmosphere emits a counter ra-
diation to the surface, which implies that the effective flux is less
than Igr;BB . According to Bernard (1962), this counter-radiation
5.2. Radiative heating from the ground
can be assumed to be proportional to Igr;BB , and expressed by the
The part of the incident flux which is not reflected causes surface semi-empirical formula
temperature oscillations. This effect was studied in meteorologi-
Iatm “ atm S T gr
4
, (194)
cal works throughout the twentieth century. We will follow here
the theoretical approach proposed by Bernard (1962) for its sim- which allows us to take it into account without studying
plicity and the physical landmarks that it brings. The ground is the whole coupled system ground-atmosphere. The factor atm
A107, page 19 of 44
A&A 603, A107 (2017)
z z
and finally,
0.9 0.9
1.0 1.0 ż
αλ eτλ e dλ.
´x
0.5
0.7
0.5
0.7 I rad “ Igr;λ
rad
(201)
λ
0.0 x 0.5 0.0 x 0.5
The temperature of the ground T gr depends on the power reach-
- 0.5 0.3 - 0.5 0.3
ing the surface,
- 1.0 0.1 - 1.0 0.1
inc
Igr “ I inc pR, ψq cos ψ. (202)
- 1.0 - 0.5 0.0 0.5 1.0 - 1.0 - 0.5 0.0 0.5 1.0
z z
inc inc
0.9 0.9 Both quantities are linearized near the equilibrium, Igr “ Igr;0 `
1.0 1.0
0.7 0.7
δIgr and T gr “ T gr;0 ` δT gr , and the perturbation is expanded in
inc
0.5 0.5
series
0.0 x 0.5 0.0 x 0.5 ÿ
δIgr
inc
“ δIgr
inc;m,σ ipσt`mϕq
e ,
- 0.5 0.3 - 0.5 0.3
σ,m
ÿ (203)
- 1.0 0.1 - 1.0 0.1
δT gr “ δT gr
m,σ ipσt`mϕq
e .
- 1.0 - 0.5 0.0 0.5 1.0 - 1.0 - 0.5 0.0 0.5 1.0 σ,m
inc
Fig. 10. Left, from top to bottom: power per unit mass (J , Eq. (184)) The spatial functions δIgrinc;m,σ
and δT gr
m,σ
are related by a transfer
and power per unit volume (J inc , Eq. 183) of a monochromatic in-
cident flux normalized by their maxima and plotted as functions of
function of the tidal frequency, denoted Bgr , that is given explic-
the normalized spatial coordinates X (the direction of the star is along itly in the next subsection
the horizontal axis) and Z (the spin axis is along the vertical axis). σ inc;m,σ
The planet is represented by a green disk of unit radius. The flux, of δT gr
m,σ
“ Bgr δIgr . (204)
wavelength λ, propagates in an atmosphere assumed to be homogeneous
in composition (CG “ C0 ) and characterized by the ratio H{R “ 0.1. So, the perturbation of the spectral radiance emitted by the
The optical depth is given by τλ “ 0.2. Right, from top to bottom: power ground can be written
per mass unit (J rad , Eq. (206)) and power per volume unit (J rad ) of a ˇ
rad ˇ
monochromatic flux radiated by the ground, normalized by their max- BIgr;λ
τλ e´x σ inc;m,σ
δIλ
rad;m,σ
“ αλ e Bgr δIgr , (205)
ˇ
ima, and plotted as functions of X and Z, with τλ “ 0.5. ˇ
BT gr ˇ
T gr “T gr;0
corresponds to an effective emissivity. So, introducing the spec- and the flux and heat power per mass unit
tral radiance of the atmosphere Iatm;λ , the albedo of the ground
ˇ
in the infrared Agr;IR can be defined by the relationship
$ rad ˇ
ż BIgr;λ
’ δI σ τ ´x
rad;m,σ
“ Bgr δIgr
inc;m,σ
αλ e e
’ λ
dλ,
ż ’ ˇ
’ ˇ
’ λ BT gr ˇ
Agr;IR Iatm “ Agr;λ Iatm;λ dλ, (195)
&
T gr “T gr;0
λ rad ˇ
ˇ
ελ σ inc;m,σ BIgr;λ
’ ż
’ δJ τ ´x
rad;m,σ
“ Bgr δIgr αλ e e
’ λ
dλ.
ˇ
and the effective flux emitted by the telluric surface can be
’ ˇ
’
% M λ BT gr ˇ
written T gr “T gr;0
Igr “ Igr;BB ´ IR
Agr Iatm . (196) (206)
rad
IR
Given that Agr « 1 ´ gr , Eq. (196) becomes The partial derivative of Igr;λ is explicitly given by
ˇ
ehc{pkB λTgr;0 q
rad ˇ
rad
Igr “ S T gr
4
, (197) BIgr;λ 2πh2 c3 λ´6
“ (207)
ˇ
ˇ 2 ” ı2 ¨
where “ gr p1 ´ atm q ď 1 is the effective emissivity of the
BT gr ˇ kB T gr;0 hc{pkB λT gr;0 q
T gr “T gr;0 e ´1
ground. The terrestrial atmosphere being rather thin and trans-
parent, « 1 for the Earth. This expression will be used further Here, we note that the response of the ground to the tidal forcing
to establish the heating by turbulent diffusion. We also introduce induces a dependence on the tidal frequency through the oscilla-
σ
here the corresponding spectral radiance of the ground, tions of T gr and the transfer function Bgr . Thus, the spatial func-
tions describing the contribution of the ground are parametrized
2hc2 πλ´5
rad
Igr;λ “ ¨ (198) by σ contrary to those associated with the incident flux.
ehc{pkB λTgr q ´ 1
The radiative flux emitted by the ground, denoted I rad , is the so- 5.3. Boundary layer turbulent heat diffusion
lution of the absorption equation (similar to the one used to com-
pute I inc ), Near the ground, heat transfers are dominated by turbulent mech-
anisms and diffusion. This is known as the planetary boundary
dIλrad layer and it is illustrated in Fig. 12. Thus, for x ! 1, thermal
` τλ e´x Iλrad “ 0, (199) diffusion cannot be ignored as it is in Sect. 2 and drives the os-
dx
cillations of temperature. However, we note that it is only valid
with the boundary condition Iλrad “ Igr;λ
rad
at x “ 0. Therefore, at the vicinity of the surface, where friction with the ground is
important. After CL70 and Siebert (1961), we neglect as a first
Iλrad “ αλ eτλ e Igr;λ
´x
rad
, (200) step the heat transport through the troposphere due to advection
A107, page 20 of 44
to provide a first analytical treatment of the problem. At the in-

/ bl
terface, the diffusive fluxes in the ground and in the atmosphere 0.4
write
ˆ ˙ ˆ ˙
kgr BT katm BT
Qgr ptq “ and Qatm ptq “ ´ ,
Im(Bgr) / B0gr
H Bx x“0´ H Bx x“0` 0.2
(208)
the parameters kgr and katm representing the thermal conductivi- 0
ties of the ground and of the lowest layer of the atmosphere. In
addition, denoting the associated thermal capacities cgr and catm ,
we introduce the thermal diffusivities
−0.2
kgr katm
Kgr “ and Katm “ , (209)
ρ0 p0 ´ q cgr ρ0 ` q catm
p0
−0.4
the parameter Katm being the vertical turbulent thermal diffusiv-
ity. Therefore, temperature variations near the surface are de-
scribed by the equations of heat transport (e.g. CL70), 0 0.2 0.4 0.6 0.8 1
Re(Bgr) / B0gr
$
K 2
’ BT “ gr B T for x ă 0,
’
H 2 Bx2
’
& Bt
(210) σ
2 Fig. 11. Nyquist diagram of the boundary layer transfer function Bgr in
’ BT K atm B T
the frequency range ´10 ď σ{σbl ď 10 (σ being the tidal frequency
’ 3 3
for x ą 0,
’
% “ 2
Bt H Bx2 and σbl the frequency characterizing the boundary layer introduced in
σ
where only the diffusive terms are taken into account. The power Eq. (217)). The parametrized complex transfer function Bgr (Eq. (215))
reduced by its value at σ “ 0 is plotted as a function of the tidal fre-
balance of the surface is deduced from Eqs. (208) and (197),
quency (σ) in the complex plane, where (the horizontal axis corresponds
σ
to the real part of the function, < Bgr 0
{Bgr , and the vertical axis to
inc
Igr ´ S T gr
4
´ Qgr ´ Qatm “ 0, (211) σ
the imaginary part = Bgr {Bgr . The arrow indicates the direction of
( 0
and provides the temperature T gr;0 of the ground at equilib- increasing σ. The gain of the transfert function is given by the distance
rium. Linearized with variables of the form given by Eq. (21), of a point P of the curve to the origin O of coordinates p0, 0q, and its
Eq. (210) can be written phase by the angle of the vector OP to the horizontal axis. This plot
shows that the response of the ground is always delayed with ` respect
to the thermal forcing except at synchronization, where arg Bgr “ 0
˘
σ r1`signpσqisx{δσgr inc;m,σ
& δT “ Bgr δIgr
$ m,σ
e for x ă 0,
and the gain is maximum. It also highlights the two identified regimes
(212) of the ground response: the lag tends to 0 (response in phase with the
σ ´r1`signpσqisx{δσatm inc;m,σ
δT “ Bgr δIgr
% m,σ
e for x ą 0, perturbation) and the gain increases at σ Ñ 0, while the lag tends to
˘π{4 and the gain to 0 at σ Ñ ˘8, the critical transition frequency
where δσgr and δσatm are the tidal frequency-dependent skin thick- being σbl .
nesses of diffusive heat transport in the solid and fluid parts, re-
spectively. Their expressions are given by
d d an expression that can be decomposed in its modulus and
1 2K 1 2Katm argument
δσgr “
gr
and δσatm “ ¨ (213) $ˇ
H |σ| H |σ| ˇB σ ˇ “ 1
,
’ ˇ
’
’
’ gr c b
It should be noted that δσgr and δσatm both decay when the tidal fre- ςgr 1 ` 2 σbl ` 2 σbl
|σ| |σ|
’
’
’
quency increases. Since δσatm corresponds to the typical depth of
&
» fi (216)
the boundary layer, it has to satisfy δσatm ! 1, otherwise the diffu- ’
’ ` σ˘ 1
’ arg Bgr “ ´sign pσq arctan –
’ — ffi
sive effects cannot be ignored in the dynamical core of the model ’
’ b fl ¨
σbl
(Sect. 2). Let us denote ςgr “ 4S T gr;0 3
the radiative impedance 1`
’
%
|σ|
of the ground considered as a perfect black body and submitted σ
to a perturbation in temperature. The transfer function Bgr σ
in- The frequency σbl , which parametrizes Bgr , is a characteris-
troduced in Eq. (204) can be deduced from the linearized power tic frequency reflecting the thermal properties of the diffusive
balance, boundary layer, and is expressed
ςgr
ˇ ˇ ˆ ˙2
kgr BδT m,σ ˇˇ katm BδT m,σ ˇˇ σbl “ 2 , (217)
δIgr
inc;m,σ
´ ςgr δT gr ´
m,σ
` “ 0, βatm ` βgr
H Bx ˇ0´ H Bx ˇ0`
(214) where the parameters βgr and βatm are the thermal conductive
capacities of the ground and of the atmosphere near the telluric
in which the expressions of Eq. (212) are substituted. We finally surface
get (see Fig. 11) ?
βgr “ ρ0 p0´ q cgr Kgr and βatm “ ρ0 p0` q catm Katm . (218)
a
σ 1 1
Bgr “ , (215) According to Bernard (1962), turbulent diffusion plays a more
ςgr
b
|σ|
1 ` r1 ` sign pσq is σbl important role than thermal conduction in the ground, which
A107, page 21 of 44
A&A 603, A107 (2017)
expression. Indeed, by writing the heat transport equation as
BδT
catm “ δJ diff (220)
Bt
and linearizing it, we compute the expression of the tidal diffu-

sive forcing term as
σ
σ ´r1`signpσqisτatm x inc;m,σ
δJ diff;m,σ pxq “ iσcatm Bgr e δIgr . (221)
5.4. Simplified bichromatic case

The expressions of the thermal forcings obtained in the previous
section may appear of some complexity because of the depen-
dence of albedos and optical depths on the wavelength. They can
be simplified notably if each source is assumed to be monochro-
matic. In this case, the star emits in the frequency range of vis-
ible light, and the planet in the range of infrared. Consequently,
the wavelength-dependent Agr;λ , Aatm;λ , ελ , λ , αλ , and τλ are re-
duced to the constant coefficients Agr;V , Aatm;V , εV , V , αV , and
τV in the range of visible light and Agr;IR , Aatm;IR , εIR , IR , αIR ,
and τIR in the infrared. Hence, the bolometric flux is given by
R2S
IS “ S T S4 , (222)
d2
where T S is the temperature of the star. This flux is partly re-

flected back to space by the atmosphere so that the effective
Fig. 12. Bichromatic case. Top: heat transfers between the atmo- flux I0 getting through the layer can be written
sphere, space and the ground. Bottom: power balance of the interface
ground-atmosphere. I0 “ p1 ´ Aatm;V q IS . (223)
The power of the incident flux I inc decays along the path of a
means βatm " βgr . However, the turbulent diffusivity Katm may light beam, absorbed by the gas. The power absorbed per unit
vary over a wide range of orders of magnitude, taking extremal mass is denoted J inc and the expressions of Eq. (185) become
values above oceans (typically βatm „ 104 J m´2 K´1 s´1{2 ).
For these reasons, Bernard (1962) prescribes for the Earth ş
,
$ inc ´εV l C0 dl
I “ I 0 e
the effective mean value Katm „ 10 m2 s´1 , which had ’
&
been prescribed before by Wilkes (1949) and which leads to (224)
σbl „ 10´6 s´1 . % J inc “ εV I e´εV l C0 dl .
’ ş
0
As shown by Eq. (216), the frequency ratio |σ| {σbl deter- M
mines the angular delay of the ground temperature variations.
If |σ| ! σbl , the response is in phase with the excitation. It From I inc , we deduce the flux reflected by the ground I ref and the
corresponds to low conductive capacities. On the contrary, if power absorbed per mass unit J ref , given by
|σ| " σbl , then the delay tends to its asymptotical limit, π{4.
τV e´x
By using the frequency σbl „ 10´6 s´1 derived from the pre- & I “ Agr;V αV I p0, ψq cos ψe ,
$ ref inc
’
scriptions of Wilkes (1949) and Bernard (1962), we note that the (225)
Earth’s solar diurnal tide belongs to the second category. This % J ref “ A α εV I inc p0, ψq cos ψeτV e´x .
’
explains the position of the diurnal peak observed in surface tem- gr;V V
M
perature oscillations, three hours late with respect to the subso-
σ
lar (midday) point. The gain of Bgr is damped by the conductive Finally, the expressions of Eq. (206) are simply replaced by
capacities of the material at the interface. Therefore, a very dif-
fusive interface tends to attenuate the temperature oscillations of σ inc;m,σ
αIR eτIR e ,
´x
& δI “ ςgr Bgr δIgr

$ rad;m,σ
the ground, as expected. ’
We now consider the profile of temperature variations in the (226)
boundary layer obtained in Eq. (212) % δJ rad;m,σ “ εIR ς B σ δI inc;m,σ α eτIR e´x .
’
gr gr gr IR
M
σ
σ ´r1`signpσqisτatm x inc;m,σ
δT bl pxq “ Bgr e δIgr , (219) The expressions of the powers per unit mass of the incident and
radiated flux (J inc and J rad ), and the corresponding powers per
where τσatm “ 1{δσatm is the vertical damping ratio of the diffu- volume unit (J inc and J rad ), are plotted in Fig. 10. The heat-
sive perturbation. Chapman & Lindzen (1970) shows that a heat ing due to the turbulent boundary layer remains unchanged (see
power due to diffusion J diff can be deduced from the previous Eq. (221)).
A107, page 22 of 44
6. Application to the Earth and Venus 5

5_Args
4 2_Args
To illustrate the regimes identified and the dependence of the
3 data
atmospheric response on the forcing frequency, we apply the
model to the cases of the semidiurnal thermal tides on Earth 2
temp (°C)
and Venus assuming first for both planets isothermal stably strat- 1
ified atmospheres, as described by the equations in Sect. 3. 0
We note here that the atmosphere of Venus is not stably strat- -1
ified but convective in regions close to the surface (Seiff et al. -2
1980; Baker et al. 2000) where the tidal mass redistribution is the -3
most important (Dobrovolskis & Ingersoll 1980; Shen & Zhang -4
1990). Nevertheless, examining the case of a stably stratified 0 2 4 6 8 10 12 14 16 18 20 22 24
isothermal atmosphere for a Venus-like planet will allow us to 0.6
better understand in an academic framework the role of stratifi- 5_Args
0.4 2_Args
cation and radiative losses in the tidal response. To make a clear data
difference between Venus and the studied stably stratified Venus- 0.2
p (mbar)
like planet, we call the latter “VenusX”. For the sake of simplic-
ity, we assume that planets orbit circularly in their equatorial 0
plane. Hence, the tidal frequency8 is given by σ “ 2 pΩ ´ norb q.
-0.2
One of the most crucial parameters of the model is the Newto-
nian cooling frequency (σ0 ). This parameter varies over several -0.4
orders of magnitude with the altitude (Pollack & Young 1975).
However, our purpose here is not to compute a quantitative tidal -0.6
perturbation, but to illustrate qualitatively the non-dissipative 0 2 4 6 8 10 12 14 16 18 20 22 24
(Earth) and dissipative (Venus) regimes. Therefore, following solar time (hour)
Lindzen & McKenzie (1967), we consider the interesting case
where the radial profile σ0 is assumed to be constant. For both Fig. 13. Average daily variation in temperature in degrees (top) and
pressure in mbar (bottom) over one year, plotted over solar time (in
planets, we arbitrarily set σ0 “ 7.5 ˆ 10´7 s´1 , which is the hours), the subsolar point corresponding to 12 h. The data was recorded
effective value of σ0 computed by Leconte et al. (2015) with a every minute over the full year 2013 with a Vantage Pro 2 weather sta-
GCM for Venus. This value is such that σ „ σ0 for Venus and tion at latitude 48.363˝ N. The time is converted to solar time and the
σ " σ0 for the Earth. In the case of an optically thin atmo- data averaged over an exact count of 365 days. The average values of
sphere, the Newtonian cooling frequency can be estimated using temperature (10.8973˝ C) and pressure (1015.83 mbar) have been re-
Eq. (17). Moreover, we impose academic quadrupolar perturba- moved to display only the variable part. In both figures, the raw data (in
tions of the form green) has been decomposed in a Fourier series over the 24 h period,
limited to two harmonics (blue), or five harmonics (red). Temperature
U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq , is dominated by the diurnal component, with a maximum value around
(227) 14h26mn. The largest harmonics of the pressure variations is the semid-
iurnal term, followed by the diurnal term. Two pressure maxima can be
J px, θ, ϕ, tq “ J2 P22 pcos θq e
ipσt`2ϕq
, observed, at 9h38mn and 22h07mn.
where U2 and J2 are fixed constant radial profiles. The tidal po-
tential U2 is computed from the Kaula multipole expansion de- considering that the scale of the vertical variations is defined by
tailed in Appendix D, and explicitly given by the expression of the module of the complex wavenumber given in Eq. (107).
Eq. (74). A zero-order approximation of the thermal power J2 For each planet, we give here the spatial distribution of per-
can be quantified by writing the total surface power absorbed by turbed quantities and the evolution of the tidal torque with the
the atmosphere as a function of θ and ϕ, which is then expanded tidal frequency. The numerical values used in these simulations
in Fourier series of longitude and associated Legendre polyno- are summarized in Table 1.
mials. Given that atmospheric tides are considered in this model
as linear perturbations around the equilibrium state, the ampli-
tudes of the perturbed quantities are proportional to the forcing. 6.1. The Earth
The horizontal structure equation is solved using the spectral
method described in Appendix B. The vertical structure equation The Earth’s semidiurnal thermal tide corresponds to one of the
is integrated numerically on the domain x P r0, xatm s using a cases detailed in CL70. Here, the radiation frequency is clearly
regular mesh with element of size δx. As pointed out by CL70, negligible compared to the tidal frequency σ « 2Ω. Therefore,
the number of points in the mesh, denoted N, must be sufficiently the tidal response of the Earth’s atmosphere is driven by dy-
large to obtain vertical profiles of the perturbation with a good namical effects only and the radiative losses added in the heat
accuracy (see Appendix G). To fix N correctly in the case of the transport equation can be ignored. This case is typical of a pure
Earth, CL70 suggests using a criterion that we adapt for other dynamic behaviour (see Figs. 5 and 6). Given that ν is slightly
configurations, i.e. larger than 1, the horizontal structure of the perturbations is es-
sentially described by gravity modes (Fig. 2). Rossby modes are
trapped at the poles. Figures 15 and 16 show maps of the per-
ˇ ˇ
xatm ˇk̂n ˇ
N ą 100 , (228) turbed quantities in latitude (denoted δ), longitude (ϕ), and al-
2π titude (x), the subsolar point being indicated by the coordinates
8
This frequency corresponds to the term defined by pl, m, j, p, qq “ pδ, ϕq “ p0, 0q. In these figures, we can observe that the pressure
p2, 2, 2, 0, 0q in the multipole expansion of the forcings detailed in and density variations present very similar spatial distributions.
Appendix C. Particularly, they are in phase with each other and the phase lag
A107, page 23 of 44
A&A 603, A107 (2017)
6.00 8.00
6.00
4.00
4.00
N.m]
T [1016 N.m]
2.00
2.00
0.00 0.00
16
T [10
−2.00
−2.00
T −4.00 T
−4.00 V V
−6.00
H H
−6.00 CL01 −8.00 CL01
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
Fig. 14. Tidal torque applied to the atmosphere of Earth (left panel) and Venus (right panel) by the thermal semidiurnal tide and its components as
functions of the reduced frequency χ “ pΩ ´ norb q {norb (horizontal axis). The forcings consist of academic quadrupolar perturbations expressed
as U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´0.985 m2 s´2 and J2 “ 1.0 ˆ 10´2 m2 s´3 for the
Earth and U2 “ ´2.349 m2 s´2 and J2 “ 1.0 ˆ 10´4 m2 s´3 for Venus. Labels T, H, and V refer to the total response and its horizontal and
vertical components, respectively (we note that TH „ Tconv ). The label CL01 refers to the model by Correia & Laskar (2001), given by Eq. (175).
Table 1. Values of physical parameters used in simulations. torque is computed using the analytical formulae of Eqs. (141)
and (149). We also compute the horizontal and vertical compo-
Parameters Earth Venus nents isolated in Sect. 3. The results are plotted in Fig. 14.
R [km] 6371.0 6051.8 At first, we note that the horizontal and vertical components
g rm s´2 s 9.81 8.87 of the torque are of the same order of magnitude and in oppo-
H [km] 8.5 15.9 sition to one another, as shown by Eq. (141) taken at the limit
M rg mol´1 s 28.965 43.45 σ Ñ 0. This behaviour was identified previously in Sect. 3.4. It
Γ1 1.4 1.4 results from the fact that vertical displacement of fluid induced
σ0 r10´7 s´1 s 7.5 7.5
by the stable stratification compensates the local density varia-
tions in this frequency regime and strongly flattens the total tidal
P0 p0q r105 Pas 1.0 93
torque. As demonstrated in Sect. 4, this is not the case in general
Ω r10´7 s´1 s 729.21 ´2.9927
and it depends on the strength of the stratification. In the case
norb r10´7 s´1 s 1.9910 3.2364
of a slowly rotating convective atmosphere (N 2 « 0) forced by
U2 rm2 s´2 s ´0.985 ´2.349 a heating located at the ground, only the horizontal component
J2 rW kg´1 s 10´2 10´4 remains. This leads to a much stronger tidal torque, similar to
those obtained by Leconte et al. (2015) with a GCM (given by
Notes. Most of the parameters are given by NASA fact sheets9 and IM-
Eq. (174)) and Correia & Laskar (2001) (see Eq. (175)).
CCE databases. For both planets, the radiation frequency is arbitrarily
fixed at σ0 “ 7.5 ˆ 10´7 s´1 , which is the effective Newtonian cooling In Fig. 17, we plot the surface variations of pressure, den-
frequency obtained by Leconte et al. (2015) for Venus. This order of sity, temperature, and their components as functions of the tidal
magnitude can be computed using Eq. (17). The quadrupolar tidal po- frequency. For each quantity, the horizontal component varies
tentials are computed from the expression of the gravitational potential smoothly while the vertical one is discontinuous at the synchro-
given by Mathis & Le Poncin-Lafitte (2009); see Appendix C. nization. This discontinuity is a consequence of the dissymmetry
between prograde and retrograde Hough modes due to the Cori-
olis acceleration (Fig. A.1), particularly as regards the gravity
is approximately the same in the whole layer, which causes the mode of the lowest degree, which is the most important one.
tidal bulge to materialize. The lag angle γ “ π{4 of the bulge Assuming that the tidal torque is entirely transmitted to the
with respect to the subsolar point is given by Eq. (137) taken telluric core of the planet, it is possible to estimate the timescale
at x “ 0 for the gravity mode of lowest degree (n “ 0) and is of the spin evolution induced by the atmospheric semidiurnal
not very sensitive to temperature structure (Lindzen 1968). For tide. Introducing the moment of inertia of the planet, I, we write
the Earth, these semidiurnal pressure peaks are measured around
9h38mn and 22h07mn (see Fig. 13). 1 dΩ T
The patterns of the velocities Vθσ,2 and Vϕσ,2 are also ex- Ω dt
“
IΩ
¨ (229)
plained well by the first horizontal Hough functions, L2,ν θ Θ0
2,ν
Thus, the timescale of the spin evolution is given by
and Lϕ2,ν Θ2,ν
0 (see Fig. 2, middle and bottom left). In contrast, the ˇ ˇ
behaviour of the temperature involves other modes. ˇ IΩ ˇ
T “ˇ ˇ ˇ¨ (230)
We then compute the evolution of the tidal torque with T ˇ
the tidal frequency by changing the value of the spin fre- For the Earth’s semidiurnal tide, χ « 365 and the horizontal
quency (Ω) in simulations. Hence, the reduced tidal frequency component of the corresponding torque is T « 1015 N m. With
χ “ pΩ ´ norb q {norb varies within the interval r´15, 15s. The I “ 8.02 ˆ 1037 kg m2 (NASA fact sheets10 ), TC „ 180 Gyr.
9 10
http://nssdc.gsfc.nasa.gov/planetary/factsheet/ http://nssdc.gsfc.nasa.gov/planetary/factsheet/
venusfact.html earthfact.html
A107, page 24 of 44
Fig. 15. Total tidal response of the Earth’s atmosphere to the solar semidiurnal perturbation caused by the quadrupolar academic forcings
U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´0.985 m2 s´2 and J2 “ 1.0 ˆ 10´2 m2 s´3 . Left,
from top to bottom: amplitudes of δp, δρ, δT , Vθ , Vϕ , and Vr in logarithmic scales as functions of the reduced latitude δ{π (horizontal axis) and
altitude x “ z{H in logarithmic scale (vertical axis). The colour function c is given by c “ log p|δ f |q for any quantity δ f . Right, from top to
bottom: Sine of the arguments of the same quantities as functions of the reduced latitude and altitude. In these plots, c “ sin rarg pδ f qs.
Therefore, the spin frequency of the Earth is not affected by at- are some important differences compared to the case of the
mospheric tides. Earth. The Venusian atmosphere does not have the same prop-
erties as the Earth’s atmosphere. First, the fluid layer is a hun-
6.2. Venus dred times more massive (denoting M the mass of the atmo-
sphere, MB „ 4.8 ˆ 1020 kg and MC „ 5.1 ˆ 1018 kg).
Geometrically speaking, Venus can be seen as an Earth-like Thus, the surface pressure is around 93 bar. Second, it is opti-
planet, with scales of the same magnitude order for the tel- cally thicker. The major part of the heating flux is thus deposited
luric core and the thickness of the atmosphere. However, there at high altitudes. The absorbed flux near the planet surface at
A107, page 25 of 44
A&A 603, A107 (2017)
Fig. 16. Total tidal response of the Earth’s atmosphere to the solar semidiurnal perturbation caused by the quadrupolar academic forcings
U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´0.985 m2 s´2 and J2 “ 1.0 ˆ 10´2 m2 s´3 . Left,
from top to bottom: real parts of δp, δρ, δT , Vθ , and Vϕ taken at the ground as functions of the reduced longitude ϕ{π (horizontal axis) and lati-
tude δ{π (vertical axis). The colour function c is given by c “ < tδ f u for any quantity δ f . Right, from top to bottom: imaginary parts of the same
quantities as functions of the reduced longitude and latitude. In these plots, c “ = tδ f u.
the subsolar point Fabs is estimated at Fabs „ 100´200 W m´2 Earth is far higher than its mean motion and the radiation fre-
(Avduevsky et al. 1970; Lacis 1975; Dobrovolskis & Ingersoll quency of its atmosphere, these three frequencies are of the same
1980), which implies J2 „ 10´4 W kg´1 . Finally, layers below order of magnitude for Venus. Therefore, the Venusian semidiur-
60 km are characterized by a strongly negative temperature gra- nal thermal tide is characterized by a tidal frequency |σ| „ σ0 ,
dient (Seiff et al. 1980) and are, therefore, weakly stratified or which is typical of the thermal regime. The term describing ra-
convective. As discussed in Sects. 3 and 4, these properties have diative losses in the heat transport equation (Eq. (18)) plays an
a strong impact on the nature of the tidal response. Particularly, important role in this regime by damping the profiles in altitude
the tidal torque may be much stronger in the convective case than of the tidal response; see Fig. 5. This figure shows that the imag-
in the stably stratified one. In the present section, we study a inary part of k̂n , which is responsible for the damping, is always
Venus-like planet with a stably stratified isothermal atmosphere comparable to the real part in this frequency range.
(VenusX) in order to understand the effects of stratification and Contrary to the case of the Earth, the semidiurnal thermal
radiative losses in a general context. tide of VenusX belongs to the family of super-inertial waves,
The other important difference between the Earth and Venus with ν « 0.48, which means that the horizontal component
is in their rotational dynamics. While the spin frequency of the of the tidal response is composed of gravity modes only. We
A107, page 26 of 44
0.00 1.00
−0.50
0.50
log10(| p|) [mbar]
−1.00
sin[arg( p)]
−1.50 0.00
−2.00
−0.50
T
−2.50
V
H
−3.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
−1.00 1.00
−1.50
0.50
|) [kg.m 3]
−2.00
)]
sin[arg(
−2.50 0.00
log10(|
−3.00
−0.50
−3.50
−4.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
1.00
2.00
1.50
0.50
log10(| T|) [K]
1.00
sin[arg( T)]
0.50 0.00
0.00
−0.50
−0.50
−1.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15 −10 −5 0 5 10 15
Fig. 17. Tidal surface oscillations of pressure, density, temperature, and their components at the equator of the Earth as functions of the re-
duced tidal frequency χ “ pΩ ´ norb q {norb (horizontal axis). The perturbation is defined by the academic quadrupolar forcings U px, θ, ϕ, tq “
U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´0.985 m2 s´2 and J2 “ 1.0 ˆ 10´2 m2 s´3 . Labels T, H, and V refer
to the total response and its horizontal and vertical components, respectively.
note that the surface variations of perturbed quantities plotted ground (Fig. 20) around the synchronization for χ P r´15, 15s.
in Fig. 19 are very well represented by the first mode. The pat- The behaviour of the torque and its components is the same as
terns of the Hough functions Θ2,ν 0 , Lθ Θ0 , and Lϕ Θ0 (see
2,ν 2,ν 2,ν 2,ν previously observed for the Earth. The maxima of horizontal and
Fig. 3, red curves) clearly appear on the maps. Moreover, the vertical components are located at χ “ ˘χB with χB ă χC ow-
lags of δp and δρ are different from that observed in the Earth’s ing to the difference of mean motions between the two planets.
semidiurnal tide. In particular, it is interesting to note that the Since the tidal torque is proportional to the surface density of the
pressure and density peaks are not superposed in this case. The atmosphere, it is a hundred times stronger in the case of Venus
effect of the damping caused by radiative losses can be observed than in the case of the Earth for a given thermal forcing. This is
in vertical cross-sections in Fig. 18, where the variations of pres- also verified by the surface variations of the perturbed quantities,
sure and density are located close to the ground. Given that the which – except for their amplitudes – are comparable to those of
real and imaginary part of the vertical wavenumber are both very the Earth.
high, vertical component is both strongly oscillating and strongly To conclude this section, we compute the timescale of the
damped. spin evolution. The Venusian semidiurnal tide is characterized
by the frequency χ « ´1.92. With the moment of inertia I “
As was previously done for the Earth, we draw the varia- 5.88 ˆ 1037 kg m2 (NASA fact sheets) and the horizontal com-
tions of the tidal torque (Fig. 14) and perturbed quantities at the ponent of the torque given by Fig. 14, i.e. T „ ´5 ˆ 1016 N m,
A107, page 27 of 44
A&A 603, A107 (2017)
Fig. 18. Total tidal response of the atmosphere of Venus to the solar semidiurnal perturbation caused by the quadrupolar academic forcings
U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´2.349 m2 s´2 and J2 “ 1.0 ˆ 10´4 m2 s´3 . Left, from
top to bottom: amplitudes of δp, δρ, δT , Vθ , Vϕ , and Vr in logarithmic scales as functions of the reduced latitude δ{π (horizontal axis) and altitude
x “ z{H in logarithmic scale (vertical axis). The colour function c is given by c “ log p|δ f |q for any quantity δ f . Right, from top to bottom: Sine
of the arguments of the same quantities as functions of the reduced latitude and altitude. In these plots, c “ sin rarg pδ f qs.
we obtain TB „ 10 Myr. The thermal forcing corresponds ap- given by this early work, T „ ´1.8 ˆ 1016 N m. We note here
proximately to the mean level of the heating profile used by that the amplitude of the response is directly proportional to the
Dobrovolskis & Ingersoll (1980) (J2 „ 10´4 W kg´1 ) so the amplitude of the excitation if one of its two components, J or U,
obtained torque is of the same order of magnitude as the value can be neglected (Sect. 2).
A107, page 28 of 44
Fig. 19. Total tidal response of the atmosphere of Venus to the solar semidiurnal perturbation caused by the quadrupolar academic forcings
U px, θ, ϕ, tq “ U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´2.349 m2 s´2 and J2 “ 1.0 ˆ 10´4 m2 s´3 . Left, from
top to bottom: real parts of δp, δρ, δT , Vθ , and Vϕ taken at the ground as functions of the reduced longitude ϕ{π (horizontal axis) and latitude δ{π
(vertical axis). The colour function c is given by c “ < tδ f u for any quantity δ f . Right, from top to bottom: imaginary parts of the same quantities
as functions of the reduced longitude and latitude. In these plots, c “ = tδ f u.
6.3. Discussion of physical assumptions stratified fluids where 2Ω ă |σ| ! N, but is less appropri-
ate if |σ| „ 2Ω, as discussed in Mathis (2009). Its valid-
We have developed a new analytic model for atmospheric tides ity domain, not well defined, remains essentially limited to
by generalizing the classical tidal theory detailed in CL70 to dis- the regime of super-inertial waves (Prat et al. 2016). It is
sipative atmospheres. represented in Fig. 21, which shows the cases that can be
This work leans on several assumptions which contribute to treated analytically (regions a and b). Concerning the im-
make computations simpler. The most important of them are portance of the neglected Coriolis terms, we refer the reader
listed here. to Gerkema & Shrira (2005), who explored the physics of
– The traditional approximation. This hypothesis, commonly gravito-inertial waves without the traditional approximation
used in the literature of geophysics, is very useful be- (see also Tort & Dubos 2014). The traditional approximation
cause it allows us to separate the horizontal and verti- appears to be a relatively strong hypothesis because it does
cal dependences in the Navier-Stokes equation. Hence, the not take into consideration the case of rotating convective
horizontal and vertical structure can be computed sepa- atmospheres (N 2 « 0, e.g. Ogilvie & Lin 2004). Convec-
rately. This approximation is relevant in strongly stably tion is an important feature of the Earth’s nearly adiabatic
A107, page 29 of 44
A&A 603, A107 (2017)
1.00 1.00
0.50
0.50
log10(| p|) [mbar]
0.00
sin[arg( p)]
−0.50 0.00
−1.00
−0.50
T
−1.50
V
H
−2.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
−1.00 1.00
−1.50
0.50
|) [kg.m 3]
−2.00
)]
sin[arg(
−2.50 0.00
log10(|
−3.00
−0.50
−3.50
−4.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
0.00 1.00
−0.50
0.50
log10(| T|) [K]
−1.00
sin[arg( T)]
−1.50 0.00
−2.00
−0.50
−2.50
−3.00 −1.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
Fig. 20. Tidal surface oscillations of pressure, density, temperature, and their components at the equator of Venus as functions of the re-
duced tidal frequency χ “ pΩ ´ norb q {norb (horizontal axis). The perturbation is defined by the academic quadrupolar forcing U px, θ, ϕ, tq “
U2 P22 pcos θq eipσt`2ϕq and J px, θ, ϕ, tq “ J2 P22 pcos θq eipσt`2ϕq with U2 “ ´2.349 m2 s´2 and J2 “ 1.0 ˆ 10´4 m2 s´3 . Labels T, H, and V refer
to the total response and its horizontal and vertical components, respectively.
layer (altitudes below 2 km) as well as the Venus’ atmo- perturbation with respect to the equilibrium state. This con-
sphere (Linkin et al. 1986). It has strong implications on the dition is verified for the Earth and Venus.
propagations of waves: gravity waves cannot exist in con- – The solid rotation approximation. We assumed that the at-
vective layers, which may significantly modify the tidal re- mosphere rotates with the solid core uniformly. Thus, the
sponse, particularly if the convective turnover time is close to atmospheric circulation (both meridional and zonal) is ig-
the period of the tidal forcing. In the general case, the effect nored. This approximation is justified by the fact that the
of convection on the structure of waves can be studied ei- most important pressure and density variations are concen-
ther numerically with GCM (Fig. 21, d) or analytically with trated in the densest layers, near the ground. Given that
simplified models (e.g. based on a Cartesian geometry). The these layers are bound to the solid core by viscous friction,
only exception for a global analytical study of a convective they rotate at the same spin frequency. However, tidal waves
atmosphere is the case treated in Sect. 4 where stable stratifi- might interact with the mean flow in cases where σ « 0
cation and rotation are too weak to be neglected (Fig. 21, b). (Goldreich & Nicholson 1989) and it will be interesting to
– The linearization around the equilibrium. In Sects. 2 and 3, take into account the global atmospheric circulation in fur-
the equations have been linearized around the equilibrium. ther works, as is done in GCM simulations.
This method is only relevant in cases where the tidal re- – The use of simplified atmospheric vertical structures. In
sponse of the atmosphere can be considered as a small order to lighten calculations, we considered the cases of
A107, page 30 of 44
– the involved characteristic length scales (e.g. planetary ra-

dius, pressure height scale of the atmosphere, length scales
of the vertical variations of tidal waves);
– the physical parameters of the planet (e.g. the molar mass
of the atmosphere, the adiabatic exponent, the surface gra-
vity, the surface pressure).
Using the equations of hydrodynamics, we have adapted the

theoretical approach detailed in CL70 to thick dissipative at-
mospheric shells. We computed the analytic expressions of the
horizontal and vertical structure equations, of the perturbed
quantities (pressure, density, temperature, velocity, displace-
Fig. 21. Asymptotic domains of parameters where the analytical ap- ment), and of Love numbers and tidal torques exerted on the
proach presented in this work can be applied (green) and where the atmosphere. These expressions have been used to identify the
numerical approach is necessary (red) as functions of the frequency ra- possible regimes of atmospheric tides: a non-dissipative Earth-
tios N 2 {σ2 (horizontal axis) and p2Ωq2 {σ2 (vertical axis). Green re- like dynamical regime characterizing fast rotating bodies, and a
gions: stably stratified atmospheres satisfying the hierarchy of frequen- dissipative Venus-like thermal regime for planets near synchro-
cies p2Ωq2 ! N 2 necessary to apply the traditional approximation a); nization. The behaviours involved in the two cases are very dif-
weakly stratified or convective atmospheres of slowly rotating plan- ferent. The dynamical regime corresponds to the case studied by
ets b). Red regions: stably stratified atmospheres where the hierarchy CL70. In this regime, tidal waves are either oscillating or evanes-
of frequencies p2Ωq2 ! N 2 is not satisfied c); super-adiabatic rotating cent depending of the equivalent depth (hn ) associated with the
convective atmospheres d). mode. In the radiative regime, they are oscillating and evanes-
cent, and the perturbation can be strongly damped.
The tidal response of the atmosphere will also be affected
by the nature of the tidal excitation. A predominating thermal
uniform background temperature (isothermal stably strati- forcing generates thermal tides. On the contrary, if the thermal
fied atmosphere) and temperature gradient (convective at- energy brought to the fluid layer is negligible compared to the ef-
mosphere), and supposed that some parameters, such as σ0 , fect of the tidal gravitational potential, the perturbation belongs
do not vary with altitude. These simplified structures are a to the family of gravitational tides. This case has not been stud-
first step towards more sophisticated ones. Owing to the im- ied in this work, but the vertical structure equation and polar-
portance of the stratification in the tidal response, it will be ization relations of the model clearly attest to behaviour that is
necessary to treat the cases of observed structures in future qualitatively different from that observed for thermal tides. Fi-
works, for instance those of the atmospheres of Earth, Venus, nally, the tidal response of the atmospheric shell is obviously af-
and Mars. fected by lower and upper boundary conditions. We showed for
– The optically thin layer hypothesis. Radiative processes are instance that a free-surface condition and a no-material-escape
introduced in the dynamics of atmospheric tides through condition would not give the same vertical profiles of the pertur-
Newtonian cooling. The linear dependence of the corre- bation. Similarly, the ground may affect temperature oscillations
sponding radiative term on the temperature perturbation can by introducing a delayed thermal forcing locally.
be interpreted as an optically thin layer approximation. In- Finally, the results obtained for the tidal torque are in agree-
deed, this term corresponds to the thermal behaviour of an at- ment with the GCM simulations by Leconte et al. (2015). This
mosphere that would emit grey body radiation towards space torque appears to be strongly dependent both on the tidal fre-
and the planet surface. Actually, the flux emitted is partly quency and on the atmospheric structure (Fig. 22). In the case
absorbed by the atmosphere itself, which induces a com- of a stably stratified atmosphere, the tidal bulge vanishes at the
plex thermal coupling between the different layers of the vicinity of synchronization. As a consequence, the tidal torque
atmosphere. This is true particularly for optically thick at- exerted on the atmosphere is very weak and cannot balance the
mospheres. In these cases, the Newtonian cooling should be solid tidal torque, which should lead the planet to synchroniza-
seen as a convenient approximation of thermal radiation for tion. On the contrary, if the stable stratification is sufficiently
an analytical approach where the dissipative processes can- weak (weakly stratified or convective atmosphere), the atmo-
not be treated without approximations. This approach seems spheric tidal torque can be strong. In this case we recover the
to be well validated by numerical simulations using GCM Maxwellian behaviour obtained numerically by Leconte et al.
(Leconte et al. 2015). (2015) and described by the early model of Correia & Laskar
(2001). This would lead planets to non-synchronized states.
These results highlight the predictive interest of the modelling
7. Conclusions of atmospheric tides by showing that it is possible to constrain
the atmospheric structure of terrestrial planets from their rotation
We have set the bases of a new theoretical framework to study
states of equilibrium.
the atmospheric tides of Earth-like exoplanets analytically. This
work is complementary to numerical simulations. The main ob- Acknowledgements. We thank the anonymous referee for comments that helped
jective is to explain the qualitative behaviour of an atmosphere to improve the paper. We thank Richard Lindzen for the very interesting discus-
submitted to both thermal and gravitational tidal perturbations sions we had about the analytic modelling of the Earth’s atmospheric tides. We
are also very grateful to Umin Lee for his help in understanding how to compute
and to explore the domain of key physical parameters. The pa- Hough functions efficiently. Finally, we thank Jeremy Leconte, who shared with
rameters highlighted by the model are us his helpful approach to the physical modelling of super-Earth atmospheres
from the point of view of numerical simulations. P. Auclair-Desrotour and S.
– the characteristic frequencies of the system (e.g. inertia, Mathis acknowledge funding by the European Research Council through ERC
Brunt-Väisälä, radiative, boundary layer, tidal frequencies); grant SPIRE 647383. This work was also supported by the Programme National
A107, page 31 of 44
The&predicFve&aspect&
A&A 603, A107 (2017)
INTERNAL&STRUCTURE& strong&atmospheric&Fdal&torque& FINAL&ROTATION&STATES&
NEUTRAL&CONVECTIVE& NONJSYNCHRONIZED&STATES&
ATMOSPHERE& OF&EQUILIBRIUM&
weak&atmospheric&Fdal&torque&
STABLYJSTRATIFIED& SPINJORBIT&
ATMOSPHERE& SYNCHRONIZATION&
AuclairJDesrotour,&Laskar,&Mathis&(2016a)&
Fig. 22. Predictive aspects related to the dependence of the atmospheric tidal torque on stratification. The neutral convective case implies a strong
torque, which can lead to non-synchronized states of equilibrium. The stably stratified case implies a weak torque, which lets the planet evolve
In&agreement&with&the&early&results&by&Ioannou&&&Lindzen&(1993),&Arras&&&Socrates&(2010)&
towards spin-orbit synchronization.
DissipaFon&of&atmospheric&thermal&Fdes&in&superJEarths& 28&
de Planétologie (CNRS/INSU) and CoRoT/Kepler and PLATO CNES grant at Haurwitz, B. 1965, Archives for Meteorology Geophysics and Bioclimatology
CEA-Saclay. Series A Meteorology and Atmopsheric Physics, 14, 361
Henning, W. G., O’Connell, R. J., & Sasselov, D. D. 2009, ApJ, 707, 1000
Hough, S. S. 1898, Phil. Trans. R. Soc. Lond. Ser. A, 191, 139
Hut, P. 1980, A&A, 92, 167
References Hut, P. 1981, A&A, 99, 126
Ingersoll, A. P., & Dobrovolskis, A. R. 1978, Nature, 275, 37
Abramowitz, M., & Stegun, I. A. 1972, Handbook of Mathematical Functions Ioannou, P. J., & Lindzen, R. S. 1993a, ApJ, 406, 252
(New York: Dover) Ioannou, P. J., & Lindzen, R. S. 1993b, ApJ, 406, 266
Arras, P., & Socrates, A. 2010, ApJ, 714, 1 Ioannou, P. J., & Lindzen, R. S. 1994, ApJ, 424, 1005
Auclair Desrotour, P., Mathis, S., & Le Poncin-Lafitte, C. 2015, A&A, 581, A118 Kato, S. 1966, J. Geophys. Res., 71, 3201
Avduevsky, V. S., Marov, M. Y., Noykina, A. I., Polezhaev, V. I., & Zavelevich, Kaula, W. M. 1962, AJ, 67, 300
F. S. 1970, J. Atmos. Sci., 27, 569 Kaula, W. M. 1964, Reviews of Geophysics and Space Physics, 2, 661
Baker, R. D., Schubert, G., & Jones, P. W. 2000, J. Atmos. Sci., 57, 184 Kelvin, L. 1863, Phil. Trans. Roy. Soc. Lond., Treatise on Natural Philosophy, 2,
Beer 1852, Annalen der Physik, 162, 78 837
Bernard, E. A. 1962, Archives for Meteorology Geophysics and Bioclimatology Kelvin, L. 1882, Proc. Roy. Soc. Edinb., 11, 396
Series A Meteorology and Atmopsheric Physics, 12, 502 Klett, Witwe, E., Detleffsen, Peter, C., et al. 1760, IH Lambert... Photometria
Bouguer, P. 1729, Essai d’Optique sur la gradation de la Lumière (Paris: sive de mensura et gradibus luminis, colorum et umbrae (sumptibus viduae
C. Jombert) Eberhardi Klett)
Bruce, G. H., Peaceman, D., Rachford Jr, H., et al. 1953, Journal of Petroleum Konopliv, A. S., & Yoder, C. F. 1996, Geophys. Res. Lett., 23, 1857
Technology, 5, 79 Lacis, A. A. 1975, J. Atmos. Sci., 32, 1107
Chapman, S., & Lindzen, R. 1970, Atmospheric tides. Thermal and gravitational Lainey, V., Dehant, V., & Pätzold, M. 2007, A&A, 465, 1075
(Dordrecht: Reidel) Lamb, H. 1911, Proc. Roy. Soc. Lond. Ser. A, 84, 551
Correia, A. C. M., & Laskar, J. 2001, Nature, 411, 767 Lamb, H. 1932, Hydrodynamics (New York: Dover)
Correia, A. C. M., & Laskar, J. 2003, J. Geophys. Res. (Planets), 108, 5123 Laplace, P. S. 1798, Traité de mécanique céleste (Duprat J. B. M.)
Correia, A. C. M., Laskar, J., & de Surgy, O. N. 2003, Icarus, 163, 1 Leconte, J., Wu, H., Menou, K., & Murray, N. 2015, Science, 347, 632
Correia, A. C. M., Levrard, B., & Laskar, J. 2008, A&A, 488, L63 Lee, U., & Saio, H. 1997, ApJ, 491, 839
Correia, A. C. M., Boué, G., Laskar, J., & Rodríguez, A. 2014, A&A, 571, A50 Lighthill, J. 1978, Waves in fluids (Cambridge: Cambridge University Press)
Covey, C., Dai, A., Marsh, D. R., & Lindzen, R. S. 2009, AGU Fall Meeting Lindzen, R. S. 1966, Monthly Weather Review, 94, 295
Abstracts Lindzen, R. S. 1967a, Nature, 215, 1260
Cowling, T. G. 1941, MNRAS, 101, 367 Lindzen, R. S. 1967b, Quart. J. Roy. Meteor. Soc., 93, 18
Dickinson, R. E., & Geller, M. A. 1968, J. Atmos. Sci., 25, 932 Lindzen, R. S. 1968, Proc. Roy. Soc. Lond. Ser. A, 303, 299
Dobrovolskis, A. R., & Ingersoll, A. P. 1980, Icarus, 41, 1 Lindzen, R. S., & McKenzie, D. J. 1967, Pure and Applied Geophysics, 66, 90
Eckart, C. 1960, Physics of Fluids, 3, 421 Linkin, V. M., Kerzhanovich, V. V., Lipatov, A. N., et al. 1986, Science, 231,
Efroimsky, M., & Lainey, V. 2007, J. Geophys. Res. (Planets), 112, 12003 1417
Egbert, G. D., & Ray, R. D. 2000, Nature, 405, 775 Longuet-Higgins, M. S. 1968, Phil. Trans. R. Soc. Lond. Ser. A, 262, 511
Fabrycky, D. C., Ford, E. B., Steffen, J. H., et al. 2012, ApJ, 750, 114 Love, A. E. H. 1911, Some Problems of Geodynamics (Cambridge University
Gastineau, M., & Laskar, J. 2014, TRIP 1.3.8, TRIP Reference manual, IMCCE, Press)
Paris Observatory, http://www.imcce.fr/trip/ Marov, M. Y., Avduevskij, V. S., Kerzhanovich, V. V., et al. 1973, J. Atmos. Sci.,
Gerkema, T., & Shrira, V. I. 2005, J. Fluid Mech., 529, 195 30, 1210
Gerkema, T., & Zimmerman, J. 2008, Lecture Notes, Royal NIOZ, Texel Mathis, S. 2009, A&A, 506, 811
Gold, T., & Soter, S. 1969, Icarus, 11, 356 Mathis, S., & Le Poncin-Lafitte, C. 2009, A&A, 497, 889
Goldreich, P., & Nicholson, P. D. 1989, ApJ, 342, 1075 Mathis, S., Talon, S., Pantillon, F.-P., & Zahn, J.-P. 2008, Sol. Phys., 251, 101
Goldreich, P., & Soter, S. 1966, Icarus, 5, 375 Mayor, M., & Queloz, D. 1995, Nature, 378, 355
Green, J. S. A. 1965, Proc. of the Royal Society of London, Series A, Mignard, F. 1979, Moon and Planets, 20, 301
Mathematical and Physical Sciences, 288, 564 Mignard, F. 1980, Moon and Planets, 23, 185
Greenberg, R. 2009, ApJ, 698, L42 Mohr, P. J., Taylor, B. N., & Newell, D. B. 2012, J. Phys. Chem. Ref. Data, 41,
Hagan, M., Forbes, J., & Richmond, A. 2003, Encyclopedia of Atmospheric 043109
Sciences, 1, 159 Neron de Surgy, O., & Laskar, J. 1997, A&A, 318, 975
A107, page 32 of 44
Ogilvie, G. I. 2014, ARA&A, 52, 171 Siebert, M. 1961, Advances in Geophysics, 7, 105
Ogilvie, G. I., & Lin, D. N. C. 2004, ApJ, 610, 477 Taylor, G. I. 1929, Proc. Roy. Soc. Lond. Ser. A, 126, 169
Pekeris, C. L. 1937, Proc. Roy. Soc. Lond. Ser. A, 158, 650 Taylor, G. I. 1936, Proc. Roy. Soc. Lond. Ser. A, 156, 318
Perryman, M. 2011, The Exoplanet Handbook (Cambridge University Press) Tort, M., & Dubos, T. 2014, Quart. J. Roy. Meteor. Soc., 140, 2388
Planck, M. 1901, Annalen der Physik, 309, 553 Townsend, R. H. D. 2003, MNRAS, 340, 1020
Pollack, J. B., & Young, R. 1975, J. Atmos. Sci., 32, 1025 Unno, W., Osaki, Y., Ando, H., Saio, H., & Shibahashi, H. 1989, Nonradial os-
Prat, V., Lignières, F., & Ballot, J. 2016, A&A, 587, A110 cillations of stars (Tokyo: University of Tokyo Press)
Press, W. H. 1981, ApJ, 245, 286 Webb, D. J. 1980, Geophys. J., 61, 573
Press, W. H., Flannery, B. P., & Teukolsky, S. A. 1986, Numerical recipes. The White, A. A., Hoskins, B. J., Roulstone, I., & Staniforth, A. 2005, Quart. J. Roy.
art of scientific computing (Cambridge University Press) Meteor. Soc., 131, 2081
Ray, R. D., Eanes, R. J., & Lemoine, F. G. 2001, Geophys. J. Int., 144, 471 Wilkes, M. V. 1949, Oscillations of the Earth’s Atmosphere (University Press
Remus, F., Mathis, S., Zahn, J.-P., & Lainey, V. 2012, A&A, 541, A165 Cambridge)
Rossby, C.-G. et al. 1939, Journal of Marine Research, 2, 38 Williams, J. G., Konopliv, A. S., Boggs, D. H., et al. 2014, J. Geophys. Res.
Seiff, A., Kirk, D. B., Young, R. E., et al. 1980, J. Geophys. Res.: Space Phys., (Planets), 119, 1546
85, 7903 Zahn, J. P. 1966, Annales d’Astrophysique, 29, 313
Shen, M., & Zhang, C. Z. 1990, Icarus, 85, 129 Zahn, J.-P., Talon, S., & Matias, J. 1997, A&A, 322, 320
A107, page 33 of 44
A&A 603, A107 (2017)
5
Appendix A: Hough functions n= 6
4 n= 5
Hough functions, named for Hough’s initial work (Hough 1898), n= 4
sign( n)log10(1 + | n|)

are the solutions of Eq. (35) integrated as an eigenvalue problem 3 n= 3
of Λm,ν
n with appropriate boundary conditions (Longuet-Higgins n= 2
2
1968; Chapman & Lindzen 1970; Lee & Saio 1997). It is pos- n= 1
sible to compute these functions in several ways. Here, we 1 n=0
present two possible approaches: the finite differences method n=1
0 n=2
proposed in the Numerical recipes (Press et al. 1986) and de-
n=3
tailed in Lee & Saio (1997), and a spectral method developed −1
n=4
from the works of Hough (1898), Longuet-Higgins (1968), and n=5
−2
Chapman & Lindzen (1970).
−3
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
A.1. Finite differences method sign( )log10(1 + | |)
1
The operator Lm,ν of Eq. (30) writes n= 6
n= 5
1 ´ µ2 d
„ 
m,ν d 0.8 n= 4
L ”
dµ 1 ´ µ2 ν2 dµ n= 3
(A.1) n= 2
1 ` µ2 ν 2
ˆ ˙
1 m2
>2
´ ` mν , 0.6 n= 1
1 ´ µ2 ν 2 1 ´ µ2 1 ´ µ2 ν 2 2, n=0
n n=1
where µ “ cos θ. The new Laplacian tidal equation is not modi-
<P22,
0.4
n=2
fied when µ is replaced by ´µ. Therefore, the domain over which n=3
this equation is integrated can be reduced to 0 ď µ ď 1. Like 0.2 n=4
the associated Legendre polynomials, the solutions are either n=5
even or odd with respect to the equatorial plane µ “ 0. The
even solutions satisfy Θm,ν n p´µq “ Θn pµq, the odd solutions
m,ν 0
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
Θn p´µq “ ´Θn pµq. For the sake of simplicity, the parity of
m,ν m,ν
sign( )log10(1 + | |)
the subscript used in the notations corresponds to the parity of
the associated function. The solutions belonging to one of these Fig. A.1. Top: eigenvalues (Λm,ν n ) of the gravity and Rossby modes
two families are all defined by the boundary condition applied of lowest degrees (´6 ď n ď 5) as functions of the spin param-
at µ “ 0: dΘm,ν n {dµ “ 0 for even solutions and Θn
m,ν
“ 0 for eter (ν). The horizontal axis represents the function of ν given by
m,ν
odd solutions. The operator L has two regular singular points, sign pνq log p1 ` |ν|q, and the vertical axis the function of Λm,ν
n given
a permanent one at µ “ 1 and another one at µ “ 1{ |ν| in the by sign pΛm,ν m,ν
n q log p1 ` |Λn |q. Bottom: projection coefficients c2,n,2 “
sub-inertial regime, characterized by |ν| ą 1 and, therefore, de- xP22 , Θ2,ν 2 2
n y of the normalized Legendre polynomial P2 on normalized
pending on the spin parameter. Both are sources of numerical Hough functions associated with these modes as functions of the spin
difficulties. To tackle the problem of the singular point at µ “ 1, parameter (see Eq. (70)).
we introduce the variable y pµq defined by Press et al. (1986) and
Lee & Saio (1997):
Θm,ν
n pµq “ 1 ´ µ
2 |m|{2
y pµq . theory, which requires`that Θm,ν
n˘ should be`regular˘at µ “ 1, and
` ˘
(A.2)
is given explicitly by 1 ´ µ2 Pdy{dµ ` 1 ´ µ2 Qy “ 0 (see
Laplace’s tidal equation becomes Lee & Saio 1997).
In order to solve Eq. (A.3) numerically, the domain is dis-
d2 y dy
` P ` Qy “ 0, (A.3) cretized. Hence, the equation is reduced to a linear system of
dµ dµ dimension NS “ 2N, where N is the number of elements of the
where mesh. In the regime of sub-inertial waves, the singular point at
µ “ 1{ |ν| corresponds to a turning point, in the vicinity of which
ν2
$ ˆ ˙
1 ` |m| the resolution of the mesh must be sufficiently high. Lee & Saio
P 2µ ,
’
’ “ ´
1 ´ µ2 ν 2 1 ´ µ2 (1997) showed that a regular
ř solution could be locally expanded
’
’
’
in series of the form y “ `8 x “ µ ´ 1{ |ν|, and
’ j
’
’ a
j“0 j x , where
1 “ m,ν `
&
Λn 1 ´ µ2 ν2 ´ |m| ´ m2
˘
Q“ (A.4) satisfying the condition
’
’ 1´µ 2
˘ ff
µ ν µ ν
’ 2 2
` 2 2
’
’
’ 2 |m| ` mν 1 ` |ν| p|m| ` mνq
’ ´ ¨ a1 ` a0 “ 0. (A.5)
1 ´ µ2 ν 2 1 ´ ν2
’
%
The boundary condition applied at µ “ 0 depends on the parity This allows us to compute the solution properly in the sub-
of the sought solution, i.e. dy{dµ “ 0 at the equator for even inertial regime. To construct the set of Hough functions, we use
modes and y “ 0 for odd modes. We also apply at µ “ 0 the as predictors the asymptotic results given by Longuet-Higgins
normalization condition y “ 1 for even modes and dy{dµ “ 1 (1968) and Townsend (2003). Two types of modes can be
for odd modes. Since P and Q are not defined at µ “ 1, the identified.
boundary condition at the poles is applied at µ “ 1 ´ , where The modes of the first type, called gravity modes, are defined
0 ă ! 1. This condition is derived from the Sturm-Liouville for ν P R and are characterized by positive eigenvalues (Fig. A.1,
A107, page 34 of 44
top panel). In the non-rotating case, where ν “ 0, they corre- the roots of a high-degree polynomial (typically, the degree N of
spond to the associated Legendre polynomials (Fig. A.1, bottom this polynomial corresponds to the number of coefficients of the
panel). The gravity modes of Hough functions are ordered with series given by Eq. (A.7)). With the current computing resources,
positive n, by analogy with this particular case, for which the N “ 500 can be reached easily. However, this method rapidly
degrees l of Legendre polynomials are given by l “ |m| ` n. In becomes tricky to apply beyond N „ 500 because of very large
the asymptotic cases where |ν| Ñ `8, analytic solutions using polynomial coefficients. Since the size of the Λns,ν range depends
Hermite polynomials can be computed (Townsend 2003). In the on N, this can limit the number of gravity modes or Rossby
sub-inertial regime, the oscillations of gravity modes are trapped modes found. For instance, in the case of the Earth’s semidiurnal
in an equatorial belt in the interval given by 0 ď µ ď 1{ |ν| tide, where ν is very close to 1, the first Rossby modes are char-
(see Fig. A.2). This explains why these modes are not well rep- acterized by strongly negative eigenvalues. Therefore, only a few
resented by associated Legendre polynomials for high values functions of this type can be computed. To address this issue, the
of |ν|. previous problem may be written as an explicit eigenvalue prob-
The modes of the second type have various denominations lem. Hence, following Hough (1898) and CL70, we expand the
in literature (see Longuet-Higgins 1968). We call them Rossby non-normalized Hough functions Θ̃ns,ν in non-normalized associ-
waves in reference to the early studies by Rossby et al. (1939). ated Legendre Polynomials, denoted P̃ks :
These modes are defined for |ν| ą 1 only. Most of them are 8
characterized by negative Λm,ν
ÿ
n and are identified in this work Θ̃ns,ν “ P̃k pµq .
s,ν s
An,k (A.8)
by subscripts n ă 0, e.g. Λ´1 ě Λ´2 ě Λ´3 ě . . ., the par- k“s
ity of n corresponding to the parity of the associated function. s,ν
Longuet-Higgins (1968) established the analytic expressions of To lighten the expressions, the An,k are simply denoted Aks in the
following analytic development. Let us introduce X “ 1/ ν2 Λ
` ˘
the functions in the vicinity of the boundaries ν “ ˘1 using
Laguerre polynomials. In contrast with gravity modes, the os- and the coefficients
cillations of Rossby modes are concentrated around the poles k´s k´s´1
in a region defined by 0 ď µ ď 1{ |ν|. As pointed out by Kk “ , Lk “ ,
p2k ´ 1q rk pk ´ 1q ´ sνs 2k ´ 3
Lindzen (1966), Rossby modes have to be taken into account
in the regime of sub-inertial waves, otherwise the set of eingen- pk ´ 1q2 pk ` sq k pk ` 1q ´ sν
functions is not complete. Mk “ , Nk “ ,
In Fig. (A.1), we can observe that the eigenvalues (Λm,ν k2 p2k ` 1q ν2 k2 pk ` 1q2
n )
strongly depend on the spin parameter. This dependence will af- k`s`1 k`s`2
fect the propagation of tidal waves along the vertical direction Sk “ , Uk “ ,
p2k ` 3q rpk ` 1q pk ` 2q ´ sνs 2k ` 5
through their vertical wavenumbers. In addition, the Coriolis ac-
celeration induces a dissymmetry between prograde (ν ă 0) and pk ` 2q2 pk ´ s ` 1q
retrograde (ν ą 0) modes. Particularly, the gravity mode of low- Tk “ ¨
est degree present two different asymptotic behaviours in the pk ` 1q2 p2k ` 1q
sub-inertial regime. The projection of the associated Legendre (A.9)
polynomial P22 , which is representative of the semidiurnal tidal Hough (1898) demonstrated the following equation, which cor-
forcing, is also directly related to ν. We retrieve for xP22 , Θ2,ν
n y the responds to Eq. (60) in CL70
dissymmetry observed for the eigenvalues and note that the vari-
XAks “ Rk Aks ` Qk Ak´2
s s
` Pk Ak`2 (A.10)
ations of the decomposition with ν are not simple to describe. In
the present study, this decomposition is computed numerically with
with the spectral method presented below. Pk “ ´S k Uk ,
Qk “ ´Kk Lk , (A.11)
A.2. Spectral method
Rk “ Nk ´ Kk Mk ´ S k T k .
The early studies by Longuet-Higgins (1968) and CL70 provide
a detailed theoretical approach of Hough functions and present The system of equations then reduces to the two eigenvalue
the very powerful general method introduced by Hough (1898) problems
to compute the eigenvalues and eigenvectors of Lm,ν . In this
» s fi » s fi
As As
method, which is not detailed here, Hough functions are ex- s
— A s`2 ffi — A ss`2 ffi
— s ffi — s ffi
panded in series of normalized associated Legendre polynomi- — A s`4 ffi — A s`4 ffi
als (Abramowitz & Stegun 1972). In the previous subsection, .
X — . ffi “ Πeven — . ffi
— .. ffi ,
— ffi — ffi
(A.12)
.
— ffi
Eq. (A.1) highlights the symmetry between the two parameters — s ffi — s ffi
defining these functions, m and ν. Hence, for m ą 0 and ν P R, —A s`2p ffi —A s`2p ffi
.. ..
– fl – fl
Θ´m,ν
n “ Θm,´ν
n and Λ´m,ν
n “ Λm,´ν
n . (A.6) . .
Therefore, Hough functions and eigenvalues for m ă 0 are read- and
ily deduced from those obtained for m ą 0. This allows us to A ss`1 A ss`1
» fi » fi
write — A ss`3
ffi — ffi A ss`3
A ss`5 A ss`5
— ffi — ffi
8
ÿ — ffi — ffi
Θns,ν pθq “ Cn,k Pk pcos θq ,
s,ν s
(A.7) ..
ffi “ Πodd — ffi , ..
— ffi — ffi
X— (A.13)
k“s
—
— s .
ffi
ffi
—
— s
ffi
ffi .
—A s`2p`1 ffi —A s`2p`1 ffi
where s “ |m|. A first possible way to compute the coefficients –
..
fl –
..
fl
s,ν
Cn,k and eigenvalues Λns,ν , described by CL70, consists in finding . .
A107, page 35 of 44
A&A 603, A107 (2017)
2 2
n=0 n=0
1.5 n=1 1.5 n=1
n=2 n=2
1 1
n=3 n=3
0.5 n=4 0.5 n=4
2,0.8
2,0
0 n=5 0 n=5
n
n
−0.5 −0.5
−1 −1
−1.5 −1.5
−2 −2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
2 2
n=0 n=0
1.5 n=1 1.5 n=1
n=2 n=2
1 1
n=3 n=3
0.5 0.5 n=4
n=4
2,1.2
2,2
n=5 0 n=5
n
0
n
−0.5 −0.5
−1 −1
−1.5 −1.5
−2 −2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
3 4
n=0 n=0
n=1 3 n=1
2
n=2 n=2
2
1 n=3 n=3
n=4 1 n=4
2,12
2,4
0 n=5 0 n=5
n
−1
−1
−2
−2
−3
−3 −4
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
Fig. A.2. Normalized Hough functions Θ2,ν
computed as functions of the colatitude (θ) for 0 ď n ď 5 (gravity modes) and various values of
n
ν “ p2Ωq {σ. Top-left: ν “ 0; top-right: ν “ 0.8; middle-left: ν “ 1.2; middle-right: ν “ 2; bottom-left: ν “ 4; bottom-right: ν “ 12. The case
ν “ 0 corresponds to the normalized associated Legendre polynomials P2l pcos θq. The two plots at the top represent the regime of super-inertial
waves (|ν| ď 1) where Hough functions look like Legendre polynomials. In the regime of sub-inertial waves (|ν| ą 1), functions become ´ more
¯
and more evanescent when |ν| increases, and oscillations are trapped around the equator in the interval rθν , π ´ θν s, where θν “ arccos |ν|´1 .
with the matrices » fi

R s`1 P s`1 0 ¨¨¨ 0 ¨¨¨
—Q s`3 R s`3 P s`3 ¨ ¨ ¨ 0 ¨ ¨ ¨ffi
— 0 Q 0
— s`5 R s`5 ¨¨¨ ¨ ¨ ¨ffi
ffi
Πodd “ — .. .. .. . (A.15)
— ffi
— . . . P s`2p´1 ¨ ¨ ¨ffi
ffi
— 0 Q R ¨ ¨ ¨fl
ffi
»
Rs Ps 0 ¨¨¨ 0 ¨¨¨
fi s`2p`1 s`2p`1
.. ..
–
—Q s`2 R s`2 P s`2 ¨¨¨ 0 ¨ ¨ ¨ffi . .
— 0 Q 0
— s`4 R s`4 ¨¨¨ ¨ ¨ ¨ffi
ffi
— . .. ..
Πeven “— . , (A.14) They are solved numerically by linear algebra algorithms with
ffi
— . . . P s`2p´2 ¨ ¨ ¨ffi
ffi great efficiency. For example, the case N “ 1000 is computed
— 0 Q s`2p R s`2p ¨ ¨ ¨fl
ffi
–
.. .. on a laptop instantaneously. To make the comparison with
. . Lee & Saio (1997) easier (see the previous subsection), we set
A107, page 36 of 44
3
n= 6 n= 6
4
n= 5 n= 5
2
n= 4 n= 4
2 n= 3 n= 3
1
n= 2 n= 2
2,1.2
2,2
0 n= 1 0 n= 1
n
n
−1
−2
−2
−4
−3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
2 3
n= 6 n= 6
1.5 n= 5 n= 5
2
n= 4 n= 4
1
n= 3 n= 3
1
0.5 n= 2 n= 2
2,12
2,4
0 n= 1 n= 1
n
n
−0.5
−1
−1
−2
−1.5
−2 −3
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ /
Fig. A.3. Normalized Hough functions Θ2,νn computed as functions of the colatitude (θ) for ´6 ď n ď ´1 (Rossby modes) and various values of
ν “ p2Ωq {σ. Top-left: ν “ 1.2; top-right: ν “ 2; bottom-left: ν “ 4; bottom-right: ν “ 12. In the regime of sub-inertial waves (|ν| ą 1), Rossby
modes are trapped around the poles when |ν| Ñ 1.
Θns,ν ą 0 at the equatorial point (θ “ π{2) for even solutions Rossby modes are trapped around the poles and are thus well
and dΘns,ν {dθ ă 0 for odd solutions. Hence, the coefficients of represented by highly oscillating associated Legendre polyno-
Eq. (A.7) are explicitly given by mials (Fig. A.3). At |ν| Ñ `8, gravity modes behave similarly
b and are trapped around the equatorial belt (Figs. 2, A.2).
2pk`sq!
A s,ν
p2k`1qpk´sq! n,k
Since the spectral method has been proved to be very effi-
s,ν s,ν
Cn,k “ Sn c , (A.16) cient at computing a large number of eigenvalues and eigenfunc-
ř`8 2p j`sq! ´ s,ν ¯2 tions at once, it is used for all calculations in this work. We note
j“s p2 j`1qp j´sq! An, j that some particular cases cannot be treated using Eq. (A.10).
These cases have been discussed by CL70 and will not be devel-
where
oped in this work.
Sns,ν “ sign Θ̃ns,ν θ “ π2
` ˘(
for even values of n,
Appendix B: Numerical scheme used to integrate
# + (A.17)
dΘ̃ns,ν ˇˇ
ˇ
spatial differential equations
Sns,ν “ ´sign for odd values of n.
dθ ˇθ“π{2
We detail here the numerical scheme used to integrate Laplace’s
The solutions and projection matrices given by the spectral tidal equation (with the finite differences method) and the verti-
method are plotted in Figs. 2, 3, A.2, A.3. In the super-inertial cal structure equation. This method, introduced by Bruce et al.
regime (|ν| ď 1), the Hough functions basis is composed of (1953) and described by Chapman & Lindzen (1970) for a par-
gravity modes alone. Therefore, the projection matrix is a band- ticular case, is very convenient when solving second-order differ-
matrix. In this regime Hough functions are very similar to the as- ential equations with non-constant coefficients. Those equations
sociated Legendre polynomials. In the case ν “ 0 (top left), the are written
bases are exactly the same and the projection matrix is diagonal. d2 y dy
The sub-inertial regime is characterized by modes of both 2
` P ` Qy “ R, (B.1)
dx dx
types, the transition being identified by Hough functions of the
lowest degrees: Rossby modes on the left of the map (lowest “where P,‰ Q, and R are given functions of x. The domain
eigenvalues) and gravity modes on the right (highest eigenval- xinf , xsup on which Eq. (B.1) is integrated is divided into N in-
ues). The maps reveal the side effect inherent to the spectral tervals, i.e. N ` 1 points, identified by the index n. Hence, for a
method. Hough functions of the highest degrees are obviously regular mesh with elements of size δx, the first- and second-order
not well represented by the family of associated Legendre Poly- derivative of a solution f are expressed
nomials; more polynomials are necessary. This is explained by df fn`1 ´ fn´1 d2 f fn`1 ` fn´1 ´ 2 fn
the asymptotic behaviour of Hough functions. At |ν| Ñ 1` , “ and “ (B.2)
dx 2δx dx2 δx2
A107, page 37 of 44
A&A 603, A107 (2017)
with n P ~1, N ´ 1. Substituting Eq. (B.2) in Eq. (B.1), we

ZR
obtain the numerical relations Zo
ZE
An fn`1 ` Bn fn ` Cn fn´1 “ Dn , (B.3) IB
with the coefficients εA

XE Orbital
δxPn plane
An “ 1 ` , Bn “ δx2 Qn ´ 2, r
"A A ΘA B
2 Reference ωB
(B.4) φA
plane IB XR ΩB*
δxPn
Cn “ 1 ´ , Dn “ δx2 Rn , !
2 Xo εA
Equatorial
the Pn , Qn , and Rn coefficients being the components of the nu- plane
merical vectors corresponding to P, Q, and R. Then, we intro-
duce the coefficients αn and βn , such that periastron
fn “ αn fn`1 ` βn . (B.5)
These coefficients are defined straightforwardly by the recursive Fig. C.1. Inertial reference, orbital and equatorial rotating frames,
relations and the associated Euler’s angles of orientation. Figure taken from
$ Mathis & Le Poncin-Lafitte (2009).
An
& αn “ ´ Bn ` αn´1Cn
’
’
’
(B.6)
Dn ´ βn´1Cn ‚ RO : tOA , XO , Y O , Z O u, the orbital frame linked to RR by the
% βn “
’
’
’ ¨
Bn ` αn´1Cn Euler angles:
– IB , the inclination of the orbital frame with respect to
The first terms (α0 and β0 ) are deduced from the boundary con- pOA , XR , Y R q;
dition at x “ xinf , which can be written – ωB , the argument of the pericentre;
ˇ
d f ˇˇ – Ω˚B , the longitude of the ascending node.
a0 ` b0 f pxinf q “ c0 (B.7) ‚ RE;T : tOA , XE , Y E , Z E u, the spin equatorial frame rotating
dx ˇ xinf with the angular velocity ΩA , and linked to RR by three Eu-
where a0 , b0 , and c0 are real or complex coefficients. Thus, we ler angles:
obtain – εA , the obliquity, i.e. the inclination of the equatorial
plane with respect to pOA , XR , Y R q;
$ a0
’ α0 “ ´ b δx ´ a
’ – ΘA , the mean sidereal angle such that ΩA “ dΘA {dt,
& 0 0 which is the angle between the minimal axis of inertia
(B.8) and the straight line due to the intersection of the planes
% β0 “ c0 δx
’
pOA , XE , Y E q and pOA , XR , Y R q;
’
b0 δx ´ a0
– φA , the general precession angle.
and, step by step, all the following terms. At x “ xsup , we apply
Finally, we denote a the semi-major axis and M̃B the mean
the condition
ˇ anomaly with M̃B « nB t, nB being the mean motion. Neglecting
d f ˇˇ ` ˘ the action of the centrifugal acceleration, the hydrostatic struc-
aN ` bN f xsup “ cN (B.9) ture of the atmosphere presents a spherical symmetry, p0 , ρ0 , T 0
dx ˇ
xsup
being functions only of the radial coordinate. Hence, any heat
with paN , bN , cN q P C3 , which gives forcing caused by the star can be written
cN δx ` βN´1 aN 1
fN “ ¨ (B.10) J pr, ψq “ h pr, ψq , (C.1)
bN δx ` aN p1 ´ αN´1 q a2
where h is a function of r and ψ, the angle between the points
The solution is finally obtained by computing the fn backwards M pr, θ, ϕq and B pa, θAB , ϕAB q. We decompose J in normalized
using Eq. (B.5). Legendre polynomials
1 ÿ
Appendix C: Decomposition of the thermal forcing J“ 2 Jl prq Pl pcos ψq (C.2)
a lě0
in Keplerian elements
We establish here the general multipole expansion in spherical with Jl prq “ xh, Pl y. Using the addition theorem
harmonics of the thermal power given in Eq. (72) using the l
notations of Mathis & Le Poncin-Lafitte (2009). We consider a 4π ÿ
Pl pcos ψq “ a Yl,m pθ, ϕq Yl,m
˚
pθAB , ϕAB q ,
planet (A) orbiting circularly around its star (B). Three reference 2 p2l ` 1q m“´l
frames, all centred on the centre of mass of the planet and repre-
sented in Fig. C.1, must be defined: (C.3)
‚ RR : tOA , XR , Y R , Z R u, time independent inertial frame, with with

Z R the direction of the total angular momentum of the sys- 1
tem. Yl,m pθ, ϕq “ ? Pml pcos θq e
imϕ
, (C.4)
2π
A107, page 38 of 44
we obtain where the spatial functions U m,σ and J m,σ are expressed
ÿ
l
U pr, θq “ Ulm,σ prq Pml pcos θq ,
$ m,σ
ÿ ÿ 4π Jl prq
Yl,m pθ, ϕq Yl,m
˚
pθAB , ϕAB q .
’
J“ a 2
’
& lě|m|
lě0 m“´l 2 p2l ` 1q a ÿ (C.13)
% J pr, θq “ l pcos θq ,
Jlm,σ prq Pm
’ m,σ
(C.5) ’
lě|m|
˚
Kaula’s transform is finally introduced to express Yl,m pθAB , ϕAB q
as a function of Keplerian elements (Kaula 1962; with
Mathis & Le Poncin-Lafitte 2009). Since the motion of the $
ÿl ÿl ÿ
Ul,m, j,p,q prq δl´2p`q,m ,
’ m,σ
planet is circular (e “ 0), we get ’
’ U prq “
& l
’
’
j“´l p“0 qPZ
Yl,m pθAB , ϕAB q “
ÿ
κl, j dlj,m pεA q Fl, j,p pIB q Gl,p,q p0q eiΨl,m, j,p,q ptq , (C.14)
’ l
ÿ l ÿ
ÿ
m,σ
Jl,m, j,p,q prq δl´2p`q,m ,
’
j,p,q ’
’ J
% l
’ prq “
(C.6) j“´l p“0 qPZ
with p j, p, qq P ~´l, l ˆ ~0, l ˆ Z, where the κl, j coefficients are the notation δ standing for the Kronecker symbol.
given by
d
2l ` 1 pl ´ | j|q! Appendix D: Radiated power as a function
κl, j “ ¨ (C.7) of physical parameters
4π pl ` | j|q!
The power per unit mass emitted by an optically thin atmo-
The functions denoted dlj,m , Fl, j,p , and Gl,p,q are called the obliq- sphere (Jrad ) can be expressed as an analytic function of physical
uity, inclination, and eccentricity functions (for more details, parameters in our simplified framework. Indeed, assuming that
see Mathis & Le Poncin-Lafitte 2009). The angle Ψl,m, j,p,q can the flux radiated by a layer of thickness dx, denoted δI, propa-
be written gates along the vertical direction without being absorbed by the
other layers, we write
Ψl,m, j,p,q ptq “ ´σl,m,p,q t ` ψl,m, j,p,q , (C.8)
1
the parameter σl,m,p,q “ mΩA ´ pl ´ 2p ` qq nB being the tidal δI “ φrad ρ0 dr, (D.1)
2
frequency of˘ the mode pl, m, p, qq and ψl,m, j,p,q “ pl ´ 2pq ωB `
j Ω˚B ´ φA ` pl ´ mq π{2 the phase lag. Substituting Eq. (C.6) where φrad is the total power emitted per unit mass. Then, the
`
in Eq. (C.5), we finally get atmosphere is treated as a grey body of molar emissivity per unit
mass a , which – using the molar concentration C0 introduced in
l
ÿ ÿ ÿ Jl prq m Sect. 2 – gives
J “ J0 ` Wl,m, j,p,q Pl pcos θq eirσl,m,p,q t`mϕs ,
a2
lě1 m“´l j,p,q δI “ aC0 S T 4 dr. (D.2)
(C.9)
Combining Eqs. (D.1) and (D.2) allows us to get
where
2a
φrad “ S T 4.
d
pl ´ | j|q! (D.3)
Wl,m, j,p,q pε, I, eq “ M
”pl ` | j|q! ı Finally, we derive the perturbed radiative loss Jrad associated
ˆ dlj,m pεq Fl, j,p pIq Gl,p,q peq eímϕl,m, j,p,q , with the tidal perturbation as a function of δT by linearizing the
(C.10) previous expression around the equilibrium state,
with ϕl,m, j,p,q “ ψl,m, j,p,q {m. 8a

Jrad “ S T 03 δT, (D.4)
With the same notations, the gravitational potential is written M
which gives the radiation frequency introduced in Eq. (17):
l
G MB l m
8κa S 3
ÿ ÿ ÿ
U “ U0 ´ Wl,m, j,p,q r Pl pcos θq eirσl,m,p,q t`mϕs .
al`1 σ0 prq “ T . (D.5)
lě1 m“´l j,p,q RGP 0
(C.11)
The tidal frequency, written σ “ mΩ ´ snorb , is defined by the Appendix E: Analytic solution for constant profiles
doublet pm, sq P Z2 . and upper boundary condition
The perturbing tidal potential and thermal power finally write
We show here that the solution of the equation describing the
vertical structure strongly depends on the choice of the upper
U σs,m ,m pr, θq eipσs,m t`sϕq ,
ÿ
$ boundary condition. As discussed in Sect. 3,
’U “
’
& ps,mqPZ2
(C.12) Ψm,σ
n p0q “ 0 (E.1)
J σs,m ,m pr, θq eipσs,m t`sϕq ,
ÿ
%J “
’
’
ps,mqPZ2 at the ground level x “ 0.
A107, page 39 of 44
A&A 603, A107 (2017)
Instead of the energy-bounded condition prescribed in the

model, at the upper limit (xatm ) we set the commonly used stress- 15.00
free condition,
10.00
1 dp0
δpn ` ξr;n “ 0. (E.2) 5.00
H dx
T [PN.m]
We obtain, at x “ xatm , 0.00
dΨm,σ
n
−5.00
` PΨm,σ
n “ Q, (E.3)
dx −10.00
with the complex coefficients −15.00 n(+ ) bounded

stress free
’ P “ A ´ p1 ´ ε q σ s;n ,
$ 2
’
n s;n −300 −200 −100 0 100 200 300
Γ1 σ2
’
’
&
(E.4)
κR2 HΛm,ν
„ 
’ xatm
n
´ m,σ m,σ
. Fig. E.1. Tidal torques exerted on the atmosphere (in peta-Newton me-
’
’Q “ e
’ 2 J ` Un
% g piσ ` σ0 q n σ2 x“xatm tres (PN.m)) induced by the first gravity mode of the Earth’s semid-
iurnal tide (n “ 0) as a function of the reduced tidal frequency (χ)
The parameter xatm must be fixed so that ρ0 pxatm q ! ρ0 p0q, obtained with two different conditions at the upper boundary: the
which corresponds to xatm " 1. A simple analytic solution can be energy-bounded condition chosen for the model (red line) and a stress-
computed assuming that Jnm,σ and Unm,σ are constant profiles. The free condition (green dashed line). The equation for the structure
second member of the vertical equation, C, given by Eq. (99), in the vertical (resp. horizontal) direction has been integrated with
10 000 mesh points (resp. 400 associated Legendre polynomials).
becomes
κΛm,ν
n
C“ tKn Jnm,σ ´ iσUnm,σ u , (E.5)
Hσ2 piσ ` σ0 q
and substituting the analytic solution (Eq. E.6) in Eq. (117), we
the solution is written get
"ˆ ˙ *
H 2C 1 “ ` ˘ ´x‰ c
8π 2 a3 H
Ψn “ ,
m,σ
` ˘
Z´ sin k̂n x ` k̂n cos k̂n x é 2 2,σSD
k2 „ Ξatm J2 (E.11)
k̂n an bn 2 5 3 MS R
(E.6)
where Ξatm is explicitly given by
with an “ ik̂n ` 1{2 and bn “ ik̂n ´ 1{2, and the complex inte-
κKn piσ ` Γ1 σ0 q
ż xatm
gration constant
Ξatm “ Meb0 x
ˆ ˙ 0 λ0 a0 b0 c2s piσ ` σ0 q2 p1 ´ ε s;n q
k̂n an bn 2 1 ` ˘
2
Q ` k̂ n ` P sin k̂n xatm „
1 Γ1 a0 b0 piσ ` σ0 q ´x
 *
H C 2 `Ne á0 x
` λ0 ´ Bn ´ e ,
Z“ ` ˘ ` ˘ 2 Kn piσ ` Γ1 σ0 q
P sin k̂n xatm ` k̂n cos k̂n xatm
(E.7) (E.12)
`1 ˘” ` ˘ x
´ atm
ı
k̂n 2 ´ P cos k̂n xatm ´ e 2
with
` ` ˘ ` ˘ ¨
P sin k̂n xatm ` k̂n cos k̂n xatm
λ
$ ˆ ˙ „ˆ ˙ 
1 i 1
’ M “ 0 Z ´ ` Bn ´
’ Z ´ Bn ´ λ 0 ,
2
Now, the condition of Eq. (108) characterizing thermal tides, i.e. 2 2 2 2
’
&
λ
ˆ ˙ „ˆ ˙ 
’ 1 i 1
% N “ 0 Z ´ ` Bn ` Bn ´ λ20 .
’
ˇ m,σ ˇ ˇ ˇ
ˇ Un ˇ ˇ H ´ hn ˇ
’ Z´
2 2 2 2
ˇ J m,σ ˇ ˇ σH ˇ , (E.8)
ˇ ˇ!ˇ ˇ
n (E.13)
is assumed. The second-order Love number reduced to the main Similarly, the torque of Eq. (88) associated with the Θ2,ν
0 -
contribution of the thermal semidiurnal tide, given by Eq. (76), contribution of the semidiurnal tide can be written
can be expressed
c T 2,σSD „ 2πR2 H= tΞatm u U2 J2 . (E.14)
8π 2 a3 H xatm 2,σSD
ż
2,σSD
k2 „ δρ0,2 pxq dx (E.9) To illustrate the difference induced by the choice of the upper
5 3 MS R 0
boundary condition, we compute the variation of the tidal torque
for a thin isothermal atmosphere. As a zero-order approxima- generated by the Earth’s semidiurnal tide with the reduced tidal
tion, we consider that the forcings are the constant profiles frequency (χ “ pΩ ´ norb q {norb ) for two conditions: the energy-
U2 “ U2,2,2,0,0 and J2 “ J2,2,2,0,0 used in Sect. 2 (with J2 P R). bounded condition used in this work and the free-surface con-
Therefore, introducing the complex impedance Ξatm , defined by dition studied here. To simplify calculations, we focus on the
ż xatm gravity mode of lowest degree (n “ 0). Results are plotted in
Fig. E.1. In the first case, we obtain the smooth evolution ob-
δρ2,σ
0,2 pxq dx “ Ξatm J2 ,
SD
(E.10) served in Fig. 14 (with slight differences in the vicinity of the
0
A107, page 40 of 44
synchronization because of the absence of Rossby modes), but of ν shows that this parameter strongly varies around the syn-
the free-surface condition obviously leads to a highly resonant chronization. The transition at σ “ 0 is discontinuous, ν switch-
behaviour. This can be explained by the form of the solution ing from ´8 to `8.
given by Eq. (E.6), where the integration constant Z depends
on k̂n and xatm . By fixing xatm , we implicitly define values of k̂n Appendix G: Consequences of an insufficient
for which Z reaches a maximum. On the contrary, the integra- spatial resolution on the numerical computation
tion constant obtained by applying the energy-bounded condi- of the vertical structure
tion, given by Eq. (128), does not depend on the upper limit far
beyond the critical altitude given by Eq. (127). As pointed out by CL70 and detailed in Sect. 5, solving numeri-
cally the vertical structure equation (Eq. (107)) with the numeri-
cal scheme of Appendix B requires using a sufficiently high spa-
Appendix F: Supplementary materials regarding tial resolution. This condition is mathematically expressed by
the dependence of the tidal response the criterion given by Eq. (228) for a regular mesh and depends
on the tidal frequency on the vertical wavenumbers that characterize Hough modes.
To compute the response of a mode, it is necessary to use a
As a complement to the tidal torques (Figs. 14) and perturbed spatial step δx lower than the corresponding variations length
pressure, density, and temperature (Figs. 17) resulting from the scale (LV;n , defined by Eq. (48)), otherwise the signal does not
Earth’s solar semidiurnal tide, we plot several quantities as a appear. We illustrate this point in here by considering the simpli-
function of the tidal frequency that can be helpful for the di- fied forced wave equation
agnostic of results (Fig. F.1). These quantities are the coeffi-
cients c2,n,2 “ xP22 , Θ2,ν 2
n y introduced in Eq. (81) which weight
d2 Ψ
` kV2 Ψ “ e´x{2 , (G.1)
the contribution of Hough modes to the tidal torque; the eigen- dx2
values of Laplace’s tidal equation (Λm,ν n ); the length scale of
vertical variations associated to Hough modes (LV;n ); the spin where kV P C is such that = tkV u ą 0, with the previously cho-
parameter (ν); and the real and imaginary parts of the complex sen boundary conditions,
vertical wavenumber (k̂n denoted kn in plots), which determines ˇ
dΨ ˇˇ
Ψ p0q “ 0 and ´ ikV Ψ xsup “ 0.
` ˘
the nature (propagative or evanescent) of tidal waves. (G.2)
These plots highlight the difference between prograde (ν ă dx xsup
ˇ
0) and retrograde (ν ą 0) modes. Indeed, it can be noted that the
projection of the associated Legendre polynomial P22 on Hough We arbitrarily set kV “ 100 p1 ` iq in order to satisfy the condi-
functions is not the same before and after the synchronization: tion LV ! 1 (keeping in mind that xsup “ 10). This corresponds
the weight of the gravity mode of degree 0, which is the most to a strongly damped tidal mode similar to those predicted by the
representative one, varies and Rossby modes are predominating model in a frequency range close to synchronization. The criti-
in a very short interval of positive tidal frequencies. This dissym- cal number of points necessary to compute the response with
metry is retrieved in the variations of the horizontal eigenvalues, our scheme is then Ncrit „ 400. We solve Eq. (G.1) for vari-
where Rossby modes only appear for σ ą 0. ous resolutions. The corresponding results are plotted in Fig. G.1
As predicted by its analytic expression (Eq. (102)), the real where they are also compared to the analytical solution given by
and imaginary part of the vertical wavenumber diverge at the Eq. (129). The flattening observed for N ă Ncrit gives an idea
synchronization. Therefore, Hough modes become both highly of the kind of numerical difficulties that need to be addressed to
oscillating and strongly damped in this region. Their spatial compute tidal low-frequency waves with a GCM. Particularly,
variation scale along the vertical direction decays, which corre- this shows that if we compute the tidal torque numerically by
sponds to what is observed in the variations of LV;n . Actually, be- using the scheme of Appendix B with an insufficient resolution,
low a given typical length scale, these modes are damped by dis- we will only be able to get the large-scale contribution associated
sipative processes such as thermal diffusion. Finally, the graph with the horizontal component (see Figs. 14).
A107, page 41 of 44
A&A 603, A107 (2017)
1.00 10.00
8.00
sign( {kn})log10(1+ | {kn}|)

n= 6
n= 5 6.00
0.80
n= 4 4.00
n= 3
2.00
>2
0.60 n= 2
n= 1 0.00
2,
n
n=0 −2.00
<P22,
0.40 n=1 −4.00

n=2
−6.00
n=3
0.20 −8.00
n=4
n=5 −10.00
0.00 −12.00
−15 −10 −5 0 5 10 15 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
5.00 12.00
sign( {kn})log10(1+ | {kn}|)

4.00
10.00
n|)
3.00
+|
8.00
2.00
n)log10(1
1.00 6.00
0.00
4.00
sign(
−1.00
2.00
−2.00
−3.00 0.00
−15 −10 −5 0 5 10 15 −15.00 −10.00 −5.00 0.00 5.00 10.00 15.00
2.00
0.0
1.50
−1.0
1.00
sign( )log10(1+| |)
0.50
log10(LV;n)
−2.0
0.00
−3.0
−0.50
−1.00
−4.0
−1.50
−5.0 −2.00
−15.00 −10.00 −5.00 0.00 5.00 10.00 15.00 −15 −10 −5 0 5 10 15
Fig. F.1. Tidal response of the Earth’s atmosphere to a quadrupolar forcing (U 9 P22 pcos θq e2iϕ and J 9 P22 pcos θq e2iϕ ) as a function of the reduced
tidal frequency χ “ pΩ ´ norb q/norb (horizontal axis). For each quantity qn associated with the Hough mode of degree n, except the coefficients
cσ,2
2,n,2 (top left) and the scales of variations LV;n , the function f pχq “ sign pqn q log p1 ` |qn |q is plotted (vertical axis) for n P ~´6, 5. Left, from
top to bottom: a) coefficients cσ,2 2 σ,2 2
2,n,2 “ xP2 , Θn y (Eq. (70)); b) eigenvalues of ` Laplace’s tidal equation (Λm,νn defined by Eq. (35)); c) normalized
ˇ ˇ˘
variation scales of the vertical structure of the vertical component, LV;n “ 2π/ xatm ˇk̂n ˇ (Eq. (48)), plotted in logarithmic scale. Right, from top
to bottom: d) real part of the vertical wavelengths (k̂n , Eq. (103)); e) imaginary part of the vertical wavelengths; f) spin parameter ν “ 2Ω{σ
(Eq. (23)).
A107, page 42 of 44
0.20 Table H.1. continued.
0.00 Symbol Designation

Vr , Vθ , Vϕ components of V in spherical coordinates
−0.20 ξ displacement vector
ξr , ξθ , ξϕ components of ξ in spherical coordinates
δp tidal pressure perturbation
Re( )
−0.40 δρ tidal density perturbation

δT tidal temperature perturbation
−0.60 ana cs sound velocity
N = 20 κ atmospheric parameter
N = 200 φrad radiated power per unit mass
−0.80
N = 2000 σ0 radiation frequency
N = 20000 S Stefan-Boltzmann constant
−1.00 a effective molar attenuation coefficient
−4 −3 −2 −1 0 1 f any perturbed quantity
log(x) σ tidal frequency
ν spin parameter
Fig. G.1. Real part of the solution of Eq. (G.1) as a function of x in m longitudinal degree
logarithmic scale computed numerically for various mesh resolutions. y pressure perturbation divided by ρ0
The number of points of the spatial discretization is set at N “ 20 (green ξH horizontal displacement
dashed line), N “ 200 (blue dashed line), N “ 2000 (pink dashed Lm,ν
θ operator associated with θ-components
line), and N “ 20 000 (cyan dashed line). These numerical solutions Lm,ν
ϕ operator associated with ϕ-components
are compared to the analytical solution given by Eq. (129) (red line). m,ν
L Laplace’s tidal operator
n meridional degree
Appendix H: Nomenclature Θm,ν
n Hough functions
Λm,ν
n eigenvalues of Laplace’s tidal equation
Jnm,σ components of J
Table H.1. List of the symbols used throughout this work and their Unm,σ components of U
designations in order of introduction. ξr;n
m,σ
components of ξr
ξθ;n
m,σ
components of ξθ
Symbol Designation ξϕ;n
m,σ
components of ξϕ
m,σ
i imaginary number Vr;n components of Vr
m,σ
< real part Vθ;n components of Vθ
m,σ
= imaginary part Vϕ;n components of Vϕ
˚
conjugate δpm,σ
n components of δp
C associated with the Earth δρm,σ
n components of δρ
B associated with Venus δT nm,σ components of δT
RET planetary equatorial rotating frame ym,σ
n components of y
XE , Y E , Z E vectorial basis associated with RET A1 , A2 , A3 coefficients of the equation r2 ξr;n
r radial coordinate B1 , B2 , B3 coefficients of the equation yn
θ colatitude A, B, C coefficients of the vertical structure
ϕ east longitude equation
δ latitude / fluctuations F function of the variable change
er , eθ , eϕ spherical vectorial basis k̂n vertical wavenumber
t time LV;n scale of variation
g gravity ε s;n acoustic coefficient
Ω planet spin vector σ s;n acoustic frequency
Ω angular velocity An , Bn , Cn coefficients of polarization relations
R planet radius Kcurv curvature term in polarization relations
atm associated with the atmosphere Υm,σ
n linear operator associated with the vertical
gr associated with the ground structure
Ratm external radius of the atmosphere Υm,σ
f ;n components of the operator
Hatm typical thickness of the atmosphere associated with f
M molar mass of the atmosphere Xn vector of the forcings
p pressure Yn vector of the response
ρ density σdiff;n diffusive frequency
T temperature G gravitational constant
p0 background pressure distribution l degrees of Legendre polynomials
ρ0 background density distribution P̃ml associated Legendre functions
T0 background temperature distribution Pml normalized associated Legendre functions
C0 background molar concentration distribution U potential induced by the tidal response
Γ1 adiabatic exponent Ũlm,σ components of U projected on the P̃ml
N Brunt-Väisälä frequency Ulm,σ components of U projected on the Pml
RGP perfect gas constant Flm,σ first component of Ulm,σ
Rs specific gas constant of the atmosphere Hlm,σ second component of Ulm,σ
U gravitational tidal potential klm,σ Love numbers for the gravitational potential
J tidal heating power per unit mass Wlm coefficients of Kaula’s expansion
V tidal velocity field
A107, page 43 of 44
A&A 603, A107 (2017)
Table H.1. continued. Table H.1. continued.
Symbol Designation Symbol Designation

δρm,σ
l density expansion coefficient for the perturbation τλ optical depth
projected on the Pml αλ global absorption coefficient
δρm,σ
n,k components of δρm,σ l Agr;λ albedo of the ground for the wavelength λ
m,σ
cl,n,k weighting coefficients of modes in T gr surface temperature
the expansion of the tidal torque ψ angle to the star-planet direction
MS mass of the star Igr;BB black-body emission of the ground
a semi-major axis gr emissivity of the ground
ε obliquity atm emissivity of the atmosphere
norb mean motion effective emissivity of the ground
σSD semidiurnal tidal frequency T gr;0 surface equilibrium temperature
U2,2,2,0,0 semidiurnal gravitational tidal potential δT gr surface temperature fluctuations
σ
J2,2,2,0,0 semidiurnal thermal forcing Bgr transfer function of the ground
J2 expansion function of the thermal forcing Qatm diffusive power flux in the atmosphere
projected on P2 Qgr diffusive power flux in the ground
V volumetric domain of the atmosphere katm thermal conductivity of the atmosphere
T tidal torque kgr thermal conductivity of the ground
T m,σ frequency components of the tidal torque Katm vertical turbulent thermal diffusivity
∆ϕ phase lag of the tidal response Kgr thermal diffusivity of the ground
H associated with the horizontal component catm thermal capacity of the atmosphere
V associated with the vertical component cgr thermal capacity of the ground
grav caused by the gravitational forcing δσatm skin thickness in the atmosphere
therm caused by the thermal forcing δσgr skin thickness in the ground
TH torque due to the horizontal component ςgr radiative impedance of the ground
TV torque due to the vertical component βatm thermal conductive capacity of the atmosphere
H factor in the expressions of the torque βgr thermal conductive capacity of the ground
z altitude σbl typical frequency of the atmospheric boundary
x altitude normalized by the pressure height scale layer
xatm normalized upper boundary τσatm vertical damping coefficient in the boundary layer
aRE;T local acceleration in the reference frame δT bl temperature fluctuations in the boundary layer
V,
V
ac centrifugal acceleration associated with visible light
hn equivalent depth IR , IR
associated with the infrared
hc characteristic depth of the vertical wavenumber N number of points of the spatial discretization
ςn scale ratio in the vertical
( wavenumber T timescale of the spin evolution
∆ M
(
discriminant in < k̂n and = k̂n mass of the atmosphere
A ,B functions generating the analytic solution χ tidal frequency divided by the mean motion
P, Q coefficients of the upper boundary equation
xcrit typical damping depth
Dn , En parameters
FP parametrized function of the analytic solution
Torb orbital period
γ lag angle
W dimensionless constant
a, b empirical parameters in the model of
Correia et al. (2003)
k̂CL;n vertical wavenumber in CL70
η relative difference between k̂n and kCL;n
λ wavelength of light
IS;λ spectral radiance of the star
IS black-body stellar flux
RS star radius
TS surface temperature of the star
d star-planet distance
c speed of light in vacuum
h Planck constant
kB Boltzmann constant
Aatm;λ albedo of the atmosphere for the wavelength λ
I0;λ effective incident flux
l linear spatial coordinate
inc
caused by the incident flux
ref
caused by the reflected flux
rad
caused by the flux emitted by the surface
CG molar concentration of an absorbing gas
εG;λ molar extinction density coefficient
Iλ heat power flux
J heat power per unit volume
A107, page 44 of 44

[desrotour17] Atmospheric tides in Earth-like planets

Uploaded by

Copyright:

Available Formats

[desrotour17] Atmospheric tides in Earth-like planets

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

[desrotour17] Atmospheric tides in Earth-like planets

Uploaded by

Copyright:

Available Formats

A&A 603, A107 (2017)

DOI: 10.1051/0004-6361/201628252 Astronomy

Atmospheric tides in Earth-like planets

1. Introduction as well as solid tides, and the energy dissipated is even

2.1. Forced dynamics equations

The parameters m and ν determine the horizontal operators separated coordinates,

Mathis et al. 2008). Hence, the traditional approximation allows n

corresponding horizontal component ξm,σ

Venus in the case of the solar semidiurnal tide defined by m “ 2,

Under the traditional approximation, solutions of Eqs. (31)´(33) 2

with the coefficients and

/ 2 dρ0 iσ N 2 iσ ` Γ1 σ0 gρ0 1 dB1

−1.0 We now transform Eq. (40) in a Schrödinger-like equation by

With the general variable change, Eq. (40) becomes

to (59) can be reduced to

with the scale ratio ςn “ 4H 2 Λm,ν 2

The term ςn pε s;n ´ 1q comes from the radial acceleration in the

sidering the case where σ0 ! |σ|, we recover the wavenumber i

no material escape when x Ñ `8, i.e. Vr | xÑ8 “ 0. This means i

H 2C ” ´ x ı From the analytical solution and the polarization relation of den-

where the expressions of Bn and Kn are given by Eqs. (120)

(132) Assuming |ε s;n | ! 1, we identify the different contributions. It

allows us to write T 2,σ under the form

with This difference, written

Therefore, the case N 2 “ 0 corresponds to the adiabatic temper-

with H p0q “ p0 p0q { rgρ0 p0qs.

4.2. Tidal response

of the horizontal velocities

Therefore, they can be easily written as a product of separable

leads us to CG “ C0 “ ρ0 {M. Therefore, the incident power flux (190)

Iinc / max(Iinc) Irad / max(Irad)

Jinc / max(Jinc) Jrad / max(Jrad)

to provide a first analytical treatment of the problem. At the in-

expression. Indeed, by writing the heat transport equation as

and linearizing it, we compute the expression of the tidal diffu-

5.4. Simplified bichromatic case

where T S is the temperature of the star. This flux is partly re-

& δI “ ςgr Bgr δIgr

6. Application to the Earth and Venus 5

– the involved characteristic length scales (e.g. planetary ra-

Using the equations of hydrodynamics, we have adapted the

INTERNAL&STRUCTURE& strong&atmospheric&Fdal&torque& FINAL&ROTATION&STATES&

sign( n)log10(1 + | n|)

with the matrices » fi

with n P ~1, N ´ 1. Substituting Eq. (B.2) in Eq. (B.1), we

An fn`1 ` Bn fn ` Cn fn´1 “ Dn , (B.3) IB

with the coefficients εA

‚ RR : tOA , XR , Y R , Z R u, time independent inertial frame, with with

with ϕl,m, j,p,q “ ψl,m, j,p,q {m. 8a

Instead of the energy-bounded condition prescribed in the

with the complex coefficients −15.00 n(+ ) bounded

sign( {kn})log10(1+ | {kn}|)

0.40 n=1 −4.00

sign( {kn})log10(1+ | {kn}|)

0.20 Table H.1. continued.

0.00 Symbol Designation

−0.40 δρ tidal density perturbation

Table H.1. continued. Table H.1. continued.

Symbol Designation Symbol Designation

You might also like

& δI “ ςgr Bgr δIgr

with n P ~1, N ´ 1. Substituting Eq. (B.2) in Eq. (B.1), we

with ϕl,m, j,p,q “ ψl,m, j,p,q {m. 8a