Time-Space Fractional Coupled Generalized Zakharov-Kuznetsov Equations Set for Rossby Solitary Waves in Two-Layer Fluids

Abstract

In this paper, the theoretical model of Rossby waves in two-layer fluids is studied. A single quasi-geostrophic vortex equation is used to derive various models of Rossby waves in a one-layer fluid in previous research. In order to explore the propagation and interaction of Rossby waves in two-layer fluids, from the classical quasi-geodesic vortex equations, by employing the multi-scale analysis and turbulence method, we derived a new (2+1)-dimensional coupled equations set, namely the generalized Zakharov-Kuznetsov(gZK) equations set. The gZK equations set is an extension of a single ZK equation; they can describe two kinds of weakly nonlinear waves interaction by multiple coupling terms. Then, for the first time, based on the semi-inverse method and the variational method, a new fractional-order model which is the time-space fractional coupled gZK equations set is derived successfully, which is greatly different from the single fractional equation. Finally, group solutions of the time-space fractional coupled gZK equations set are obtained with the help of the improved $(G^{\prime }/G)$-expansion method.

1. Introduction

Although the existence of solitary waves has been known in hydrokinetics for about a century, it was not until recently that the theory was applied to wave phenomena in the atmospheric, ocean [1,2,3,4,5,6] and large lake dynamics system, such as solitary waves [7], internal gravity waves [8], internal Kelvin waves [9] and so on. Since Long(1964) derived the Korteweg de-Vries(KdV) equation on the positive pressure system, the isolated Rossby wave theory has been gradually developed [10,11,12], but in many complex atmospheric and oceanic systems, waves interact with each other. The problem of wave–wave interaction is a very important one in the earth fluid, and further study can deepen people’s understanding of large-scale motion phenomena in the atmosphere, ocean and large lakes. There are many motion equations describing Rossby waves in a layer of fluid, such as one-dimensional KdV equation [13], MKdV equation [14], Boussinesq equation [15] and high-dimensional ZK equation [16], mZK equation [17], KP equation [18] and so on. The actual atmospheric ocean system is very complex, compared with the single equation of low dimension, the coupled equations set [19,20,21] of high dimension are more practical. Therefore, this paper derives the (2+1)-dimensional coupled generalized ZK(gZK) equations set from the classical quasi-geostrophic vorticity equations [22,23]. The gZK equation is a class of important high-dimensional nonlinear evolution equation in mathematical physics, the gZK equations set as the extension of a single equation can be used to describe the interaction of nonlinear Rossby waves in two-layer fluids [24,25,26].

Fractional calculus have been used to model atmospheric and oceanic systems, and these models have been found to be suitable to be described by fractional differential equations [27,28,29]. Standard mathematical models of integer derivatives, including nonlinear models, do not work well in many cases. For fractional-order equations set, most researchers only choose the time fractional-order equations set [30,31], but few study the time-space fractional-order equations set, especially the time-space coupled fractional-order equations set. So, for the first time, we transform integral order coupled gZK equations set into fractional order coupled gZK equations set of space-time, and find that it has great research value.

Given the exact solution of nonlinear equations set [32,33] plays an important role in explaining some complex phenomena in physics, dynamics and other scientific fields, mathematicians and physicists have done a lot of researches on solving nonlinear equations set [34,35,36,37,38,39,40,41]. So far, researchers have proposed many effective solutions, such as the $(G^{\prime }/G)$-expansion method [42], the exp-function method [43], the Kudryashov method [21], the sun-equation method [44], the functional variable method [45], the modified extended Tanh method [46], Khater method [47], Godunov-type method [48], Lie group analysis method [49,50] and so on. We choose the modified $(G^{\prime }/G)$-expansion method to solve the time-space fractional coupled gZK equations set, and get several different kinds of solutions.

2. Derivation of Coupled gZK Equations Set

A single set of quasi-geostrophic vortex equation is usually used to derive various equations which can be used to research Rossby solitary waves [51] in a single layer of the atmosphere and ocean. For the waves in two-layer fluids, they are divided into upper and lower layers as shown in Figure 1, but the isolated waves between the two layers do not propagate separately but interact with each other, this is where the coupled equations set model comes in.

In the first place, for the exploration of the Rossby waves propagation and action between two-layer fluids, the coupled gZK equations set of the objective function is derived from the following two layers of quasi-geostrophic vortex equations set by using the multi-scale analysis and turbulence method

$$\begin{array}{c}{q}_{At}+J({\psi }_A,q_A)+\beta {\psi }_{Ax}=0,\hfill \\ q_{Bt}+J({\psi }_B,q_B)+\beta {\psi }_{Bx}=0,\hfill \end{array}$$

(1)

where ${\psi }_A$, ${\psi }_B$ are the stream functions of the upper fluid and the lower fluid respectively, $J[a,b]$ is the Jacobi operator and $J(a,b)=a_x{b}_y-b_x{a}_y$ and $\beta$ represents the Coriolis parameter, as well as

$$\begin{array}{c}{q}_A={\psi }_{Axx}+{\psi }_{Ayy}+F({\psi }_B-{\psi }_A),\hfill \\ q_B={\psi }_{Bxx}+{\psi }_{Byy}+F({\psi }_A-{\psi }_B).\hfill \end{array}$$

(2)

Then the Equation (1) colorredcan be expanded to the following form

$$\begin{array}{c}{\psi }_{Axxt}+{\psi }_{Ayyt}+F{({\psi }_B-{\psi }_A)}_t+{\psi }_{Ax}[{\psi }_{Axxy}+{\psi }_{Ayyy}+F{({\psi }_B-{\psi }_A)}_y]\hfill \\ -[{\psi }_{Axxx}+{\psi }_{Ayyx}+F{({\psi }_B-{\psi }_A)}_x]{\psi }_{Ay}+\beta {\psi }_{Ax}=0,\hfill \\ {\psi }_{Bxxt}+{\psi }_{Byyt}+F{({\psi }_A-{\psi }_B)}_t+{\psi }_{Bx}[{\psi }_{Bxxy}+{\psi }_{Byyy}+F{({\psi }_A-{\psi }_B)}_y]\hfill \\ -[{\psi }_{Bxxx}+{\psi }_{Byyx}+F{({\psi }_A-{\psi }_B)}_x]{\psi }_{By}+\beta {\psi }_{Bx}=0,\hfill \end{array}$$

(3)

with F represents the weak coupling coefficient between two-layer fluids [22].

In the next place, so as to derive the gZK type equations set, we take the long wave approximation in the x direction, then the stream functions ${\psi }_A$ and ${\psi }_B$ can be divided into the basic stream functions and the disturbance stream functions, written as

$$\begin{array}{c}{\psi }_A={\varphi }_{A_0}(y)+{\varphi }_A(x,y,t)=(U_{A_0}+c_{0}y)+{\varphi }_A(x,y,t),\hfill \\ {\psi }_B={\varphi }_{B_0}(y)+{\varphi }_B(x,y,t)=(U_{B_0}+c_{0}y)+{\varphi }_B(x,y,t).\hfill \end{array}$$

(4)

Suppose the coupling between the two-layer fluids is weak, and the rotation effect of the Earth is very small, therefore we can adopt the following space-time transformation

$$F=\epsilon F_0,\beta ={\epsilon }^2\beta ,$$

(5)

$$X=\epsilon (x-c_{0}t),Y=\epsilon y,T={\epsilon }^{3}t,$$

(6)

where $\epsilon$ is a small parameter.

Under the premise of Equations (5) and (6), Equation (3) can be rewritten as the following form

$$\begin{array}{c}-c_0{\epsilon }^3{\varphi }_{AXXX}+{\epsilon }^5{\varphi }_{AXXT}-c_0{\epsilon }^3{\varphi }_{AYYX}-2c_0{\epsilon }^2{\varphi }_{AYyX}-c_0\epsilon {\varphi }_{AyyX}+{\epsilon }^5{\varphi }_{AYYT}+2{\epsilon }^4{\varphi }_{AYyT}\hfill \\ +{\epsilon }^3{\varphi }_{AyyT}+{\epsilon }^4{F}_0({\varphi }_{BT}-{\varphi }_{AT})-c_0{\epsilon }^2{F}_0({\varphi }_{BX}-{\varphi }_{AX})+\epsilon {\varphi }_{AX}[{\epsilon }^3{\varphi }_{AXXY}+{\epsilon }^2{\varphi }_{AXXy}+\hfill \\ {\varphi }_{A_{0}yyy}+{\epsilon }^3{\varphi }_{AYYY}+3{\epsilon }^2{\varphi }_{AYYy}+3\epsilon {\varphi }_{AYyy}+{\varphi }_{Ayyy}+\epsilon F_0({\varphi }_{B_{0}y}-{\varphi }_{A_{0}y})+{\epsilon }^2{F}_0({\varphi }_{BY}-{\varphi }_{Ay})\hfill \\ +\epsilon F_0({\varphi }_{By}-{\varphi }_{Ay})]-[{\epsilon }^3{\varphi }_{AXXX}+{\epsilon }^3{\varphi }_{AYYX}+2{\epsilon }^2{\varphi }_{AYyX}+{\epsilon }^2{F}_0({\varphi }_{BX}-{\varphi }_{AX})]({\varphi }_{A_{0}y}\hfill \\ +\epsilon {\varphi }_{AY}+{\varphi }_{Ay})+{\epsilon }^4{\beta }_1{\varphi }_{AX}=0,\hfill \\ -c_0{\epsilon }^3{\varphi }_{BXXX}+{\epsilon }^5{\varphi }_{BXXT}-c_0{\epsilon }^3{\varphi }_{AYYX}-2c_0{\epsilon }^2{\varphi }_{BYyX}-c_0\epsilon {\varphi }_{ByyX}+{\epsilon }^5{\varphi }_{BYYT}+2{\epsilon }^4{\varphi }_{BYyT}\hfill \\ +{\epsilon }^3{\varphi }_{ByyT}+{\epsilon }^4{F}_0({\varphi }_{AT}-{\varphi }_{BT})-c_0{\epsilon }^2{F}_0({\varphi }_{AX}-{\varphi }_{BX})+\epsilon {\varphi }_{BX}[{\epsilon }^3{\varphi }_{BXXY}+{\epsilon }^2{\varphi }_{BXXy}+\hfill \\ {\varphi }_{B_{0}yyy}+{\epsilon }^3{\varphi }_{BYYY}+3{\epsilon }^2{\varphi }_{BYYy}+3\epsilon {\varphi }_{BYyy}+{\varphi }_{Byyy}+\epsilon F_0({\varphi }_{A_{0}y}-{\varphi }_{B_{0}y})+{\epsilon }^2{F}_0({\varphi }_{AY}-{\varphi }_{By})\hfill \\ +\epsilon F_0({\varphi }_{Ay}-{\varphi }_{By})]-[{\epsilon }^3{\varphi }_{BXXX}+{\epsilon }^3{\varphi }_{BYYX}+2{\epsilon }^2{\varphi }_{BYyX}+{\epsilon }^2{F}_0({\varphi }_{AX}-{\varphi }_{BX})]({\varphi }_{B_{0}y}\hfill \\ +\epsilon {\varphi }_{BY}+{\varphi }_{By})+{\epsilon }^4{\beta }_1{\varphi }_{BX}=0,\hfill \end{array}$$

(7)

where the perturbation stream functions have the following series expansion forms

$$\begin{array}{c}{\varphi }_A=\epsilon {\varphi }_{A_1}(X,Y,T)+{\epsilon }^2{\varphi }_{A_2}(X,Y,T)+{\epsilon }^3{\varphi }_{A_3}(X,Y,T)+o({\epsilon }^4),\hfill \\ {\varphi }_B=\epsilon {\varphi }_{B_1}(X,Y,T)+{\epsilon }^2{\varphi }_{B_2}(X,Y,T)+{\epsilon }^3{\varphi }_{B_3}(X,Y,T)+o({\epsilon }^4).\hfill \end{array}$$

(8)

By substituting Equation (8) into Equation (7), we obtained the following equations about small parameter $\epsilon$

$${\epsilon }^2:\left\{\begin{array}{c}-{\varphi }_{A_{0yyy}}{\varphi }_{A_{1X}}-c_0{\varphi }_{A_{1yyX}}+{\varphi }_{A_{0y}}{\varphi }_{A_{1yyX}}=0,\hfill \\ -{\varphi }_{B_{0yyy}}{\varphi }_{B_{1X}}-c_0{\varphi }_{B_{1yyX}}+{\varphi }_{B_{0y}}{\varphi }_{B_{1yyX}}=0,\hfill \end{array}\right.$$

(9)

$${\epsilon }^3:\left\{\begin{array}{c}-{\varphi }_{A_{0yyy}}{\varphi }_{A_{2X}}-c_0{\varphi }_{A_{2yyX}}+{\varphi }_{A_{0y}}{\varphi }_{A_{2yyX}}-2{\varphi }_{A_{1YyX}}({\varphi }_{A_{0y}}-c_0)\hfill \\ +{\varphi }_{A_{1X}}{\varphi }_{A_{1yyy}}-{\varphi }_{A_{1yyX}}{\varphi }_{A_{1y}}+F_0{\varphi }_{B_{1X}}({\varphi }_{A_{0y}}-c_0)+F_0{\varphi }_{A_{1X}}({\varphi }_{B_{0y}}-c_0)=0,\hfill \\ -{\varphi }_{B_{0yyy}}{\varphi }_{B_{2X}}-c_0{\varphi }_{B_{2yyX}}+{\varphi }_{B_{0y}}{\varphi }_{B_{2yyX}}-2{\varphi }_{B_{1YyX}}({\varphi }_{B_{0y}}-c_0)\hfill \\ +{\varphi }_{B_{1X}}{\varphi }_{B_{1yyy}}-{\varphi }_{B_{1yyX}}{\varphi }_{B_{1y}}+F_0{\varphi }_{A_{1X}}({\varphi }_{B_{0y}}-c_0)+F_0{\varphi }_{B_{1X}}({\varphi }_{A_{0y}}-c_0)=0,\hfill \end{array}\right.$$

(10)

$${\epsilon }^4:\left\{\begin{array}{c}-{\varphi }_{A_{0yyy}}{\varphi }_{A_{3X}}-c_0{\varphi }_{A_{3yyX}}+{\varphi }_{A_{0y}}{\varphi }_{A_{3yyX}}-2{\varphi }_{A_{2YyX}}({\varphi }_{A_{0y}-c_0})+{\varphi }_{A_{2X}}{\varphi }_{A_{1yyy}}\hfill \\ +{\varphi }_{A_{1X}}{\varphi }_{A_{2yyy}}-{\varphi }_{A_{1yyX}}{\varphi }_{A_{2y}}+F_0{\varphi }_{B_{2X}}({\varphi }_{A_{0y}}-c_0)+F_0{\varphi }_{A_{2X}}({\varphi }_{B_{0y}}-c_0)\hfill \\ +{\varphi }_{A_{1XXX}}({\varphi }_{A_{0y}}-c_0)+{\varphi }_{A_{1yyT}+{\varphi }_{A_{1YYX}}}({\varphi }_{A_{0y}}-c_0)+{\varphi }_{A_{1X}}{\varphi }_{A_{0yyy}}+3{\varphi }_{A_{1X}}{\varphi }_{A_{0yyy}}\hfill \\ +3{\varphi }_{A_{1X}}{\varphi }_{A_{1Yyy}}-2c_0{\varphi }_{B_{1YyX}}{\varphi }_{B_{1y}}-{\varphi }_{B_{1yyX}}{\varphi }_{B_{1Y}}+F_0{\varphi }_{B_{1X}}{\varphi }_{A_{1y}}-F_0{\varphi }_{B_{1y}}{\varphi }_{A_{1X}}=0,\hfill \\ -{\varphi }_{B_{0yyy}}{\varphi }_{B_{3X}}-c_0{\varphi }_{B_{3yyX}}+{\varphi }_{B_{0y}}{\varphi }_{B_{3yyX}}-2{\varphi }_{B_{2YyX}}({\varphi }_{B_{0y}}-c_0)+{\varphi }_{B_{2X}}{\varphi }_{B_{1yyy}}\hfill \\ +{\varphi }_{B_{1X}}{\varphi }_{B_{2yyy}}-{\varphi }_{B_{1yyX}}{\varphi }_{B_{2y}}+F_0{\varphi }_{A_{2X}}({\varphi }_{B_{0y}}-c_0)+F_0{\varphi }_{B_{2X}}({\varphi }_{A_{0y}}-c_0)\hfill \\ +{\varphi }_{B_{1XXX}}({\varphi }_{B_{0y}}-c_0)+{\varphi }_{B_{1yyT}+{\varphi }_{B_{1YYX}}}({\varphi }_{B_{0y}}-c_0)+{\varphi }_{B_{1X}}{\varphi }_{B_{0yyy}}+3{\varphi }_{B_{1X}}{\varphi }_{B_{0yyy}}\hfill \\ +3{\varphi }_{B_{1X}}{\varphi }_{B_{1Yyy}}-2c_0{\varphi }_{B_{1YyX}}{\varphi }_{B_{1y}}-{\varphi }_{B_{1yyX}}{\varphi }_{B_{1Y}}+F_0{\varphi }_{B_{1X}}{\varphi }_{A_{1y}}-F_0{\varphi }_{B_{1y}}{\varphi }_{A_{1X}}=0.\hfill \end{array}\right.$$

(11)

In Equation (9), it is easy to find that ${\varphi }_{A_1}$ and ${\varphi }_{A_2}$ have the following separable variables form solutions

$$\begin{array}{c}{\varphi }_{A_1}=A_1(X,Y,T)B_1(y)\equiv A_1{B}_1,\hfill \\ {\varphi }_{B_1}=A_2(X,Y,T)B_2(y)\equiv A_2{B}_2.\hfill \end{array}$$

(12)

Substituting (12) into Equation (9), we get the following equations about variable y

$$\begin{array}{c}{U}_{A_{0y}}{B}_{1y}-U_{A_{0yy}}{B}_1+C_1=0,\hfill \\ U_{B_{0y}}{B}_{2y}-U_{B_{0yy}}{B}_2+C_2=0,\hfill \end{array}$$

(13)

where $C_1$, $C_2$ are arbitrary constants. Applying Equations (12) and (13) to Equation (10), integrating with respect to X once, cancel out the integral function, and we get the following equations

$$\begin{array}{c}2U_{A_{0y}}(B_1{\partial }_{yy}-B_1){\varphi }_{A_2}+B_1[A_1^2(B_1{B}_{1yyy}-B_{1y}{B}_{1yy})\hfill \\ -4U_{A_{0y}}{A}_{1Y}{B}_{1y}-2F_0(U_{B_{0y}}{B}_1{A}_1-U_{A_{0y}}{B}_2{A}_2)]=0,\hfill \\ 2U_{B_{0y}}(B_2{\partial }_{yy}-B_2){\varphi }_{B_2}+B_2[A_2^2(B_2{B}_{2yyy}-B_{2y}{B}_{2yy})\hfill \\ -4U_{B_{0y}}{A}_{2Y}{B}_{2y}-2F_0(U_{A_{0y}}{B}_2{A}_2-U_{B_{0y}}{B}_1{A}_1)]=0.\hfill \end{array}$$

(14)

It’s easy to see from this set of equations that

$$\begin{array}{c}{\varphi }_{A_2}=(a_1{A}_1^2+a_2{A}_{1Y}+a_3{A}_1+a_4{A}_2)B_1,\hfill \\ {\varphi }_{B_2}=(b_1{A}_2^2+b_2{A}_{2Y}+b_3{A}_2+b_4{A}_1)B_2,\hfill \end{array}$$

(15)

where $a_i,b_i,i=1,2,3,4$ are functions of y

$$\begin{array}{c}{a}_1=\frac{B_1{B}_{1yyy}-B_{1y}{B}_{1yy}}{4U_{A_{0y}}{B}_{1yy}},a_2=-\frac{B_{1y}}{B_{1yy}},a_3=-\frac{F_0{U}_{B_{0y}}}{2U_{A_{0y}}{B}_{1yy}},a_4=\frac{F_0{B}_2}{2B_{1yy}},\hfill \\ b_1=\frac{B_2{B}_{2yyy}-B_{2y}{B}_{2yy}}{4U_{B_{0y}}{B}_{2yy}},a_2=-\frac{B_{2y}}{B_{2yy}},b_3=-\frac{F_0{U}_{A_{0y}}}{2U_{B_{0y}}{B}_{2yy}},b_4=\frac{F_0{B}_1}{2B_{2yy}}.\hfill \end{array}$$

(16)

In the end, after substituting Equation (12), Equation (15) and ${\psi }_{A_3}={\psi }_{B_3}=0$ into Equation (9), the integrating from 0 to $y_0$ of the resulting, the following coupled gZK equations set can be obtained though simple calculation.

$$\begin{array}{c}{A}_{1T}+c_1{(A_1{A}_2)}_X+c_2{(A_1^2)}_X+c_3{(A_2^2)}_X+c_4{A}_{1XY}+c_5{A}_{2XY}+c_6{(A_1^2)}_{XY}+c_7{A}_{1XXX}+c_8{A}_{1XYY}=0,\hfill \\ A_{2T}+d_1{(A_1{A}_2)}_X+d_2{(A_1^2)}_X+d_3{(A_2^2)}_X+d_4{A}_{1XY}+d_5{A}_{2XY}+d_6{(A_2^2)}_{XY}+d_7{A}_{2XXX}+d_8{A}_{2XYY}=0,\hfill \end{array}$$

(17)

where

$$\begin{array}{c}{c}_1={\int }_0^{y_0}(\frac{a_4{B}_1{B}_{1yyy}+F_0{B}_1{B}_{1y}}{B_{1yy}}-a_4{B}_1)dy,c_2={\int }_0^{y_0}(\frac{F_0{U}_{B_{0y}}}{B_{1yy}}-2a_3{B}_1-2B_{1y})dy,\hfill \\ c_3=2{\int }_0^{y_0}\frac{b_1{F}_0{U}_{A_{0y}}{B}_2}{B_{1yy}}dy,c_4={\int }_0^{y_0}\frac{a_3{B}_{1y}+a_2{F}_0{U}_{B_{0y}}{B}_1}{B_{1yy}}dy,c_5={\int }_0^{y_0}\frac{a_4{B}_{1y}+b_2{F}_0{U}_{A_{0y}}{B}_2}{B_{1yy}}dy,\hfill \\ c_6={\int }_0^{y_0}(\frac{a_2{B}_1{B}_{1yy}-2B_{1y}^2+2a_1{B}_{1y}}{B_{1y}}-a_2{B}_1)dy,c_7={\int }_0^{y_0}\frac{U_{A_{0y}}{B}_1}{B_{1yy}}dy,c_8={\int }_0^{y_0}\frac{a_2{B}_{1y}+U_{A_{0y}}{B}_1}{B_{1yy}},\hfill \\ d_1={\int }_0^{y_0}(\frac{b_4{B}_2{B}_{2yy}+F_0{B}_2{B}_{2y}}{B_{2yy}}-b_4{B}_2)dy,d_3={\int }_0^{y_0}(\frac{b_1{F}_0{U}_{A_{0y}{B}_2}}{B_{2yy}}-2a_3{B}_2-2a_3{B}_{2y})dy,\hfill \\ d_2=2{\int }_0^{y_0}\frac{a_1{F}_0{U}_{B_{0y}}{B}_1}{B_{2yy}},d_4={\int }_0^{y_0}\frac{b_4{B}_{2y}+a_2{F}_0{U}_{B_{0y}}{B}_1}{B_{2yy}}dy,d_5={\int }_0^{y_0}\frac{b_3{B}_{2y}+b_2{F}_0{U}_{B_{0y}}{B}_2}{B_{2yy}}dy,\hfill \\ d_6={\int }_0^{y_0}(\frac{b_2{B}_2{B}_{2yyy}-2B_{2y}^2+2b_1{B}_{2y}}{B_{2yy}}-b_2{B}_2{B}_{2yy})dy,d_7={\int }_0^{y_0}\frac{U_{B_{0}y}{B}_2}{B_{2yy}}dy,d_8={\int }_0^{y_0}\frac{b_2{B}_{2y}+U_{B_{0}y}{B}_2}{B_{2yy}}.\hfill \end{array}$$

(18)

Remark 1.

The coupled gZK equations set is the extension of a single ZK equation and a class of important nonlinear evolution equations of high dimension. They describe two kinds of weakly nonlinear waves interaction with each other and the interaction between two waves is reflected in multiple coupling terms ${(A_1{A}_2)}_X$, ${(A_1^2)}_{XY}$ and ${(A_2^2)}_{XY}$.

3. The Time-Space Fractional Coupled gZK Equations set

In previous work, we only derived a single fractional-order equation, but here we will apply the semi-inverse method and the variational method [52,53,54] to derive the coupled fractional-order equations set for the first time, and obtain a new fractional-order coupled equations set, namely time-space fractional coupled gZK equations set. colorredFor the ease of understanding, some definitions and properties of fractional order are introduced before demonstrating the specific derivation process.

Definition 1

([52]). Modified Rieman-Liouville derivative

$$D_t^{\alpha }f(t)=\left\{\begin{array}{c}\frac{1}{\mathrm{\Gamma }(1-\alpha )}\frac{d}{dt}{\int }_0^t{(t-\delta )}^{-\alpha }(f(\delta )-f(0))d\delta ,0<\alpha <1,\hfill \\ {(f^{(n)(x)})}^{(\alpha -n)},n\le \alpha <n+1,n\ge 1,\hfill \end{array}\right.$$

where $f(t)$ is a continuous function.

Definition 2

([42]). Assume that $f(t)$ denotes a continuous $R\to R$ function, we use the following equality for the integral

$$D_t^{\alpha }f(t)=\frac{1}{\mathrm{\Gamma }(\alpha )}{\int }_0^t{(t-\zeta )}^{-1}f(\zeta )d\zeta =\frac{1}{\mathrm{\Gamma }(1+\alpha )}\frac{d}{dt}{\int }_0^{t}f(\zeta ){(d\zeta )}^{\alpha },0<\alpha \le 1.$$

Property 1

([54]). Integral property of fractional order equation

$${\int }_a^t{(d\tau )}^{\alpha }f(\tau )=\alpha {\int }_a^{t}d\tau {(t-\tau )}^{\alpha }f(\tau ).$$

Property 2

([54]). Integration by parts property of fractional order equation

$$\begin{array}{cc}\hfill {\int }_a^b{(d\tau )}^{\alpha }f(t)D_t^{\alpha }g(t)& =\frac{1}{\mathrm{\Gamma }(1-\alpha )}[g(t)f(t){|}_a^b-{\int }_a^b{(d\tau )}^{\alpha }g(t)D_t^{\alpha }f(t)],\hfill \\ f(t),g(t)\in [a,b].\hfill \end{array}$$

Introducing two potential functions $U(X,Y,T)$ and $V(X,Y,T)$, and their relationship to $A_1$ and $A_2$ is that $A_1=U_X,A_2=V_X$, substituting these two expressions into Equation (17) separately, the potential equations of the coupled gZK equations set have the form as

$$\begin{array}{c}{U}_{XT}+c_1{(v_{X}V)}_X+c_2{(U_X^2)}_X+c_3{(V_X^2)}_X+c_4{U}_{XXY}\hfill \\ +c_5{V}_{XXY}+c_6{(U_X^2)}_{XY}+c_7{U}_{XXXX}+c_8{U}_{XXYY}=0,\hfill \\ V_{XT}+d_1{(V_{X}U)}_X+d_2{(U_X^2)}_X+d_3{(V_X^2)}_X+d_4{U}_{XXY}\hfill \\ +d_5{V}_{XXY}+d_6{(V_X^2)}_{XY}+d_7{V}_{XXXX}+d_8{V}_{XXYY}=0.\hfill \end{array}$$

(19)

Further, the semi-inverse method was applied to derive the Lagrangian equations of coupled gZK equations, functional of the Equation (19) could be written as

$$\begin{array}{ll}J(U,V)=& {\int }_{R}dX{\int }_{R}dY{\int }_{T}dT\left\{U\right[m_1{U}_{XT}+m_2{c}_1{(U_{X}V)}_X+m_3{c}_2{U}_X{U}_{XX}+m_4{c}_3{V}_X{V}_{XX}\\ & +m_5{c}_4{U}_{XXY}+m_6{c}_5{V}_{XXY}+m_7{c}_6{(U_X^2)}_{XY}+m_8{c}_7{U}_{XXXX}+m_9{c}_8{U}_{XXYY}]\\ & +V[n_1{V}_{XT}+n_2{d}_1{(U_{X}V)}_X+n_3{d}_2{U}_X{U}_{XX}+n_4{d}_3{V}_X{V}_{XX}+n_5{d}_4{U}_{XXY}\\ & +n_6{d}_5{V}_{XXY}+n_7{d}_6{(V_X^2)}_{XY}+n_8{d}_7{V}_{XXXX}+n_9{d}_8{V}_{XXYY}\left]\right\},\end{array}$$

(20)

with $m_i,n_i$, $i=1,\cdots ,9$ are Lagrange coefficients and will be calculated later to determine the exact values. Carrying out the integration by parts in Equation (20) and taking $U_X{|}_R=U_Y{|}_R=U_T{|}_T=V_X{|}_R=V_Y{|}_R=V_T{|}_T=0$, the functional be rewritten as

$$\begin{array}{ll}J(U,V)=& {\int }_{R}dX{\int }_{R}dY{\int }_{T}dT\left\{\right[-m_1{U}_X{U}_T-m_2{c}_1{U}_X^2{V}_X-\frac{1}{2}{m}_3{c}_2{U}_X^3-\frac{1}{2}{m}_4{c}_3{U}_X{V}_X^2\\ & -m_5{c}_4{U}_X{U}_{XY}-m_6{c}_5{U}_X{V}_{XY}-m_7{c}_6(U_X^2)U_{XY}-m_8{c}_7{U}_{XX}^2-m_9{c}_8{U}_{XY}^2]\\ & -[n_1{V}_X{V}_T+n_2{d}_1{V}_X^2{U}_X+\frac{1}{2}{n}_3{d}_2{V}_X^3-\frac{1}{2}{n}_4{d}_3{V}_X{U}_X^2+n_5{d}_4{V}_X{V}_{XY}\\ & +n_6{d}_5{V}_X{U}_{XY}+n_7{d}_6{V}_X^2{V}_{XY}+n_8{d}_7{V}_{XX}^2+n_9{d}_8{V}_{XY}^2\left]\right\}.\end{array}$$

(21)

Using the variational method for this functional equation, integrating by parts to optimize this variational, the resulting forms are expressed as

$$\begin{array}{c}-2m_1{U}_{XT}-2m_2{c}_1{(U_{X}V)}_X-3m_3{c}_2{(U_X^2)}_X-3m_4{c}_3{(V_X^2)}_X-2m_5{c}_4{U}_{XXY}\hfill \\ -m_6{c}_5{V}_{XXY}-4m_7{c}_6{(U_X^2)}_{XY}+m_8{c}_7{U}_{XXXX}-m_9{c}_8{U}_{XXYY}=0,\hfill \\ -2n_1{V}_{XT}-2n_2{d}_1{(V_{X}U)}_X-3n_3{d}_2{(U_X^2)}_X-3n_4{d}_3{(V_X^2)}_X-2n_5{d}_4{U}_{XXY}\hfill \\ -n_6{d}_5{V}_{XXY}-4n_7{d}_6{(V_X^2)}_{XY}+n_8{d}_7{V}_{XXXX}-n_9{d}_8{V}_{XXYY}=0.\hfill \end{array}$$

(22)

Since the Equations (22) and (19) are equal, the values of all the Lagrangian constants in the equations can be obtained, $m_1=m_2=m_5=n_1=n_2=n_5=-\frac{1}{2}, m_3=m_4=n_3=n_4=-\frac{1}{3}, m_7=n_7=-\frac{1}{4}, m_8=n_8=1, m_6=m_9=n_6=n_9=-1$. Thus, the Lagrangian forms [55,56] of the integer order coupled gZK equations set is,

$$\begin{array}{c}{I}_1=\frac{1}{2}{U}_X{U}_T+\frac{1}{2}{c}_1{U}_X^2{V}_X+\frac{1}{3}{c}_2{U}_X^3+\frac{1}{3}{c}_3{U}_X^2{V}_X+c_4{U}_X{U}_{XY}\\ +\frac{1}{2}{c}_5{U}_X{V}_{XY}+\frac{1}{4}{c}_6(U_X^2)U_{XY}-c_7{U}_{XX}^2+c_8{U}_{XY}^2=0,\\ I_2=\frac{1}{2}{V}_X{V}_T+\frac{1}{2}{d}_1{V}_X^2{U}_X+\frac{1}{3}{d}_2{V}_X^3+\frac{1}{3}{d}_3{V}_X^2{U}_X+d_4{V}_X{V}_{XY}\\ +\frac{1}{2}{d}_5{V}_X{U}_{XY}+\frac{1}{4}{d}_6(V_X^2)V_{XY}-d_7{V}_{XX}^2+d_8{V}_{XY}^2=0.\end{array}$$

(23)

The fractional variational problem of Lagrange was obtained [57]. A natural generalization of Agrawal’s approach [53,58,59,60], was applied to the fractional calculus of constrained systems. In order to obtain time-space fractional gZK equation set, we use the Lagrangian to minimize certain functionals which will naturally contain fractional derivative terms. Analogously, based on the Definition 1 and Agrawal’s method, the Lagrangian forms of the time-space fractional coupled gZK equations are given as

$$\begin{array}{ll}{F}_1& =\frac{1}{2}{D}_T^{\alpha }U\times D_X^{\beta }U+\frac{1}{2}{c}_1{(D_X^{\beta }U)}^2{D}_X^{\beta }V+\frac{1}{3}{c}_2{(D_X^{\beta }U)}^3+\frac{1}{3}{c}_3{(D_X^{\beta }V)}^2{D}_X^{\beta }U+c_4{D}_X^{\beta }U\times D_{XY}^{\beta \omega }U\\ & +\frac{1}{2}{c}_5{D}_X^{\beta }U\times D_{XY}^{\beta \omega }V+\frac{1}{4}{c}_6{(D_X^{\beta }U)}^2{D}_{XY}^{\beta \omega }V-c_7{(D_X^{2\beta }U)}^2+c_8{(D_{XY}^{\beta \omega }U)}^2=0,\\ F_2& =\frac{1}{2}{D}_T^{\alpha }V\times D_X^{\beta }V+\frac{1}{2}{d}_1{(D_X^{\beta }V)}^2{D}_X^{\beta }U+\frac{1}{3}{d}_2{(D_X^{\beta }V)}^3+\frac{1}{3}{d}_3{(D_X^{\beta }U)}^2{D}_X^{\beta }V+d_4{D}_X^{\beta }V\times D_{XY}^{\beta \omega }V\\ & +\frac{1}{2}{d}_5{D}_X^{\beta }V\times D_{XY}^{\beta \omega }U+\frac{1}{4}{d}_6{(D_X^{\beta }V)}^2{D}_{XY}^{\beta \omega }U-d_7{(D_X^{2\beta }V)}^2+d_8{(D_{XY}^{\beta \omega }V)}^2=0,\end{array}$$

(24)

where $D_{XY}^{\beta \omega }f=D_Y^{\omega }\left[D_X^{\beta }f\right],D_X^{2\beta }f=D_X^{\beta }\left[D_X^{\beta }f\right]$.

It’s similar to what we did for the integral order equation, the functional of the time-space coupled gZK equations set has the form

$$J_F(U,V)={\int }_R{(dX)}^{\beta }{\int }_R{(dY)}^{\omega }{\int }_T{(dT)}^{\alpha }(F_1+F_2).$$

(25)

and the variation of functional Equation (24) leads to

$$\begin{array}{ll}\delta J_F(U,V)=& {\int }_R{(dX)}^{\beta }{\int }_R{(dY)}^{\omega }{\int }_T{(dT)}^{\alpha }[D_T^{\alpha }(\frac{\partial F_1}{\partial D_T^{\alpha }U})+D_X^{\beta }(\frac{\partial F_1}{\partial D_X^{\beta }U})+D_Y^{\omega }(\frac{\partial F_1}{\partial D_X^{\beta }U})\\ & -D_X^{2\beta }(\frac{\partial F_1}{\partial D_X^{2\beta }}U)]\delta JU+{\int }_R{(dX)}^{\beta }{\int }_R{(dY)}^{\omega }{\int }_T{(dT)}^{\alpha }[D_T^{\alpha }(\frac{\partial F_2}{\partial D_T^{\alpha }V})\\ & +D_X^{\beta }(\frac{\partial F_2}{\partial D_X^{\beta }V})+D_Y^{\omega }(\frac{\partial F_2}{\partial D_X^{\beta }V})-D_X^{2\beta }(\frac{\partial F_2}{\partial D_X^{2\beta }}V)]\delta JV.\end{array}$$

(26)

According to the properties introduced at the beginning, integrating the Equation (26) by parts and making $\delta J_F(U,V)=0$, optimizing the variation of the function, the following form Euler-Lagrange equations [53] for the time-space fractional coupled gZK equations set can be given

$$\begin{array}{c}{D}_T^{\alpha }(\frac{\partial F_1}{\partial D_T^{\alpha }U})+D_X^{\beta }(\frac{\partial F_1}{\partial D_X^{\beta }U})+D_Y^{\omega }(\frac{\partial F_1}{\partial D_X^{\beta }U})-D_X^{2\beta }(\frac{\partial F_1}{\partial D_X^{2\beta }}U)=0,\hfill \\ D_T^{\alpha }(\frac{\partial F_2}{\partial D_T^{\alpha }V})+D_X^{\beta }(\frac{\partial F_2}{\partial D_X^{\beta }V})+D_Y^{\omega }(\frac{\partial F_2}{\partial D_X^{\beta }V})-D_X^{2\beta }(\frac{\partial F_2}{\partial D_X^{2\beta }}V)=0.\hfill \end{array}$$

(27)

The last step is to plug expressions for $F_1,F_2$ given by Equation (24) and fractional potential functions $D_X^{\beta }U(X,Y,T)=u(X,Y,T),D_X^{\beta }V(X,Y,T)=v(X,Y,T)$ in this equation, the final equations set is

$$\begin{array}{c}{D}_T^{\alpha }u+c_1{D}_X^{\beta }(uv)+c_2{D}_X^{\beta }(u^2)+c_3{D}_X^{\beta }(v^2)+c_4{D}_X^{\beta }{D}_Y^{\omega }u\hfill \\ +c_5{D}_X^{\beta }{D}_Y^{\omega }v+c_6{D}_X^{\beta }{D}_Y^{\omega }(u^2)+c_7{D}_X^{3\beta }u+c_8{D}_X^{\beta }{D}_Y^{2\omega }v=0,\hfill \\ D_T^{\alpha }v+d_1{D}_X^{\beta }(uv)+d_2{D}_X^{\beta }(u^2)+d_3{D}_X^{\beta }(v^2)+d_4{D}_X^{\beta }{D}_Y^{\omega }u\hfill \\ +d_5{D}_X^{\beta }{D}_Y^{\omega }v+d_6{D}_X^{\beta }{D}_Y^{\omega }(v^2)+d_7{D}_X^{3\beta }v+d_8{D}_X^{\beta }{D}_Y^{2\omega }v=0.\hfill \end{array}$$

(28)

It is the time-space fractional coupled gZK equations set. This new set of fractional-order equations will promote the study of fractional-order nonlinear equations and has great significance for the future research.

4. Solutions of Time-Space Fractional Coupled gZK Equations Set

In the previous section, the integral order coupled equations set is transformed into the fractional order coupled equations set. To further explore the Rossby solitary waves interaction between two-layer fluids, we solved the time-space fractional coupled gZK equations set by improved $(G^{\prime }/G)$-expansion method [42,43] in this section.

Firstly, by using the following fractional traveling wave transformations

$$u(X,Y,T)={\varphi }_1(\xi ),v(X,Y,T)={\varphi }_2(\xi )$$

(29)

$$\xi =\frac{k_1{X}^{\beta }}{\mathrm{\Gamma }(1+\beta )}+\frac{k_2{Y}^{\omega }}{\mathrm{\Gamma }(1+\omega )}-\frac{\sigma T^{\alpha }}{\mathrm{\Gamma }(1+\alpha )}$$

(30)

where the $k_1,k_2,\sigma$ are constants, and using chain rule, we have the equations

$$\begin{array}{c}{D}_T^{\alpha }u={\rho }_T^{\prime }\frac{\partial {\varphi }_1}{\partial \xi }{D}_T^{\alpha },D_T^{\alpha }v={\rho }_T^{\prime }\frac{\partial {\varphi }_2}{\partial \xi }{D}_T^{\alpha },\hfill \\ D_X^{\beta }u={\rho }_X^{\prime }\frac{\partial {\varphi }_1}{\partial \xi }{D}_X^{\beta },D_X^{\beta }v={\rho }_X^{\prime }\frac{\partial {\varphi }_2}{\partial \xi }{D}_X^{\beta },\hfill \\ D_Y^{\omega }u={\rho }_Y^{\prime }\frac{\partial {\varphi }_1}{\partial \xi }{D}_Y^{\omega },D_Y^{\omega }v={\rho }_Y^{\prime }\frac{\partial {\varphi }_2}{\partial \xi }{D}_Y^{\omega },\hfill \end{array}$$

(31)

with ${\rho }_T,{\rho }_X,{\rho }_Y$ are the fractal indexes, without loss of generality we can make ${\rho }_T={\rho }_X={\rho }_Y=l$, thereinto l is a constant.

Put the Equation (29) with Equations (30) and (31) into the Equation (28), the time-space fractional coupled equations set can be reduce to the ordinary differential coupled equations set

$$\begin{array}{c}-\sigma {\varphi }_1^{\prime }+k_1(c_1{\varphi }_2{\varphi }_1^{\prime }+c_1{\varphi }_1{\varphi }_2^{\prime }+c_2{\varphi }_1{\varphi }_1^{\prime }+c_3{\varphi }_2{\varphi }_2^{\prime })+k_1{k}_{2}l[c_4{\varphi }_1^{″}+c_5{\varphi }_2^{″}\hfill \\ +c_6{\varphi }_1{\varphi }_1^{″}+c_6{({\varphi }_1^{\prime })}^2]+(c_7{k}_1^3{l}^2+c_8{k}_1{k}_2^2{l}^2){\varphi }_1^{‴}=0,\hfill \\ -\sigma {\varphi }_2^{\prime }+k_1(d_1{\varphi }_2{\varphi }_1^{\prime }+d_1{\varphi }_1{\varphi }_2^{\prime }+d_2{\varphi }_1{\varphi }_1^{\prime }+d_3{\varphi }_2{\varphi }_2^{\prime })+k_1{k}_{2}l[d_4{\varphi }_1^{″}+d_5{\varphi }_2^{″}\hfill \\ +d_6{\varphi }_2{\varphi }_2^{″}+d_6{({\varphi }_2^{\prime })}^2]+(d_7{k}_1^3{l}^2+d_8{k}_1{k}_2^2{l}^2){\varphi }_2^{‴}=0.\hfill \end{array}$$

(32)

Secondly, suppose Equation (32) have the solutions in relation to $(G^{\prime }/G)$ as follows

$${\varphi }_1(\xi )=e_0+e_1(\frac{G^{\prime }}{G}),{\varphi }_2(\xi )=f_0+f_1(\frac{G^{\prime }}{G}).$$

(33)

where $e_0,e_1,f_0,f_1$ are computed later, $G=G(\xi )$ satisfies the second ordinary differential equation

$$G^{″}+m{G}^{\prime }+nG=0.$$

(34)

thereinto the apostrophe represents derivative with respect to $\xi$ and $m,n$ are parameters.

By substituting Equation (33) with Equation (34) into (32), collecting all terms with the same order of $(G^{\prime }/G)$, equating each coefficient of the resulting polynomial to zero, we can obtain a set of algebraic equations for $k_1,k_2,\sigma ,l,e_0,e_1,f_0,f_1,m$ and n.

The last, due to the very complicated coefficients in the equations, the calculation process is extremely complicated, in order to calculate accurate results, take $c_1=d_1=c_6=d_6=2,c_2=c_3=d_2=d_3=6,c_4=c_5=c_7=c_8=d_4=d_5=d_7=d_8=1$. Solving the algebraic equations system and subsequently substituting these constants $k_1,k_2,\sigma ,l,e_0,e_1,f_0,f_1,m$ and n, we get the solutions of the time-space fractional gZK coupled equations set that we want in following as

Case 1:

$$\begin{array}{c}\sigma =\frac{(H_{1}l{k}_1+8k_1)(I_1{H}_1+H_3)-4(2H_2{H}_1+H_3)k_1-I_1{H}_{1}l+H_4{H}_1{k}_{2}l}{4k_2^{2}l},e_0=\frac{I_1{H}_1+H_2}{4},\hfill \\ e_1=e_1,f_0=-\frac{I_1{H}_1+H_3}{4k_2^{2}l},f_1=e_1,l=l,m=-\frac{H_1}{4k_2},n=0,\hfill \end{array}$$

(35)

$$\left\{\begin{array}{c}{u}_1=\frac{I_1{H}_1+H_3}{4}\pm e_1\frac{H_2}{8k_2}\left[\frac{C_{1}sinh(\pm \frac{H_2}{8k_2}\xi )+C_{2}cosh(\pm \frac{H_2}{8k_2}\xi )}{C_{1}cosh(\pm \frac{H_2}{8k_2}\xi )+C_{2}sinh(\pm \frac{H_2}{8k_2}\xi )}\right]+\frac{H_1}{8k_2},\hfill \\ v_1=\frac{I_1{H}_1+H_2}{4}\pm f_1\frac{H_2}{8k_2}\left[\frac{C_{1}sinh(\pm \frac{H_2}{8k_2}\xi )+C_{2}cosh(\pm \frac{H_2}{8k_2}\xi )}{C_{1}cosh(\pm \frac{H_2}{8k_2}\xi )+C_{2}sinh(\pm \frac{H_2}{8k_2}\xi )}\right]+\frac{H_1}{8k_2},\hfill \end{array}\right.$$

(36)

where $C_1,C_2$ are arbitrary constants, $H_1=2k_2{e}_1-3(k_1^2+k_2^{2}l),I_1=-(k_1^2+k_2^2)l^2,H_2=2e_1{k}_{2}l+5I_1,H_3=k_2^{2}l-4e_1{k}_2,H_4=l{k}_1{k}_2-4e_1{k}_1$.

Case 2:

$$\begin{array}{c}\sigma =\frac{-16H_5{k}_1+I_2{k}_1{k}_2{l}^2+I_3{k}_{2}l-4l^2{k}_1{k}_2^2}{k_{2}l},e_0=\frac{H_5}{k_{2}l},e_1=-\frac{H_6}{2k_2},\hfill \\ f_0=-\frac{H_5}{k_{2}l},f_1=-\frac{H_6}{2k_2},l=l,m=0,n=\frac{1}{2},\hfill \end{array}$$

(37)

$$\left\{\begin{array}{c}{u}_2=\frac{H_5}{k_{2}l}-\frac{H_6\sqrt{2}}{4k_2}\left[\frac{-C_{1}sin(\frac{\sqrt{2}}{2}\xi )+C_{2}cos(\frac{\sqrt{2}}{2}\xi )}{C_{1}cos(\frac{\sqrt{2}}{2}\xi )+C_{2}sin(\frac{\sqrt{2}}{2}\xi )}\right]-\frac{\sqrt{2}}{2},\hfill \\ v_2=-\frac{H_5}{k_{2}l}-\frac{H_6\sqrt{2}}{4k_2}\left[\frac{-C_{1}sin(\frac{\sqrt{2}}{2}\xi )+C_{2}cos(\frac{\sqrt{2}}{2}\xi )}{C_{1}cos(\frac{\sqrt{2}}{2}\xi )+C_{2}sin(\frac{\sqrt{2}}{2}\xi )}\right]-\frac{\sqrt{2}}{2},\hfill \end{array}\right.$$

(38)

where $C_1,C_2$ are arbitrary constants, $H_5=\frac{k_1{k}_2^{2}l-2[3(k_1^2+k_2^2)l+4k_2]k_1}{k_2},I_2=-(2k_1^2+3k_2^2)k_1{l}^2+4k_1{k}_{2}l,I_3=4(k_1^3+k_1{k}_2^2)l^2,H_6=-3(k_1^2+k_2^2)l+4k_2$.

Case 3:

$$\begin{array}{c}\sigma =\frac{I_{4}n}{H_7^2},e_0=-12\frac{H_8}{H_7},e_1=\frac{4k_2^2}{H_7},f_0=36\frac{H_8}{H_7},\hfill \\ f_1=0,l=\frac{12k_2}{H_7},m=-2n+1,n=n,\hfill \end{array}$$

(39)

(i)

When $4n^2-8n+2>0$, the hyperbolic solutions as:

$$\left\{\begin{array}{c}{u}_{31}=-12\frac{H_8}{H_7}+\frac{4k_2^2\sqrt{4n^2-8n+2}}{2H_7}\left[\frac{C_{1}sinh(\sqrt{4n^2-8n+2}\xi )+C_{2}cosh(\sqrt{4n^2-8n+2}\xi )}{C_{1}cosh(\sqrt{4n^2-8n+2}\xi )+C_{2}sinh(\sqrt{4n^2-8n+2}\xi )}\right]-\frac{1-2n}{2},\hfill \\ v_{31}=36\frac{H_8}{H_7}-\frac{1-2n}{2},\hfill \end{array}\right.$$

(40)

(ii)

When $4n^2-8n+2>0$, the trigonometric solutions as:

$$\left\{\begin{array}{c}{u}_{32}=-12\frac{H_8}{H_7}-\frac{4k_2^2\sqrt{4n^2+2}}{2H_7}\left[\frac{-C_{1}sin(\sqrt{4n^2-8n+2}\xi )+C_{2}cos(\sqrt{4n^2-8n+2}\xi )}{C_{1}cos(\sqrt{4n^2-8n+2}\xi )+C_{2}sin(\sqrt{4n^2-8n+2}\xi )}\right]-\frac{1-2n}{2},\hfill \\ v_{32}=36\frac{H_8}{H_7}-\frac{1-2n}{2},\hfill \end{array}\right.$$

(41)

(iii)

When $4n^2-8n+2>0$, the solutions as:

$$\left\{\begin{array}{c}{u}_{33}=\left(\frac{C_2}{C_1+C_2\xi }-\frac{1-2n}{2}\right),\hfill \\ v_{33}=\left(\frac{C_1}{C_2+C_1\xi }-\frac{1-2n}{2}\right),\hfill \end{array}\right.$$

(42)

where $C_1,C_2$ are arbitrary constants, $I_4=288k_1{k}_2^2(k_1^2+k_2^2),H_7=9k_1^2+7k_2^2,H_8=2k_1^{2}n+2k_2^{2}n-k_1^2-k_2^2$.

The exact solutions obtained by the research show that the fluctuation relationship of each wave not only contains its own wave number and amplitude, but also contains the amplitude of another wave, which explains the main characteristics of nonlinear waves interaction [61]. On the other hand, the interaction between Rossby waves has a great influence on the propagation stability of waves. When both waves are unstable, and they are still unstable after the interaction. When at least one wave is stable, the two waves may be stable or unstable through the interaction, which is related to the values of the coupling term coefficients.

5. Conclusions

In this paper, based on the quasi-geostrophic vortex equation set, we obtain the (2+1)-dimensional coupled gZK equations set for the first time, which can describe Rossby solitary waves interactions in two-layer fluids. Next, according to the new model and using the semi-inverse method and the fractional variational principle, a new (2+1)-dimensional time-space fractional coupled gZK equations set is obtained. Then, we solved the (2+1)-dimensional time-space fractional coupled gZK equations set. The coupled gZK equations set is the evolution of a single gZK equation in two-layer fluids, which is of great significance for the study of Rossby waves propagation and interaction. How Rossby solitary waves described by coupled equations set interacts specifically and how the energy changes during the interaction, which are our research aim in the future.