3. Determination of the Hertz Potential

2005GI000131-fig02
Figure 2
[6]  To solve the above formulated problem, we use the conformal mapping x(u) transforming the z leq h region of the complex plane u = x + i z into some one cohesive region z1 leq f(x1) of the x = x1 + i z1 plane (see Figure 2). Under such transformation, the z = h plane is transformed into a curved surface z1 = f(x1), and the boundary problem (2)-(3) is rewritten in the (x1, z1) coordinate system in the form

eq006.gif(4)

eq007.gif(5)

Here partial/partial n is the derivative along the normal to the line of the curved boundary, and the values

eq008.gif

and

eq009.gif

are, respectively, modulus of the derivative of the mapping x(u) and of the inverse to it transformation u(x) (it is evident that b = 1/n ).

[7]  Using the uncertainty in the b(x1, z1) function, we choose it in such a way that the equality

eq010.gif(6)

(where g = i k0 h0 = const ) would be fulfilled. Therefore, at the boundary

eq011.gif

Finally we come to the problem on electromagnetic wave propagation in the semispace with the known inhomogeneous refraction index and one-dimensionally curved boundary having a constant impedance.

[8]  To solve the boundary problem (4), (6), we apply the operator

eq012.gif(7)

to the right-hand and left-hand sides of the equation (4). Here partial /partial s is the derivative along the direction of the tangent line to the s(x1,z1) curve (see Figure 2). We put on the family of curves s the following requirements:

[9]  1. Only one line s passes through every point of the (x1,z1) plane.

[10]  2. In the point of the boundary z1 = f(x1) the direction of the tangent line to the corresponding curve of the s family coincides with the direction of the normal n along which the differentiation in the boundary condition (6) is performed (see Figure 2 where the s(x1,z1) curves are shown by dashed curves). Therefore at the points of the curved boundary, the following equality partial/partial s = partial/partial n is true.

[11]  For the sake of certainty, we choose particular family s as a set of curves that are transformed into the lines x = const under the inverse conformal mapping u(x), which transforms the considered curved semispace z1 leq f(x1) of the complex plane x = x1 + i z1 into the z leq h region of the u = x + i z plane. Under this transformation, the curve z1 = f(x1) is transformed into the line z = h.

[12]  In an analogy with Zaboronkova et al. [2003] we perform conformal transformation in variables (x1, z1) and then come back to the initial coordinates (x,z). Zaboronkova et al. [2003] demonstrated that for the function

eq013.gif

satisfying the Dirichlet's zero condition at the upper boundary, it is true

eq014.gif(8)

Therefore we come to the following equation in the initial variables (x,z):

eq015.gif(9)

In the same way as Al'pert et al. [1953] and Bezrodny et al. [1984], we will consider the case of a grazing propagation when the horizontal distance x from the source to the observational point is much larger than the vertical distance h from this point to the disturbed boundary. In this case the grazing angle of the falling wave is qs ll 1. In such case, with the accuracy of the terms sim qs2, the following relation is true:

eq016.gif

Then we are able to rewrite the equation (9) in the form

eq017.gif(10)

where

eq018.gif

Here and further the function 1+N(x, z) is interpreted as the dielectric permeability of the effective "medium". Such "medium" is interpreted as an internal factor connected with the method of solution of the problem. To each type of the impedance, inhomogeneity corresponds its own dependence N(x, z). To solve the auxiliary boundary problem (8) and (10), we apply the geometric optic method (because l/lg ll 1 ). Using the methodology of Al'pert et al. [1953], we assume that in the zgeq h region there exists an effective medium with the dielectric permeability e+ = 1 + N(x,z - h) symmetric relative to the permeability e- = 1 + N(x,z) of the medium at z leq h. We designate the corresponding refraction indices as npm = (e<img src="latex006.gif" alt="pm">)1/2. We assume that in the point M0'(0, 2h) symmetrical to M0(0,0) (the place of the location of the source (1) relative to the plane z = h, there is an imaginary source j', the auxiliary function G=[n (partial /partial (- z)) + g ]P corresponding to this source [see Al'pert et al., 1953]. It is evident that at the z = h level the sum of the fields of these two sources is equal to zero. So the boundary condition (8) is automatically satisfied.

[13]  We designate as G1 fal and G1 ref the auxiliary function for the wave from the source (1) falling from below onto the boundary z =h and for the corresponding reflected wave, respectively. Then

eq019.gif

eq020.gif

eq021.gif

eq022.gif(11)

Here M1 = M1(x/2,h) is the central point of the region important for reflection from the weak boundary z = h [see Ginzburg, 1960], R1 = R1 (x/2,h), and Rh = Rh (x,0). The value A is the amplitude of the function G1 fal, multiplier r and multiplicand T are the refraction and transmission indices, respectively, and the factor A determines the divergence of the ray reflected from the boundary [see Kravtsov and Orlov, 1980].

[14]  In a similar way we introduce an auxiliary function G2 tr for the waves arriving from above from the source j':

eq023.gif

eq024.gif

eq025.gif(12)

Here R0' = R0'(0, 2h) is the radius vector of the point M0' determining the coordinates of the symmetrical source j'. We can also write A( Rh - R0' ) =A( R1 - R0' ) A( Rh - R1). It follows from the problem symmetry that

eq026.gif

(that is why, in particular, the phases of the functions G1 ref and G2 tr coincide) and A( R1 - R0' ) = A( R1 - R0).

[15]  Finally, we have

eq027.gif

eq028.gif(13)

Here it is taken into account that 1 + r = T. Thus the formulated problem is reduced to a calculation of the amplitude and phase of the ray propagating in a smoothly inhomogeneous medium with the dielectric permeability epm.

[16]  Taking into account grazing propagation, one can easily obtain from the expression G= gP an approximate formula for calculation of the Hertz potential P with the accuracy to the terms of the order of qs2:

eq029.gif(14)

[17]  On the basis of the derived formulae (13) and (14), we will study the influence of different kinds of IGW on propagation of electromagnetic waves. Let us consider two cases that correspond to the models with imaginary and real impedance values at the upper reflecting boundary. In particular, we will consider the impact of the IGW of periodical and soliton types.


AGU

Powered by TeXWeb (Win32, v.2.0).