Skip to main content

Stability of weak solutions for the large-scale atmospheric equations

Abstract

In this paper, we consider the Navier-Stokes equations and temperature equation arising from the evolution process of the atmosphere. Under certain assumptions imposed on the initial data, we show the L 1 -stability of weak solutions for the atmospheric equations. Some ideas and delicate estimates are introduced to prove these results.

MSC:35Q35, 76D03.

1 Introduction and main results

In this paper, we consider the atmospheric motion model under the constant external force and without the effects of topography, and the aerosphere is regarded as a spherical shell encompassing the earth. We introduce a moving frame running with the earth, (λ,θ,p), where λ∈[0,2π] is the longitude, θ∈[0,π] is the colatitude, and p∈[ p 0 , p s ] is the atmospheric pressure, which can be used instead of the geocentric distance r because it is strictly monotonically decreasing function for r, where p s is the atmospheric pressure on the surface of the earth, and p 0 >0 is the atmospheric pressure at a certain isobaric surface. In the coordinate system consisting of the moving frame and time, the atmospheric state functions are defined by the atmospheric horizontal velocity V=( v λ , v θ ), the rate of pressure w= d p d t , the temperature T, and the geopotential Φ. All of them satisfy the following system:

{ ∂ V ∂ t + ( V ⋅ ∇ ) V + w ∂ V ∂ p + ( 2 ω cos θ + cot θ a v λ ) β V + ∇ Φ = μ 1 Δ V + ν 1 ∂ ∂ p ( α 2 ( p ) ∂ V ∂ p ) , ∂ T ∂ t + ( V ⋅ ∇ ) T + w ∂ T ∂ p − c 0 2 w R p = c 0 2 μ 2 R 2 Δ T + c 0 2 ν 2 R 2 ∂ ∂ p ( α 2 ( p ) ∂ T ∂ p ) + Ψ c p , ∇ ⋅ V + ∂ w ∂ p = 0 , ∂ Φ ∂ p + R T p = 0 ,
(1.1)

with the initial data

V | t = 0 = V 0 ,T | t = 0 = T 0 ,ω | t = 0 = ω 0 ,
(1.2)

where ω is the angular velocity of the earth; c 0 , c p , and R are the thermodynamics parameters; μ i and ν i , i=1,2, are the diffusion coefficients; α(p)∈C[ p 0 , p s ] satisfying α(p)≥ C α >0, namely the diffusion is related to the atmospheric pressure; Ψ is the diabatic heating of the atmosphere, which is a function of (λ,θ,p) and stands for the effect of the constant external force on the atmospheric system. We have β= ( 0 1 − 1 0 ) in 2ωcosθβV=2ωcosθ k → ∧V, which denotes the Coriolis force on the atmosphere. The differential operators grad=∇ and div=∇⋅ on the spherical surface have the following form:

{ ∇ = ( 1 a sin θ ∂ ∂ λ , 1 a ∂ ∂ θ ) , ∇ ⋅ V = 1 a sin θ ∂ v λ ∂ λ + 1 a sin θ ∂ ( sin θ v θ ) ∂ θ , Δ = 1 a 2 sin θ ∂ ∂ θ ( sin θ ∂ ∂ θ ) + 1 a 2 sin 2 θ ∂ 2 ∂ λ 2 ,
(1.3)

where a is the radius of the earth. The vertical scale of the atmosphere is very much smaller compared with the radius of the earth, so the geocentric distance r is replaced by the radius of the earth in the differential operators. The above equations are studied on Ω×[0,M]:=[0,2π]×[0,π]×[ p 0 , p s ]×[0,M], where M>0.

The boundary conditions without the relief are: All the functions are 2π periodical w.r.t. λ, π periodical w.r.t. θ, and

{ v λ ∂ p | p = p 0 = v θ ∂ p | p = p 0 = T ∂ p | p = p 0 = w | p = p 0 = 0 , v λ | p = p s = v θ | p = p s = T ∂ p | p = p s = w | p = p s = Φ | p = p s = 0 .
(1.4)

There are many important results achieved on the atmospheric problem. Zeng [1], Li and Chou [2] have made important progress on the formulation and the analysis of the models. For different research purposes, different atmospheric models have been investigated by Pedlosky [3], Washington and Parkinson [4], Lions et al. [5–8] and references therein. Recently, Chepzhov and Vishik [9] introduced the atmospheric equations considered in this paper. Huang and Guo [10] proved the existence of the weak solutions to the atmospheric equations by the basic differential equation theory and the existence of the corresponding trajectory attractors, from which the existence of the atmospheric global attractors follows. Furthermore, Huang and Guo [11] studied the model of the climate for weather forecasts in which the pressing force of topography on atmosphere and the divergent effect of airflow are included, and they proved the existence and the asymptotic behaviors of the weak solution.

The rest of the paper is as follows. In Section 2, the main results about the L 1 -stability of weak solutions to the Navier-Stokes equations and temperature equation are stated. In Section 3, we will give several important a priori estimates. Then we will justify the stability of the weak solutions in Section 4. Finally, in Section 5, the conclusion will be given.

2 Main results

The L 1 -stability theory of weak solutions to (1.1) will be considered, and there is a simple version of system (1.1). From (1.1)3,4 and the boundary conditions (1.4), we have

Φ(p)=R ∫ p p s 1 s T(s)ds
(2.1)

and

w(p)=∇⋅ ∫ p p s V(s)ds,
(2.2)

which implies

∇⋅ ∫ p 0 p s V(s)ds=0.
(2.3)

Substitute (2.1) and (2.2) into (1.1)1,2 and define the unknown function U:=(V,T), then we have the simplification of system (1.1):

{ ∂ V ∂ t + ( V ⋅ ∇ ) V + ∇ ⋅ ∫ p p s V ( s ) d s ∂ V ∂ p + ( 2 ω cos θ + cot θ a v λ ) β V + R ∇ ∫ p p s 1 s T ( s ) d s = μ 1 Δ V + ν 1 ∂ ∂ p ( α 2 ( p ) ∂ V ∂ p ) , ∂ T ∂ t + ( V ⋅ ∇ ) T + ∇ ⋅ ∫ p p s V ( s ) d s ∂ T ∂ p − c 0 2 R p ∇ ⋅ ∫ p p s V ( s ) d s = c 0 2 μ 2 R 2 Δ T + c 0 2 ν 2 R 2 ∂ ∂ p ( α 2 ( p ) ∂ T ∂ p ) + Ψ c p , U | t = 0 = ( v λ , v θ , T ) | t = 0 = U 0 = ( v λ 0 , v θ 0 , T 0 ) , U ( λ , θ , p ) = U ( λ + 2 π , θ , p ) = U ( λ , θ + π , p ) , ∂ U ∂ p | p = p 0 = 0 , V | p = p s = 0 , ∂ T ∂ p | p = p s = 0 .
(2.4)

Denote

L(U):= ( − μ 1 Δ V − ν 1 ∂ ∂ p ( α 2 ( p ) V ∂ p ) , − c 0 2 μ 2 R 2 Δ T − c 0 2 ν 2 R 2 ∂ ∂ p ( α 2 ( p ) ∂ T ∂ p ) ) ,
(2.5)
B ( U , U ) : = ( ( V ⋅ ∇ ) V + ∇ ⋅ ∫ p p s V ( s ) d s ∂ V ∂ p B ( U , U ) : = + ( 2 ω cos θ + cot θ a v λ ) β V + R ∇ ∫ p p s 1 s T ( s ) d s , B ( U , U ) : = ( V ⋅ ∇ ) T + ∇ ⋅ ∫ p p s V ( s ) d s ∂ T ∂ p B ( U , U ) : = − c 0 2 R p ∇ ⋅ ∫ p p s V ( s ) d s ) T
(2.6)

and

F:= ( 0 , Ψ c p ) ,
(2.7)

then we show the definition of weak solutions of system (2.4).

Definition 2.1 (Definition of weak solution)

For any M>0, U is said to be a weak solution of system (2.4) on Ω×[0,M], if U has the following regularities:

U∈ L ∞ ( 0 , M ; L 2 ( Ω ) ) ∩ L 2 ( 0 , M ; H 1 ( Ω ) ) ,
(2.8)

and satisfies the equations in the sense of distributions

U t +L(U)+B(U,U)=F,in  D ′ ( ( 0 , M ) × Ω ) .
(2.9)

Namely, we have for all φ=( φ v λ , φ v θ , φ T )=( φ V , φ T )∈ C ∞ (0,M; C 0 ∞ (Ω)) and φ(M,⋅)=0,

( U 0 , φ ( 0 , ⋅ ) ) + ∫ 0 M (U, φ t )dt− ∫ 0 M ( a ( U , φ ) + b ( U , U , φ ) − ( F , φ ) ) dt=0,
(2.10)

where (⋅,⋅) is the inner product of L 2 (Ω),

a ( U , φ ) = μ 1 ∫ Ω ∇ V ⋅ ∇ φ V d σ d p + c 0 2 μ 2 R 2 ∫ Ω ∇ T ⋅ ∇ φ T d σ d p + ν 1 ∫ Ω α 2 ( p ) ∂ V ∂ p ⋅ ∂ φ V ∂ p d σ d p + c 0 2 ν 2 R 2 ∫ Ω α 2 ( p ) ∂ T ∂ p ∂ φ T ∂ p d σ d p
(2.11)

and

b ( U , U , φ ) = ∫ Ω ( ( V ⋅ ∇ ) V ⋅ φ V + ( V ⋅ ∇ ) T φ T ) d σ d p + ∫ Ω ( ( ∇ ⋅ ∫ p p s V ( s ) d s ) ∂ V ∂ p ⋅ φ V + ( ∇ ⋅ ∫ p p s V ( s ) d s ) ∂ T ∂ p φ T ) d σ d p + ∫ Ω ( R ∇ ∫ p p s 1 s T ( s ) d s ⋅ φ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V ( s ) d s ) φ T ) d σ d p + ∫ Ω ( 2 ω cos θ + cot θ a v λ ) ( v θ φ v λ − v λ φ v θ ) d σ d p ,
(2.12)

where

∫ Ω fdσdp:= ∫ p 0 p s ∫ 0 π ∫ 0 2 π f a 2 sinθdλdθdp.
(2.13)

Then we can state the main results of the present paper as follows.

Theorem 2.1 (Stability of weak solutions)

For any M>0, let U n =( v λ n , v θ n , T n )=( V n , T n ) be a sequence of weak solution of system (2.4) subject to the initial data

U n | t = 0 = ( v λ n , v θ n , T n ) | t = 0 = U 0 n = ( v λ 0 n , v θ 0 n , T 0 n ) ,
(2.14)

and U 0 n be such that

U 0 n → U 0 =( v λ 0 , v θ 0 , T 0 )∈ L 1 (Ω),
(2.15)

where U 0 ∈ L 2 (Ω) and satisfies the following upper bound uniformly with respect to n∈N:

∫ Ω | U 0 n | 2 dσdp= ∫ Ω ( | v λ 0 n | 2 + | v θ 0 n | 2 + | T 0 n | 2 ) dσdp<C,
(2.16)

where C>0 denotes a constant, and we assume that Ψ∈ H − 2 (Ω).

Then, up to a subsequence, still denoted by the same symbol, we have

U n →U∈ L 2 ( 0 , T ; L 2 ( Ω ) ) ,
(2.17)

where U=( v λ , v θ ,T) is a weak solution of system (2.4) with the initial data U 0 =( v λ 0 , v θ 0 , T 0 ).

Remark 2.1 Note that from (2.17), we can find that if

U 0 n → U 0 ∈ L 1 (Ω),
(2.18)

then

U n →U∈ L 1 ( 0 , T ; L 1 ( Ω ) ) ,
(2.19)

which is the L 1 -stability of weak solutions for the system.

Remark 2.2 Furthermore if

U 0 n → U 0 a.e.
(2.20)

and

U 0 n , U 0 ∈ L 2 (Ω),
(2.21)

from the Egorov theorem, we have for ϵ>0 the following. Let δ= ϵ 2 , then there exists a domain Ω δ ⊂Ω, such that |Ω/ Ω δ |<δ, and for all (λ,θ,p)∈ Ω δ , ∃N>0, for ∀n>N, we have

| U 0 n → U 0 |<ϵ.
(2.22)

Then for ∀n>N we have

∫ Ω | U 0 n − U 0 |dσdp= ∫ Ω δ | U 0 n − U 0 |dσdp+ ∫ Ω / Ω δ | U 0 n − U 0 |dσdp,
(2.23)

where we have

∫ Ω δ | U 0 n − U 0 |dσdp≤| Ω δ |ϵ≤Cϵ
(2.24)

and

∫ Ω / Ω δ | U 0 n − U 0 | d σ d p ≤ ( ∫ Ω / Ω δ 1 d σ d p ) 1 2 ( ∫ Ω / Ω δ | U 0 n − U 0 | 2 d σ d p ) 1 2 ≤ C δ 1 2 = C ϵ ,
(2.25)

and from (2.24) and (2.25), we can find

U 0 n → U 0 ∈ L 1 (Ω),
(2.26)

thus, we have from Remark 2.1

U n →U∈ L 1 ( 0 , T ; L 1 ( Ω ) ) ,
(2.27)

which implies that

U n →Ua.e.,
(2.28)

which means the weak solutions are stable almost everywhere.

3 The a priori estimates

Next, we will give the a priori estimates for the weak solution U n to system (2.4). Firstly, from a direct calculation, we can establish the following lemma; we omit the proof.

Lemma 3.1

∫ Ω ∇ϕ⋅∇ψdσdp=− ∫ Ω Δϕ⋅ψdσdp,
(3.1)

where ϕ and ψ can be vector-valued functions, or the scalar functions,

∫ Ω ϕ⋅∇ψdσdp=− ∫ Ω ∇⋅ϕψdσdp,
(3.2)

where ϕ is a vector-valued function, and ψ is a scalar function.

Then we have the usual energy inequality as follows.

Lemma 3.2 Let T>0. Under the assumptions of Theorem  2.1, we have for the weak solution U n to system (2.4)

∥ U n ∥ L 2 ( Ω ) 2 + ∫ 0 t ∥ U n ∥ H 1 ( Ω ) 2 dτ≤C ( 1 + ∥ Ψ ∥ H − 1 ( Ω ) 2 ) ,t∈[0,M],
(3.3)

where C>0 denotes a constant dependent on the initial data and time M and independent of n.

Proof Take the inner product of (2.4) with U n , integrating on Ω, we have

( U t n , U n ) + ( L ( U n ) , U n ) + ( B ( U n , U n ) , U n ) = ( F , U n ) ,
(3.4)

and using the boundary conditions, we have

d d t ∫ Ω ( V n 2 + T n 2 ) d σ d p + μ 1 ∫ Ω | ∇ V n | 2 d σ d p + ν 1 ∫ Ω α 2 ( p ) ( ∂ V n ∂ p ) 2 d σ d p + c 0 2 μ 2 R 2 ∫ Ω | ∇ T n | 2 d σ d p + c 0 2 ν 2 R 2 ∫ Ω α 2 ( p ) ( ∂ T n ∂ p ) 2 d σ d p = ∫ Ω Ψ c p T n d σ d p ,
(3.5)

which implies

d d t ∥ U n ∥ L 2 ( Ω ) 2 + C ∥ U n ∥ H 1 ( Ω ) ≤ C ∥ Ψ ∥ H − 1 ( Ω ) 2 + ϵ C ∥ T n ∥ H 1 ( Ω ) 2 ≤ C ∥ Ψ ∥ H − 1 ( Ω ) 2 + ϵ C ∥ U n ∥ H 1 ( Ω ) 2 ,
(3.6)

where ϵ>0 is a small constant such that we have

d d t ∥ U n ∥ L 2 ( Ω ) 2 +C ∥ U n ∥ H 1 ( Ω ) 2 ≤C ∥ Ψ ∥ H − 1 ( Ω ) 2 ,
(3.7)

after the integration with respect to t∈[0,M], we have

∥ U n ∥ L 2 ( Ω ) 2 + ∫ 0 t ∥ U n ∥ H 1 ( Ω ) 2 d τ ≤ ∥ U 0 n ∥ L 2 ( Ω ) 2 + C ∥ Ψ ∥ H − 1 ( Ω ) 2 ≤ C ( 1 + ∥ Ψ ∥ H − 1 ( Ω ) 2 ) ,
(3.8)

where C>0 denotes a constant dependent of the initial data and time M and independent of n. □

4 Proof of main results

With the help of the a priori estimates in (3.3), we have the following estimates for the sequence of weak solutions U n :

U n ∈ L ∞ ( 0 , M ; L 2 ( Ω ) ) ∩ L 2 ( 0 , M ; H 1 ( Ω ) ) ,
(4.1)

then we will prove the main results, in order to address the convergence of sequence of the weak solution; a lemma of the compactness result will be given first.

Lemma 4.1 (Lion’s compactness result)

Suppose E 0 , E, E 1 are Banach spaces, E 0 ↪↪E↪ E 1 , which means E 0 is compactly embedded in E, E is embedded in E 1 , and p 1 >1. Denote

W 2 , p 1 (0,M; E 0 , E 1 )= { ψ | ψ ∈ L 2 ( 0 , M ; E 0 ) , ψ t ∈ L p 1 ( 0 , M ; E 1 ) }
(4.2)

as the Banach space with the norm

∥ ψ ∥ W 2 , p 1 = ∥ ψ ∥ L 2 ( 0 , M ; E 0 ) + ∥ ψ ∥ L p 1 ( 0 , M ; E 1 ) ,
(4.3)

then

W 2 , p 1 (0,M; E 0 , E 1 )↪↪ L 2 (0,M;E).
(4.4)

Then we will give the proof of the stability of the weak solutions.

Lemma 4.2 Let U n be the weak solution sequence of system (2.4). Then, up to a subsequence, we have

U n →U,in  L 2 ( 0 , M ; L 2 ( Ω ) )
(4.5)

and

∫ 0 M ( U n , φ t ) dt→ ∫ 0 M (U, φ t )dt,
(4.6)

for all φ=( φ v λ , φ v θ , φ T )=( φ V , φ T )∈ C ∞ (0,M; C 0 ∞ (Ω)) and φ(M,⋅)=0.

Proof From (4.1), we have the following estimates for the test function ϕ=( ϕ v λ , ϕ v θ , ϕ T )=( ϕ V , ϕ T )∈ H 2 (Ω):

( L ( U n ) , ϕ ) = μ 1 ∫ Ω ∇ V n ⋅ ∇ ϕ V d σ d p + c 0 2 μ 2 R 2 ∫ Ω ∇ T n ⋅ ∇ ϕ T d σ d p + ν 1 ∫ Ω α 2 ( p ) ∂ V n ∂ p ⋅ ∂ ϕ V ∂ p d σ d p + c 0 2 ν 2 R 2 ∫ Ω α 2 ( p ) ∂ T n ∂ p ∂ ϕ T ∂ p d σ d p ≤ C ∥ U n ∥ H 1 ( Ω ) ∥ ϕ ∥ H 1 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω ) ,
(4.7)

which implies

∥ L ( U n ) ∥ H − 2 ( Ω ) ≤C ∥ U n ∥ H 1 ( Ω ) ,
(4.8)

and we have

∫ 0 M ∥ L ( U n ) ∥ H − 2 ( Ω ) 2 dt≤C ∫ 0 M ∥ U n ∥ H 1 ( Ω ) 2 dt≤C,
(4.9)

where C>0 denotes a constant independent of n, namely,

L ( U n ) ∈ L 2 ( 0 , M ; H − 2 ( Ω ) ) .
(4.10)

Next, we can find that

( B ( U n , U n ) , ϕ ) = ∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ ϕ V + ( V n ⋅ ∇ ) T n ϕ T ) d σ d p + ∫ Ω ( ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ V n ∂ p ⋅ ϕ V + ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ T n ∂ p ϕ T ) d σ d p + ∫ Ω ( R ∇ ∫ p p s 1 s T n ( s ) d s ⋅ ϕ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ϕ T ) d σ d p + ∫ Ω ( 2 ω cos θ + cot θ a v λ n ) ( v θ n ϕ v λ − v λ n ϕ v θ ) d σ d p ,
(4.11)

and we have

∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ ϕ V + ( V n ⋅ ∇ ) T n ϕ T ) d σ d p ≤ C ( ∫ Ω | ∇ U n | 2 d σ d p ) 1 2 ( ∫ Ω | U n | 2 | ϕ | 2 d σ d p ) 1 2 ≤ C ∥ U n ∥ H 1 ( Ω ) ( ∫ Ω | U n | 3 d σ d p ) 1 3 ( ∫ Ω | ϕ | 6 d σ d p ) 1 6 ≤ C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 ∥ ϕ ∥ H 1 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 ,
(4.12)
∫ Ω ( ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ V n ∂ p ⋅ ϕ V + ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ T n ∂ p ϕ T ) d σ d p = ∫ Ω ∇ ⋅ V n V n ⋅ ϕ V d σ d p + ∫ Ω ∇ ⋅ V n T n ϕ T d σ d p − ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V n ⋅ ∂ ϕ V ∂ p d σ d p − ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T n ∂ ϕ T ∂ p d σ d p ≤ C ( ∫ Ω | ∇ U n | 2 d σ d p ) 1 2 ( ∫ Ω | U n | 2 | ϕ | 2 d σ d p ) 1 2 + C ( ∫ Ω | ∇ U n | 2 d σ d p ) 1 2 ( ∫ Ω | U n | 2 | ∂ ϕ ∂ p | 2 d σ d p ) 1 2 ≤ C ∥ U n ∥ H 1 ( Ω ) ( ∫ Ω | U n | 3 d σ d p ) 1 3 ( ∫ Ω | ϕ | 6 d σ d p ) 1 6 + C ∥ U n ∥ H 1 ( Ω ) ( ∫ Ω | U n | 3 d σ d p ) 1 3 ( ∫ Ω | ∂ ϕ ∂ p | 6 d σ d p ) 1 6 ≤ C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 ∥ ϕ ∥ H 1 ( Ω ) + C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 ∥ ϕ ∥ H 2 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 ,
(4.13)
∫ Ω ( R ∇ ∫ p p s 1 s T n ( s ) d s ⋅ ϕ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ϕ T ) d σ d p ≤ C ( ∫ Ω | ∇ U n | 2 d σ d p ) 1 2 ( ∫ Ω | ϕ | 2 d σ d p ) 1 2 ≤ C ∥ U n ∥ H 1 ( Ω ) ∥ ϕ ∥ H 1 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω )
(4.14)

and

∫ Ω ( 2 ω cos θ + cot θ a v λ n ) ( v θ n ϕ v λ − v λ n ϕ v θ ) d σ d p ≤ C ( ∫ Ω ( 1 + | U n | 2 ) d σ d p ) 1 2 ( ∫ Ω | U n | 2 | ϕ | 2 d σ d p ) 1 2 ≤ C ( 1 + ∥ U n ∥ L 2 ( Ω ) ) ( ∫ Ω | U n | 3 d σ d p ) 1 3 ( ∫ Ω | ϕ | 6 d σ d p ) 1 6 ≤ C ∥ U n ∥ H 1 ( Ω ) 1 2 ( 1 + ∥ U n ∥ L 2 ( Ω ) 3 2 ) ∥ ϕ ∥ H 1 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω ) 1 2 ( 1 + ∥ U n ∥ L 2 ( Ω ) 3 2 ) ,
(4.15)

from (4.12)-(4.15), we have

∥ B ( U n , U n ) ∥ H − 2 ( Ω ) ≤ C ∥ U n ∥ H 1 ( Ω ) 3 2 ∥ U n ∥ L 2 ( Ω ) 1 2 + C ∥ U n ∥ H 1 ( Ω ) + C ∥ U n ∥ H 1 ( Ω ) 1 2 ( 1 + ∥ U n ∥ L 2 ( Ω ) 3 2 )
(4.16)

and

∫ 0 M ∥ B ( U n , U n ) ∥ H − 2 ( Ω ) 4 3 d t ≤ C ∫ 0 M ( ∥ U n ∥ H 1 ( Ω ) 2 ∥ U n ∥ L 2 ( Ω ) 2 3 + ∥ U n ∥ H 1 ( Ω ) 4 3 + ∥ U n ∥ H 1 ( Ω ) 2 3 ( 1 + ∥ U n ∥ L 2 ( Ω ) 3 2 ) 4 3 ) d t ≤ C ∫ 0 M ( ∥ U n ∥ H 1 ( Ω ) 2 + ∥ U n ∥ H 1 ( Ω ) 4 3 + ∥ U n ∥ H 1 ( Ω ) 2 3 ) d t ≤ C ∫ 0 M ∥ U n ∥ H 1 ( Ω ) 2 t + C ≤ C ,
(4.17)

where C>0 denotes a constant independent of n, namely,

B ( U n , U n ) ∈ L 4 3 ( 0 , M ; H − 2 ( Ω ) ) .
(4.18)

Finally, we have

(F,ϕ)= ∫ Ω Ψ c p ϕ T dσdp≤C ∥ Ψ ∥ H − 2 ( Ω ) ∥ ϕ ∥ H 2 ( Ω ) ≤C ∥ Ψ ∥ H − 2 ( Ω )
(4.19)

and

F∈ L ∞ ( 0 , M ; H − 2 ( Ω ) ) .
(4.20)

Then we have from (4.10), (4.18), and (4.20)

U t n ∈ L 4 3 ( 0 , M ; H − 2 ( Ω ) ) ,
(4.21)

which together with Lemma 4.1 and U n ∈ L 2 (0,M; H 1 (Ω)) gives

U n →U∈ L 2 ( 0 , M ; L 2 ( Ω ) ) ,
(4.22)

and using (4.22), we can prove (4.6) holds. □

Lemma 4.3 Let U n be the weak solution sequence of system (2.4). Then, up to a subsequence, we have

∫ 0 M a ( U n , φ ) dt→ ∫ 0 M a(U,φ)dt,
(4.23)

for all φ=( φ v λ , φ v θ , φ T )=( φ V , φ T )∈ C ∞ (0,M; C 0 ∞ (Ω)) and φ(M,⋅)=0.

Proof As ∂ U n ∂ λ , ∂ U n ∂ θ and ∂ U n ∂ p ∈ L 2 (0,M; L 2 (Ω)), thus, we have

∂ U n ∂ λ ⇀ ∂ U ∂ λ , ∂ U n ∂ θ ⇀ ∂ U ∂ θ , ∂ U n ∂ p ⇀ ∂ U ∂ p ∈ L 2 ( 0 , M ; L 2 ( Ω ) ) ,
(4.24)

which means that the sequences converge weakly; then we have

∫ 0 M a ( U n , φ ) d t = μ 1 ∫ 0 M ∫ Ω ∇ V n ⋅ ∇ φ V d σ d p d t + c 0 2 μ 2 R 2 ∫ 0 M ∫ Ω ∇ T n ⋅ ∇ φ T d σ d p d t + ν 1 ∫ 0 M ∫ Ω α 2 ( p ) ∂ V n ∂ p ⋅ ∂ φ V ∂ p d σ d p d t + c 0 2 ν 2 R 2 ∫ 0 M ∫ Ω α 2 ( p ) ∂ T n ∂ p ∂ φ T ∂ p d σ d p d t → μ 1 ∫ 0 M ∫ Ω ∇ V ⋅ ∇ φ V d σ d p d t + c 0 2 μ 2 R 2 ∫ 0 M ∫ Ω ∇ T ⋅ ∇ φ T d σ d p d t + ν 1 ∫ 0 M ∫ Ω α 2 ( p ) ∂ V ∂ p ⋅ ∂ φ V ∂ p d σ d p d t + c 0 2 ν 2 R 2 ∫ 0 M ∫ Ω α 2 ( p ) ∂ T ∂ p ∂ φ T ∂ p d σ d p d t = ∫ 0 M a ( U , φ ) d t .
(4.25)

 □

Lemma 4.4 Let U n be the weak solution sequence of system (2.4). Then, up to a subsequence, we have

∫ 0 M b ( U n , U n , φ ) dt→ ∫ 0 M b(U,U,φ)dt,
(4.26)

for all φ=( φ v λ , φ v θ , φ T )=( φ V , φ T )∈ C ∞ (0,M; C 0 ∞ (Ω)) and φ(M,⋅)=0.

Proof We have

∫ 0 M b ( U n , U n , φ ) d t = ∫ 0 M ∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ φ V + ( V n ⋅ ∇ ) T n φ T ) d σ d p d t + ∫ 0 M ∫ Ω ( ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ V n ∂ p ⋅ φ V + ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ T n ∂ p φ T ) d σ d p d t + ∫ 0 M ∫ Ω ( R ∇ ∫ p p s 1 s T n ( s ) d s ⋅ φ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V n ( s ) d s ) φ T ) d σ d p d t + ∫ 0 M ∫ Ω ( 2 ω cos θ + cot θ a v λ n ) ( v θ n φ v λ − v λ n φ v θ ) d σ d p d t ,
(4.27)

and we have from (4.5) and (4.24)

∫ 0 M ∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ φ V + ( V n ⋅ ∇ ) T n φ T ) d σ d p d t = ∫ 0 M ∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ φ V + ( V n ⋅ ∇ ) T n φ T ) d σ d p d t − ∫ 0 M ∫ Ω ( ( V ⋅ ∇ ) V n ⋅ φ V + ( V ⋅ ∇ ) T n φ T ) d σ d p d t + ∫ 0 M ∫ Ω ( ( V ⋅ ∇ ) V n ⋅ φ V + ( V ⋅ ∇ ) T n φ T ) d σ d p d t → ∫ 0 M ∫ Ω ( ( V ⋅ ∇ ) V ⋅ φ V + ( V ⋅ ∇ ) T φ T ) d σ d p d t ,
(4.28)

where we use the fact

∫ 0 M ∫ Ω ( ( V n ⋅ ∇ ) V n ⋅ φ V + ( V n ⋅ ∇ ) T n φ T ) d σ d p d t − ∫ 0 M ∫ Ω ( ( V ⋅ ∇ ) V n ⋅ φ V + ( V ⋅ ∇ ) T n φ T ) d σ d p d t ≤ C ( ∫ 0 M ∫ Ω | U n − U | 2 d p d t ) 1 2 ( ∫ 0 M ∫ Ω | ∇ U n | 2 d p d t ) 1 2 → 0 .
(4.29)

Using |∇ U n |∈ L 2 (0,M; L 2 (Ω)), we have

∇⋅ ∫ p p s V n (s)ds,∇ ∫ p p s 1 s T n (s)ds∈ L 2 ( 0 , M ; L 2 ( Ω ) ) ,
(4.30)

which implies

∇⋅ ∫ p p s V n (s)ds⇀∇⋅ ∫ p p s V(s)ds,∇ ∫ p p s 1 s T n (s)ds⇀∇ ∫ p p s 1 s T(s)ds,
(4.31)

and we have

∫ 0 M ∫ Ω ( ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ V n ∂ p ⋅ φ V + ( ∇ ⋅ ∫ p p s V n ( s ) d s ) ∂ T n ∂ p φ T ) d σ d p d t = ∫ 0 M ∫ Ω ∇ ⋅ V n V n ⋅ φ V d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V n T n φ T d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V n ⋅ ∂ φ V ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T n ∂ φ T ∂ p d σ d p d t = ∫ 0 M ∫ Ω ∇ ⋅ V n V n ⋅ φ V d σ d p d t − ∫ 0 M ∫ Ω ∇ ⋅ V n V ⋅ φ V d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V n T n φ T d σ d p d t − ∫ 0 M ∫ Ω ∇ ⋅ V n T φ T d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V n V ⋅ φ V d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V n T φ T d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V n ⋅ ∂ φ V ∂ p d σ d p d t + ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V ⋅ ∂ φ V ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T n ∂ φ T ∂ p d σ d p d t + ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T ∂ φ T ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V ⋅ ∂ φ V ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T ∂ φ T ∂ p d σ d p d t → ∫ 0 M ∫ Ω ∇ ⋅ V V ⋅ φ V d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V T φ T d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V ( s ) d s ) V ⋅ ∂ φ V ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V ( s ) d s ) T ∂ φ T ∂ p d σ d p d t = ∫ 0 M ∫ Ω ( ( ∇ ⋅ ∫ p p s V ( s ) d s ) ∂ V ∂ p ⋅ φ V + ( ∇ ⋅ ∫ p p s V ( s ) d s ) ∂ T ∂ p φ T ) d σ d p d t ,
(4.32)

where we use the fact

∫ 0 M ∫ Ω ∇ ⋅ V n V n ⋅ φ V d σ d p d t − ∫ 0 M ∫ Ω ∇ ⋅ V n V ⋅ φ V d σ d p d t + ∫ 0 M ∫ Ω ∇ ⋅ V n T n φ T d σ d p d t − ∫ 0 M ∫ Ω ∇ ⋅ V n T φ T d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V n ⋅ ∂ φ V ∂ p d σ d p d t + ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) V ⋅ ∂ φ V ∂ p d σ d p d t − ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T n ∂ φ T ∂ p d σ d p d t + ∫ 0 M ∫ Ω ( ∇ ⋅ ∫ p p s V n ( s ) d s ) T ∂ φ T ∂ p d σ d p d t ≤ C ( ∫ 0 M ∫ Ω | U n − U | 2 d p d t ) 1 2 ( ∫ 0 M ∫ Ω | ∇ U n | 2 d p d t ) 1 2 → 0 ,
(4.33)

applying (4.31), we also have

∫ 0 M ∫ Ω ( R ∇ ∫ p p s 1 s T n ( s ) d s ⋅ φ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V n ( s ) d s ) φ T ) d σ d p d t → ∫ 0 M ∫ Ω ( R ∇ ∫ p p s 1 s T ( s ) d s ⋅ φ V − c 0 2 R p ( ∇ ⋅ ∫ p p s V ( s ) d s ) φ T ) d σ d p d t .
(4.34)

Finally, by means of (4.5), we can prove that

∫ 0 M ∫ Ω ( 2 ω cos θ + cot θ a v λ n ) ( v θ n φ v λ − v λ n φ v θ ) d σ d p d t → ∫ 0 M ∫ Ω ( 2 ω cos θ + cot θ a v λ ) ( v θ φ v λ − v λ φ v θ ) d σ d p d t .
(4.35)

Summing (4.28), (4.32), (4.34), and (4.35), we complete the proof of (4.26). □

5 Conclusion

In this paper, the stability of weak solutions for the atmospheric equations is investigated with the constant external force and without the effects of topography; from Theorem 2.1 and Remark 2.1 and Remark 2.2, we show that if U 0 n → U 0 ∈ L 1 (Ω), then U n →U∈ L 1 (0,T; L 1 (Ω)); if U 0 n → U 0 a.e., then U n →U a.e., which means that if the difference of the initial data of two different weak solutions is small almost everywhere, then the difference of this two weak solutions is small almost everywhere as time increases. Furthermore, in the future we will consider the stability of weak solutions to the atmospheric models with the effects of topography, a non-constant external force, radiation heating, and the moist phase transformation, etc.

References

  1. Zeng Q: Mathematical and Physical Foundations of Numerical Weather Prediction. Science Press, Beijing; 1979. (in Chinese)

    Google Scholar 

  2. Li JP, Chou JF: The existence of the atmosphere attractor. Sci. China, Ser. D, Earth Sci. 1996,40(2):89–96.

    Google Scholar 

  3. Pedlosky J: Geophysical Fluid Dynamics. 2nd edition. Springer, Heidelberg; 1987.

    Book  MATH  Google Scholar 

  4. Washington WM, Parkinson CL: An Introduction to Three-Dimensional Climate Modeling. Oxford University Press, Oxford; 1986.

    MATH  Google Scholar 

  5. Lions JL, Temam R, Wang SH: New formulations of the primitive equations of the atmosphere and applications. Nonlinearity 1992,5(2):237–288. 10.1088/0951-7715/5/2/001

    Article  MathSciNet  MATH  Google Scholar 

  6. Lions JL, Temam R, Wang SH: Models of the coupled atmosphere and ocean (CAO I). Comput. Mech. Adv. 1993, 1: 5–54.

    MathSciNet  MATH  Google Scholar 

  7. Lions JL, Temam R, Wang SH: Models of the coupled atmosphere and ocean (CAO II). Comput. Mech. Adv. 1993, 1: 55–119.

    MathSciNet  MATH  Google Scholar 

  8. Lions JL, Temam R, Wang SH: Mathematical theory for the coupled atmosphere-ocean models (CAO III). J. Math. Pures Appl. 1995,74(2):105–163.

    MathSciNet  MATH  Google Scholar 

  9. Chepzhov VV, Vishik MI: Attractors for Equations of Mathematical Physics. Am. Math. Soc., Providence; 2002.

    Google Scholar 

  10. Huang HY, Guo BL: Existence of the solutions and the attractors for the large-scale atmospheric equations. Sci. China, Ser. D, Earth Sci. 2006,49(6):650–660. 10.1007/s11430-006-0650-8

    Article  MathSciNet  Google Scholar 

  11. Huang HY, Guo BL: The existence of weak solutions and the trajectory attractors to the model of climate dynamics. Acta Math. Sci. Ser. A 2006,27(6):1098–1110.

    Google Scholar 

Download references

Acknowledgements

The research is supported by NNSFC No. 11101145, China Postdoctoral Science Foundation No. 2012M520360.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Ruxu Lian.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

RXL organized and wrote this paper. QCZ examined all the steps of the proofs in this research and gave some advices. All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Lian, R., Zeng, Qc. Stability of weak solutions for the large-scale atmospheric equations. J Inequal Appl 2014, 455 (2014). https://doi.org/10.1186/1029-242X-2014-455

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/1029-242X-2014-455

Keywords