sign in sign up
<<
Home , Primitive equations

arXiv:1902.03186v3 [math.AP] 11 Nov 2020 rnpr qain nvria ieto ersne yteter the by represented direction vertical in equations transport qain ihol oiotlvsoiyadtepyia aea Dir lateral physical fo the problem and conditions. time-periodic horizontal and value only initial dy with the hydrost equations atmospheric are a work and assuming this of oceanic equations Navier-Stokes subject describing the for from used derived are are they and flows, em r lovr neetn ic hycmiefaue fbot term of the partia by features represented only combine directions with horizontal they in models since v equations such interesting sion vertical very view the also of are and point terms dominant analytical the be From to glected. considered is viscosity horizontal nuto losu oietf lse fiiildt o hc hsproble this which Fo for well-posed. data directions. initial globally horizontal of even the classes or identify in to p smoothed us is is allows regularity intuition that it expects while thus direction one vertical speaking, Roughly below. (1.2) eidcsolutions. periodic 51,7D3 60,86A10. 86A05, 76D03, 35M10, H NTA AU N H IEPROI PROBLEM TIME-PERIODIC THE AND VALUE INITIAL THE h 3 The h oiaint td hspolmi hti aygohsclmode geophysical many in that is problem this study to motivation The 2010 e od n phrases. and words Key RMTV QAIN IHHRZNA VISCOSITY: HORIZONTAL WITH EQUATIONS PRIMITIVE ahmtc ujc Classification. Subject Mathematics samdlwt yeblcadprblcfaue o hc c which self-evident. time-peri not for the are for features results parabolic such small applicable, and directly hyperbolic not with are model a is eidcstig o h ieproi rbe,existenc problem, time-periodic the z For setting. periodic ii(.Fn.Aa.221) 6644,21)friiild initial for 2017) resul 4606-4641, well-posedness 272(11): global Anal. the Func. beyond (J. Titi goes extend This thus and solution. instantaneously regularizes solution weak in o nta aain data initial for au rbe,w baneitneaduiuns flocal of uniqueness and existence obtain botto we and top problem, on value conditions appro boundary direct unnecessary a Inste avoids apply lar we sides. limit, the viscosity on vertical vanishing conditions boundary Dirichlet physical iee naclnrcldmi ( = Ω domain cylindrical a on sidered Abstract. D wa n togtm eidcsltosi rvnfrsmall for proven is solutions periodic time strong and -weak piiieeutosaeoeo h udmna oesfrgeophy for models fundamental the of one are equations -primitive H 1 Ω.If (Ω). O HSCLBUDR CONDITIONS BOUNDARY PHYSICAL FOR MUHSEN ATNSA,ADMR WRONA MARC AND SAAL, MARTIN HUSSEIN, AMRU h 3 The v 0 1. ∈ H D rmtv qain,hrzna icst,iiilvalue initial viscosity, horizontal equations, primitive nrdcinadmi results main and Introduction H 1 piiieeutoswt nyhrzna icst r co are viscosity horizontal only with equations -primitive (( 1 − (( ,h h, − ,h h, ) L , ) L , 2 ( 2 G ( )adlclsrn ouin o nta data initial for solutions strong local and )) G rmr:3Q5 eodr:3A1 56,35Q86, 35K65, 35A01, Secondary: 35Q35; Primary: )), 1 − ∂ ,h h, z v 0 ) × ∈ G L , q Ω for (Ω) G ⊂ dcpolmee for even problem odic c hc nparticu- in which ach R n nqeesof uniqueness and e oagoa strong global a to s 2 do osdrn a considering of ad t near ata > q yCo iand Li Cao, by t moh ihthe with smooth, .Frteinitial the For m. ocs ic this Since forces. z wa solutions -weak − asclresults lassical ,te the then 2, ∆ m tcblne The balance. atic H aaoi diffu- parabolic h H cltboundary ichlet w∂ 1 n hyperbolic and eevdi the in reserved h primitive the r ais They namics. soiyi ne- is iscosity nthe in rbe,time- problem, z lwn this llowing v slocally is m compare , n- viscosity l z - sthe ls sical 2 HUSSEIN, SAAL, AND WRONA

Many forces acting on geophysical flows such as the attraction by the moon, which becomes visible in the falling and rising , are time-periodic. Moreover, in some models the wind is described as a perturbation of a periodic function. A time-periodic adds in each period energy to the system, and since there is only partial viscosity it is not self-evident whether the system remains stable enough to have time-periodic solutions. However, it turns out that at least for forces being small over one period of time, there are unique small time-periodic solutions. This work is part of the third author’s PhD thesis [37], and therein some of the computations are elaborated in more detail. 1.1. Primitive equations with only horizontal viscosity. To be precise, here time intervals (0,T ) for T (0, ) and a cylindrical domain Ω are considered, ∈ ∞ (1.1) Ω := ( h,h) G R3, for h> 0 and a smooth domain G R2, − × ⊂ ⊂ where the boundary ∂Ω decomposes into a lateral, upper and bottom part Γ := ∂G [ h,h], Γ := G +h , Γ := G h . l × − u ×{ } b × {− } The primitive equations describe the velocity u = (v, w): Ω R3 and the → p: Ω R of a fluid, where v = (v1, v2) denotes the horizontal components and w stands→ for the vertical one. The primitive equations with horizontal viscosity are

∂tv + v H v + w∂zv ∆H v + H p = f, in Ω (0,T ), · ∇ − ∇∂ p =0, in Ω × (0,T ), (1.2)  z div v + ∂ w =0, in Ω × (0,T ),  H z ×  v(t =0) = v0, in Ω which are supplemented by the boundary conditions  (1.3) v =0 onΓ (0,T ) and w =0 on Γ Γ (0,T ). l × u ∪ b × The first boundary condition is a lateral no-slip boundary condition and the latter is due to the divergence free condition div u = 0 and ν u = 0 for the outer normal derivative ν on ∂Ω. Here, x, y G are the horizontal· coordinates and ∈ T ∗ z ( h,h) the vertical coordinate, H = (∂x, ∂y) , divH = H and ∆H = 2∈ −2 ∇ ∇ ∂x + ∂y denote the horizontal gradient, divergence and Laplacian, respectively and v = v ∂ + v ∂ . Note, that for the primitive equations the nonlinear term · ∇H 1 x 2 y w∂zv is stronger compared to the nonlinearity of the Navier-Stokes equation since w = w(v) given by (2.2) below involves first order derivatives, while the pressure here is only two-dimensional. For simplicity we have formulated the equations without the , but being a zero order term it does not alter the well-posedness results discussed here. Moreover, we consider only the velocity equation without or salinity focusing on the mathematical difficulties. The general anisotropic primitive equa- tions are given if one replaces in (1.2) the term ∆H by ν1∆H + ν2∂zz for horizontal viscosity ν 0 and vertical viscosity ν 0. Here, physical constants are normal- 1 ≥ 2 ≥ ized to one, thus we consider the case ν1 = 1 and ν2 = 0. 1.2. Previous results. Cao, Li and Titi, see [5,6], have been the first to study the primitive equations with only horizontal viscosity analytically. They tackled this problem in a periodical setting by considering a vanishing vertical viscosity limit, i.e., ∆ ε∂2 for ε 0, − H − z → PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 3 and by this strategy they obtained remarkable global strong well-posedness results for the initial value problem for initial data with regularity near H1(Ω), and local well-posedness for initial data in H1(Ω). Recently, the second author [30] applied a more direct approach considering the system without vanishing viscosity limit. Thereby local well-posedness results even for less partial has been proven, and for only horizontal viscosity unnecessary boundary conditions on bottom and top have been avoided. Note that for the Navier-Stokes equations with only horizontal viscosity there are also some local well-posedness results, cf. [1, Chapter 6]. The mathematical analysis of the initial value problem for the primitive equa- tions with full viscosity, i.e., with viscosity term ν1∆H + ν2∂zz where ν1,ν2 > 0, has been started by Lions, Temam and Wang [22–24] which launched a lot of ac- tivity in the analysis of these equations. In difference to the 3D Navier-Stokes equations the primitive equations are known to be time-global well-posed for ini- tial data in H1(Ω) by the breakthrough result of Cao and Titi [7], see also [19] for different boundary conditions and non-cylindrical domains. Refinements of this 2 include global well-posendess for initial data with v0, ∂zv0 L (Ω), see [17], or v L1(( h,h),L∞(G)), see [12]. ∈ 0 ∈ − For the inviscid 3D-primitive equations, i.e., ν1 = ν2 = 0, blow-up results are known by Wong [36], see also [4], and there are ill-posedness results for Sobolev spaces by Han-Kwan and Nguyen [16]. Local well-posedness has been proven only for analytical data by Kukavica et al. [18]. The primitive equations with partial viscosity are an intermediate model between these well- and ill-posed situations. For more information on previous results on the primitive equations we refer to the works of Washington and Parkinson [35], Pedlosky [28], Majda [26] and Vallis [34]; see also the recent surveys by Li and Titi [21] and by Hieber and the first author [13]. 1.3. Main results and discussion. Our main results are stated in the following. Below, the notions of weak and z-weak solutions are made precise in Definitions 2.1 and 2.2, where function spaces are introduced in Subsection 2.1. Theorem 1.1 (Local solutions for the initial value problem). 2 1 −1 2 1 2 2 (a) Let f L ((0,T ),H (( h,h),H (G)) ) and v0 H (( h,h),L (G)) ∈ − ′ ∈ − with divH v0 =0. Then there exists a time T (0,T ] and a unique z-weak solution to the initial boundary value problem ∈(1.2), (1.3) on (0,T ′), i.e., a weak solution with v L∞((0,T ′),L2(Ω)2), v L2((0,T ′),L2(Ω)2×2), z ∈ ∇H z ∈ and this z-weak solution satisfies v C0([0,T ′],L2(Ω)2). One has T ′ = ∈ T if v0 H1((−h,h),L2(Ω)) and f L2((0,T ),H1((−h,h),H−1(G))) are sufficiently small.k k k k 1 2 (b) If v0 H (Ω) with v0( ,z) ∂G =0 for almost all z ( h,h), divH v0 =0, and f∈ L2((0,T ),H1((· h,h| ),L2(G))2), then there∈ − exists a time T ′ (0,T ] and∈ a unique strong− solution to (1.2), (1.3) on (0,T ′), i.e. a z-weak∈ solution where in addition v L∞((0,T ),L2(Ω)2×2), ∆ v, ∂ v L2((0,T ),L2(Ω)2). ∇H ∈ H t ∈ For v 1 and f 2 1 2 sufficiently small, one has k 0kH k kL ((0,T ),H ((−h,h),L (G))) T ′ = T . 4 HUSSEIN, SAAL, AND WRONA

Remark 1.2. Continuous dependence on the data can be proven as well by adapt- ing the estimates obtained in the proof of Theorem 1.1. Theorem 1.3 (Global solutions for the initial value problem). 1 2 (a) Let v0 Hη (Ω) with v0( ,z) ∂G =0 for almost all z ( h,h), divH v0 =0 and f ∈ 0, where · | ∈ − ≡ 1 H (Ω) := v : v0 1 = v0 1 + ∂zv0 2+η + v0 ∞ < , η { k kHη k kH k k k k ∞} for some η > 0, then there exists a unique global strong solution to (1.2), (1.3) on (0,T ) for any T > 0. Moreover, T 2 2+η 2 sup ( v + ∂zv )+ H v Cη,h,T ( v0 1 ), k∇ k2 k k2+η k∇ ∇ k2 ≤ k kHη t∈[0,T ] Z0 for an increasing function Cη,h,T depending on η, h, T . 1 2 2 2+η 2 (b) Let v0 H (( h,h),L (G) ) with ∂zv0 L (Ω) for η> 0 with divH v0 = 0 and ∈f 0.− Then the unique z-weak∈ solution from Theorem 1.1 (a) ex- tends to≡ a unique global strong solution to (1.2), (1.3) on (0,T ) for any T > 0. Moreover, for any δ (0,T ) ∈ T sup ( v 2 + ∂ v 2+η)+ v 2 C k∇ k2 k z k2+η k∇H ∇ k2 ≤ T,h,δ,v0 t∈[δ,T ] Zδ

for a constant CT,h,δ,v0 , depending on T,h,δ,v0. Note that the regularity of the initial value in Theorem 1.1 (a) is similar to the one obtained for the Navier-Stokes equation with horizontal viscosity, compare [1, Theorem 6.2]. It is also the same condition obtained by Ju for the existence and uniqueness of global z-weak solutions for the primitive equations with full viscosity, see [17]. Theorem 1.1 (b) and Theorem 1.3 (a) correspond to the result by Cao, Li and Titi in [6, Theorem 1.1]. However, they consider a cubical domain with periodic boundary conditions in all three directions. As already pointed out in [30], vertical boundary conditions are not necessary, but they are preserved by the equation. Here, we consider the more physical Dirichlet boundary conditions on the sides and no boundary condition on top and bottom. The proof of the a priori bounds in [6] uses a vanishing vertical viscosity limit. Here, we follow a more direct approach considering the case of horizontal viscosity without such limits. For the global a priori bound, we have been able to adapt the overall strategy of Cao, Li and Titi, but due to the boundary conditions here, controlling the pressure terms becomes more involved. Note also that in [6] the a priori bound is proven for periodic boundary conditions in all three directions, here we do not require any boundary conditions on the top and bottom part of the boundary. Moreover, using that regularity is preserved in the vertical directions while being smoothed in the horizontal directions, we show in Theorem 1.3 (b) that a z-weak solution with slightly more integrability of the initial data regularizes to reach the setting of Theorem 1.3 (a) for t> 0. Thus existence and uniqueness of global strong solutions holds even for a larger class of initial conditions. It is remarkable that the regularity of the initial conditions required in 1.3 (b) is very close to the one obtained by Ju for the case of full viscosity, cf. [17]. PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 5

Furthermore, for the local well-posedness results a force term is included here which allows us to analyze the time-periodic problem. The notion of T -periodic z-weak solutions is explained in Definition 2.3 below. Theorem 1.4 (Time-periodic problem). There exists ε> 0 such that 2 1 −1 2 (a) if f L ((0,T ),H (( h,h),H (G)) ) and f L2((0,T ),H1((−h,h),H−1(G)) < ε, then∈ there exists a− unique z-weak T -periodick k solution v, where v lies in the regularity class given in Theorem 1.1 (a); 2 1 2 2 (b) if f L ((0,T ),H (( h,h),L (G)) ) with f L2((0,T ),H1((−h,h),L2(G))) < ε, then∈ there exists a unique− strong T -periodick solutionk v, i.e., v lies in the regularity class given in Theorem 1.1 (b).

Here, existence of a T -periodic z-weak solution means that there exists a v0 such that there exists a z-weak solution to the initial boundary value problem (1.2), (1.3) with initial condition v0 and force f which is T -periodic. It seems that so far, there has been no result on the time-periodic problem for partial viscosities. For the primitive equations with full viscosity there are several results, see Hsia and Shiue [15] and Tachim Medjo [32], on the existence of unique global strong time-periodic solutions for periodic forces assuming a smallness condition on the force. In contrast, in [10] existence of strong periodic solutions for possibly large periodic forces has been shown, but the solutions are possibly non-unique. Here, we have adapted the strategy by Galdi, Hieber and Kashiwabara in [10] to consider the Poincar´emap for the construction on time-periodic solutions where we take advantage of the a priori estimates obtained for the initial value problem. A crucial ingredient in our proof is that due to the lateral Dirichlet boundary conditions, there is a Poincar´einequality of the type

v 2 C v 2 . k kL (Ω) ≤ k∇H kL (Ω) Note that for parabolic problems there are quite a few results on time-periodic solutions. For instanceLukaszewicz et al. [25] treat the the case certain of semilinear parabolic equations in Hilbert spaces. There are also the maximal Lp-regularity approaches in Banach spaces for time-periodic solutions by Geissert et al. [11], and by Kyed and co-authors, cf. [3,9,20] and the references therein. More concretely, for the Navier-Stokes equations there are also many results on periodic solutions going back to the work of Serrin [31], see e.g. also [27] and references therein for a more recent survey. We would like to emphasis that for the case of partial viscosity considered here the system is not purely parabolic anymore, and in particular since the vertical derivatives in the non-linearity cannot be controlled by the linear part all these approaches are not applicable. Instead one has to extract additional information for these particular equations. 1.4. Organization of the paper. In the subsequent Section 2 basic definitions and notations are introduced. In particular, the spaces of hydrostatic-solenoidal functions and the notions of weak and z-weak solutions to the initial value and time-periodic problem and their regularity properties are discussed. In Section 3 the existence and uniqueness of local z-weak and local strong solutions is proven, respectively. The time-periodic problem is discussed in Section 4 including the proof of Theorem 1.4. In Section 5, global a priori bounds are proven, and the proof of Theorem 1.3 is given. Some auxiliary results are collected in Section 6. 6 HUSSEIN, SAAL, AND WRONA

2. Preliminaries 2.1. Function spaces and notations. By L2(Ω) we denote the standard real Lebesgue space with scalar product

f,g := f(x,y,z)g(x,y,z) d(x,y,z), h iΩ ZΩ 2 where L (G) and f,g are defined analogously. By f 2 and f 2 we h iG k kL (Ω) k kL (G) denote the induced norm dropping the subscripts Ω and G in the notation if there is no ambiguity. For f,g L2((0,T ),L2(Ω)) we write ∈ T f,g := f(t,x,y,z)g(t,x,y,z) d(x,y,z) dt h iΩ,T Z0 ZΩ for the scalar product in space and time. For a function f L∞((0,T ),L∞(Ω)) we use the abbreviation ∈

f ∞ := sup f(t) L∞(Ω) . k k t∈(0,T ) k k For s N the space Hs(Ω) consists of f L2(Ω) such that α f L2(Ω) for α s endowed∈ with the norm ∈ ∇ ∈ | |≤ α 2 1/2 f s = f 2 , k kH (Ω) k∇ kL (Ω) |α|≤s X  and Hs(Ω) := f Hs(Ω): f = 0 . Here we used the multi-index notation 0 { ∈ |∂Ω } α = ∂α1 ∂α2 ∂α3 for α N3. The spaces Hs(G) and Hs(G) are defined analogously, ∇ x y z ∈ 0 0 and we will again just write f Hs if there is no ambiguity. For non-integer s 0, the spaces Hs are defined byk complexk interpolation, and one sets by duality H−≥s = s ′ (H0 ) , compare also [33, Chapter 3]. Moreover, we set (2.1) H1 (Ω) := f H1(Ω): f =0 0,l { ∈ |Γl } −1 1 ′ and Hl (Ω) := H0,l(Ω) is its dual space. Analogous definitions hold for Sobolev spaces Hs,p for p [1, ], where Hs = Hs,2, and for Sobolev spaces of functions with values in Banach∈ spaces∞ such as H1((0,T ),Lp(Ω)) and Hs,q(( h,h),Lp(G)). Sometimes we use the short hand notation Hs,pHr,q for Hs,p(( h,h−),Hr,q(G)). z xy − 2.2. Hydrostatic-solenoidal vector fields. Now let us reformulate the primitive equations (1.2) and (1.3). The divergence free condition ∂zw + divH v = 0 and the boundary condition w(z = h) = 0 are equivalent to ± z h (2.2) w(t,x,y,z)= div v(t,x,y,ξ) dξ and div v(t,x,y,ξ) dξ =0 − H H Z−h Z−h for v sufficiently smooth, e.g., div v L1(Ω). This means, that v – the mean value H ∈ of v in the vertical direction – is divergence free, i.e., divH v = 0, where the vertical average and its complement are 1 h (2.3) v(t,x,y) := v(t,x,y,ξ) dξ andv ˜ := v v. 2h − Z−h Hence one identifies a suitable hydrostatic-solenoidal space as

||·|| 2 L2 (Ω) := v C∞(Ω)2 : div v =0 L , σ { ∈ c H } PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 7

∞ where Cc stands for smooth compactly supported functions. Note that this space admits the decomposition (2.4) L2 (Ω) = v L2(Ω)2 : v =0 L2 (G), σ { ∈ }⊕ σ and the hydrostatic Helmholtz projection thereon is (2.5) P: L2(Ω)2 L2 (Ω), Pv =v ˜ + P v, → σ G ||·|| where L2 (G) = v C∞(G)2 : div v =0 L2 is the space of solenoidal vector σ { ∈ c H } fields over G, and PG the corresponding (classical) Helmholtz projection. More precisely, since due to the product structure L2(Ω) = L2(G) L2( h,h), one ob- tains by applying P that ⊗ − (2.6) L2 (Ω) = L2(G)2 L2( h,h) L2 (G) span 1 , σ ⊗ 0 − ⊕ σ ⊗ { } where L2( h,h) = v L2( h,h): h v(z)dz = 0 and 1 L2( h,h) is a 0 − { ∈ − −h } ∈ − constant function. R 2.3. Weak and z-weak solutions. Next we give a precise notion of weak solu- tions. Definition 2.1 (Weak solution). Let f L2((0,T ),L2(( h,h),H−1(G))2) and v L2 (Ω). A function v is called a weak∈ solution of the primitive− equations (1.2) 0 ∈ σ with boundary conditions (1.3) on (0,T ) with initial condition v0 and force f if (i) One has that v : [0,T ] L2 (Ω) is weakly continuous with v(0) = v and → σ 0 v L∞((0,T ),L2 (Ω)), v L2((0,T ),L2(Ω)2×2) ∈ σ ∇H ∈ 1 2 with v(t, z, , ) H0 (G) almost everywhere for z ( h,h); (ii) For some constant· · ∈ c> 0 it satisfies ∈ −

v ∞ 2 + v 2 2 k kL ((0,T ),L (Ω)) k∇H kL ((0,T ),L (Ω))

c v 2 + f 2 −1 ; ≤ k 0kL (Ω) k kL ((0,T ),H (G))   (iii) v satisfies (1.2) and (1.3) in the weak sense, i.e.,

v, ϕ + v(T ), ϕ(T ) v , ϕ(0) + v, ϕ − h tiΩ,T h iΩ − h 0 iΩ h∇H ∇H iΩ,T + v v, ϕ wv,ϕ w v, ϕ = f, ϕ , h · ∇H iΩ,T − h ziΩ,T − h z iΩ,T h iΩ,T where w = w(v) is given by (2.2), holds for any

ϕ H1,1((0,T ),L2 (Ω)) C0([0,T ],L2 (Ω)) ∈ σ ∩ σ L2((0,T ),L2(( h,h),H1(G))2 H1(( h,h),L∞(G))2). ∩ − 0 ∩ − Note that there are different notions of weak solutions for the primitive equations, compare [10] or [32]. The notion of z-weak solutions for the primitive equations has been introduced by Bresch et al. [2] as solutions for the 2D-case. It plays also an important role in the study of the 3D-case with full viscosity, see [17] and the references therein. This is adapted here to the case of only horizontal viscosity. Definition 2.2 (z-weak solution). Let f L2((0,T ),H1(( h,h),H−1(G))2) and v L2 (Ω) with ∂ v L2(Ω)2. A weak solution∈ v of the primitive− equations (1.2) 0 ∈ σ z 0 ∈ 8 HUSSEIN, SAAL, AND WRONA with boundary conditions (1.3) on (0,T ) with initial condition v0 and force f is called a z-weak solution if additionally

v L∞((0,T ),L2(Ω)2) and v L2((0,T ),L2(Ω)2×2). z ∈ ∇H z ∈

Definition 2.3 (T -periodic z-weak solutions). A z-weak solution v is called T - periodic if v(0) = v(T ).

Remark 2.4. (a) z-weak solutions are additionally in C0([0,T ],L2(Ω)), in this sense v(0) = v(T ) has to be understood in Definition 2.3. (b) For a weak solution one has for the non-linear terms w v, wz v, v H v L2((0,T ),L1(Ω)2), and this guarantees that each term· in the· weak· ∇ formu-∈ lation is well-defined. A z-weak solution is regular enough to assure that even w v L2((0,T ),L1(Ω)2) and therefore wv,ϕ w v, ϕ = · z ∈ − h z iΩ,T −h z iΩ,T wv , ϕ for test functions ϕ as in Definition 2.1. h z iΩ,T

2.4. Regularity of z-weak solutions. In the following proposition we show that for z-weak solutions the class of admissible test functions for which especially the nonlinear terms are well-defined is much larger than for weak solutions. This turns out to be useful when testing a z-weak solution with itself, and in particular when proving the uniqueness of z-weak solutions.

Proposition 2.5 (Class of test functions for z-weak solutions). For 2 1 −1 2 2 2 f L ((0,T ),H (( h,h),H (G)) ), v0 Lσ(Ω) with ∂zv0 L (Ω) let v be a z-weak∈ solution to (1.2)− , (1.3) on (0,T ). Then∈ ∈

v, ϕ , v v, ϕ , w∂ v, ϕ , and f, ϕ h∇H ∇H iΩ,T h · ∇H iΩ,T h z iΩ,T h iΩ,T are well-defined for all

ϕ L∞((0,T ),L2 (Ω)) L2((0,T ),L2(( h,h),H1(G))2). ∈ σ ∩ − 0

(n) (n) Proof. Let (ϕ )n be a sequence of smooth functions such that ϕ ϕ for n in L∞((0,T ),L2 (Ω)) L2((0,T ),L2(( h,h),H1(G))2). Then we have→ → ∞ σ ∩ − 0

(n) (n) H v, H ϕ H v, H ϕ Ω,T and f, ϕ f, ϕ Ω,T ∇ ∇ Ω,T → h∇ ∇ i Ω,T → h i D E D E for n . → ∞ (n) (n) (n) The nonlinear terms wv,ϕz + ∂zwv,ϕ = w∂zv, ϕ and Ω,T Ω,T − Ω,T (n) v H v, ϕ haveD to be toE handled with more care. Using Lemma 6.2 a) · ∇ Ω,T

PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 9 with f = v(t), g = ϕ(n)(t) and h = v(t) we obtain ∇H T (n) v(t) H v(t), ϕ (t) dt 0 · ∇ Ω Z D E T 1/2 1/2 1/2 1/2 (n) (n) c H v(t) L2(Ω) v(t) L2(Ω) H ϕ (t) ϕ (t) ≤ k∇ k k k ∇ L2(Ω) L2(Ω) Z0 1/2 1/2 ∂ v(t) v (t) + v( t) 2 dt · k z∇H kL2(Ω)k∇H kL2(Ω) k∇ H kL (Ω)  1/2  1/2 (n) c v ∞ ϕ L ((0,T ),L2(Ω)) ∞ ≤ k k L ((0,T ),L2(Ω)) T 1/2 1/2 (n) 1 2 H v(t) L2(Ω) H ϕ (t) H v(t) H ((−h,h),L (G)) dt · k∇ k ∇ L2(Ω) k∇ k Z0 1/2 1 /2 c v ∞ v v 2 1 2 ≤ k kL ((0,T ),L2(Ω)) k∇H kL 2((0,T ),L2(Ω)) k∇H kL ((0,T ),H ((−h,h),L (G))) 1/2 1/2 (n) (n) H ϕ ϕ , · ∇ L2((0,T ),L2(Ω)) L∞((0,T ),L2(Ω))

(n) and analogously we get by using Lemma 6.2 a) with f = ∂zv(t), g = ϕ (t) and h = w(t) T (n) w(t)∂z v(t), ϕ (t) dt Ω Z0 D 1/2 E 1/2 c ∂ v ∞ ∂ v w 2 1 2 ≤ k z kL ((0,T ),L2(Ω)) k∇H z kL2((0,T ),L2(Ω)) k kL ((0,T ),H ((−h,h),L (G))) 1/2 1/2 (n) (n) H ϕ ϕ . · ∇ L2((0,T ),L2(Ω)) L∞((0,T ),L2(Ω))

Considering these estimates for ϕ(n) ϕ( m) gives the convergence − (n) v(t) H v(t), ϕ (t) v(t) H v(t), ϕ(t) Ω · ∇ Ω → h · ∇ i and D E (n) w(t)∂z v(t), ϕ (t) w(t)∂zv(t), ϕ(t) Ω Ω → h i in L1((0,T )). D E  Moreover, z-weak solutions preserve certain Lq-regularity vertically reflecting the transport-like behavior in this direction. Proposition 2.6 (Lq-norm of v remains bounded for ∂ v Lq.). Let v be z z 0 ∈ a z-weak solution to (1.2), (1.3) on (0,T ), T > 0, with initial condition v0 1 2 2 q ∈2 H (( h,h),L (Ω) ) with divH v0 =0 and force f 0. If in addition ∂zv0 L (Ω) for q>− 2, then v L∞((0,T ),Lq(Ω))2. ≡ ∈ z ∈ Proof. We multiply the equation for ∂zv, ∂ v ∆ ∂ v + ∂ v v + v ∂ v div v∂ v + w∂ v =0, z t − H z z · ∇H · ∇H z − H z zz by ∂ v q−2∂ v and get | z | z 1 d q (q−2)/2 2 ∂ v q + (q 1) ∂ v ∂ v 2 q dt k z kL − k| z | ∇H z kL = ∂ v v + v ∂ v div v∂ v + w∂ v, ∂ v q−2∂ v − z · ∇H · ∇H z − H z zz | z | z Ω = ∂ v v, ∂ v q−2∂ v + div v∂ v, ∂ v q−2∂ v . − z · ∇H | z | z Ω H z | z | z Ω

10 HUSSEIN,SAAL,ANDWRONA

Using Lemma 6.2 a) with f = ∂ v q/2 = g and h = v, we obtain | z | ∇H q−2 q/2 q/2 ∂zv v, ∂zv ∂zv c H ∂zv ∂zv H v H1L2 | · ∇ | | Ω |≤ ∇ | | L2 | | L2 k∇ k z xy (q−2)/2 q/2 c ∂zv H ∂ zv ∂ zv Lq H v H1L2 ≤ | | ∇ L2 k k k∇ k z xy

1 (q−2)/2 2 q ∂zv H ∂zv + c H v H1L2 ∂zv Lq ≤ 2 | | ∇ L2 k∇ k z xy k k and

q−2 1 (q−2)/2 2 q divH v∂zv, ∂zv ∂zv ∂zv H ∂zv + c H v H1L2 ∂zv Lq . | | | Ω |≤ 2 | | ∇ L2 k∇ k z xy k k

Thus,

1 d q (q−2)/2 2 2 q ∂zv q + (q 2) ∂zv H ∂zv 2 c H v 1 2 ∂zv q q dt k kL − k| | ∇ kL ≤ k∇ kHz Lxy k kL and this implies

2 k∇H vk 2 1 2 q q L ((0,T ),Hz Lxy) ∂ v(t) q ∂ v q e , k z kL ≤k z 0kL so ∂ v L∞((0,T ),Lq(Ω)).  z ∈ 3. Local solutions with force Theorem 1.1 (a) and (b) correspond to Proposition 3.1 and 3.2, respectively.

3.1. Local z-weak solutions. We work in the spaces H := v L2(Ω)2 ∂ v L2(Ω)2 and H := H L2 (Ω) { ∈ | z ∈ } σ ∩ σ equipped with the scalar product u, v := u, v + ∂ u, ∂ v and h iH h iΩ h z z iΩ V := v H ∂ v, ∂ v H H1 (Ω)2 and V := V L2 (Ω) { ∈ | x y ∈ } ∩ 0,l σ ∩ σ with the scalar product u, v := u, v + u, v . Note that H = h iV h iH h∇H ∇H iH H1(( h,h),L2(G))2 and V = H1(( h,h),H1(G))2. By V ′ we denote the space − − 0 V ′ = H1(( h,h),H−1(G))2. − We also denote the dual pairing in V V ′ by , to keep the notation simple. × h· ·iH Proposition 3.1 (Existence and uniqueness of local z-weak solutions). Let v0 Hσ and f L2((0,T ), V ′), then there exists a T ′ (0,T ] such that there is a unique∈ z-weak∈ solution of the primitive equations on (0∈,T ′) with v C0([0,T ′],L2(Ω)). If ′ ∈ v and f 2 ′ are sufficiently small, then T = T . || 0||H || ||L ((0,T ),V ) Proof. We subdivide the proof of the existence and uniqueness into several steps. Step 1 (Galerkin approximation). To define a suitable basis for a Galerkin ∞ 2 scheme, one can take advantage of (2.6). To this end, let (ϕm)m C (G) 1 2 ⊂ ∩ H0 (G) be an orthonormal basis of eigenfunctions to the eigenvalues (µm)m of the Dirichlet Laplacian in L2(G)2, and (˜ϕ ) C∞(G)2 H1(G)2 L2 (G) an or- m m ⊂ ∩ 0 ∩ σ thonormal basis of eigenfunctions to the eigenvalues (˜µm)m of the Stokes operator in L2 (G). Moreover, cos kπ (z + h) for k N defines a basis of eigenfunctions to σ 2h ∈ 0 the Neumann Laplacian on L2( h,h) and by the first representation theorem even  a basis on H1( h,h). − − PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 11

Hence, we define for m N, k N the functions Φ C∞(Ω) by ∈ ∈ 0 m,k ∈ 1 kπ (3.1) Φ (x,y,z) := ϕ (x, y)cos (z + h) for k> 0, m,k h m 2h   1 (3.2) Φ (x,y,z) := ϕ˜ (x, y). m,0 2h m Then span Φ m N, k N is dense in H , in particular div Φ = 0, { m,k| ∈ ∈ 0} σ H m,k because for k> 0 we have already Φm,k = 0 and divH Φm,0 = divH ϕ˜m = 0. We set (3.3) H := span Φ m, k n and P : H H σ,n { m,k| ≤ } n σ → σ,n to be the orthogonal projection onto it. We project the primitive equations onto the finite dimensional space Hσ,n and we are looking for a solution v = (v , v ): [0,T ] H n 1,n 2,n → σ,n of the system of ordinary differential equations

(3.4) d v , Φ + v v , Φ + w ∂ v , Φ ∆ v , Φ dt h n m,kiH h n · ∇H n m,kiH h n z n m,kiH − h H n m,kiH = f, Φ h m,kiH for m, k n, where we already used H p, Φm,k Ω = 0, with initial condition ≤ z h∇ i v (0) = P v and w (t,x,y,z)= div v (t,x,y,s) ds. Note that the proper- n n 0 n − −h H n ties of Φ imply div v = 0 and thus we have w (z = h) = 0. Now, we can m,k H n R n represent ±

v (t)= g(mk)(t)Φ for some g(mk) : [0,T ] R. n n m,k n → m,kX≤n 1 The existence of a solution vn H ((0,T ),Hσ,n) follows from classical ODE theory. 2 ∈ 2 Step 2 (L -estimate). Next, we prove an estimate for vn in Lσ(Ω). Integrating by parts, for k> 0 it holds for any function g V ′ that ∈ k2π2 g, Φ = g, Φ + g, ∂ Φ = 1+ g, Φ , h m,kiH h m,kiΩ h − zz m,kiΩ 4h2 h m,kiΩ   and for k = 0 the corresponding equality holds, because ∂zΦm,0 = 0. Thus (3.4) yields

(3.5) ∂ v , Φ + v v , Φ + w ∂ v , Φ ∆ v , Φ h t n m,kiΩ h n · ∇H n m,kiΩ h n z n m,kiΩ − h H n m,kiΩ = f, Φ . h m,kiΩ (mk) We multiply this equation by gn and sum over m, k n. It then follows that ≤ d 2 2 v 2 +2 v 2 =2 f, v 2 v v , v 2 w ∂ v , v dt k nkL (Ω) k∇H nkL (Ω) h niΩ − h n · ∇H n niΩ − h n z n niΩ =2 f, v h niΩ 1 2 2 f 2 −1 + ε v 2 1 , ≤ ε k kL ((−h,h),H (G)) k nkL ((−h,h),H (G)) for ε > 0, where we used wn( h) = 0 when integrating by parts in the vertical direction to show that v ±v , v + w ∂ v , v = 0. Thus, for ε small h n · ∇H n niΩ h n z n niΩ 12 HUSSEIN,SAAL,ANDWRONA enough, using the Poincar´einequality of Lemma 6.1, there exists a constant C > 0 independent of vn such that

(3.6) v ∞ 2 + v 2 2 k nkL ((0,T ),L (Ω)) k∇H nkL ((0,T ),L (Ω)) C v 2 + C f 2 2 −1 . ≤ k 0kL (Ω) k kL ((0,T ),L ((−h,h),H (G))) Step 3 (H-estimate). To derive now an estimate for vn in H we multiply (3.4) (mk) by gn and sum over m, k n. This gives ≤ 1 d 2 1 d 2 2 2 v 2 + ∂ v 2 + v 2 + ∂ v 2 2 dt k nkL (Ω) 2 dt k z nkL (Ω) k∇H nkL (Ω) k z∇H nkL (Ω) = f, v + ∂ f, ∂ v ∂ (v v ), ∂ v ∂ (w ∂ v ), ∂ v h niΩ h z z niΩ − h z n · ∇H n z niΩ − h z n z n z niΩ = f, v + ∂ f, ∂ v (∂ v ) v , ∂ v + div v ∂ v , ∂ v . h niΩ h z z niΩ − h z n · ∇H n z niΩ h H n z n z niΩ Here one has used the cancellation property for the non-linear term with respect to , Ω. Note that with respect to , H one does not have such cancellation in general.h· ·i Now, Lemma 6.2 a) yields withh· ·if = g = ∂ v and h = v that z n ∇H n (∂ v ) v , ∂ v | h z n · ∇H n z niΩ | 1/2 1/2 c ∂ v v ∂ v 2 ∂ v 2 ≤ k z∇H nkL2(Ω)k∇H nkL2(Ω)k z∇H nkL (Ω)k z nkL (Ω) + c ∂ v 2 v 2 ∂ v 2 k z∇H nkL (Ω)k∇H nkL (Ω)k z nkL (Ω) 2 −1 2 2 4 ε ∂ v 2 + cε v 2 ∂ v 2 + ∂ v 2 ≤ k z∇H nkL (Ω) k∇H nkL (Ω) k z nkL (Ω) k z nkL (Ω)   for any ε > 0. An analogous estimate holds for the term divH vn∂zvn, ∂zvn Ω . 1 | h i | So, we obtain by choosing ε = 4 (3.7) d 2 2 2 1 2 2 2 4 v + v f ′ + v + c v 2 v + v dt k nkH k∇H nkH ≤k kV 2 k nkV k∇H nkL (Ω) k nkH k nkH and integrating with respect to time gives for t> 0  t t 2 1 2 2 2 1 2 vn(t) H + H vn(s) H ds v0 H + f L2((0,t),V ′) + vn(s) H ds k k 2 0 k∇ k ≤k k k k 2 0 k k Z t Z 2 2 4 + c v (s) 2 v (s) + v (s) ds. k∇H n kL (Ω) k n kH k n kH Z0 Using Poincar´e’s inequality, see Lemma 6.1, leads to a situation where a non-linear version of Gr¨onwall’s Lemma – recapped here in Lemma 6.3 – is applicable, i.e., t 2 2 2 2 (3.8) v (t) v + f 2 ′ + c Ψ(s)ω( v (s) ) ds k n kH ≤k 0kH k kL ((0,t),V ) k n kH Z0 2 2 where ω(s)=1+ s + s , Ψ(s) := v (s) 2 , and k∇H n kL (Ω) u ds 2 1+2u Φ(u)= = arctan + c0, ω(s) √3 √3 Z0   t 2 where c0 is such that Φ(0) = 0. Due to the boundedness of 0 H vn(s) L2(Ω) ds 2 k∇ k by the L -estimate in Step 2, Lemma 6.3 implies R t 2 −1 2 2 2 v (t) Φ Φ( v + f 2 ′ )+ v (s) 2 ds . k n kH ≤ k 0kH k kL ((0,t),V ) k∇H n kL (Ω)  Z0  2 2 t 2 This is well-defined provided that Φ( v + f 2 ′ )+ v (s) 2 ds k 0kH k kL ((0,t),V ) 0 k∇H n kL (Ω) lies in the range of Φ which can be assured for small times 0 < t T ′, where R ≤ PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 13

T ′ (0,T ] is sufficiently small. Using the monotonicity of Φ which implys the one of Φ∈−1 and the energy inequality (3.6) one even obtains that

2 −1 2 2 2 2 v (t) Φ Φ( v + f 2 ′ )+ v 2 + c f 2 ′ . k n kH ≤ k 0kH k kL ((0,t),V ) k 0kL (Ω) k kL ((0,t),V )  ′ 2 2  Hence for small times T (0,T ] or for data v + f ′ being suffi- ∈ k 0kL2(Ω) k kL2((0,t),V ) ciently small when T ′ = T , one obtains using the continuity of Φ−1 that for some c> 0 t 2 2 2 2 ′ (3.9) v (t) + v (s) ds c( v + f 2 ′ ), t (0,T ]. k n kH k∇H n kH ≤ k 0kH k kL ((0,t),V ) ∈ Z0 Step 4 (Convergence). On this interval [0,T ′] we can deduce the weak conver- 2 gence of a subsequence of (vn)n in L ((0,T ),Hσ) (which we do not rename) to 2 some limit v L ((0,T ),Hσ). The energy estimate (3.9) for the sequence gives 2 ∈ that v ∞ and v 2 remain bounded, and hence v is in the k kL ((0,T ),H) k∇H kL ((0,T ),H) regularity class of weak and z-weak solutions. To show that the limit is in fact a weak solution, one takes into account that 2 2 especially the full gradient of vn is uniformly bounded in L ((0,T ),L (Ω)) and from the compact embedding in the Rellich-Kondrachov theorem the strong convergence 2 2 1 1 of (vn)n in L ((0,T ),L (Ω)) follows. This implies that in L ((0,T ),L (Ω)) v v ⇀ v v, w v ⇀ wv and ∂ w v ⇀ ∂ wv. n · ∇H n · ∇H n n z n n z Let now ϕ C1([0,T ],H V ) be of the form ϕ (t)= h (t)Φ with n ∈ n ∩ n m,k≤n m,k m,k h C1([0,T ], R). From w(z = h) we conclude m,k ∈ ± P w ∂ v , ϕ = ∂ w v , ϕ w v , ∂ ϕ , h n z n niΩ − h z n n niΩ − h n n z niΩ and using (3.5) we get

T v , ∂ ϕ + v v , ϕ ∂ w v , ϕ w v , ∂ ϕ − h n t niΩ h n · ∇H n niΩ − h z n n niΩ − h n n z niΩ Z0 T + v , ϕ dt + v (T ), ϕ (T ) v (0), ϕ (0) = f, ϕ dt h∇H n ∇H niΩ h n n iΩ − h n n iΩ h niΩ Z0 and passing to the limit n gives → ∞ v, ∂ ϕ + v v, ϕ ∂ wv,ϕ wv, ∂ ϕ − h t iΩ,T h · ∇H iΩ,T − h z iΩ,T − h z iΩ,T + v, ϕ + v(T ), ϕ(T ) v(0), ϕ(0) = f, ϕ . h∇H ∇H iΩ,T h iΩ − h iΩ h iΩ,T

Showing that v(0) = v0 and v(T ) are well-defined follows from the next step which only uses the Galerkin approximation and the convergence. 1 Step 5 (Continuity in time). For l>n we extend vn H ((0,T ),Hσ,n) to 1 0 2 (mk) ∈ H ((0,T ),H ) C ([0,T ],L (Ω)) by setting gn = 0 if m l or k l. For σ,l ⊂ ≥ ≥ n

∂ u, Φ + v v v v , Φ + w ∂ v w ∂ v , Φ h t m,kiH h n · ∇H n − l · ∇H l m,kiH h n z n − l z l m,kiH ∆ u, Φ = f, Φ . − h H m,kiH h m,kiH 14 HUSSEIN,SAAL,ANDWRONA

Hence, 1 d 2 2 u 2 + u 2 2 dt k kL (Ω) k∇H kL (Ω) = v v v v ,u + w ∂ v w ∂ v ,u + f,u h l · ∇H l − n · ∇H n iΩ h l z l − n z n iΩ h iΩ = (w w )∂ v ,u + w ∂ (v v ),u h l − n z n iΩ h l z l − n iΩ + (v v ) v ,u + v (v v ),u + f,u h l − n · ∇H n iΩ h l · ∇H l − n iΩ h iΩ = w ∂ v ,u + u v ,u + f,u . h u z n iΩ h · ∇H n iΩ h iΩ As in the proof of Proposition 2.5 it follows from Lemma 6.2 a) with f = u, g = ∂zvn and h = wu that w ∂ v ,u c u 1/2 u 1/2 ∂ v 1/2 ∂ v 1/2 | h u z n iΩ |≤ k∇H kL2(Ω) k kL2(Ω) k∇H z nkL2(Ω) k z nkL2(Ω) 1/2 1/2 ∂ w w + w 2 · k z ukL2(Ω)k ukL2(Ω) k ukL (Ω) 1/2 1/2  1/2 1/2  c u u ∂ v ∂ v u 2 ≤ k∇H kL2(Ω) k kL2(Ω) k∇H z nkL2(Ω) k z nkL2(Ω) k∇H kL (Ω) 1 2 2 2 2 u 2 + c v v u 2 ≤ 4 k∇H kL (Ω) k∇H nkH k nkH k kL (Ω) and similarly from Lemma 6.2 a) with f = g = u and h = v that ∇H n u v ,u c u 2 u 2 | h · ∇H n iΩ |≤ k∇H kL (Ω) k kL (Ω) 1/2 1/2 ∂ v v + v 2 · k z∇H nkL2(Ω)k∇H nkL2(Ω) k∇H nkL (Ω) 1 2  2 2  u 2 + c v u 2 . ≤ 4 k∇H kL (Ω) k∇H nkH k kL (Ω) Altogether, we have

t t 2 2 u(t) L2(Ω) + H u(s) L2(Ω) ds f(s),u(s) Ω ds k k 0 k∇ k ≤ 0 h i Z t Z 2 2 2 2 + c 1+ H vn(s) H + H vn(s) H vn(s) H u(s) L2(Ω) ds, 0 k∇ k k∇ k k k k k Z   and because of t 2 2 2 1+ H vn(s) H + H vn(s) H vn(s) H ds 0 k∇ k k∇ k k k Z 2 2 2 C(T + v 2 + v ∞ v 2 ) < ≤ k∇H nkL ((0,T ),H) k nkL ((0,T ),H) k∇H nkL ((0,T ),H) ∞ uniformly in n, we get T 2 2 u(t) 2 c u(0) 2 + f(s),u(s) ds . k kL (Ω) ≤ k kL (Ω) |h iΩ| Z0 ! The weak- -convergence of v v in L2((0,T ), V ) yields T f(s),u(s) ds 0 ∗ n → 0 |h iΩ| → (n,l ) and with u(0) 2 0 (n,l ) follows → ∞ k kL (Ω) → → ∞ R T 2 2 sup u(t) 2 c u(0) 2 + f(s),u(s) ds 0 k kL (Ω) ≤ k kL (Ω) |h iΩ| → t∈[0,T ] Z0 ! for n,l . Thus v C0([0,T ],L2(Ω)). Step→ 6 ∞(Uniqueness).∈ In this step the estimates on the non-linear terms are similar to the above. Let v1, v2 be two z-weak solutions to the same initial datum PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 15 v0 Hσ. Set u := v1 v2 and wu := w1 w2 for w1 = w(v1) and w2 = w(v2). Then∈ we have u(t = 0)− = 0 in L2(Ω) and − ∂ u ∆ u + P(v v v v + w ∂ v w ∂ v )=0. t − H 1 · ∇H 1 − 2 · ∇H 2 1 z 1 − 2 z 2 ∞ 2 2 2 1 2 We have u L ((0,T ),Lσ(Ω)) L ((0,T ),L (( h,h),H0 (G)) ). Hence, Proposi- tion 2.5 yields∈ that we can test∩ the above equation− with u, and similar to Step 5

1 d 2 2 u 2 + u 2 = w ∂ v ,u + u v ,u , 2 dt k kL (Ω) k∇H kL (Ω) h u z 1 iΩ h · ∇H 1 iΩ where

1 2 2 2 2 w ∂ v ,u u 2 + v v u 2 and | h u z 1 iΩ |≤ 4 k∇H kL (Ω) k∇H 1kH k 1kH k kL (Ω) 1 2 2 2 u v ,u u 2 + c v u 2 . | h · ∇H 1 iΩ |≤ 4 k∇H kL (Ω) k∇H 1kH k kL (Ω) It follows

t 2 2 (3.10) u(t) L2(Ω) + H u(s) L2(Ω) ds k k 0 k∇ k t Z 2 2 2 2 2 H v1(s) H v1(s) H + H v1(s) H v1(s) H u(s) L2(Ω) ds, ≤ 0 k∇ k k k k∇ k k k k k Z   and we can apply Gr¨onwall’s inequality to obtain u = 0. 

3.2. Local strong solutions. Proposition 3.2 (Local strong well-posedness). Let v H1 (Ω)2 L2 (Ω) and 0 ∈ 0,l ∩ σ f L2((0,T ),H). Then the local z-weak solution on (0,T ′) for T ′ (0,T ] given by∈ Proposition 3.1 has the additional regularity ∈ v L∞((0,T ′),L2(Ω)), ∆ v, ∂ v L2((0,T ′),L2(Ω)). ∇H ∈ H t ∈ Proof. Recall that we have obtained in the proof of Proposition 3.1 a solution vn = 1 (v1n, v2n) H ((0,T ),Hn,σ) of the system (3.4) of ordinary differential equations ∈ 2 with vn(0) = Pnv0. These satisfy the L -estimate (3.6) on [0,T ] and the H-estimate (3.9) on some small time interval [0,T ′] for T ′ (0,T ], where T = T ′ if the data are sufficiently small. ∈ Step 1 (L2-L2 -estimate on ∆ v and L∞-L2 -estimate on v). Here we use the t x H t x ∇H higher regularity of the initial data and the fact, that the functionsΦm,k defined in (3.1) and (3.2) are a basis of eigenfunctions to certain operators. Multiplication of (3.5) first with the eigenvalue µm of ϕm or respectivelyµ ˜m ofϕ ˜m and second with (mk) gn , and then summing over both m and k gives

∂ v , ∆ v + v v , ∆ v + w ∂ v , ∆ v + ∆ v , ∆ v h t n − H niΩ h n · ∇H n − H niΩ h n z n − H niΩ h H n H niΩ = f, ∆ v , h − H niΩ and it follows that

d 2 2 v 2 + ∆ v 2 dt k∇H nkL (Ω) k H nkL (Ω) 2 2 v v , ∆ v +2 w ∂ v , ∆ v + f 2 . ≤ h n · ∇H n H niΩ h n z n H niΩ k kL (Ω) 16 HUSSEIN,SAAL,ANDWRONA

Using Lemma 6.2 b) with f = ∆ v , g = v and h = v we get H n n ∇H n v v , ∆ v | h n · ∇H n H niΩ | 2 1/2 1/2 c ∆ v 2 v v ≤ k H nkL (Ω) ∇H n L2(Ω) k∇H nkL2(Ω) 1/2 1/2 v ( v 2 + v 2 ) ·k nkH k n kL (Ω) k∇H nkL (Ω) 3/2 1/2 1/2 c ∆ v ( v v + v 2 v ) ≤ k H nkL2Ω) k∇H nkL2(Ω) k nkH k∇H nkL (Ω) k nkH 1 2 4 2 2 2 ∆ v 2 + c( v + v 2 v ) v 2 , ≤ 8 k H nkL (Ω) k nkH k∇H nkL (Ω) k nkH k∇H nkL (Ω) and by Lemma 6.2 b) with f = ∆H vn, g = wn and h = ∂zvn

w ∂ v , ∆ v | h n z n H niΩ | 1/2 1/2 c ∆ v 2 ∂ v ∂ v ≤ k H nkL (Ω) k∇H z nkL2(Ω) k z nkL2(Ω) 1/2 1/2 ( w 2 + ∂ w 2 ) ( w 2 + w 2 ) · k nkL (Ω) k z nkL (Ω) k nkL (Ω) k∇H nkL (Ω) 1/2 1/2 1/2 c ∆ v 2 ∂ v ∂ v v ≤ k H nkL (Ω) k∇H z nkL2(Ω) k z nkL2(Ω) k∇H nkL2(Ω) 2 1/2 ( v 2 + v ) · k∇H nkL (Ω) ∇H n L2(Ω) c v 1/2 v 1/2 ≤ k nkV k nkH 1/2 3/2 ∆ v 2 v 2 + v ∆ v · k H nkL (Ω) k∇H nkL (Ω) k∇H nkL2(Ω) k H nkL2(Ω) 1  2 2 2 2 ∆ v 2 + c v 2 v v + v v . ≤ 8 k H nkL (Ω) k∇H nkL (Ω) k nkV k nkH k nkV k nkH   So, for some c> 0 we have the estimate

d 2 1 2 2 (3.11) v 2 + ∆ v 2 f 2 dt k∇H nkL (Ω) 2 k H nkL (Ω) ≤k kL (Ω) 2 2 4 2 2 2 + c v v + v v + v + v 2 v v 2 . k nkV k nkH k nkV k nkH k nkH k∇H nkL k nkH k∇H nkL   2 Our previous estimate (3.9) shows, that the pre-factor of H vn L2 on the right hand side has a bounded time integral for small times ork∇ smallk data and thus Gr¨onwall’s lemma implies that

2 2 (3.12) v ∞ 2 + ∆ v 2 2 k∇H nkL ((0,T ),L (Ω)) k H nkL ((0,T ),L (Ω)) 2 2 c v 2 + f 2 2 . ≤ k∇H 0kL (Ω) k kL ((0,T ),L (Ω))   2 2 Step 2 (Lt -Lx-estimate on ∂tv). To obtain a better regularity in time we multiply (mk) (3.5) with ∂tgn and sum over m and k. It follows that

∂ v , ∂ v + v v , ∂ v + w ∂ v , ∂ v ∆ v , ∂ v h t n t niΩ h n · ∇H n t niΩ h n z n t niΩ − h H n t niΩ = f, ∂ v . h t niΩ PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 17

Similar to the above we get by Lemma 6.2 b) with f = ∂tvn, g = vn and h = H vn that ∇ 2 1/2 1/2 v v , ∂ v c ∂ v 2 v v | h n · ∇H n t niΩ |≤ k t nkL (Ω) ∇H n L2(Ω) k∇H nkL2(Ω) 1/2 1/2 v ( v 2 + v 2 ) ·k nkH k nkL (Ω) k∇H nkL (Ω) 1/2 3/2 c ∂ v 2 ∆ v v ≤ k t nkL (Ω) k H nkL2(Ω) k nkH1 (Ω) 1 2 3 ∂ v 2 + c ∆ v 2 v 1 , ≤ 8 k t nkL (Ω) k H nkL (Ω) k nkH (Ω) where we used that due to the ellipticity of the Laplacian the second derivatives 2v can be bounded in L2(G) and hence also in L2(Ω) by the horizontal Lapla- ∇H n cian ∆H vn. By Lemma 6.2 b) with f = ∂tvn, g = wn and h = ∂zvn we obtain that 1/2 1/2 1/2 w ∂ v , ∂ v c ∂ v 2 ∂ v ∂ v v | h n z n t niΩ |≤ k t nkL (Ω) k∇H z nkL2(Ω) k z nkL2(Ω) k∇H nkL2(Ω) 1/2 ( v 2 + ∆ v 2 ) · k∇H nkL (Ω) k H nkL (Ω) 1/2 c ∂ v 2 v v 1 ≤ k t nkL (Ω) k nkV k nkH (Ω) 1/2 ( v 1 + ∆ v 2 ) · k nkH (Ω) k H nkL (Ω) 1 2 ∂ v 2 ≤ 8 k t nkL (Ω) 2 + c v v 1 ( v 1 + ∆ v 2 ) k nkV k nkH (Ω) k nkH (Ω) k H nkL (Ω) which leads to

d 2 2 2 v 2 + ∂ v 2 2 f 2 dt k∇H nkL (Ω) k t nkL (Ω) ≤ k kL (Ω) 2 3 + c v v 1 ( v 1 + ∆ v 2 )+ ∆ v 2 v 1 . k nkV k nkH (Ω) k nkH (Ω) k H nkL (Ω) k H nkL (Ω) k nkH (Ω) The time integral of the right hand side is bounded by Step 1 and (3.9) assuming smallness of time or data, there is a c> 0 with 2 2 2 ∂ v 2 2 c( v 2 + f 2 2 ). k t nkL ((0,T ),L (Ω)) ≤ k∇H 0kL (Ω) k kL ((0,T ),L (Ω)) Now we pass to the limit as in the proof of the existence of a z-weak solution and we get, that for a subsequence (which we do not rename) also ∂tvn and ∆H vn converge 2 2 ∞ 2 weakly in L ((0,T ),L (Ω)) and that the limit of H vn is in L ((0,T ),L (Ω)). ∇ 

4. Time-periodic solutions for small forces The methods used to prove global existence and uniqueness results for the initial value problem for small data can be adapted for the construction of time-periodic solutions. This will be done first for z-weak and then for strong solutions.

Proposition 4.1 (Existence and uniqueness of time-periodic solutions). For εb > 0 there exists εa > 0 such that 2 ′ (a) if f L ((0,T ), V ) with f L2((0,T ),V ′) <εa, then there exists a T -periodic ∈ k k 2 2 z-weak solution with sup0

Proof of Proposition 4.1. Consider as in the proof of Theorem 3.1 the finite dimen- sional spaces Hn. Adapting the strategy of [10], we consider the Poincar´emap

H H , v (0) v (T ), n → n n 7→ n where vn is the solution to (3.4). Step 1 (Existence of a z-weak solution). Note that for f L2((0,T ), V ′) the differential energy inequality (3.7) can be modified using the∈ Poincar´e inequality from Lemma 6.1 to become for some c> 0

d 2 2 2 2 2 4 v + c v c f ′ + c v 2 v + v , dt k nkH k nkH ≤ k kV k∇H nkL (Ω) k nkH k nkH and multiplying by esc and integrating with respect to t this becomes 

t tc 2 2 sc 2 e vn(t) H vn(0) H + c e f(s) V ′ ds k k ≤k k 0 k k Z t sc 2 2 4 + c e v (s) 2 v (s) + v (s) ds. k∇H n kL (Ω) k n kH k n kH Z0  Assuming that vn(0) H and f L2((0,T ),V ′) are sufficiently small, one has from (3.9) for t = T andk (3.6)k that k k

Tc 2 2 Tc 4 2 (4.1) e v (T ) v (0) + ce ( v (0) + f(s) 2 ′ ). k n kH ≤k n kH k n kH k kL ((0,T ),V ) 2 Assume now that vn(0) H R for R (0, 1) being small enough to satisfy the k k ≤ −Tc ∈ smallness condition and (1 e )/c > R, moreover let f(s) L2((0,T ),L2(Ω)) be sufficiently small for (3.9) to− hold and k k

2 1 cT 2 f(s) 2 ′ (e 1)R R . k kL ((0,T ),V ) ≤ cecT − − Then

eTc v (T ) R + ceTcR2 + (ecT 1)R eTcR2 = ReTc. k n kH ≤ − − Hence, for this R> 0, and B := v H : v R and given f, the map R,n { n ∈ σ,n k nkH ≤ } B B , v(0) v(T ), R,n → R,n 7→ where vn is the solution to (3.4) is a continuous self-mapping. By Brouwer’s fixed point theorem, for any n N , there is a fixed point, i.e., v H with ∈ 0 n,0 ∈ n,σ vn,0 = vn(0) = vn(T ). Since the vn,0 are uniformly bounded in H by R there is a convergent subsequence in L2(Ω), the limit of which is in H. Following the proof of Proposition 3.1, the approximate solutions vn converge to a z-weak solution with v(0) = v(T ). By (3.9)

2 2 2 2 v (s) ∞ + v (s) 2 c( v + f 2 ′ ) k n kL ((0,T ),H) k∇H n kL ((0,T ),H) ≤ k 0kH k kL ((0,T ),V ) c(R2 + ε2) ≤ a which for R and εa sufficiently small is smaller εb. PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 19

Step 2 (Existence of a strong solution). Using Lemma 6.1, one can modify (3.11) to become after multiplying esc and integrating with respect to t

t tc 2 2 sc 2 e H vn(t) L2(Ω) H vn(0) L2(Ω) + c e f(s) L2(Ω) ds k∇ k ≤ k∇ k 0 k k t Z + c esc v (s) v (s) + v (s) 2 v (s) 2 k n kV k n kH k n kV k n kH Z0  4 2 2 2 + v (s) + v (s) 2 v (s) v (s) 2 ds. k n kH k∇H n kL k n kH k∇H n kL  Assuming that v0 H1(Ω) and f L2((0,T ),H) are sufficiently small, one can combine this with the previouslyk k obtainedk k estimate (4.1) to obtain Tc Tc 2 e v (T ) 1 v (0) 1 + ce ( v (0) 1 + f(s) 2 2 ). k n kH (Ω) ≤k n kH (Ω) k n kH (Ω) k kL ((0,T ),L (Ω)) Proceeding now analogously to the above, one proves the existence of a small T - 1 2 2 periodic solution with v(0) = v(T ) H (Ω) Lσ(Ω) sufficiently small. By Propo- sition 3.2 this is a strong solution. ∈ ∩ The norm estimate follows as above, but now by combining (3.9) with (3.12). 

Proof of Theorem 1.4. The existence of T -periodic solutions in Theorem 1.4 (a) and (b) follows directly from Proposition 4.1 (a) for forces f 2 ′ <ε and k kL ((0,T ),V ) a (b) for forces f 2 <ε , respectively. k kL ((0,T ),H) a Now, to prove the uniqueness let v1 be the T -periodic z-weak solution con- structed in Proposition 4.1 satisfying the required smallness assumption, and let and v be another T -periodic z-weak solutions for the same f L2((0,T ), V ′). As 2 ∈ in Step 6 in the proof of Theorem 3.1, one considers u := v1 v2, and then it holds that − d 2 2 2 2 2 2 u 2 + u 2 c( v + v v ) u 2 , dt k kL (Ω) k∇H kL (Ω) ≤ k∇H 1kH k∇H 1kH k 1kH k kL (Ω) compare (3.10), and hence by Poincar´e’s inequality, cf. Lemma 6.1,

d 2 2 2 2 2 u 2 c( v + v v C) u 2 . dt k kL (Ω) ≤ k∇H 1kH k∇H 1kH k 1kH − k kL (Ω) By the differential form of Gr¨onwall’s inequality

2 2 u(0) 2 = u(T ) 2 k kL (Ω) k kL (Ω) T 2 2 2 2 c R k∇H v1(s)kH +k∇H v1(s)kH kv1(s)kH ds−TC u(0) 2 e 0 . ≤k kL (Ω) and for c T v (s) 2 + v (s) 2 v (s) 2 ds

5. Global strong solutions The main idea is to establish first a global a priori bound for some smooth data, and second to use both the partial parabolic smoothing in the horizontal directions and the conservation of regularity in vertical direction to show that some z-weak solutions reach this setting for t> 0. 20 HUSSEIN,SAAL,ANDWRONA

5.1. Global a priori bound. For the global bound on strong solutions (cf. Propo- sition 5.9), we prove a differential inequality of the form ′ 2 f v ∞ f, ≤k kL where f contains certain Sobolev-norms of the solution, via performing the first order estimates (cf. Proposition 5.7 and 5.8). As in [6], to control the v ∞- coefficient, we use the logarithmic Sobolev inequality (cf. Proposition 5.2)k k and show that the Lq-norm of the solutions grow asymptotically at most as √q (cf. Proposition 5.5). Then the classical Gr¨onwall lemma gives the desired bound. This implies global existence via a standard contradiction argument. To prove the logarithmic Sobolev inequality in our setting we need the following extension result.

Lemma 5.1. Let q , q (1, ) and f LqH (( h,h),H1,qH (G)), such that ∂ f H z ∈ ∞ ∈ − 0 z ∈ Lqz (Ω). Then there exists an extension f˜ : R3 R, such that → ∞ f f˜ ∞ R3 , f˜ q R3 C f q , ∂ f˜ qi R3 C ∂ f qi , k kL ≤k kL ( ) k kL ( ) ≤ k kL k i kL ( ) ≤ k i kL for all q (1, ) and (q , ∂ ) (q , ), (q , ∂ ) . ∈ ∞ i i ∈{ H ∇H z z } Proof. The idea is, vertically, to reflect f on z = h, extend it periodically to a function fˆ : G R R and cut it off thereafter. Horizontally, we simply extend it by zero. More× explicitly→ φ(z)fˆ(x,y,z) (x, y) G, f˜(x,y,z)= 0 else, ∈  where f(x,y,z +4kh) k Z : z +4kh [ h,h], fˆ(x,y,z)= f(x, y, 2h (z +4kh)) ∃k ∈ Z : z +4kh ∈ [h,− 3h],  − ∃ ∈ ∈ for some φ C∞(R), such that φ 1 on ( h,h), and 0 φ 1 on R.  ∈ 0 ≡ − ≤ ≤ 3 Proposition 5.2 (Logarithmic Sobolev inequality). Let p = (p1,p2,p3) (1, ) 3 −1 pH ∈1,pH ∞ with pH = p1 = p2 and i=1 pi < 1. Then for any F L (( h,h),H0 (G)) p3 ∈ − such that ∂zF L (Ω), we have ∈ P 3 F Lq λ F ∞ C max 1, sup k k log e + ( F pi + ∂ F pi ) , k kL ≤ p,λ,Ω qλ k kL k i kL q≥2 i=1 !   X for any λ> 0 when all the norms are finite. Proof. By the previous Lemma 5.1, there exists an extension F˜ of F to the whole space such that

∞ ∞ F F˜ , F˜ q R3 C F q , ∂ F˜ pi R3 C ∂ F . k kL ≤k kL k kL ( ) ≤ k kL k i kL ( ) ≤ k i kpi Thus, it follows from the logarithmic Sobolev inequality on the whole space (cf. e.g. [6, Lemma 5.1]) that

F ∞ F˜ ∞ R3 k kL ≤k kL ( ) ˜ 3 F Lq(R3) λ C max 1, sup k k log e + ( F˜ pi R3 + ∂ F˜ pi R3 ) ≤ p,λ qλ k kL ( ) k i kL ( ) ( q≥2 ) i=1 ! X 3 F Lq λ C max 1, sup k k log e + ( F pi + ∂ F pi ) , ≤ p,λ,Ω qλ k kL k i kL q≥2 i=1 !   X PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 21

finishing the proof. 

Assume from now on that v is a strong and sufficiently smooth solution to the primitive equations on (0,T ) with initial condition v0 and f 0. On the way of showing, that the Lq-norm of the solution grows asymptotically≡ at most of order 2 2 (√q) we need to prove that the term v H v lies in L ((0,T ),L (Ω)). To this O · ∇ end we need estimates on H v L2 . As initial step, recall that by testing with v one obtains that the energyk∇ equalityk for strong solutions of the primitive equations holds for almost all t (0,T ) ∈ t 2 2 2 (5.1) v(t) 2 +2 v(s) 2 ds = v(0) 2 . k kL (Ω) k∇H kL (Ω) k kL (Ω) Z0 Lemma 5.3 (L2- and L4-estimates). It holds

T T 2 4 2 2 sup ( v 2 + v˜ 4 )+ ∆ v 2 dt + v˜ v˜ 2 dt k∇H kL k kL k H kL k| |∇H kL 0≤t≤T Z0 Z0 4 2 K(T )(1 + v˜ 4 + v 2 ), ≤ k 0kL k∇H 0kL R 4 where K : [0, ) is a continuously increasing function determined by h, v˜0 L4 2∞ → k k and v 2 . k∇H 0kL Proof. We shall only give a sketch of the proof. Recall, the equation of the problem splits into an equation for v on G

∂ v¯ ∆ v¯ + p = v¯ v¯ 1 h (˜v v˜ + (div v˜)˜v) dz, t − H ∇H − · ∇H − 2h −h · ∇H H (5.2) div v¯ =0, H R v¯(0) =¯v0, and an equation forv ˜ on Ω ∂ v˜ ∆ v˜ = v˜ v˜ w(˜v)˜v v¯ v˜ v˜ v¯ t − H − · ∇H − z − · ∇H − · ∇H (5.3) + 1 h (˜v v˜ + (div v˜)˜v) dz, 2h −h · ∇H H v˜(0)=v ˜ . 0 R Similarly to [14, Section 6, Step 1], one first multiplies (5.2) by PG∆H v, and then integrates over G. When integrating by parts the vanishes, and applying a compensation argument yields

1 d 2 2 v 2 + P ∆ v 2 = 2 dt k∇H kL (G) k G H kL (G) 1 h v¯ v¯ (˜v v˜ + (div v˜)˜v) dz P ∆ v d(x,y,z) − · ∇H − 2h · ∇H H G H ZΩ Z−h ! 2 2 1 2 C v v 2 +4 v˜ v˜ 2 + P ∆ v 2 ≤ k| |∇H kL (G) k| | · |∇H |kL (Ω) 4 k G H kL (G) 2 2 2 1 2 C v 4 v 4 +4 v˜ v˜ 2 + P ∆ v 2 ≤ k kL (G) k∇H kL (G) k| |∇H kL (Ω) 4 k G H kL (G) 2 2 2 1 2 2 C v 2 v 2 v 2 + P ∆ v 2 +4 v˜ v˜ 2 , ≤ k kL (G) k∇H kL (Ω) k∇H kL (G) 2 k G H kL (G) k| |∇H kL (Ω) where one uses H¨older’s and Young’s inequality along with Ladyzhenskaya’s in- equality and ellipticity of the 2-D Stokes operator PG∆H with domain contained 22 HUSSEIN,SAAL,ANDWRONA in H2(G)2. Note that C > 0 depends only on h. Hence,

t 2 P 2 2 (5.4) H v(t) L2(G) + G∆v(s) L2(G) ds H v(0) L2(G) k∇ k 0 k k ≤ k∇ k t Z t 2 2 2 2 + C v 2 v 2 v 2 ds +8 v˜ v˜ 2 ds. k kL (G) k∇H kL (Ω) k∇H kL (G) k| |∇H kL (Ω) Z0 Z0 Next one follows [14, Section 6, Step 3], i.e., testing (5.3) withv ˜ v˜ 2 yields | | 1 d 4 2 2 2 2 v˜ L4 + v˜ H v˜ +2 H v˜ d(x,y,z)= v˜ H v v˜ v˜ d(x,y,z) 4 dt k k Ω | | |∇ | ∇ | | − Ω ·∇ · | | Z h i Z 1 h + (˜ v v˜ + (div v˜)˜v) dzv˜ v˜ 2 d(x,y,z). 2h · ∇H H | | ZΩ Z−h Analogously to [14, Section 6, Step 3, estimates on I7 and I8] one obtains

2 2 4 1 2 2 v˜ H v v˜ v˜ d(x,y,z) C H v¯ L2(G) v˜ L4(Ω) + H v˜ . · ∇ · | | ≤ k∇ k k k 4 ∇ | | L2(Ω) ZΩ and 1 h (˜v v˜ + (div v˜)˜v) dzv˜ v˜ 2 d(x,y,z) 2h · ∇H H | | ZΩ Z−h 2 4 1 2 2 C H v L2(G) v˜ L4(Ω) + H v˜ . ≤ k∇ k k k 4 ∇ | | L2(Ω)

Hence, t 4 2 2 (5.5) v˜(t) L4 +2 v˜ H v˜ L2 + v˜ H v˜ L2 ds k k 0 k| ||∇ |k k| |∇ | |k Z t 4 2 4 v˜ 4 + C v 2 v˜ 4 ds. ≤k 0kL k∇H kL (G)k kL (Ω) Z0 1 2 2 Adding now (5.5) to -times (5.4) gives with v(t) 2 v 2 16 k kL (G) ≤k 0kL (Ω) t t 4 2 2 2 v˜(t) 4 + v 2 + P∆ v 2 ds + v˜ v˜ 2 ds k kL k∇H kL (G) k H kL k| |∇H kL Z0 Z0 4 2 v˜(0) 4 +8 v(0) 2 ≤k kL k∇H kL (G) t 2 2 2 4 + C ( v 2 + 1) v 2 ( v 2 + v˜ 4 ) ds. k 0kL (Ω) k∇H kL (Ω) k∇H kL (G) k kL Z0 Using Gr¨onwall’s inequality the claim follows with 4 ′ K(T ) = ( v˜(0)) 4 +8 v(0) 2 )K (T ), where k kL k∇H kL (G) 2 T 2 2 2 C max{1,kv(0)k } R k∇H vk ds C max{1,kv0k }kv(0)k K′(T )= e L2(Ω) 0 L2(Ω) e L2(Ω) L2(Ω) .  ≤ Now, using the decomposition v = v +˜v one shows the integrability of 2 2 2 div v v 2 C v˜ v˜ 2 + C v v 2 , k H ⊗ kL ≤ k| |∇H kL k| |∇H kL where the first addend is integrable by Lemma 5.3, and 2 2 2 2 2 v v 2 v v v 2 v . k| |∇H kL ≤k k∞ k∇H k2 ≤k kH (G) k∇H k2 Hence, Lemma 5.3 implies the following corollary. PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 23

Corollary 5.4 (L4-estimate). It holds T 2 4 2 (5.6) div v v 2 dt K(T )(1 + v˜ 4 + v 2 ) k H ⊗ kL ≤ k 0kL k∇H 0kL Z0 R 4 where K : [0, ) is a continuously increasing function determined by h, v˜0 L4 2∞ → k k and v 2 . k∇H 0kL Proposition 5.5 (Lq-estimate). Let q 4, then it holds ≥ sup v˜ Lq K1(T )(1 + v˜0 Lq )√q, 0≤t≤T k k ≤ k k where K1 : [0, ) R is a continuously increasing function determined by h, 4 ∞ →2 v 4 and v 2 . k 0kL k∇H 0kL For the proof we need the following estimate on the pressure. Lemma 5.6 (Estimate on the pressure for the 2-D Stokes equations). Let v,p be a solution to the two-dimensional Stokes equations ∂ v ∆ v + p = f, div v =0. t − H ∇H H Then for almost every s (0,T ) ∈ p(s) 2 f(s) 2 + ∆ v(s) 2 . k∇H kL ≤ || ||L || H ||L provided each term is finite.

Proof. Applying the complement of the 2-D Helmholtz projection PG, i.e., 1 PG to the Stokes equations gives − p = (1 P )(f(s) + ∆ v). ∇H − G H 2 2 2 2 Since this is an orthogonal projection in L (G) one has g 2 = P g 2 + k kL (G) k G kL (G) 2 2 2 (1 P )g 2 for any g L (G) , and hence the claim follows.  k − G kL (G) ∈ Proof of Proposition 5.5. We shall only give a sketch of the proof. Recall the mo- mentum equation of the problem z ∂ v ∆ v = p (v )v v dz ∂ v . t − H −∇H − · ∇H − ∇H · z  Z−h   Multiplying the above equation by v v q−2, integrating over Ω yields by integration by parts | |

1 q−2 d 2 q−2 2 2 (5.7) v q v q + v v + (q 2) v d(x,y,z) 2k kL dt k kL | | |∇H | − ∇H | | ZΩ h i = p v q −2v d(x,y,z) =: I. − ∇H · | | ZΩ Using a series of standard integral inequalities, one can show

q(q−2) 2 2 q−2 q−1 I C(1 + H p L2(G))(1 + v L2 ) q v Lq + v Lq ≤ k∇ k k k k k k k ! q 2 −1 + (q 2) v 2 H v , − | | ∇ | | L2

24 HUSSEIN,SAAL,ANDWRONA where C > 0 is independent of q. Note that from (5.2) and Lemma 5.6 it follows that 2 2 2 p 2 div v v 2 + ∆ v 2 . k∇H kL (G) ≤k H ⊗ kL (G) k H kL (G) Combining this with the above we end up with

d 2 d 2 (q +1+ v q )= v q dt k kL dt k kL 2 2 2 2 C(1 + v 2 )(1 + div v v 2 + ∆ v 2 ) q +1+ v q . ≤ k kL k H ⊗ kL (G) k H kL (G) k kL The Gr¨onwall inequality implies now  2 2 2 2 2 sup v Lq K1 (T )(q +1+ v0 Lq ) K1 (T )(1 + v0 Lq )q, 0≤t≤T k k ≤ k k ≤ k k

C T (1+kvk2 )(1+kdiv v⊗vk2 )+k∆ vk2 ds 2 R0 L2 H L2(G) H L2(G) where K1 (T ) = e , which is finite due to Corollary 5.4. 

Proposition 5.7 (Lq-estimate for ∂ v). For q [2, ) it holds z ∈ ∞ d q q−2 2 q ∂ v q + ∂ v ∂ v d(x,y,z) C ( v +1)( ∂ v q + 1) . dt k z kL | z | |∇H z | ≤ q k k∞ k z kL ZΩ Note that this bound differs from the local one in Proposition 2.6 by the assump- tions on v.

Proof. Differentiating the momentum equation with respect to z gives z ∂ ∂ v ∆ ∂ v = ( v)∂ v (∂ v )v + ( v dz)∂2v (v )∂ v. t z − H z ∇H · z − z · ∇H ∇H · z − · ∇H z Z−h Note that the last two summands of the above equation vanish after multiplication by ∂ v q−2∂ v, and integration over Ω, since | z | z z ( v dz)∂2v (v )∂ v ∂ v q−2∂ v d(x,y,z) ∇H · z − · ∇H z · | z | z ZΩ  Z−h  1 z = ( v dz)∂ ∂ v q (v ) ∂ v q d(x,y,z)=0. q ∇H · z| z | − · ∇H | z | ZΩ  Z−h  Therefore, integrating by parts and Young’s inequality imply

1 d q q−2 2 2 ∂zv Lq + ∂zv H ∂zv + (q 2) H ∂zv d(x,y,z) q dt k k Ω | | |∇ | − |∇ | || Z h i = [( v)∂ v (∂ v )v] ∂ v q−2∂ v d(x,y,z) ∇H · z − z · ∇H · | z | z ZΩ = v ∂ v q ∂ v∂ vT ∂ v q−2 v d(x,y,z) − · ∇H | z | − ∇H · z z | z | · ZΩ    C v ∂ v q−1 ∂ v d(x,y,z) ≤ q | || z | |∇H z | ZΩ 1 ∂ v q−2 ∂ v 2 d(x,y,z)+ C v 2 ∂ v q d(x,y,z). ≤ 2 | z | |∇H z | q | | | z | ZΩ ZΩ 1 q−2 2 So, subtracting 2 Ω ∂zv H ∂zv d(x,y,z) from the above inequality and mul- tiplying it by q finishes| this| proof.|∇ |  R PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 25

Proposition 5.8 (L2-estimate for v). It holds for q> 2 that ∇H 4q d 2 2 2 2 q−2 v 2 + ∆ v 2 C v ∞ v 2 + C( ∂ v q + 1). dt k∇H kL k H kL ≤ k kL k∇H kL k z kL Proof. Multiplying the momentum equation by ∆H v, and integrating over Ω, it follows from integrating by parts that −

1 d 2 2 H v L2 + ∆H v d(x,y,z) 2 dt k∇ k Ω | | Z z = (v )v div v ∂ v ∆ v d(x,y,z) · ∇H − H z · H ZΩ  Z−h   h h C v ∞ v 2 ∆ v 2 + v dz ∂ v ∆ v dz d(x,y,z). ≤ k kL k∇H kL k H kL |∇H | | z || H | ZG Z−h Z−h Using a series of standard integral inequalities, one can show that the later summand of the right hand side can be estimated by suitable terms. More specifically, h h 1 − 1 v ∂ v ∆ v C(1 + ∆ v 2 q ) ∂ v q ∆ v 2 . |∇H | | z || H |≤ k H k k z kL k H kL ZG Z−h Z−h The above and Young’s inequality imply

1 d 2 2 v 2 + ∆ v 2 2 dt k∇H kL k H kL 1 1 2 − q C v ∞ v 2 +(1+ ∆ v ) ∂ v q ∆ v 2 ≤ k kL k∇H kL k H kL2 k z kL k H kL  4q  1 2 2 2 q−2 ∆ v 2 + C( v ∞ v 2 + ∂ v q + 1). ≤ 2k H kL k kL k∇H kL k z kL 1 2 So, subtracting ∆ v 2 finishes the proof.  2 k H kL Proposition 5.9 (Uniform a priori bound). For any finite time T , we have T 2 2+η 2 sup ( v 2 + ∂zv )+ H v 2 dt Cη,h,T ( v0 1 ), k∇ kL k k2+η k∇ ∇ kL ≤ k kHη t∈[0,T ] Z0 for an increasing function Cη,h,T depending only on η, h, T and with v0 1 = k kHη v 1 + ∂ v + v ∞ . k 0kH k z 0k2+η k 0kL Recall that the definition of 1 is given in Theorem 1.3. k·kHη Proof. Summing up Proposition 5.7 and 5.8, one can show

d 2 A(t)+ B(t) C(1 + v ∞ )A(t), dt ≤ k kL where λ A = A1 + A1 + A2, B = A1 + B1 + B2, λ =4/η 2 2+η 2 A = ∂ v 2 + ∂ v + e, B = ∂ v 1 k z kL k z k2+η 1 k∇H z k2 2 2 A = v 2 + e, B = ∆ v 2 + e. 2 k∇H kL 2 k H kL We will show 2 (1 + v ∞ ) C log B. k kL ≤ 26 HUSSEIN,SAAL,ANDWRONA

Thus a logarithmic type Gr¨onwall inequality (cf. e.g. [6, Lemma 2.5]) will imply the desired bound. By Proposition 5.2, Proposition 5.5 and the Sobolev and horizontal Poincar´einequalities, one has 2 2 v Lq v ∞ C max 1, sup k k log(e + v + v + ∂ v 2 + v 2 ) k kL ≤ q k∇H k6 k k6 k z kL k kL  q≥2  C log(e + v 1 + v 1 + ∂ v 2 ) ≤ k∇H kH k kH k z kL C log(e + v + v + ∂ v 2 ) ≤ k∇H k2 k∇∇H k2 k z kL C log(e + ∆ v + ∂ v + ∂ v 2 ) C log B, ≤ k H k2 k∇H z k2 k z kL ≤ finishing the proof. 

Proof of Theorem 1.3 (a). Note first that a local solution can be constructed as in Proposition 3.1 and 3.2 by a Galerkin scheme using the a priori bound from Proposition 5.9. Now, to prove global existence, let Tmax be the supremum over the existence times of the strong solution, and assume that T < . Choose max ∞ ∗ δ < max Tmax,T (C v0 1 ) , { k kHη } where T ∗(K) denotes the minimal existence time given by Theorem 1.1 (b) of the strong solution of the problem with initial data of norm at most K, and C is the constant of the previous Proposition. Since v ∞ 1 C v0 1 there is a Tˆ (Tmax δ, Tmax), such that k kL ((0,Tmax),H ) ≤ k kHη ∈ − v(Tˆ) 1 C v0 1 . By the local existence of strong solutions there exists a k kH ≤ k kHη strong solutionv ˆ to the problem with initial data v(Tˆ) with an existence time ∗ ˆ ∗ + + T ( v(T ) H1 ). Since T : R R is monotone decreasing, it holds by the last propositionk k that → ∗ ∗ T ( v(Tˆ) 1 ) T (C v0 1 ) > δ. k kH ≥ k kHη Note that v ˆ =v ˆ ˆ . By uniqueness of the solution, v can be ex- |[T ,Tmax] |[0,Tmax−T ] ∗ tended to Tˆ +T ( v(Tˆ) 1 ) > Tˆ +δ>T . A contradiction to the maximality of k kH max Tmax. So, our assumption of Tmax < was wrong, finishing the global existence proof. ∞ 

5.2. Extension of z-weak solutions to global strong solutions.

Proof of Theorem 1.3 (b). By Theorem 1.1 (a), there is a z-weak solution on (0,T ′) for some T ′ > 0, and v L2((0,T ′),L2(Ω)2×2) and hence v(t) H1H1 ∇H z ∈ ∈ z xy ⊂ H1(Ω) for almost every t (0,T ′). This has to be understood in the sense that if one has a smooth approximating∈ sequence (v ) C0((0,T ′),H1H1 ), then there n ⊂ 0 z xy exists a subsequence (v ) such that v (t) converges for almost every t (0,T ′). nk nk ∈ In particular there exists t > 0 with v(t ) H1 (Ω) and div v(t ) = 0. Taking 1 1 ∈ 0,l H 1 v(t1) as new initial value one obtains by Theorem 1.1 (b) a strong solution on ′′ ′′ ′ ′′ (t1,T ) for some T (t1,T ). This strong solution agrees on (t1,T ) with the ∈ ′ 2 original z-weak solution defined on (0,T ) since vnk (t1) converges also in L and 0 ′ 2 2 since v C ([0,T ],L ) its L -limit is in fact v(t1). Hence by the uniqueness of ∈ ′′ z-weak solutions, cf. Proposition 3.1, both agree on (t1,T ). Note that v(t1) H1 is finite, but there is no explicit control on its norm. k k PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 27

η η ∂ v L ∂zv(t) L z 0 ∈ ∈ z weak solution strong solution global strong solution − 1 1 v0 H v(t ) H v(t ) H ∈ 1 ∈ 2 ∈ η t t0 =0 t1 t2

Figure 1. From z-weak to global solution

Recall that the strong solution has the regularity v L2((t ,T ′′),L2H2 ) and ∈ 1 z xy as z-weak solution v L2((0,T ′),H1H1 ). Consider ∆ and ∆ = ∂2 on ∈ z xy H z z D(∆ )= L2(H1 L2H2 ) and D(∆ )= v H2L2 : v (z = h)=0 , H z 0,xy ∩ z xy z { ∈ z xy z ± } respectively. These are commuting self-adjoint operators in Hilbert spaces, and hence one obtains by the mixed derivative theorem, cf. [29, Corollary III.4.5.10], for θ (0, 1) ∈ H1H1 L2H2 L2H1 = D(∆1/2 ∆ 1/2) D(∆ ) z 0,xy ∩ z xy ∩ z 0,xy z ◦ H ∩ H D(∆θ/2 ∆ θ/2 ∆ 1−θ) ⊂ z ◦ H ◦ H = D(∆θ/2 ∆ 1−θ/2) HθH2−θ. z ◦ H ⊂ z xy 3/4 5/4 ∞ ∞ ∞ In particular for θ = 3/4 one has Hz Hxy Lz Lxy = L (Ω). Moreover, by ∞ ′ ⊂ η 2 Proposition 2.6 one has that vz L ((0,T ),L (Ω) ) since by assumption here ∂ v Lη(Ω). Putting the pieces∈ together, one deduces that z 0 ∈ v L2((t ,T ′′),H1(Ω)2 v L2(Ω)2 : v Lη(Ω)2 L∞(Ω)2) ∈ 1 ∩{ ∈ z ∈ } ∩ and therefore for almost every t (t ,T ′′) one has ∈ 1 v(t) H1 = H1(Ω)2 v L2(Ω)2 : v Lη(Ω)2 L∞(Ω)2. ∈ η ∩{ ∈ z ∈ } ∩ ′′ Following the previous arguments one takes now such t2 (t1,T ) as new initial time, and one ends up in the situation of Theorem 1.3 (a) which∈ gives that v extends ′′ to a global strong solution on (t2,T ) for any T > 0. In fact 0

6. Some inequalities Lemma 6.1 (Poincar´einequality for lateral vanishing trace). There exists a con- stant C > 0 such that 2 1 v 2 C v 2 for v L (( h,h),H (G)), || ||L (Ω) ≤ ||∇H ||L (Ω) ∈ − 0 2 2 1 v 2 C ∆ v 2 for v L (( h,h),H (G) H (G)). ||∇H ||L (Ω) ≤ || H ||L (Ω) ∈ − ∩ 0 Proof. Note that by the classical 2D-Poincar´einequality for some C2D > 0 2 2 2 v( ,z) 2 C v( ,z) 2 for allmost every z ( h,h), || · ||L (G) ≤ 2D||∇H · ||L (G) ∈ − and integrating with respect to z gives the first estimate with C = C2D√2h. 28 HUSSEIN,SAAL,ANDWRONA

For the second inequality, note that there is a C2D > 0 such that 2 2 2 v( ,z) 2 C ∆ v( ,z) 2 for allmost every z ( h,h), || · ||L (G) ≤ 2D|| H · ||L (G) ∈ − 2 2 and integrating with respect to z gives v 2 C ∆ v 2 . Using this, || ||L (Ω) ≤ || H ||L (Ω) Cauchy-Schwartz and Young’s inequality yields

1 2 1 2 (C + 1) 2 v, v 2 = v, ∆ v 2 v + ∆ v ∆ v .  h∇H ∇H iL (Ω) h H iL (Ω) ≤ 2 k k 2 k H k ≤ 2 k H k The following inequalities is helpful for proving local a priori estimates. Lemma 6.2 (Tri-linear estimates). a) Let f,g L2(( h,h),H1(G)) and h H1(( h,h),L2(G)). Then ∈ − 0 ∈ − 1/2 1/2 1/2 1/2 1/2 1/2 fg,h c f f g g ∂ h h + h 2 . | h i|≤ k∇H kL2 k kL2 k∇H kL2 k kL2 k z kL2 k kL2 k kL b) Let f L2(Ω), g H1(Ω) and h L2(( h,h),H1(G)). Then  ∈ ∈ ∈ − 0 fg,h | h i | 1/2 1/2 1/2 1/2 c f 2 h h ( g 2 + ∂ g 2 ) ( g 2 + g 2 ) . ≤ k kL k∇H kL2 k kL2 k kL k z kL k kL k∇H kL Proof. a) We have h fg,h = f(z)g(z),h(z) dz | h iΩ | | h iG | Z−h h f(z) 4 g(z) 4 h(z) 2 dz ≤ k kL (G)k kL (G)k kL (G) Z−h h h ∞ 2 f(z) 4 g(z) 4 dz ≤k kL ((−h,h),L (G)) k kL (G)k kL (G) Z−h h ∞ 2 f 2 4 g 2 4 . ≤k kL ((−h,h),L (G))k kL ((−h,h),L (G))k kL ((−h,h),L (G)) 2 By Ladyzhenskaya’s inequality f(z) 4 c f(z) 2 f(z) 2 we ob- k kL (G) ≤ k∇H kL (G) k kL (G) tain h h 2 2 f 2 4 = f(z) 4 dz c f(z) 2 f(z) 2 dz k kL ((−h,h),L (G)) k kL (G) ≤ k∇H kL (G) k kL (G) Z−h Z−h c f 2 f 2 , ≤ k∇H kL (Ω) k kL (Ω) and by the Gagliardo-Nirenberg interpolation inequality 1/2 1/2 h ∞ 2 c( ∂ h h + h 2 ). k kL ((−h,h),L (G)) ≤ k z kL2(Ω)k kL2(Ω) k kL (Ω) Hence we get 1/2 1/2 1/2 1/2 1/2 1/2 fg,h c ∂ h h + h 2 f f g g . | h iΩ |≤ k z kL2 k kL2 k kL k∇H kL2 k kL2 k∇H kL2 k kL2 For the proof of b) see [30, Lemma 2.1 (a)]. 

Lemma 6.3 (Non-linear Gr¨onwall inequality, cf. [8]). Let x:[a,b] R+ be a continuous function that satisfies the inequality: → t x(t) M + Ψ(s)ω(x(s)) ds, t [a,b], ≤ ∈ Za where M 0, Ψ:[a,b] R+ is continuous and w : R+ R+ is continuous and monotone-increasing.≥ → → PRIMITIVE EQUATIONS WITH HORIZONTAL VISCOSITY 29

Then the estimate t x(t) Φ−1 Φ(M)+ Ψ(s) ds , t [a,b], ≤ ∈  Za  holds, where Φ: R R is given by → u ds Φ(u) := , u R. w(s) ∈ Zu0 References

[1] H. Bahouri, J. Y. Chemin and R. Danchin. Fourier analysis and nonlinear partial differential equations. Springer, Heidelberg, 2011. doi:10.1007/978-3-642-16830-7 [2] D. Bresch, F-Guill´en-Gonz´alez, N. Masmoudi, and M. A. Rodr´ıguez-Bellido. On the unique- ness of weak solutions of the two-dimensional primitive equations. Differential Integral Equa- tions, 162(1):77–94, 2003. https://projecteuclid.org/euclid.die/1356060697 [3] A. Celik and M. Kyed. Nonlinear wave equation with damping: periodic forcing and non- resonant solutions to the Kuznetsov equation. ZAMM Z. Angew. Math. Mech., 98(3):412–430, 2018. doi:10.1002/zamm.201600280 [4] Ch. Cao, S. Ibrahim, K. Nakanishi and E. S. Titi. Finite-time blowup for the inviscid primitive equations of oceanic and atmospheric dynamics. Comm. Math. Phys., 337(2):473–482, 2015. doi:10.1007/s00220-015-2365-1 [5] Ch. Cao, J. Li and E. S. Titi. Global well-posedness of the 3D primitive equations with only horizontal viscosity and diffusivity. Comm. Pure Appl. Math., 69(8):1492–1531, 2016. doi:10.1002/cpa.21576 [6] Ch. Cao, J. Li and E. S. Titi. Strong solutions to the 3D primitive equations with only horizontal dissipation: Near H1 initial data. J. Funct. Anal., 272(11):4606–4641, 2017. doi:10.1016/j.jfa.2017.01.018 [7] Ch. Cao and E. S. Titi. Global well-posedness of the three-dimensional viscous primitive equations of large scale ocean and atmosphere dynamics. Ann. of Math. (2), 166(1):245–267, 2007. doi:10.4007/annals.2007.166.245 [8] S. S. Dragomir. Some Gronwall type inequalities and applications. Nova Science Publishers, Inc., Hauppauge, NY, 2003. [9] T. Eiter and M. Kyed. Time-periodic linearized Navier-Stokes equations: an approach based on Fourier multipliers. In Particles in flows, Adv. Math. Fluid Mech., Birkh¨auser/Springer, Cham: 77–137, 2017. doi:10.1007/978-3-319-60282-0 2 [10] P. Galdi, M. Hieber and T. Kashiwabara. Strong time-periodic solutions to the 3D prim- itive equations subject to arbitrary large forces. Nonlinearity, 30(10):3979–3992, 2017. doi:10.1088/1361-6544/aa8166 [11] M. Geissert, M. Hieber and T. H. Nguyen. A general approach to time periodic incom- pressible viscous fluid flow problems. Arch. Ration. Mech. Anal., 220(3):1095–1118, 2016. doi:10.1007/s00205-015-0949-8 [12] Y. Giga, M. Gries, A. Hussein, M. Hieber and T. Kashiwabara. The Primitive Equations in the scaling invariant space L∞(L1). Preprint arXiv:1710.04434, 2017. [13] M. Hieber, A. Hussein. An approach to the primitive equations for ocean and atmospheric dy- namics by evolution equations. In: Fluids Under Pressure, T. Bodnar et al. (eds.), Advances in Math. Fluid Mech., Birkh¨auser, 2020. doi:10.1007/978-3-030-39639-8 [14] M. Hieber and T. Kashiwabara. Global strong well-posedness of the three dimensional primitive equations in Lp-spaces. Arch. Rational Mech. Anal.221(3):1077–1115, 2016. doi:10.1007/s00205-016-0979-x [15] Ch-H. Hsia and M. Ch. Shiue. On the asymptotic stability analysis and the existence of time- periodic solutions of the primitive equations. Indiana Univ. Math. J., 62(2):403–441, 2013. doi:10.1512/iumj.2013.62.4902 [16] D. Han-Kwan, T. T. Nguyen. Ill-Posedness of the Hydrostatic Euler and Sin- gular Vlasov Equations. Arch. Ration. Mech. Anal., 221(3):1317–1344, 2016. doi:10.1007/s00205-016-0985-z [17] N. Ju. On H2 solutions and z-weak solutions of the 3D primitive equations. Indiana Univ. Math. J., 66(3):973–996, 2017. doi:10.1512/iumj.2017.66.6065 30 HUSSEIN,SAAL,ANDWRONA

[18] I. Kukavica, R. Temam, V. Vicol and M. Ziane. Local existence and uniqueness for the hydrostatic Euler equations on a bounded domain. J. Differential Equations, 250(3):1719– 1746, 2011. doi:10.1016/j.jde.2010.07.032 [19] I. Kukavica and M. Ziane. On the regularity of the primitive equations of the ocean. Nonlin- earity 20(12): 2739–2753, 2007. doi:10.1088/0951-7715/20/12/001. [20] M. Kyed and J. Sauer. A method for obtaining time-periodic Lp estimates. J. Differential Equations 262(1): 633–652, 2017. doi:10.1016/j.jde.2016.09.037. [21] J. Li and E. S. Titi. Recent Advances Concerning Certain Class of Geophysical Flows. In Handbook of Mathematical Analysis in Mechanics of Viscous Fluids, Springer International Publishing, 2016. doi:10.1007/978-3-319-10151-4 22-1 [22] J. L. Lions, R. Temam and Sh. H. Wang. New formulations of the primitive equations of atmo- sphere and applications. Nonlinearity, 5(2):237–288, 1992. doi:10.1088/0951-7715/5/2/001 [23] J. L. Lions, R. Temam and Sh. H. Wang. On the equations of the large-scale ocean. Nonlin- earity, 5(5):1007–1053, 1992. doi:10.1088/0951-7715/5/5/002 [24] J. L. Lions, R. Temam and Sh. H. Wang. Models for the coupled atmosphere and ocean. (CAO I,II). Comput. Mech. Adv., 1:3–119, 1993. [25] G.Lukaszewicz, E. E. Ortega-Torres and M. A. Rojas-Medar Strong periodic solu- tions for a class of abstract evolution equations. Nonlinear Anal., 54(6):1045–1056, 2003. doi:10.1016/S0362-546X(03)00125-1 [26] A. Majda. Introduction to PDEs and Waves for the Atmosphere and Ocean. (Courant Lecture Notes in Mathematics vol 9). Providence, RI: American Mathematical Society, 2003. [27] H. Morimoto. Survey on time periodic problem for fluid flow under inhomoge- neous boundary condition Discrete Contin. Dyn. Syst. Ser. S, 5(3):631–639, 2012. doi:10.3934/dcdss.2012.5.631 [28] J. Pedlosky. Geophysical . Second Edition. Springer, New York, 1987. doi:10.1007/978-1-4612-4650-3 [29] J. Pr¨uss and G. Simonett. Moving interfaces and quasilinear parabolic evolution equations. Birkh¨auser/Springer, [Cham], 2016. doi:10.1007/978-3-319-27698-4 [30] M. Saal. Primitive Equations with half horizontal viscosity. To appear in Advances in Dif- ferential Equations Preprint arXiv:1807.05045, 2018. [31] J. Serrin. A note on the existence of periodic solutions of the Navier-Stokes equations. Arch. Rational Mech. Anal., 3:120–122, 1959. doi:10.1007/BF00284169 [32] Th. Tachim Medjo. Existence and uniqueness of strong periodic solutions of the prim- itive equations of the ocean. Discrete Contin. Dyn. Syst., 26(4):1491–1508, 2010. doi:10.3934/dcds.2010.26.1491 [33] H. Triebel. Theory of Function Spaces. (Reprint of 1983 edition) Springer AG, Basel, 2010. doi:10.1007/978-3-0346-0416-1 [34] G. K. Vallis. Atmospheric and Oceanic Fluid Dynamics. Second Edition. Cambridge Univ. Press, 2006. [35] W. M. Washington and C. L. Parkinson. An Introduction to Three Dimensional Climate Modeling. Second Edition. University Science Books, 2005. [36] T. K. Wong. Blowup of solutions of the hydrostatic Euler equations. Proc. Amer. Math. Soc.,: 143: 1119–1125, 2015. doi:10.1090/S0002-9939-2014-12243-X [37] M. Wrona. Liquid Crystals and the Primitive Equations: An Approach by Maximal Regular- ity. Dissertation. Technische Universit¨at Darmstadt, 2020. doi:10.25534/tuprints-00011551

Department of Mathematics, TU Kaiserslautern, Paul-Ehrlich-Straße 31, 67663 Kaiserslautern, Germany Email address: [email protected]

Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy Email address: [email protected]

Departement of Mathematics, TU Darmstadt, Schlossgartenstr. 7, 64289 Darmstadt, Germany Email address: [email protected]