Let us remark the horizontal motion can be further approximated by the geostrophic balance when the Rossby number (the ratio of the horizontal acceleration to the Coriolis force) is very small. By taking advantage of these assumptions and other geophysical considerations several intermediate models have been developed and used in numerical studies of weather prediction and long-time climate dynamics (see, e.g., [4] , [7] , [8] , [19] , [20] , [21] , [24] , [25] , [26] , [27] and references therein). Some of these models have also been the subject of analytical mathematical study (see, e.g., [2] , [3] , [5] , [6] , [10] , [11] , [12] , [14] , [22] , [23] , [29] and references therein).

In this paper we will focus on the $3D$  primitive equations in a cylindrical domain $$\Omega =M\times (-h,0),$$  where $M$  is a smooth bounded domain in ${\mathbb{R}}^2$  :

$$\begin{array}{ccc}& & \frac{\partial v}{\partial t}+(v\cdot \nabla )v+w\frac{\partial v}{\partial z}+\nabla p+f\vec{k}\times v+L_{1}v=0\end{array}$$

(1)

$$\begin{array}{ccc}& & {\partial }_{z}p+T=0\end{array}$$

(2)

$$\begin{array}{ccc}& & \nabla \cdot v+{\partial }_{z}w=0\end{array}$$

(3)

$$\begin{array}{ccc}& & \frac{\partial T}{\partial t}+v\cdot \nabla T+w\frac{\partial T}{\partial z}+L_{2}T=Q\end{array}$$

(4)

where the horizontal velocity field $v=(v_1,v_2)$  , the velocity field $(v_1,v_2,w)$  , the temperature $T$  and the pressure $p$  are the unknowns. $f=f_0(\beta +y)$  is the Coriolis parameter, $Q$  is a given heat source. The viscosity and the heat diffusion operators $L_1$  and $L_2$  are given by

$$\begin{array}{ccc}& & L_1=-\frac{1}{R{e}_1}\Delta -\frac{1}{R{e}_2}\frac{{\partial }^2}{\partial z^2},\end{array}$$

(5)

$$\begin{array}{ccc}& & L_2=-\frac{1}{R{t}_1}\Delta -\frac{1}{R{t}_2}\frac{{\partial }^2}{\partial z^2},\end{array}$$

(6)

where $R{e}_1,R{e}_2$  are positive constants representing the horizontal and vertical Reynolds numbers, respectively, and $R{t}_1,R{t}_2$  are positive constants which stand for the horizontal and vertical heat diffusivity, respectively. We set $\nabla =({\partial }_x,{\partial }_y)$  to be the horizontal gradient operator and $\Delta ={\partial }_x^2+{\partial }_y^2$  to be the horizontal Laplacian. We partition the boundary of $\Omega$  into:

$$\begin{array}{ccc}& & {\Gamma }_u=\left\{\right(x,y,z)\in \overline{\Omega }:z=0\},\end{array}$$

(7)

$$\begin{array}{ccc}& & {\Gamma }_b=\left\{\right(x,y,z)\in \overline{\Omega }:z=-h\},\end{array}$$

(8)

$$\begin{array}{ccc}& & {\Gamma }_s=\left\{\right(x,y,z)\in \overline{\Omega }:(x,y)\in \partial M,-h\le z\le 0\}.\end{array}$$

(9)

We equip the system ( 1 )–( 4 ) with the following boundary conditions – with wind–driven on the top surface and non-slip and non-heat flux on the side walls and bottom (see, e.g., [17] , [18] , [19] , [21] , [24] ,[25] , [26] ):

$$\begin{array}{ccc}& & \text{on}{\Gamma }_u:\frac{\partial v}{\partial z}=h\tau ,w=0,\frac{\partial T}{\partial z}=-\alpha (T-T^{*});\end{array}$$

(10)

$$\begin{array}{ccc}& & \text{on}{\Gamma }_b:\frac{\partial v}{\partial z}=0,w=0,\frac{\partial T}{\partial z}=0;\end{array}$$

(11)

$$\begin{array}{ccc}& & \text{on}{\Gamma }_s:v\cdot \vec{n}=0,\frac{\partial v}{\partial \vec{n}}\times \vec{n}=0,\frac{\partial T}{\partial \vec{n}}=0,\end{array}$$

(12)

where $\tau (x,y)$  is the wind stress on ocean surface, $\vec{n}$  is the normal vector of ${\Gamma }_s$  , and $T^{*}(x,y)$  is typical temperature distribution of the top surface of the ocean. For simplicity we assume here that $\tau$  and $T^{*}$  are time independent. However, the results presented here are equally valid when these quantities are time dependent and satisfy certain bounds in space and time.

Due to the boundary conditions ( 10 )–( 12 ), it is natural to assume that $\tau$  and $T^{*}$  satisfy the compatibility boundary conditions:

$$\begin{array}{ccc}& & \tau \cdot \vec{n}=0,\frac{\partial \tau }{\partial \vec{n}}\times \vec{n}=0,\text{on}\partial M.\end{array}$$

(13)

$$\begin{array}{ccc}& & \frac{\partial T^{*}}{\partial \vec{n}}=0\text{on}\partial M.\end{array}$$

(14)

In addition, we supply the system with the initial condition:

$$\begin{array}{ccc}& & v(x,y,z,0)=v_0(x,y,z).\end{array}$$

(15)

$$\begin{array}{ccc}& & T(x,y,z,0)=T_0(x,y,z).\end{array}$$

(16)

In [17] and [18] the authors set up the mathematical framework to study the viscous primitive equations for the atmosphere and ocean circulation. Moreover, similar to the $3D$  Navier–Stokes equations, they have shown the global existence of weak solutions (without uniqueness) and the short time existence and uniqueness of strong solutions of the viscous primitive equations model. In [14] the authors proved the global existence and uniqueness of strong solutions to the viscous primitive equations in thin domains for a large set of initial data whose size depends inversely on the thickness of the domain. In this paper we show the global existence and uniqueness (regularity) of the strong solutions to the $3D$  viscous primitive equation model ( 1 )–( 16 ) in general cylindrical domain, $\Omega$  , and initial data. As in the case of $3D$  Navier–Stokes equations the question of uniqueness of the weak solutions to this model is still open.

2 Preliminaries

2.1 New Formulation

First, let us reformulate the system ( 1 )–( 16 ). We integrate the equation ( 3 ) in the $z$  direction to obtain $$w(x,y,z,t)=w(x,y,-h,t)-{\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi .$$  By virtue of ( 10 ) and ( 11 ) we have

$$\begin{array}{c}w(x,y,z,t)=-{\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi ,\end{array}$$

(17)

and

$$\begin{array}{c}{\int }_{-h}^0\nabla \cdot v(x,y,\xi ,t)d\xi =\nabla \cdot {\int }_{-h}^{0}v(x,y,\xi ,t)d\xi =0.\end{array}$$

(18)

We denote by

$$\begin{array}{c}\overline{\phi }(x,y)=\frac{1}{h}{\int }_{-h}^0\phi (x,y,\xi )d\xi ,\forall (x,y)\in M.\end{array}$$

(19)

In particular,

$$\begin{array}{c}\overline{v}(x,y)=\frac{1}{h}{\int }_{-h}^{0}v(x,y,\xi )d\xi ,\text{in}M.\end{array}$$

(20)

We will denote the fluctuation by

$$\begin{array}{c}\tilde{v}=v-\overline{v}.\end{array}$$

(21)

Notice that

$$\begin{array}{ccc}& & \overline{\tilde{v}}=0.\end{array}$$

(22)

From the above and ( 12 ) we obtain

$$\begin{array}{ccc}& & \nabla \cdot \overline{v}=0,\text{in}M,\end{array}$$

(23)

and

$$\begin{array}{ccc}& & \overline{v}\cdot \vec{n}=0,\frac{\partial \overline{v}}{\partial \vec{n}}\times \vec{n}=0,\text{on}\partial M.\end{array}$$

(24)

By integrating equation ( 2 ) we obtain $$p(x,y,z,t)=-{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi +p_s(x,y,t).$$  Substitute ( 17 ) and the above relation into equation ( 1 ) we reach

$$\begin{array}{ccc}& & \frac{\partial v}{\partial t}+(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}\end{array}$$

$$\begin{array}{ccc}& & +\nabla p_s(x,y,t)-\nabla {\int }_{-h}^{z}T(x,y,\xi ,t)d\xi +f\vec{k}\times v+L_{1}v=0.\end{array}$$

(25)

Remark 1. Notice that due to the compatibility boundary conditions ( 13 ) and ( 14 ) one can convert the boundary condition ( 10 )–( 12 ) to be homogeneous by replacing $(v,T)$  by $(v+\frac{(z+h{)}^2-h^3/3}{2}\tau ,T+T^{*})$  while ( 23 ) still true. For simplicity and without loss generality we will assume that $\tau =0,T^{*}=0.$  However, we emphasize that our results are still valid for general $\tau$  and $T^{*}$  provided they are smooth enough. In a forthcoming paper we will study the long-time dynamics and global attractors to the primitive equations with general $\tau$  and $T^{*}$  .

Therefore, under the assumption that $\tau =0,T^{*}=0$  , we have the following new formulation for system ( 1 )–( 16 ):

$$\begin{array}{ccc}& & \frac{\partial v}{\partial t}+L_{1}v+(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}\end{array}$$

$$\begin{array}{ccc}& & +\nabla p_s(x,y,t)-\nabla {\int }_{-h}^{z}T(x,y,\xi ,t)d\xi +f\vec{k}\times v=0,\end{array}$$

(26)

$$\begin{array}{ccc}& & \frac{\partial T}{\partial t}+L_{2}T+v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial T}{\partial z}=Q,\end{array}$$

(27)

$$\begin{array}{ccc}& & {\frac{\partial v}{\partial z}|}_{z=0}=0,{\frac{\partial v}{\partial z}|}_{z=-h}=0,{v\cdot \vec{n}|}_{{\Gamma }_s}=0,{\frac{\partial v}{\partial \vec{n}}\times \vec{n}|}_{{\Gamma }_s}=0,\end{array}$$

(28)

$$\begin{array}{ccc}& & {({\partial }_{z}T+\alpha T)|}_{z=0}=0;{{\partial }_{z}T|}_{z=-h}=0;{{\partial }_{n}T|}_{{\Gamma }_s}=0,\end{array}$$

(29)

$$\begin{array}{ccc}& & v(x,y,z,0)=v_0(x,y,z),\end{array}$$

(30)

$$\begin{array}{ccc}& & T(x,y,z,0)=T_0(x,y,z).\end{array}$$

(31)

2.2 Properties of $\overline{v}$  and $\tilde{v}$ 

By taking the average of equations ( 26 ) in the $z$  direction, over the interval $(-h,0)$  , and using the boundary conditions ( 28 ), we obtain

$$\begin{array}{ccc}& & \frac{\partial \overline{v}}{\partial t}+\overline{(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}}+\nabla p_s(x,y,t)-\nabla \left[\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right]\end{array}$$

$$\begin{array}{ccc}& & +f\vec{k}\times \overline{v}-\frac{1}{R{e}_1}\Delta \overline{v}=0.\end{array}$$

(32)

As a result of ( 22 ), ( 23 ) and integration by parts we have

$$\begin{array}{ccc}& & \overline{(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}}=(\overline{v}\cdot \nabla )\overline{v}+\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}.\end{array}$$

(33)

By subtracting ( 32 ) from ( 26 ) and using ( 33 ) we get

$$\begin{array}{ccc}& & \frac{\partial \tilde{v}}{\partial t}+L_1\tilde{v}+(\tilde{v}\cdot \nabla )\tilde{v}-\left({\int }_{-h}^z\nabla \cdot \tilde{v}(x,y,\xi ,t)d\xi \right)\frac{\partial \tilde{v}}{\partial z}+(\tilde{v}\cdot \nabla )\overline{v}+(\overline{v}\cdot \nabla )\tilde{v}-\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}\end{array}$$

$$\begin{array}{ccc}& & -\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right)+f\vec{k}\times \tilde{v}=0.\end{array}$$

(34)

Therefore, $\overline{v}$  satisfies the following equations and boundary conditions:

$$\begin{array}{ccc}& & \frac{\partial \overline{v}}{\partial t}-\frac{1}{R{e}_1}\Delta \overline{v}+(\overline{v}\cdot \nabla )\overline{v}+\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}+f\vec{k}\times \overline{v}\end{array}$$

$$\begin{array}{ccc}& & +\nabla \left[p_s(x,y,t)-\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right]=0,\end{array}$$

(35)

$$\begin{array}{ccc}& & \nabla \cdot \overline{v}=0,\end{array}$$

(36)

$$\begin{array}{ccc}& & \overline{v}\cdot \vec{n}=0,\frac{\partial \overline{v}}{\partial \vec{n}}\times \vec{n}=0,\text{on}\partial M,\end{array}$$

(37)

and $\tilde{v}$  satisfies the following equations and boundary conditions:

$$\begin{array}{ccc}& & \frac{\partial \tilde{v}}{\partial t}+L_1\tilde{v}+(\tilde{v}\cdot \nabla )\tilde{v}-\left({\int }_{-h}^z\nabla \cdot \tilde{v}(x,y,\xi ,t)d\xi \right)\frac{\partial \tilde{v}}{\partial z}+(\tilde{v}\cdot \nabla )\overline{v}+(\overline{v}\cdot \nabla )\tilde{v}\end{array}$$

$$\begin{array}{ccc}& & -\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}+f\vec{k}\times \tilde{v}-\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right)=0,\end{array}$$

(38)

$$\begin{array}{ccc}& & {\frac{\partial \tilde{v}}{\partial z}|}_{z=0}=0,{\frac{\partial \tilde{v}}{\partial z}|}_{z=-h}=0,{\tilde{v}\cdot \vec{n}|}_{{\Gamma }_s}=0,{\frac{\partial \tilde{v}}{\partial \vec{n}}\times \vec{n}|}_{{\Gamma }_s}=0,.\end{array}$$

(39)

Remark 2. We recall that by virtue of the maximum principle one is able to show the global well-posedness of the $3D$  viscous Burgers equations (see, for instance, [16] and references therein). Such an argument, however, is not valid for the $3D$  Navier–Stokes equations because of the pressure term. Remarkably, the pressure term is absent from equation ( 38 ). This fact allows us to obtain a bound for the $L^6$  norm of $\tilde{v}$  , which is a key estimate in our proof of the global regularity for the system ( 1 )–( 16 ).

2.3 Functional spaces and Inequalities

Let us denote by $L^2(\Omega ),L^2\left(M\right)$  and $H^m(\Omega ),H^m\left(M\right)$  the usual $L^2-$  Lebesgue and Sobolev spaces, respectively ([1] ). We denote by

$$\begin{array}{c}\parallel \phi {\parallel }_p=\{\begin{array}{cc}{\left({\int }_{\Omega }\left|\phi \right(x,y,z){|}^{p}dxdydz\right)}^{\frac{1}{p}},& \text{for every}\phi \in L^p(\Omega )\\ {\left({\int }_M\left|\phi \right(x,y){|}^{p}dxdy\right)}^{\frac{1}{p}},& \text{for every}\phi \in L^p\left(M\right).\end{array}\end{array}$$

(40)

Let

$$\begin{array}{ccc}\widetilde{{\mathcal{V}}_1}& =& \left\{v\in C^{\infty }(\Omega ):{\frac{\partial v}{\partial z}|}_{z=0}=0,{\frac{\partial v}{\partial z}|}_{z=-h}=0,{v\cdot \vec{n}|}_{{\Gamma }_s}=0,{\frac{\partial v}{\partial \vec{n}}\times \vec{n}|}_{{\Gamma }_s}=0,\nabla \cdot \overline{v}=0\right\},\end{array}$$

$$\begin{array}{ccc}\widetilde{{\mathcal{V}}_2}& =& \left\{T\in C^{\infty }(\Omega ):{\frac{\partial T}{\partial z}|}_{z=-h}=0;{\left(\frac{\partial T}{\partial z}+\alpha T\right)|}_{z=0}=0;{\frac{\partial T}{\partial \vec{n}}|}_{{\Gamma }_s}=0\right\}.\end{array}$$

Let $V_1$  and $V_2$  be the closure spaces of $\widetilde{{\mathcal{V}}_1}$  in $H^1(\Omega )$  , and $\widetilde{{\mathcal{V}}_2}$  in $H^1(\Omega )$  under $H^1-$  topology, respectively.

Definition 1. Let $v_0,T_0,Q\in L^2(\Omega )$  and let $S$  be a fixed positive time. $(v,T)$  is called a weak solution of ( 26 )–( 31 ) on the time interval $[0,S]$  if it satisfies ( 26 ) and ( 27 ) in weak sense, and also

$$\begin{array}{ccc}& & v\in C\left(\right[0,S],L^2(\Omega \left)\right)\cap L^2\left(\right[0,S],V_1),\end{array}$$

$$\begin{array}{ccc}& & T\in C\left(\right[0,S],L^2(\Omega \left)\right)\cap L^2\left(\right[0,S],V_2),\end{array}$$

$$\begin{array}{ccc}& & \frac{dv}{dt}\in L^1\left(\right[0,S],V_1^{\prime }),\end{array}$$

$$\begin{array}{ccc}& & \frac{dT}{dt}\in L^1\left(\right[0,S],V_2^{\prime }),\end{array}$$

where $V_1^{\prime }$  and $V_2^{\prime }$  are the dual spaces of $V_1$  and $V_2$  , respectively. Moreover, if $v_0\in V_1$  and $T_0\in V_2$  , a weak solution is called strong solution of ( 26 )–( 31 ) on $[0,S]$  if, in addition, it satisfies

$$\begin{array}{ccc}& & v\in C\left(\right[0,S],V_1)\cap L^2\left(\right[0,S],H^2(\Omega \left)\right),\end{array}$$

$$\begin{array}{ccc}& & T\in C\left(\right[0,S],V_2)\cap L^2\left(\right[0,S],H^2(\Omega \left)\right).\end{array}$$

For convenience, we recall the following Sobolev and Ladyzhenskaya's inequalities in ${\mathbb{R}}^2$  ([1] , [9] , [13] , [15] )

$$\begin{array}{ccc}& & \parallel \phi {\parallel }_{L^4\left(M\right)}\le C_0\parallel \phi {\parallel }_{L^2}^{1/2}\parallel \phi {\parallel }_{H^1\left(M\right)}^{1/2},\end{array}$$

(41)

$$\begin{array}{ccc}& & \parallel \phi {\parallel }_{L^8\left(M\right)}\le C_0\parallel \phi {\parallel }_{L^6\left(M\right)}^{3/4}\parallel \phi {\parallel }_{H^1\left(M\right)}^{1/4},\end{array}$$

(42)

for every $\phi \in H^1\left(M\right),$  and the following Sobolev and Ladyzhenskaya's inequalities in ${\mathbb{R}}^3$  ([1] , [9] , [13] , [15] )

$$\begin{array}{ccc}& & \parallel \psi {\parallel }_{L^3(\Omega )}\le C_0\parallel u{\parallel }_{L^2(\Omega )}^{1/2}\parallel u{\parallel }_{H^1(\Omega )}^{1/2},\end{array}$$

(43)

$$\begin{array}{ccc}& & \parallel u{\parallel }_{L^6(\Omega )}\le C_0\parallel u{\parallel }_{H^1(\Omega )},\end{array}$$

(44)

for every $u\in H^1(\Omega ).$  Here $C_0$  is a positive constant which might depend on the shape of $M$  and $\Omega$  but not on their size. And we recall the Poincaré inequalities

$$\begin{array}{ccc}& & \parallel \phi {\parallel }_2^2\le \frac{C_0}{\left|M\right|}\parallel \nabla \phi {\parallel }_2^2,\forall \phi \in V_1,\end{array}$$

(45)

$$\begin{array}{ccc}& & \parallel \psi {\parallel }_2\le h\parallel {\psi }_z{\parallel }_2,\forall \psi \in V_2,\end{array}$$

(46)

$$\begin{array}{ccc}& & \parallel \psi {\parallel }_2\le h\parallel {\psi }_z{\parallel }_2+\parallel \psi (z=0){\parallel }_2\end{array}$$

$$\begin{array}{ccc}& & \le 2{\left(h^{2}R{t}_2+\frac{1}{\alpha }\right)}^{1/2}{\left(\frac{1}{R{t}_2}\parallel {\psi }_z{\parallel }_2^2+\alpha \parallel \psi (z=0){\parallel }_2^2\right)}^{1/2},\forall \psi \in H^1(\Omega ).\end{array}$$

(47)

Here $\left|M\right|$  is the area of $M$  . Moreover, by ( 41 ) we get

$$\begin{array}{ccc}& & \parallel \phi {\parallel }_{L^{12}\left(M\right)}^{12}=\parallel |\phi {|}^3{\parallel }_{L^4\left(M\right)}^4\le C_0\parallel |\phi {|}^3{\parallel }_{L^2\left(M\right)}^2\parallel |\phi {|}^3{\parallel }_{H^1\left(M\right)}^2\end{array}$$

$$\begin{array}{ccc}& & \le C_0\parallel \phi {\parallel }_{L^6\left(M\right)}^6\left({\int }_M|\phi {|}^4{\left|\nabla \phi \right|}^{2}dxdy\right)+\parallel \phi {\parallel }_{L^6\left(M\right)}^{12},\end{array}$$

(48)

for every $\phi \in H^1\left(M\right).$  Also, we recall the integral version of Minkowsky inequality for the $L^p$  spaces, $p\ge 1$  . Let ${\Omega }_1\subset {\mathbb{R}}^{m_1}$  and ${\Omega }_2\subset {\mathbb{R}}^{m_2}$  be two measurable sets, where $m_1$  and $m_2$  are two positive integers.

Suppose that $f(\xi ,\eta )$  is measurable over ${\Omega }_1\times {\Omega }_2$  . Then,

$$\begin{array}{c}{\left[{\int }_{{\Omega }_1}{\left({\int }_{{\Omega }_2}\left|f\right(\xi ,\eta \left)\right|d\eta \right)}^{p}d\xi \right]}^{1/p}\le {\int }_{{\Omega }_2}{\left({\int }_{{\Omega }_1}\left|f\right(\xi ,\eta ){|}^{p}d\xi \right)}^{1/p}d\eta .\end{array}$$

(49)

3 A Priori estimates

In previous subsections we have reformulated the system ( 1 )–( 16 ) to be equivalent to the system ( 26 )–( 31 ). In this section we will get some useful a priori estimates about the solutions of ( 26 )–( 31 ).

The estimates presented here are formal. One can justify them rigorously by establishing them first for a Galerkin approximating system and by using the appropriate “Compactness Theorems” (see, for example, [9] ,[17] ,[18] ,[28] ) to accomplish them for the existence of solutions and the corresponding a priori estimates.

3.1 $L^2$  estimates

By taking the inner product of equation ( 27 ) with $T$  , in $L^2(\Omega )$  , we get

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel T{\parallel }_2^2}{dt}+\frac{1}{R{t}_1}\parallel \nabla T{\parallel }_2^2+\frac{1}{R{t}_2}\parallel T_z{\parallel }_2^2+\alpha \parallel T(z=0){\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }QTdxdydz-{\int }_{\Omega }\left(v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial T}{\partial z}\right)Tdxdydz.\end{array}$$

By integration by parts we get

$$\begin{array}{ccc}& & -{\int }_{\Omega }\left(v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial T}{\partial z}\right)Tdxdydz=0.\end{array}$$

(50)

As a result of the above we conclude

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel T{\parallel }_2^2}{dt}+\frac{1}{R{t}_1}\parallel \nabla T{\parallel }_2^2+\frac{1}{R{t}_2}\parallel T_z{\parallel }_2^2+\alpha \parallel T(z=0){\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }QTdxdydz\le \parallel Q{\parallel }_2\parallel T{\parallel }_2.\end{array}$$

By ( 47 ) and Cauchy–Schwarz inequality we obtain

$$\begin{array}{ccc}& & \frac{d\parallel T{\parallel }_2^2}{dt}+\frac{2}{R{t}_1}\parallel \nabla T{\parallel }_2^2+\frac{1}{R{t}_2}\parallel T_z{\parallel }_2^2+\alpha \parallel T(z=0){\parallel }_2^2\end{array}$$

(51)

$$\begin{array}{ccc}& & \le 4(h^{2}R{t}_2+\frac{1}{\alpha })\parallel Q{\parallel }_2^2.\end{array}$$

(52)

By ( 47 ) and thanks to Gronwall inequality we get

$$\begin{array}{ccc}& & \parallel T{\parallel }_2^2+{\int }_0^t\left[\frac{1}{R{t}_1}\parallel \nabla T\left(s\right){\parallel }_2^2+\frac{1}{R{t}_2}\parallel T_z\left(s\right){\parallel }_2^2+\alpha \parallel T(z=0)\left(s\right){\parallel }_2^2\right]ds\end{array}$$

$$\begin{array}{ccc}& & \le e^{-\frac{t}{4(h^{2}R{t}_2+\frac{1}{\alpha })}}\parallel T_0{\parallel }_2^2+16(h^{2}R{t}_2+\frac{1}{\alpha }{)}^2\parallel Q{\parallel }_2^2,\end{array}$$

(53)

By taking the inner product of equation ( 26 ) with $v$  , in $L^2(\Omega )$  , we reach

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla v{\parallel }_2^2+\frac{1}{R{e}_2}|v_z{\parallel }^2\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }\left[(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}\right]\cdot vdxdydz\end{array}$$

$$\begin{array}{ccc}& & +{\int }_{\Omega }\left(f\vec{k}\times v+\nabla p_s-\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi \right)\right)\cdot vdxdydz.\end{array}$$

By integration by parts we get

$$\begin{array}{ccc}& & {\int }_{\Omega }\left[(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}\right]\cdot vdxdydz=0.\end{array}$$

(54)

By ( 36 ) we have

$$\begin{array}{ccc}& & {\int }_{\Omega }\nabla p_s\cdot vdxdydz=h{\int }_M\nabla p_s\cdot \overline{v}dxdy=-h{\int }_{\Omega }{p}_s(\nabla \cdot \overline{v})dxdy=0.\end{array}$$

(55)

By direct calculation we obtain

$$\begin{array}{ccc}& & f\vec{k}\times v\cdot v=0.\end{array}$$

(56)

Thus, by ( 54 )–( 56 ) we have

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla v{\parallel }_2^2+\frac{1}{R{e}_2}\parallel v_z{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi (\nabla \cdot v)dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le h\parallel T{\parallel }_2\parallel \nabla v{\parallel }_2.\end{array}$$

By Cauchy–Schwarz and ( 53 ) we obtain

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla v{\parallel }_2^2+\frac{1}{R{e}_2}\parallel v_z{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le h^{2}R{e}_1\parallel T{\parallel }_2^2\le h^{2}R{e}_1\left(e^{-\frac{t}{4(h^{2}R{t}_2+\frac{1}{\alpha })}}\parallel T_0{\parallel }_2^2+16(h^{2}R{t}_2+\frac{1}{\alpha }{)}^2\parallel Q{\parallel }_2^2\right).\end{array}$$

By ( 45 ) and ( 46 ) and thanks to Gronwall inequality we get

$$\begin{array}{ccc}& & \parallel v{\parallel }_2^2+{\int }_0^t\left[\frac{1}{R{e}_1}\parallel \nabla v\left(s\right){\parallel }_2^2\frac{1}{R{e}_2}\parallel v_z\left(s\right){\parallel }_2^2\right]ds\end{array}$$

$$\begin{array}{ccc}& & \le e^{-\frac{t}{max\left\{h\right|M|/C_0,hR{e}_2\}}}\left(h\parallel \overline{v_0}{\parallel }_2^2+\parallel \widetilde{v_0}{\parallel }_2^2\right)\end{array}$$

$$\begin{array}{ccc}& & +h^{2}R{e}_1\left(\parallel T_0{\parallel }_2^2+16(h^{2}R{t}_2+\frac{1}{\alpha }{)}^2\parallel Q{\parallel }_2^2\right)max\left\{\right|M|/C_0,hR{e}_2\}.\end{array}$$

(57)

Therefore, by ( 53 ) and ( 57 ) we have

$$\begin{array}{ccc}& & \parallel v\left(t\right){\parallel }_2^2+{\int }_0^t\left[\frac{1}{R{e}_1}\parallel \nabla v\left(s\right){\parallel }_2^2+\frac{1}{R{e}_2}\parallel v_z\left(s\right){\parallel }_2^2\right]ds\end{array}$$

$$\begin{array}{ccc}& & +\parallel T\left(t\right){\parallel }_2^2+{\int }_0^t\left[\frac{1}{R{t}_1}\parallel \nabla T\left(s\right){\parallel }_2^2+\frac{1}{R{t}_2}\parallel T_z\left(s\right){\parallel }_2^2+\alpha \parallel T(z=0)\left(s\right){\parallel }_2^2\right]ds\le K_1,\end{array}$$

(58)

where

$$\begin{array}{ccc}& & K_1=e^{-\frac{t}{max\left\{h\right|M|/C_0,hR{e}_2\}}}\parallel v_0{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & +h^{2}R{e}_1\left(\parallel T_0{\parallel }_2^2+16(h^{2}R{t}_2+\frac{1}{\alpha }{)}^2\parallel Q{\parallel }_2^2\right)\left(1+max\left\{\right|M|/C_0,hR{e}_2\}\right).\end{array}$$

(59)

3.2 $L^6$  estimates

Taking the inner product of the equation ( 38 ) with $|\tilde{v}{|}^4\tilde{v}$  in $L^2(\Omega )$  , we get

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel \tilde{v}{\parallel }_6^6}{dt}+\frac{1}{R{e}_1}{\int }_{\Omega }\left(|\nabla \tilde{v}{|}^2|\tilde{v}{|}^4+{\left|\nabla |\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left(\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^4+{\left|{\partial }_z|\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }\{(\tilde{v}\cdot \nabla )\tilde{v}-\left({\int }_{-h}^z\nabla \cdot \tilde{v}(x,y,\xi ,t)d\xi \right)\frac{\partial \tilde{v}}{\partial z}+(\tilde{v}\cdot \nabla )\overline{v}+(\overline{v}\cdot \nabla )\tilde{v}-\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}\end{array}$$

$$\begin{array}{ccc}& & +f\vec{k}\times \tilde{v}-\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right)\}\cdot |\tilde{v}{|}^4\tilde{v}dxdydz.\end{array}$$

By integration by parts we get

$$\begin{array}{ccc}& & -{\int }_{\Omega }\left[(\tilde{v}\cdot \nabla )\tilde{v}-\left({\int }_{-h}^z\nabla \cdot \tilde{v}(x,y,\xi ,t)d\xi \right)\frac{\partial \tilde{v}}{\partial z}\right]\cdot |\tilde{v}{|}^4\tilde{v}dxdydz=0.\end{array}$$

(60)

By direct calculation we obtain

$$\begin{array}{ccc}& & \left(f\vec{k}\times \tilde{v}\right)\cdot |\tilde{v}{|}^4\tilde{v}=0.\end{array}$$

(61)

By ( 36 ) and boundary condition ( 28 ) we have

$$\begin{array}{ccc}& & {\int }_{\Omega }(\overline{v}\cdot \nabla )\tilde{v}\cdot |\tilde{v}{|}^4\tilde{v}dxdydz=0.\end{array}$$

(62)

Thus, by ( 60 )–( 62 ) we have

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel \tilde{v}{\parallel }_6^6}{dt}+\frac{1}{R{e}_1}{\int }_{\Omega }\left(|\nabla \tilde{v}{|}^2|\tilde{v}{|}^4+{\left|\nabla |\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left(\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^4+{\left|{\partial }_z|\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }\{(\tilde{v}\cdot \nabla )\overline{v}-\overline{(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}}\end{array}$$

$$\begin{array}{ccc}& & -\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right\})\cdot |\tilde{v}{|}^4\tilde{v}dxdydz.\end{array}$$

Notice that by integration by parts and boundary condition ( 28 ) we have

$$\begin{array}{ccc}& & -{\int }_{\Omega }[(\tilde{v}\cdot \nabla )\overline{v}-\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}\end{array}$$

$$\begin{array}{ccc}& & -\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right)]\cdot |\tilde{v}{|}^4\tilde{v}dxdydz\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }[(\nabla \cdot \tilde{v})\overline{v}\cdot |\tilde{v}{|}^4\tilde{v}+(\tilde{v}\cdot \nabla \left)\right(\left|\tilde{v}{|}^4\tilde{v}\right)\cdot \overline{v}-\overline{{\tilde{v}}^k{\tilde{v}}^j}{\partial }_{x_k}\left(\right|\tilde{v}{|}^4{\tilde{v}}^j)\end{array}$$

$$\begin{array}{ccc}& & -\left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi -\frac{1}{h}{\int }_{-h}^0{\int }_{-h}^{z}T(x,y,\xi ,t)d\xi dz\right)\nabla \cdot \left(\right|\tilde{v}{|}^4\tilde{v})]dxdydz.\end{array}$$

Therefore, by Cauchy–Schwarz inequality and Hölder inequality we obtain

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel \tilde{v}{\parallel }_6^6}{dt}+\frac{1}{R{e}_1}{\int }_{\Omega }\left(|\nabla \tilde{v}{|}^2|\tilde{v}{|}^4+{\left|\nabla |\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left(\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^4+{\left|{\partial }_z|\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_M\left[\left|\overline{v}\right|{\int }_{-h}^0|\nabla \tilde{v}||\tilde{v}{|}^{5}dz\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & +C{\int }_M\left[\left({\int }_{-h}^0|\tilde{v}{|}^{2}dz\right)\left({\int }_{-h}^0|\nabla \tilde{v}||\tilde{v}{|}^{4}dz\right)\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & +C{\int }_M\left[\overline{\left|T\right|}{\int }_{-h}^0|\nabla \tilde{v}||\tilde{v}{|}^{4}dz\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_M\left[\left|\overline{v}\right|{\left({\int }_{-h}^0|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dz\right)}^{1/2}{\left({\int }_{-h}^0|\tilde{v}{|}^{6}dz\right)}^{1/2}\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & +C{\int }_M\left[\left({\int }_{-h}^0|\tilde{v}{|}^{2}dz\right){\left({\int }_{-h}^0|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dz\right)}^{1/2}{\left({\int }_{-h}^0|\tilde{v}{|}^{4}dz\right)}^{1/2}\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & +C{\int }_M\left[\overline{\left|T\right|}{\left({\int }_{-h}^0|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dz\right)}^{1/2}{\left({\int }_{-h}^0|\tilde{v}{|}^{4}dz\right)}^{1/2}\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel \overline{v}{\parallel }_{L^4\left(M\right)}{\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{1/2}{\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{6}dz\right)}^{2}dxdy\right)}^{1/4}\end{array}$$

$$\begin{array}{ccc}& & +C{\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{2}dz\right)}^{4}dxdy\right)}^{1/4}{\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{1/2}{\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{4}dz\right)}^{2}dxdy\right)}^{1/4}\end{array}$$

$$\begin{array}{ccc}& & +C\parallel \overline{\left|T\right|}{\parallel }_{L^4\left(M\right)}{\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{1/2}{\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{4}dz\right)}^{2}dxdy\right)}^{1/4}.\end{array}$$

By using Minkowsky inequality ( 49 ), we get

$$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{6}dz\right)}^{2}dxdy\right)}^{1/2}\le C{\int }_{-h}^0{\left({\int }_M|\tilde{v}{|}^{12}dxdy\right)}^{1/2}dz.\end{array}$$

By ( 48 ),

$$\begin{array}{ccc}& & {\int }_M|\tilde{v}{|}^{12}dxdy\le C_0\left({\int }_M|\tilde{v}{|}^{6}dxdy\right)\left({\int }_M\left|\tilde{v}{|}^4\right|\nabla \tilde{v}{|}^{2}dxdy\right)+{\left({\int }_M|\tilde{v}{|}^{6}dxdy\right)}^2.\end{array}$$

Thus, by Cauchy–Schwarz inequality we obtain

$$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{6}dz\right)}^{2}dxdy\right)}^{1/2}\le C\parallel \tilde{v}{\parallel }_{L^6(\Omega )}^3{\left({\int }_{\Omega }\tilde{v}{|}^4|\nabla \tilde{v}{|}^{2}dxdydz\right)}^{1/2}+\parallel \tilde{v}{\parallel }_{L^6(\Omega )}^6.\end{array}$$

(63)

Similarly, by ( 49 ) and ( 42 ), we also get

$$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{4}dz\right)}^{2}dxdy\right)}^{1/2}\le C{\int }_{-h}^0{\left({\int }_M|\tilde{v}{|}^{8}dxdy\right)}^{1/2}dz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_{-h}^0\parallel \tilde{v}{\parallel }_{L^6\left(M\right)}^3\left(\parallel \nabla \tilde{v}{\parallel }_{L^2\left(M\right)}+\parallel \tilde{v}{\parallel }_{L^2\left(M\right)}\right)dz\le C\parallel \tilde{v}{\parallel }_6^3\left(\parallel \nabla \tilde{v}{\parallel }_2+\parallel \tilde{v}{\parallel }_2\right),\end{array}$$

(64)

and

$$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|\tilde{v}{|}^{2}dz\right)}^{4}dxdy\right)}^{1/4}\le C{\int }_{-h}^0{\left({\int }_M|\tilde{v}{|}^{8}dxdy\right)}^{1/4}dz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_{-h}^0\parallel \tilde{v}{\parallel }_{L^6\left(M\right)}^{3/2}\left(\parallel \nabla \tilde{v}{\parallel }_{L^2\left(M\right)}^{1/2}+\parallel \tilde{v}{\parallel }_{L^2\left(M\right)}^{1/2}\right)dz\le C\parallel \tilde{v}{\parallel }_6^{3/2}\left(\parallel \nabla \tilde{v}{\parallel }_2^{1/2}+\parallel \tilde{v}{\parallel }_2^{1/2}\right).\end{array}$$

(65)

Therefore, by ( 63 )–( 65 ) and ( 41 ), we reach

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel \tilde{v}{\parallel }_6^6}{dt}+\frac{1}{R{e}_1}{\int }_{\Omega }\left(|\nabla \tilde{v}{|}^2|\tilde{v}{|}^4+{\left|\nabla |\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left(\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^4+{\left|{\partial }_z|\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel \overline{v}{\parallel }_2^{1/2}\parallel \nabla \overline{v}{\parallel }_2^{1/2}\parallel \tilde{v}{\parallel }_6^{3/2}{\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{3/4}+C\parallel \overline{v}{\parallel }_2^{1/2}\parallel \nabla \overline{v}{\parallel }_2^{1/2}\parallel \tilde{v}{\parallel }_6^6\end{array}$$

$$\begin{array}{ccc}& & +C\parallel \tilde{v}{\parallel }_6^3\left(\parallel \nabla \tilde{v}{\parallel }_2+\parallel \tilde{v}{\parallel }_2\right){\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{1/2}\end{array}$$

$$\begin{array}{ccc}& & +C\parallel \overline{T}{\parallel }_2^{1/2}\parallel \nabla \overline{T}{\parallel }_2^{1/2}\parallel \tilde{v}{\parallel }_6^{3/2}\left(\parallel \nabla \tilde{v}{\parallel }_2^{1/2}+\parallel \tilde{v}{\parallel }_2^{1/2}\right){\left({\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz\right)}^{1/2}.\end{array}$$

By Young's inequality and Cauchy–Schwarz inequality we have

$$\begin{array}{ccc}& & \frac{d\parallel \tilde{v}{\parallel }_6^6}{dt}+\frac{1}{R{e}_1}{\int }_{\Omega }\left(|\nabla \tilde{v}{|}^2|\tilde{v}{|}^4+{\left|\nabla |\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left(\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^4+{\left|{\partial }_z|\tilde{v}{|}^2\right|}^2|\tilde{v}{|}^2\right)dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel \overline{v}{\parallel }_2^2\parallel \nabla \overline{v}{\parallel }_2^2\parallel \tilde{v}{\parallel }_6^6+C\parallel \tilde{v}{\parallel }_6^6\parallel \nabla \tilde{v}{\parallel }_2^2+C\parallel \overline{T}{\parallel }_2^2\parallel \nabla \overline{T}{\parallel }_2^2+C\parallel \tilde{v}{\parallel }_2^2\parallel \tilde{v}{\parallel }_6^6.\end{array}$$

By ( 58 ) and Gronwall inequality, we get

$$\begin{array}{ccc}& & \parallel \tilde{v}\left(t\right){\parallel }_6^6+{\int }_0^t\left(\frac{1}{R{e}_1}{\int }_{\Omega }|\nabla \tilde{v}{|}^2|\tilde{v}{|}^{4}dxdydz+\frac{1}{R{e}_2}{\int }_{\Omega }\left|{\tilde{v}}_z{|}^2\right|{\tilde{v}}_z{|}^{4}dxdydz\right)\le K_6,\end{array}$$

(66)

where

$$\begin{array}{ccc}& & K_6=e^{K_1^{2}t}\left[\parallel v_0{\parallel }_{H^1(\Omega )}^6+K_1^2\right].\end{array}$$

(67)

Taking the inner product of the equation ( 27 ) with $|T{|}^{4}T$  in $L^2(\Omega )$  , and by ( 27 ), we get

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel T{\parallel }_6^6}{dt}+\frac{5}{R{t}_1}{\int }_{\Omega }|\nabla T{|}^2|T{|}^{4}dxdydz+\frac{5}{R{t}_2}{\int }_{\Omega }\left|T_z{|}^2\right|T{|}^{4}dxdydz+\alpha \parallel T(z=0){\parallel }_6^6\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }Q|T{|}^{4}Tdxdydz-{\int }_{\Omega }\left(v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial T}{\partial z}\right)|T{|}^{4}Tdxdydz.\end{array}$$

By integration by parts and ( 36 we get

$$\begin{array}{ccc}& & -{\int }_{\Omega }\left(v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial T}{\partial z}\right)|T{|}^{4}Tdxdydz=0.\end{array}$$

(68)

As a result of the above we conclude

$$\begin{array}{ccc}& & \frac{1}{6}\frac{d\parallel T{\parallel }_6^6}{dt}+\frac{5}{R{t}_1}{\int }_{\Omega }|\nabla T{|}^2|T{|}^{4}dxdydz+\frac{5}{R{t}_2}{\int }_{\Omega }\left|T_z{|}^2\right|T{|}^{4}dxdydz+\alpha \parallel T(z=0){\parallel }_6^6\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }Q|T{|}^{4}Tdxdydz\le \parallel Q{\parallel }_6\parallel T{\parallel }_6^5.\end{array}$$

By Gronwall, again, we get

$$\begin{array}{ccc}& & \parallel T\left(t\right){\parallel }_6\le \parallel Q{\parallel }_{H^1(\Omega )}t+\parallel T_0{\parallel }_{H^1(\Omega )}.\end{array}$$

(69)

3.3 $H^1$  estimates

3.3.1 $\parallel \nabla \overline{v}{\parallel }_2$  estimates

By taking the inner product of equation ( 35 ) with $-\Delta \overline{v}$  in $L^2\left(M\right)$  , and applying ( 36 ), we reach

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel \nabla \overline{v}{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \Delta \overline{v}{\parallel }_2^2={\int }_M\left\{(\overline{v}\cdot \nabla )\overline{v}+\overline{\left[(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}\right]}\right\}\cdot \Delta \overline{v}dxdy+{\int }_{M}f\vec{k}\times \overline{v}\cdot \Delta \overline{v}dxdy.\end{array}$$

Following the proof to $2D$  Navier–Stokes equations (cf. e.g., [9] ) one would get

$$\begin{array}{ccc}& & \left|{\int }_M(\overline{v}\cdot \nabla )\overline{v}\cdot \Delta \overline{v}dxdy\right|\le C\parallel \overline{v}{\parallel }_2^{1/2}\parallel \nabla \overline{v}{\parallel }_2\parallel \Delta \overline{v}{\parallel }_2^{3/2}.\end{array}$$

By Cauchy–Schwarz inequality and Hölder inequality, we get

$$\begin{array}{ccc}& & \left|{\int }_M\overline{(\tilde{v}\cdot \nabla )\tilde{v}+(\nabla \cdot \tilde{v})\tilde{v}}\cdot \Delta \overline{v}dxdy\right|\le C{\int }_M{\int }_{-h}^0\left|\tilde{v}\right||\nabla \tilde{v}|dz|\Delta \overline{v}|dxdy\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_M\left[{\left({\int }_{-h}^0\left|\tilde{v}{|}^2\right|\nabla \tilde{v}|dz\right)}^{1/2}{\left({\int }_{-h}^0|\nabla \tilde{v}|dz\right)}^{1/2}|\Delta \overline{v}|\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & \le C{\left[{\int }_M{\left({\int }_{-h}^0\left|\tilde{v}{|}^2\right|\nabla \tilde{v}|dz\right)}^{2}dxdy\right]}^{1/4}{\left[{\int }_M{\left({\int }_{-h}^0|\nabla \tilde{v}|dz\right)}^{2}dxdy\right]}^{1/4}{\left[{\int }_M|\Delta \overline{v}{|}^{2}dxdy\right]}^{1/2}\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel \nabla \tilde{v}{\parallel }_2^{1/2}{\left({\int }_{\Omega }\left|\tilde{v}{|}^4\right|\nabla \tilde{v}{|}^{2}dxdydz\right)}^{1/4}\parallel \Delta \overline{v}{\parallel }_2.\end{array}$$

Thus, by Young's inequality and Cauchy–Schwarz inequality, we have

$$\begin{array}{ccc}& & \frac{d\parallel \nabla \overline{v}{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \Delta \overline{v}{\parallel }_2^2\le C\parallel \overline{v}{\parallel }_2^2\parallel \nabla \overline{v}{\parallel }_2^4+C\parallel \nabla \tilde{v}{\parallel }_2^2+C{\int }_{\Omega }\left|\tilde{v}{|}^4\right|\nabla \tilde{v}{|}^{2}dxdydz+C\parallel \overline{v}{\parallel }_2^2.\end{array}$$

By ( 58 ), ( 66 ) and thanks to Gronwall inequality and we obtain

$$\begin{array}{ccc}& & \parallel \nabla \overline{v}{\parallel }_2^2+\frac{1}{R{e}_1}{\int }_0^t|\Delta \overline{v}{|}_2^{2}ds\le K_2,\end{array}$$

(70)

where

$$\begin{array}{ccc}& & K_2=e^{K_1^{2}t}\left[\parallel v_0{\parallel }_{H^1(\Omega )}^2+K_1+K_6\right].\end{array}$$

(71)

3.3.2 $\parallel v_z{\parallel }_2$  estimates

Denote by $u=v_z.$  It is clear that $u$  satisfies

$$\begin{array}{ccc}& & \frac{\partial u}{\partial t}+L_{1}u+(v\cdot \nabla )u-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}\end{array}$$

$$\begin{array}{ccc}& & +(u\cdot \nabla )v-(\nabla \cdot v)u+f\vec{k}\times u-\nabla T=0.\end{array}$$

(72)

Taking the inner product of the equation ( 72 ) with $u$  in $L^2$  and using the boundary condition ( 28 ), we get

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel u{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel {\partial }_{z}u{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }\left((v\cdot \nabla )u-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}\right)\cdot udxdydz\end{array}$$

$$\begin{array}{ccc}& & -{\int }_{\Omega }\left((u\cdot \nabla )v-(\nabla \cdot v)u+f\vec{k}\times u-\nabla T\right)\cdot udxdydz.\end{array}$$

By integration by parts we get

$$\begin{array}{ccc}& & -{\int }_{\Omega }\left((v\cdot \nabla )u-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}\right)\cdot udxdydz=0.\end{array}$$

(73)

By direct calculation we obtain

$$\begin{array}{ccc}& & f\vec{k}\times u\cdot u=0.\end{array}$$

(74)

Thus, by ( 73 ) and ( 74 ) we have

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel u{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel {\partial }_{z}u{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }\left((u\cdot \nabla )v-(\nabla \cdot v)u-\nabla T\right)\cdot udxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_{\Omega }\left(\left|v\right|\right)\left|u\right||\nabla u|dxdydz+\parallel T{\parallel }_2\parallel \nabla u{\parallel }_2\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel v{\parallel }_6\parallel u{\parallel }_3\parallel \nabla u{\parallel }_2+\parallel T{\parallel }_2\parallel \nabla u{\parallel }_2\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel v{\parallel }_6\parallel u{\parallel }_2^{1/2}\parallel \nabla u{\parallel }_2^{3/2}+\parallel T{\parallel }_2\parallel \nabla u{\parallel }_2.\end{array}$$

By Young's inequality and Cauchy–Schwarz inequality, we have

$$\begin{array}{ccc}& & \frac{d\parallel u{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel {\partial }_{z}u{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel v{\parallel }_6^4\parallel u{\parallel }_2^2+C\parallel T{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le C\left(\parallel \nabla \overline{v}{\parallel }_2^4+\parallel \tilde{v}{\parallel }_6^4\right)\parallel u{\parallel }_2^2+C\parallel T{\parallel }_2^2.\end{array}$$

By ( 58 ), ( 66 ), ( 70 ), and Gronwall inequality, we get

$$\begin{array}{ccc}& & \parallel v_z{\parallel }_2^2+\frac{1}{R{e}_1}{\int }_0^t\parallel \nabla v_z\left(s\right){\parallel }_2^2+\frac{1}{R{e}_2}{\int }_0^t\parallel v_{zz}\left(s\right){\parallel }_2^{2}ds\le K_z,\end{array}$$

(75)

where

$$\begin{array}{ccc}& & K_z=e^{(K_2^2+K_6^{2/3})t}\left[\parallel v_0{\parallel }_{H^1(\Omega )}^2+K_1\right].\end{array}$$

(76)

3.3.3 $\parallel \nabla v{\parallel }_2$  estimates

By taking the inner product of equation ( 26 ) with $-\Delta v$  in $L^2(\Omega )$  , we reach

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel \nabla v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \Delta v{\parallel }_2^2+\frac{1}{R{e}_2}\parallel \nabla v_z{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & =-{\int }_{\Omega }[(v\cdot \nabla )v-\left({\int }_{-h}^z\nabla \cdot v(x,y,\xi ,t)d\xi \right)\frac{\partial v}{\partial z}\end{array}$$

$$\begin{array}{ccc}& & +f\vec{k}\times v+\nabla p_s-\nabla \left({\int }_{-h}^{z}T(x,y,\xi ,t)d\xi \right)]\cdot \Delta vdxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_{\Omega }\left[\left|v\right||\nabla v|+{\int }_{-h}^0|\nabla v|dz\left|{\tilde{v}}_z\right|+{\int }_{-h}^0|\nabla T|dz\right]|\Delta v|dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel v{\parallel }_{L^6(\Omega )}\parallel \nabla v{\parallel }_{L^3(\Omega )}\parallel \Delta v{\parallel }_2+C{\int }_M\left({\int }_{-h}^0|\nabla v|dz{\int }_{-h}^0\left|v_z\right||\Delta v|dz\right)dxdy+C\parallel \nabla T{\parallel }_2\parallel \Delta v{\parallel }_2.\end{array}$$

Notice that by applying the Proposition 2.2 in [5]  with $u=v,f=\Delta v$  and $g=v_z$  , we get

$$\begin{array}{ccc}& & {\int }_M\left({\int }_{-h}^0|\nabla v|dz{\int }_{-h}^0\left|v_z\right||\Delta v|dz\right)dxdy\le C\parallel \nabla v{\parallel }_2^{1/2}\parallel v_z{\parallel }_2^{1/2}\parallel \nabla v_z{\parallel }_2^{1/2}\parallel \Delta v{\parallel }_2^{3/2}.\end{array}$$

As a result and by ( 43 ) and ( 44 ), we obtain

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel \nabla v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \Delta v{\parallel }_2^2+\frac{1}{R{e}_2}\parallel \nabla v_z{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le C\left(\parallel v{\parallel }_{L^6(\Omega )}+\parallel \nabla v{\parallel }_2^{1/2}\parallel v_z{\parallel }_2^{1/2}\right)\parallel \nabla v{\parallel }_2^{1/2}\parallel \Delta v{\parallel }_2^{3/2}+h\parallel \nabla T{\parallel }_2\parallel \Delta v{\parallel }_2.\end{array}$$

Thus, by Young's inequality and Cauchy–Schwarz inequality, we have

$$\begin{array}{ccc}& & \frac{d\parallel \nabla v{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \Delta v{\parallel }_2^2+\frac{1}{R{e}_2}\parallel \nabla v_z{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le C\left(\parallel v{\parallel }_{L^6(\Omega )}^4+\parallel \nabla v{\parallel }_2^2\parallel v_z{\parallel }_2^2\right)\parallel \nabla v{\parallel }_2^2+C\parallel \nabla T{\parallel }_2^2.\end{array}$$

By ( 58 ), ( 66 ), ( 70 ), ( 75 ) and thanks to Gronwall inequality, we obtain

$$\begin{array}{ccc}& & \parallel \nabla v{\parallel }_2^2+{\int }_0^t\left(\frac{1}{R{e}_1}\parallel \Delta v\left(s\right){\parallel }_2^2+\frac{1}{R{e}_2}\parallel \nabla v_z\left(s\right){\parallel }_2^2\right)ds\le K_V,\end{array}$$

(77)

where

$$\begin{array}{ccc}& & K_V=e^{(K_1^2+K_2+K_6^{2/3}+K_1{K}_z)t}\left[\parallel v_0{\parallel }_{H^1(\Omega )}^2+K_1\right].\end{array}$$

(78)

3.3.4 $\parallel T{\parallel }_{H^1}$  estimates

Taking the inner product of the equation ( 27 ) with $-\Delta T-T_{zz}$  in $L^2(\Omega )$  , we get

$$\begin{array}{ccc}& & \frac{1}{2}\frac{d\left(\parallel \nabla T{\parallel }_2^2+\parallel T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\right)}{dt}\end{array}$$

$$\begin{array}{ccc}& & +\frac{1}{R{t}_1}\parallel \nabla T_z{\parallel }_2^2+\left(\frac{1}{R{t}_1}+\frac{1}{R{t}_2}\right)\left(\parallel \nabla T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\right)+\frac{1}{R{t}_2}\parallel T_{zz}{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & ={\int }_{\Omega }\left[v\cdot \nabla T-\left({\int }_{-h}^z\nabla \cdot vd\xi \right)T_z-Q\right]\left[\Delta T+T_{zz}\right]dxdydz\end{array}$$

$$\begin{array}{ccc}& & \le C{\int }_{\Omega }\left(\left|v\right||\nabla T|+\left|Q\right|\right)\left|\Delta T+T_{zz}\right|dxdydz+{\int }_M\left[{\int }_{-h}^0|\nabla v|dz{\int }_{-h}^0\left|T_z\right|\left|\Delta T+T_{zz}\right|dz\right]dxdy\end{array}$$

$$\begin{array}{ccc}& & \le C\parallel v{\parallel }_6\parallel \nabla T{\parallel }_3{\left(\parallel \Delta T{\parallel }_2^2+\parallel \nabla T_z{\parallel }_2^2+\parallel T_{zz}{\parallel }_2^2\right)}^{1/2}\end{array}$$

$$\begin{array}{ccc}& & +C\parallel \nabla v{\parallel }_2^{1/2}\parallel \Delta v{\parallel }_2^{1/2}\parallel T_z{\parallel }_2^{1/2}{\left(\parallel \Delta T{\parallel }_2^2+\parallel \nabla T_z{\parallel }_2^2+\parallel T_{zz}{\parallel }_2^2\right)}^{3/2}+\parallel Q{\parallel }_2{\left(\parallel \Delta T{\parallel }_2^2+\parallel \nabla T_z{\parallel }_2^2+\parallel T_{zz}{\parallel }_2^2\right)}^{1/2}\end{array}$$

$$\begin{array}{ccc}& & \le C\left[\parallel v{\parallel }_6\parallel \nabla T{\parallel }_2^{1/2}+\parallel \nabla v{\parallel }_2^{1/2}\parallel \Delta v{\parallel }_2^{1/2}\parallel T_z{\parallel }_2^{1/2}\right]{\left(\parallel \Delta T{\parallel }_2^2+\parallel \nabla T_z{\parallel }_2^2+\parallel T_{zz}{\parallel }_2^2\right)}^{3/2}\end{array}$$

$$\begin{array}{ccc}& & +\parallel Q{\parallel }_2{\left(\parallel \Delta T{\parallel }_2^2+\parallel \nabla T_z{\parallel }_2^2+\parallel T_{zz}{\parallel }_2^2\right)}^{1/2}.\end{array}$$

By Young's inequality and Cauchy–Schwarz inequality we have

$$\begin{array}{ccc}& & \frac{d\left(\parallel \nabla T{\parallel }_2^2+\parallel T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\right)}{dt}\end{array}$$

$$\begin{array}{ccc}& & +\frac{1}{R{t}_1}\parallel \nabla T_z{\parallel }_2^2+\left(\frac{1}{R{t}_1}+\frac{1}{R{t}_2}\right)\left(\parallel \nabla T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\right)+\frac{1}{R{t}_2}\parallel T_{zz}{\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & \le C\left(\parallel v{\parallel }_6^4+\parallel \nabla v{\parallel }_2^2\parallel \Delta v{\parallel }_2^2\right)\left(\parallel \nabla T{\parallel }_2^2+\parallel T_z{\parallel }_2^2\right)+C\parallel Q{\parallel }_2^2.\end{array}$$

By ( 66 ), ( 77 ), and Gronwall inequality, we get

$$\begin{array}{ccc}& & \parallel \nabla T{\parallel }_2^2+\parallel T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\end{array}$$

$$\begin{array}{ccc}& & +{\int }_0^t\left[\frac{1}{R{t}_1}\parallel \nabla T_z{\parallel }_2^2+\left(\frac{1}{R{t}_1}+\frac{1}{R{t}_2}\right)\left(\parallel \nabla T_z{\parallel }_2^2+\alpha \parallel \nabla T(z=0){\parallel }_2^2\right)\frac{1}{R{t}_2}\parallel T_{zz}{\parallel }_2^2\right]ds\le K_t,\end{array}$$

(79)

where

$$\begin{array}{ccc}& & K_t=e^{(K_6^2+K_V^2)t}\left[\parallel T_0{\parallel }_{H^1(\Omega )}^2+\parallel Q{\parallel }_2^2\right].\end{array}$$

(80)

4 Existence and Uniqueness of the Strong Solutions

In previous sections we have reformulated the system ( 1 )–( 16 ) and got a priori estimates ( 58 )–( 79 ). In this section we show the global existence and uniqueness of the strong solution of the system ( 26 )–( 31 ).

Theorem 2. Suppose that $Q\in H^1(\Omega )$  . Then for every $v_0\in V_1$  , $T_0\in V_2$  and $S>0,$  there is a unique strong solution $(v,p_s,T)$  of the system ( 26 )–( 31 ), which depends continuously on the initial data.

Remark: The steps of the following proof are formal in the sense that they can be made more rigorous by proving the corresponding estimates first for the Galerkin approximation system. Then the estimates for the exact solution can be established by passing to the limit in the Galerkin procedure by using the appropriate “Compactness Theorems” (see, for example, [9] ,[17] , [18] , and [28] for details).

• Proof. One can show the existence of the strong solution based on the a priori estimates ( 70 ), ( 75 ), ( 77 ) and ( 79 ). Here we only show the continuous dependence on the initial data and the the uniqueness of the strong solutions.
  Let $(v_1,(p_s{)}_1,T_1)$  and $(v_2,(p_s{)}_2,T_2)$  be two strong solutions of the system ( 26 )–( 31 ) with initial data $\left(\right(v_0{)}_1,\left(T_0{)}_1\right)$  and $\left(\right(v_0{)}_2,\left(T_0{)}_2\right)$  , respectively. Denote by $u=v_1-v_2,q_s=(p_s{)}_1-(p_s{)}_2,\theta =T_1-T_2.$  It is clear that

  $$\begin{array}{ccc}& & \frac{\partial u}{\partial t}+L_{1}u+(v_1\cdot \nabla )u+(u\cdot \nabla )v_2-\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}-\left({\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \right)\frac{\partial v_2}{\partial z}\end{array}$$

  $$\begin{array}{ccc}& & +f\vec{k}\times u+\nabla q_s-\nabla \left({\int }_{-h}^z\theta (x,y,\xi ,t)d\xi \right)=0,\end{array}$$

  (81)

  $$\begin{array}{ccc}& & \frac{\partial \theta }{\partial t}+L_2\theta +v_1\cdot \nabla \theta +u\cdot \nabla T_2\end{array}$$

  $$\begin{array}{ccc}& & -\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial \theta }{\partial z}-\left({\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \right)\frac{\partial T_2}{\partial z}=0,\end{array}$$

  (82)

  $$\begin{array}{ccc}& & u(x,y,z,t)=(v_0{)}_1-(v_0{)}_2,\end{array}$$

  (83)

  $$\begin{array}{ccc}& & \theta (x,y,z,0)=(T_0{)}_1-(T_0{)}_2.\end{array}$$

  (84)

  By taking the inner product of equation ( 81 ) with with $u$  in $L^2(\Omega )$  , and equation ( 82 ) with $\theta$  , in $L^2(\Omega )$  we get

  $$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel u{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel u_z{\parallel }_2^2\end{array}$$

  $$\begin{array}{ccc}& & =-{\int }_{\Omega }\left[(v_1\cdot \nabla )u+(u\cdot \nabla )v_2-\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}-\left({\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \right)\frac{\partial v_2}{\partial z}\right]\cdot udxdydz\end{array}$$

  $$\begin{array}{ccc}& & -{\int }_{\Omega }\left[f\vec{k}\times u+\nabla q_s-\nabla \left({\int }_{-h}^z\theta (x,y,\xi ,t)d\xi \right)\right]\cdot udxdydz,\end{array}$$

  and

  $$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel \theta {\parallel }_2^2}{dt}+\frac{1}{R{t}_1}\parallel \nabla \theta {\parallel }_2^2+\frac{1}{R{t}_2}\parallel {\theta }_z{\parallel }_2^2+\alpha \parallel \theta (z=0){\parallel }_2^2\end{array}$$

  $$\begin{array}{ccc}& & =-{\int }_{\Omega }\left[v_1\cdot \nabla \theta +u\cdot \nabla T_2-\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial \theta }{\partial z}-\left({\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \right)\frac{\partial T_2}{\partial z}\right]\theta dxdydz.\end{array}$$

  By integration by parts, and the boundary conditions ( 28 ) and ( 29 ), we get

  $$\begin{array}{ccc}& & -{\int }_{\Omega }\left((v_1\cdot \nabla )u-\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial u}{\partial z}\right)\cdot udxdydz=0,\end{array}$$

  (85)

  $$\begin{array}{ccc}& & -{\int }_{\Omega }\left(v_1\cdot \nabla \theta -\left({\int }_{-h}^z\nabla \cdot v_1(x,y,\xi ,t)d\xi \right)\frac{\partial \theta }{\partial z}\right)\cdot \theta dxdydz=0.\end{array}$$

  (86)

  By direct calculation we obtain

  $$\begin{array}{ccc}& & \left(f\vec{k}\times u\right)\cdot u=0.\end{array}$$

  (87)

  Thus, by ( 85 ), ( 86 ) and ( 87 ) we have

  $$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel u{\parallel }_2^2}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel u_z{\parallel }_2^2\end{array}$$

  $$\begin{array}{ccc}& & =-{\int }_{\Omega }(u\cdot \nabla )v_2\cdot udxdydz+{\int }_{\Omega }{\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \frac{\partial v_2}{\partial z}\cdot udxdydz.\end{array}$$

  and

  $$\begin{array}{ccc}& & \frac{1}{2}\frac{d\parallel \theta {\parallel }_2^2}{dt}+\frac{1}{R{t}_1}\parallel \nabla \theta {\parallel }_2^2+\frac{1}{R{t}_2}\parallel {\theta }_z{\parallel }_2^2+\alpha \parallel \theta (z=0){\parallel }_2^2\end{array}$$

  $$\begin{array}{ccc}& & =-{\int }_{\Omega }(u\cdot \nabla )T_2\theta dxdydz+{\int }_{\Omega }{\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \frac{\partial T_2}{\partial z}\theta dxdydz.\end{array}$$

  Notice that

  $$\begin{array}{ccc}& & \left|{\int }_{\Omega }(u\cdot \nabla )v_2\cdot udxdydz\right|\le \parallel \nabla v_2{\parallel }_2\parallel u{\parallel }_3\parallel u{\parallel }_6\le C\parallel \nabla v_2{\parallel }_2\parallel u{\parallel }_2^{1/2}\parallel \nabla u{\parallel }_2^{3/2},\end{array}$$

  (88)

  $$\begin{array}{ccc}& & \left|{\int }_{\Omega }(u\cdot \nabla )T_2\theta dxdydz\right|\le \parallel \nabla v_2{\parallel }_2\parallel \theta {\parallel }_3\parallel u{\parallel }_6\le C\parallel \nabla T_2{\parallel }_2\parallel \theta {\parallel }_2^{1/2}\parallel \nabla \theta {\parallel }_2^{1/2}\parallel \nabla u{\parallel }_2.\end{array}$$

  (89)

  Moreover,

  $$\begin{array}{ccc}& & \left|{\int }_{\Omega }{\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \frac{\partial v_2}{\partial z}\cdot udxdydz\right|\le {\int }_M\left({\int }_{-h}^0|\nabla u|dz{\int }_{-h}^0\left|{\partial }_z{v}_2\right|\left|u\right|dz\right)dxdy\end{array}$$

  $$\begin{array}{ccc}& & \le {\int }_M\left({\int }_{-h}^0|\nabla u|dz{\left({\int }_{-h}^0|{\partial }_z{v}_2{|}^{2}dz\right)}^{1/2}{\left({\int }_{-h}^0|u{|}^{2}dz\right)}^{1/2}\right)dxdy\end{array}$$

  $$\begin{array}{ccc}& & \le {\left({\int }_M{\left({\int }_{-h}^0|\nabla u|dz\right)}^{2}dxdy\right)}^{1/2}{\left({\int }_M{\left({\int }_{-h}^0|{\partial }_z{v}_2{|}^{2}dz\right)}^{2}dxdy\right)}^{1/4}{\left({\int }_M{\left({\int }_{-h}^0|u{|}^{2}dz\right)}^{2}dxdy\right)}^{1/4}.\end{array}$$

  By Cauchy–Schwarz inequality, we get

  $$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|\nabla u|dz\right)}^{2}dxdy\right)}^{1/2}\le C\parallel \nabla u{\parallel }_2.\end{array}$$

  (90)

  By using Minkowsky inequality ( 49 ) and ( 41 ), we obtain

  $$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|u{|}^{2}dz\right)}^{2}dxdy\right)}^{1/2}\le C{\int }_{-h}^0{\left({\int }_M|u{|}^{4}dxdy\right)}^{1/2}dz\end{array}$$

  $$\begin{array}{ccc}& & \le C{\int }_{-h}^0\left|u\right||\nabla u|dz\le C\parallel u{\parallel }_2\parallel \nabla u{\parallel }_2,\end{array}$$

  (91)

  and

  $$\begin{array}{ccc}& & {\left({\int }_M{\left({\int }_{-h}^0|{\partial }_z{v}_2{|}^{2}dz\right)}^{2}dxdy\right)}^{1/2}\le C{\int }_{-h}^0{\left({\int }_M|{\partial }_z{v}_2{|}^{4}dxdy\right)}^{1/2}dz\end{array}$$

  $$\begin{array}{ccc}& & \le C{\int }_{-h}^0\left|{\partial }_z{v}_2\right||\nabla {\partial }_z{v}_2|dz\le C\parallel {\partial }_z{v}_2{\parallel }_2\parallel \nabla {\partial }_z{v}_2{\parallel }_2.\end{array}$$

  (92)

  Similarly, we have

  $$\begin{array}{ccc}& & \left|{\int }_{\Omega }{\int }_{-h}^z\nabla \cdot u(x,y,\xi ,t)d\xi \frac{\partial T_2}{\partial z}\theta dxdydz\right|\le C\parallel \nabla u{\parallel }_2\parallel {\partial }_z{T}_2{\parallel }_2^{1/2}\parallel \nabla {\partial }_z{T}_2{\parallel }_2^{1/2}\parallel \theta {\parallel }_2^{1/2}\parallel \nabla \theta {\parallel }_2^{1/2}.\end{array}$$

  (93)

  Therefore, by estimates ( 88 )–( 93 ), we reach

  $$\begin{array}{ccc}& & \frac{1}{2}\frac{d\left(\parallel u{\parallel }_2^2+\parallel \theta {\parallel }_2^2\right)}{dt}+\frac{1}{R{e}_1}\parallel \nabla u{\parallel }_2^2+\frac{1}{R{e}_2}\parallel u_z{\parallel }_2^2+\frac{1}{R{t}_1}\parallel \nabla \theta {\parallel }_2^2+\frac{1}{R{t}_2}\parallel {\theta }_z{\parallel }_2^2+\alpha \parallel \theta (z=0){\parallel }_2^2\end{array}$$

  $$\begin{array}{ccc}& & \le C\left(\parallel \nabla v_2{\parallel }_2+\parallel {\partial }_z{v}_2{\parallel }_2^{1/2}\parallel \nabla {\partial }_z{v}_2{\parallel }_2^{1/2}\right)\parallel u{\parallel }^{1/2}\parallel \nabla u{\parallel }_2^{3/2}\end{array}$$

  $$\begin{array}{ccc}& & +C\parallel \nabla T_2{\parallel }_2\parallel \theta {\parallel }_2^{1/2}\parallel \nabla \theta {\parallel }_2^{1/2}\parallel \nabla u{\parallel }_2+C\parallel \nabla u{\parallel }_2\parallel {\partial }_z{T}_2{\parallel }_2^{1/2}\parallel \nabla {\partial }_z{T}_2{\parallel }_2^{1/2}\parallel \theta {\parallel }_2^{1/2}\parallel \nabla \theta {\parallel }_2^{1/2}.\end{array}$$

  By Young's inequality, we get

  $$\begin{array}{ccc}& & \frac{d\parallel u{\parallel }_2^2}{dt}\le C\left(\parallel \nabla v_2{\parallel }_2^4+\parallel \nabla T_2{\parallel }_2^4+\parallel {\partial }_z{v}_2{\parallel }_2^2\parallel \nabla {\partial }_z{v}_2{\parallel }_2^2+\parallel {\partial }_z{T}_2{\parallel }_2^2\parallel \nabla {\partial }_z{T}_2{\parallel }_2^2\right)\left(\parallel u{\parallel }_2^2+\parallel \theta {\parallel }_2^2\right).\end{array}$$

  Thanks to Gronwall inequality, we obtain

  $$\begin{array}{ccc}& & \parallel u\left(t\right){\parallel }_2^2+\parallel \theta \left(t\right){\parallel }_2^2\le \left(\parallel u(t=0){\parallel }_2^2+\parallel \theta (t=0){\parallel }_2^2\right)\times \end{array}$$

  $$\begin{array}{ccc}& & exp\left\{C{\int }_0^t\left(\parallel \nabla v_2\left(s\right){\parallel }_2^4+\parallel \nabla T_2\left(s\right){\parallel }_2^4+\parallel {\partial }_z{v}_2\left(s\right){\parallel }_2^2\parallel \nabla {\partial }_z{v}_2\left(s\right){\parallel }_2^2+\parallel {\partial }_z{T}_2\left(s\right){\parallel }_2^2\parallel \nabla {\partial }_z{T}_2\left(s\right){\parallel }_2^2\right)ds\right\}.\end{array}$$

  Since $(v_2,T_2)$  is a strong solution, we have

  $$\begin{array}{ccc}& & \parallel u\left(t\right){\parallel }_2^2+\parallel \theta \left(t\right){\parallel }_2^2\le \left(\parallel u(t=0){\parallel }_2^2+\parallel \theta (t=0){\parallel }_2^2\right)exp\left\{C\left(K_V^{2}t+K_t^{2}t+K_z{K}_V+K_t^2\right)\right\}.\end{array}$$

  The above inequality proves the continuous dependence of the solutions on the initial data, and in particular, when $u(t=0)=\theta (t=0)=0$  , we have $u\left(t\right)=\theta \left(t\right)=0,$  for all $t\ge 0$  . Therefore, the strong solution is unique.

  □

Acknowledgments

This work was supported in part by the NSF grant No. DMS–0204794, the MAOF Fellowship of the Israeli Council of Higher Education, and by the USA Department of Energy, under contract number W–7405–ENG–36 and ASCR Program in Applied Mathematical Sciences. References

1. R.A. Adams, Sobolev Spaces, Academic Press, New York, 1975.
2. A. Babin, A. Mahalov and B. Nicolaenko, Fast singular oscillating limits and global regularity for the $3D$  primitive equations of geophysics, Math. Model. Numer. Anal. 34 (2000), 201–222.
3. V. Barcilon, P. Constantin and E.S. Titi, Existence of solutions to the Stommel-Charney model of the Gulf Stream, SIAM J. Math. Anal. 19 (1988), 1355–1364.
4. G. Browning, A. Kasahara, H.-O. Kreiss, Initialization of the primitive equations by the bounded derivative method, J. Atmospheric Sci. 37 (1980), 1424–136.
5. C. Cao and E.S. Titi, Global Well–posedness and Finite Dimensional Global Attractor for a 3–D Planetary Geostrophic Viscous Model, Comm. Pure Appl. Math. 56, 198-233, 2003
6. C. Cao, E.S. Titi and M. Ziane, A “horizontal” hyper–diffusion $3-D$  thermocline planetary geostrophic model: well-posedness and long time behavior, Nonlinearity 17, 1749–1776, 2004.
7. J.G. Charney, The use of the primitive equations of motion in numerical prediction, Tellus 7 (1955), 22–26.
8. J.G. Charney, The gulf stream as an inertial boundary layer, Proc. Nat. Acad. Sci. U.S.A., 41(1955), 731–740.
9. P. Constantin and C. Foias, Navier-Stokes Equations, The University of Chicago Press, 1988.
10. P. Constantin, A. Majda and E. Tabak, Formation of strong fronts in the $2$  -D quasigeostrophic thermal active scalar, Nonlinearity 7 (1994), 1495–1533.
11. P. Constantin, A. Majda and E. Tabak, Singular front formation in a model for quasigeostrophic flow, Phys. Fluids 6 (1994), 9–11.
12. P.F. Embid and A.J. Majda, Low Froude number limiting dynamics for stably stratified flow with small or finite Rossby numbers, Geophys. Astrophys. Fluid Dynam. 87 (1998), 1–50.
13. G.P. Galdi, An Introduction to the Mathematical Theory of the Navier-Stokes Equations, Vol. I & II, Springer-Verlag, 1994.
14. C. Hu, R. Temam and M. Ziane, The primitive equations on the large scale ocean under the small depth hypothesis, Discrete Contin. Dyn. Syst. 9 (2003), 97–131.
15. O.A. Ladyzhenskaya, The Boundary Value Problems of Mathematical Physics, Springer-Verlag, 1985.
16. Ladyženskaja, O. A., Solonnikov, V. A. and Ural'ceva, N. N., Linear and Quasilinear Equations of Parabolic Type, American Mathematical Society, Providence, R.I., (1967).
17. J.L. Lions, R. Temam and S. Wang, New formulations of the primitive equations of atmosphere and applications, Nonlinearity 5 (1992), 237–288.
18. J.L. Lions, R. Temam and S. Wang, On the equations of the large scale Ocean, Nonlinearity 5 (1992), 1007–1053.
19. J. Pedlosky, Geophysical Fluid Dynamics, Springer-Verlag, New York, 1987.
20. L.F. Richardson, Weather Prediction by Numerical Process, Cambridge University Press, Cambridge 1922 (reprint, Dover, New York, 1988).
21. R. Samelson, Coastal boundary conditions and the baroclinic structure of wind–driven continental shelf currents, Journal of Physical Oceanography, 27(1997), 2645-2662.
22. R. Samelson, R. Temam and S. Wang, Some mathematical properties of the planetary geostrophic equations for large scale ocean circulation, Applicable analysis, 70(1998), 147-173.
23. R. Samelson, R. Temam and S. Wang, Remarks on the planetary geostrophic model of gyre scale ocean circulation, Differential and Integral Equation, 13 (2000), 1–14.
24. R. Samelson and G. Vallis, A simple friction and diffusion scheme for planetary geostrophic basin models, J. Phys. Oceanogr., 27 (1997), 186–194
25. R. Samelson and G. Vallis, Large-scale circulation with small diapycnal diffusion: The two-thermocline limit, J. Marine Res., 55 (1997), 223–275
26. D. Seidov, An intermediate model for large–scale ocean circulation studies, Dynamics of Atmospheres and Oceans, 25(1996), 25-55.
27. H. Stommel, The westward intensification of wind–driven ocean currents, Trans. Amer. Geophys. union, 29(1948), 291–304.
28. R. Temam, Navier-Stokes Equations, Theory and Numerical Analysis, North-Holland, 1984.
29. R. Temam and M. Ziane, Some mathematical problems in geophysical fluid dynamics, Handbook of mathematical fluid dynamics, 2003

(C. Cao) Department of Mathematics, Florida International University, University Park, Miami, FL 33199, USA. E-mail address : caoc@fiu.edu (E.S. Titi) Department of Mathematics, and Department of Mechanical and Aerospace Engineering, University of California, Irvine, CA 92697-3875, USA. Also: Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot 76100, Israel. E-mail address : etiti@math.uci.edu E-mail address : edriss.titi@weizmann.ac.il