3. Principal Equations

[17]  Using the second Green formula one can obtain an expression for function P(r,j,z) in any point of the D region via the integral over the irregularity surface Sp cup Sl:

eq037.gif

eq038.gif

eq039.gif(2)

where R(r,j,z) Sp, Sl denotes the observational point, R'(r',j',z')in Sp, Sl corresponds to the integration point, n' is the normal directed outside the wave volume, P0( R) is the field of the initial source in a regular waveguide with a thickness of h and homogeneous walls with the impedances dg and di, and P0( R, R') is the Green function. The expression for the latter may be obtained from the formula for P0( R) substituting r by r1=(r2+ r'2 - 2rr' cos(j-j'))1/2 and zt by h (see Figure 1). In the limiting case R to Sp, Sl in the right-hand side of equation (2), there appears an extra term of the form P( R)/2, this fact being related to the jump in the normal derivative of the Green function partial P0( R', R)/partial n'.

[18]  This very choice of the Green function P0( R, R') in the form of a solution for the regular waveguide makes it possible to limit the region of integration in equation (2) down to the dimensions of the irregularity surfaces Sp and Sl. The method of construction of an approximate solution of equation (2) with the accuracy up to the terms of the order of O((kr)-1) was described in detail by Soloviev [1998]. The condition of its applicability is krgg1 (that is, the observational point should be located in the wave zone from the source). There is no additional conditions, for example, on the dimensions of the irregularity. In this paper we describe only the main stages of the solution of equation (2).

[19]  Two-dimensional equation (2) is solved by asymptotic transformation into an one-dimensional one. To do that we reveal the quickly oscillating multiplier and introduce the attenuation function

eq040.gif

eq041.gif

If one introduces the ecliptic coordinates

eq042.gif

eq043.gif

eq044.gif

equation (2) for attenuation function would have the form

eq045.gif

eq046.gif

eq047.gif

eq048.gif

eq049.gif

[20]  When krgg1, the exponential multiplier in the first integral becomes quickly changing in the direction transverse to the propagation circuit on the background of the multiplier left under the integral.

eq050.gif

[21]  The latter makes it possible to perform the calculation of the integral in terms of dv over the cylinder base by the stationary phase method. As a result the integral in terms of variable v, which may be considered as a curvature integral along the irregularity boundary. The contour of integration is passed clockwise. Thus, with an accuracy up to the terms of the order of O((kr)-1), we obtain the expression for the attenuation function

eq051.gif

eq052.gif

eq053.gif(3)

where

eq054.gif

eq055.gif

eq056.gif

eq057.gif

[22]  To solve equation (3), we use the numerical-analytical method of semi-inversion [Soloviev and Agapov, 1997], which combines the direct inversion of the dominant part of the integral operator of the problem, that is a Volterra operator, with the iterative process by which the remaining part of the integral operator is inverted through successive approximations. The influence of the side surface of the cylindrical irregularity Sl is taken into account by convertional stepwise procedure. Below, all the results of numerical calculations are presented for the attenuation function V( R). This function physically demonstrate the difference of the field in the waveguide with a three-dimensional irregularity from the dipole field over an ideally conducting surface.


AGU

Powered by TeXWeb (Win32, v.2.0).