![MathJax Logo](/templates/jsp/_style2/_tandf/pb2/images/math-jax.gif)
Abstract
In this paper, we consider the two-dimensional (2D) inverse acoustic scattering by a sound-soft, infinite rough surface. Our approach to this problem is based on the nonlinear integral equation method which was developed by Kress and Rundell for bounded obstacles. By using multiple frequency data, we derive a stable and accurate reconstruction method to image the profile of the rough surface. Numerical experiments are presented to illustrate the effectiveness of the inversion method.
1 Introduction
In this paper, we study the 2D Dirichlet boundary value problem for the Helmholtz equation in a nonlocally perturbed half-plane. The Dirichlet boundary problem models time-harmonic acoustic scattering by a sound-soft, infinite rough surface as well as electromagnetic wave propagation over outdoor ground, sea surfaces or optical grating.
As in Figure , let the scattering surface be the graph of some function
, i.e.
and let the space above
be denoted by
, i.e.
Whenever we wish to express the dependence of
and
on the boundary function
, we write
for
and
for
. Let
be the incident field and
be the scattered field. The total field
satisfies the Helmholtz equation in
, and vanishes on
for a sound-soft surface. In order to ensure the uniqueness of the solution, we assume that
satisfies some radiation condition along with the following growth condition in the
-direction: for some
1.1
1.1 In this paper,
is required to satisfy the upward propagating radiation condition (UPRC) [Citation1]: for some
1.2
1.2 where
and
. The properties of UPRC are shown in Theorem 2.9 of [Citation1].
Here, we need a function space for
, which is a Banach space under the norm
. The above problem of scattering by an infinite sound-soft rough surface can now be formulated as the following Dirichlet boundary value problem (DP) for the scattered field
: given
, determine
such that:
(1) |
| ||||
(2) | |||||
(3) |
| ||||
(4) |
|
The inverse scattering problem we are concerned with is: given the incident field and the near-field
on some open subset
, determine the scattering surface
which causes such a near-field on
.
Chandler-Wilde and Ross studied the inverse problem for scattering by infinite surfaces in a lossy medium in [Citation2] where they imposed the following growth condition on
:
for some
and
. Under the growth condition, they proved that the inverse problem has at most one solution.
For numerical investigation of the inverse problem, many authors obtained the accurate reconstructions by a variety of methods. For global rough surface, Burkard and Potthast studied the inverse problem in via the Kirsch-Kress scheme [Citation11] and developed a time-domain probe method for rough surface reconstruction in
,[Citation12] and Lines used Point Source Method to reconstruct perfectly conducting unbounded rough surfaces when the incident field is a time-harmonic point source in [Citation13]. An extension of the Point Source Method is also given in [Citation13] when the incident field is not necessarily time-harmonic. For local rough surface, Bao and Lin [Citation14] developed a stable and convergent reconstruction algorithm by multiple frequency measurements of near-field, and Zhang and Zhang [Citation15] proposed a novel integral equation formulation defined on a bounded curve and obtained an accurate reconstruction of the profile from multiple frequency far-field data. The main ideas of employing multiple frequency data include overcoming the ill-posedness of the inverse problem and increasing resolution limited by Rayleigh’s criteria. For more related work on multiple frequency data, we refer the readers to [Citation16–Citation18] and the references therein. In this paper, our attention is restricted to the reconstruction of a global sound-soft rough surface by multiple frequency near-field data, corresponding to incident plane waves and point sources.
For solving the inverse scattering problem, there are many iterative methods, for example Newton-type iterations. The derivatives occurring in the iterative process can be obtained through associated boundary value problems.[Citation19] For simplicity, in [Citation20] Kress and Rundell developed an iterative method named nonlinear integral equations in which the derivatives can be explicitly expressed in terms of integral operators rather than boundary value problems. This reduces the computational costs. Recently, nonlinear integral equation method is very popular. Ivanyshyn and Johansson used it to reconstruct the shape of a planar acoustically sound-soft obstacle in [Citation21]. By the same method, Ivanyshyn and Kress reconstructed the surface impedance function of a three-dimensional acoustic scatterer with a known shape in [Citation22]. Furthermore, in [Citation23] they studied the inverse boundary value problems for inclusions and cracks via the method. For the 2D inverse electrical impedance problem in the case of piecewise constant conductivities, Eckel and Kress reconstructed the shape and conductivities accurately with a finite number of Cauchy pairs in [Citation24]. To the best of our knowledge, there is no paper to study the method for rough surface scattering. Therefore, the main purpose of this paper is to extend the nonlinear integral equation method to rough surface scattering.
The paper is organized as follows. In Section 2, we focus on the well-posedness of the direct scattering problem. This is followed by the study of the Nyström method in Section 3. After describing the nonlinear integral equation method for the inverse scattering problem in Section 4, we conclude with several numerical experiments in Section 5.
2 An integral equation formulation of the direct problem
In this section, we propose an integral equation formulation developed in [Citation6], and establish the corresponding unique solvability in . For detailed discussions, we refer to [Citation6].
For two positive constants ,
, we define
by
Let the rough surface function
with
. We consider two different kinds of incoming waves in this paper. One is an time-harmonic acoustic plane wave
, where
is the wave number satisfying
and
, and
is the incident direction given by
with the incident angle
. The other is a sum of several point sources
, where
is the location of the point source, and for
,
is the free-space Green’s function for Helmholtz equation while
is the Hankel function of the first kind of order zero for
.
When , from [Citation6] we can look for the solution to the problem (DP) in the form of a combined double- and single-layer potential
2.1
2.1 for some
, where
is the Dirichlet Green’s function for
in the upper half-plane
,
,
is an arbitrary complex number to be fixed later and
denotes the normal vector at
pointing out of
.
When , instead of (Equation2.3
2.3
2.3 ) we will find a solution in the following form [Citation25]:
2.2
2.2 where
, and
denotes the wave length. We represent the solution
in the form of (Equation2.3
2.3
2.3 ), then we can obtain the following theorem (Theorem 3.2 in [Citation6]):
Theorem 2.1
The combined double- and single-layer potential (Equation2.32.3
2.3 ) satisfies the problem (DP) with
, provided
satisfies the boundary integral equation
2.3
2.3
Remark 2.2
It is easy to see that in the case of (Equation2.3
2.3
2.3 ). Similarly, for the case of (Equation2.4
2.4
2.4 ), we can get (Equation2.5
2.5
2.5 ) with
, a sum of Dirichlet Green functions. This implies that
as
for
, which is the advantage of the new form (Equation2.4
2.4
2.4 ). We may choose suitable point source positions
to let
below
. Then the form (Equation2.4
2.4
2.4 ) satisfies the condition (3) of the problem (DP) with
.
For , we denote
. Then we can represent the boundary integral Equation (Equation2.5
2.5
2.5 ) in
by defining
and the integral kernel
:
2.4
2.4 With the kernel, we can rewrite the integral operator in (Equation2.5
2.5
2.5 ) in the simple form:
2.5
2.5 We will denote
by
to express the dependence of the rough surface
. Then (Equation2.5
2.5
2.5 ) becomes:
2.6
2.6 where
is the identity operator. For the solvability of (Equation2.8
2.6
2.6 ) in
, we have the following theorem (Theorem 3.4 in [Citation6]):
Theorem 2.3
Let . Then for all
the integral operator
is bijective (and so boundedly invertible) with
Thus, the integral Equations (Equation2.5
2.5
2.5 ) and (Equation2.8
2.6
2.6 ) have exactly one solution for every
and
, with
for some constant
depending only on
and
.
3 The Nyström method for the direct problem
For the numerical solution of the boundary integral Equation (Equation2.82.6
2.6 ), we use the Nyström method based on appropriately weighted numerical quadratures on an equidistant mesh. Colton and Kress used it to obtain the numerical solution of the scattering problem for bounded obstacles in [Citation26]. Meier et al. developed a Nyström method for the scattering problem by rough surfaces and diffraction gratings in [Citation10], where they established stability and convergence of this method and proved that the convergence rates depend on the smoothness of components of the integral kernel. For details of some results in this section, we refer to [Citation10].
In order to introduce the Nyström method, we define some function spaces and integral operators. We denote by the Banach space of all functions whose derivatives up to order
are bounded and continuous on
. Furthermore, we define
which is a closed subspace of
. In addition, let
with the weight
and
with the weight
.
Next, we define the improper integral operator by
In numerical quadrature formula, we replace
by
which is the unique trigonometric polynomial with the interpolation property
.
can be explicitly expressed by:
where the coefficients are given by
Then from [Citation27], we approximate
by
which is defined by
3.1
3.1 where
for
In order to calculate the integral in (Equation2.8
2.6
2.6 ), we introduce the next integral:
where
, for some
and
. By the trapezium rule, we approximate the integral by
3.2
3.2 From [Citation10], we know that this approximation converges rapidly as
.
In the Nyström method, the kernel in (Equation2.7
2.5
2.5 ) is required to be the following form:
3.3
3.3 Here
and
for some
. Then we can approximate (Equation2.7
2.5
2.5 ) by using (Equation3.9
3.1
3.1 ) and (Equation3.10
3.2
3.2 ).
Note thatWith
, we have
where
Here
and
have logarithmic singularity at
, and
is continuous at
. We split the kernels
and
into
where
In particular, using
and
where
denotes Euler’s constant, we can deduce the diagonal terms
Using
and
(see (5.16.3) in [Citation28]), where
denotes the gamma function, we can obtain the diagonal terms
Split off the logarithmic singularity, the kernels
,
,
and
turn out to be analytic. Set
for any
, then
3.4
3.4 Now we have written
in the form (Equation3.12
3.4
3.4 ). In order to obtain (Equation3.11
3.3
3.3 ), our approach is to express
,
by
,
using Theorem 2.1 in [Citation10]. Above all, we introduce a cut-off function
satisfying that (1)
(2)
(3)
(4)
In calculation, we let
With
and (2.6), (2.7) in [Citation10],
and
are given by:
with the diagonal terms
By Theorem 4.3 and Theorem 2.1 in [Citation10], we can deduce that when the rough surface function
,
and
. Therefore, the integral operator
can be divided into two parts as:
where
For
, define the period extension operator
by
Due to the facts that
and
is
period function with respect to
, it can be obtained that
By (Equation3.9
3.1
3.1 ) and (Equation3.10
3.2
3.2 ), we define
and
as
In the Nystr
m method, we approximate
by
. Clearly,
where
. Then the integral Equation (Equation2.8
2.6
2.6 ) is replaced by the approximating equation:
3.5
3.5
Remark 3.1
In [Citation10], the authors established the stability analysis of the Nyström method. It depends crucially on the error estimates in Theorem 3.3 which establish the convergence of to
. Besides, the convergence of the method is also proved. Here we omit the detailed results.
The multiple frequency data used in all examples in Section 5 are obtained by solving (Equation3.133.5
3.5 ) through the collocation method. Here we define parameters used in all examples in Section 5. Let the incident direction
when the incident wave is the plane wave. While the point sources as the incident wave we set
in (Equation2.4
2.4
2.4 ),
, a total of 81 point sources, and
is the symmetry point of
with respect to the plane
. Furthermore, let
in (Equation2.3
2.3
2.3 ) to guarantee that the condition
holds in Theorem 2.3. When solving (Equation3.13
3.5
3.5 ), we set the truncated interval as
and put 10 equidistant nodes in each wave length
, and then the step length
with
. Let
replace
in (Equation3.13
3.5
3.5 ),
. After obtaining the density
from (Equation3.13
3.5
3.5 ), we use the trapezoidal quadrature formula to calculate the scattered field
on the receiving plane by (Equation2.3
2.3
2.3 ) for incident plane waves and (Equation2.4
2.4
2.4 ) for incident point sources. We put the receiving points on the plane
from
to
with step length
, a total of
equidistant nodes. Unless otherwise stated we let
.
4 Nonlinear integral equations for the inverse problem
The inverse scattering problem we are considering consists of reconstructing the rough surface from the measured data
,
, corresponding to the incident field
or
. In this paper, we will solve the inverse scattering problem via the nonlinear integral equation method. The main idea of the method is to prove the equivalence between the inverse scattering problem and an integral scheme. Firstly, we derive the integral scheme: from Section 2, we know that the density
in (Equation2.3
2.3
2.3 ) or (Equation2.4
2.4
2.4 ) can be solved from
4.1
4.1 Then the scattered field
on
can be reformulated with the known density
as
4.2
4.2 Here
when
and
when
. With the integral schemes (Equation4.14
4.1
4.1 ) and (Equation4.15
4.2
4.2 ), we can immediately state the following equivalence theorem.
Theorem 4.1
Assume the rough surface can be uniquely determined by the near-field
. Then for a given near-field
, the boundary
is a solution of the inverse scattering problem if and only if the pair
solves the integral scheme (Equation4.14
4.1
4.1 ) and (Equation4.15
4.2
4.2 ).
Proof
If the pair solves the integral scheme (Equation4.14
4.1
4.1 ) and (Equation4.15
4.2
4.2 ), we define:
In view of (Equation4.14
4.1
4.1 ), it leads to
on
. Then
is the solution to the problem (DP). Because of (Equation4.15
4.2
4.2 ) and the assumption we can deduce that the boundary
is a solution to the inverse scattering problem.
Conversely, for a given near-field , we assume that
is a solution to the inverse scattering problem. Then the solution to the problem (DP) is
where
solves (Equation4.14
4.1
4.1 ). From the uniqueness of the problem (DP), it can be obtained that (Equation4.15
4.2
4.2 ) holds. Therefore, the theorem is completed.
With the equivalence Theorem 4.1, our treatment of the inverse scattering problem is to solve the integral schemes (Equation4.144.1
4.1 ) and (Equation4.15
4.2
4.2 ). We denote the right-hand side of (Equation4.15
4.2
4.2 ) as
. It is easy to see that
is nonlinear with respect to
. For a fixed
, finding
from (Equation4.15
4.2
4.2 ) requires a linearization of (Equation4.15
4.2
4.2 ). We assume that the linearization of (Equation4.15
4.2
4.2 ) at
with respect to the direction
reads
4.3
4.3 The formulations of
and
will be given in the Appendix 1.
Further analysis of the linearized Equation (Equation4.164.3
4.3 ) requires us to choose parameterizations for functions in
defined in Section 2. Since the cubic B-spline function
is twice continuously differentiable and compactly supported, we express the rough surface function
and the updated function
as a linear combination of cubic B-spline functions, i.e.
4.4
4.4
4.5
4.5 for some positive constants
and
. Here
is a piecewise function given by:
The graph of
is shown in Figure .
Define ,
and denote the receiving points on
by
. With given
in the form (Equation4.17
4.4
4.4 ), by substituting
and (Equation4.17
4.4
4.4 ), (Equation4.18
4.5
4.5 ) into (Equation4.16
4.3
4.3 ), (Equation4.16
4.3
4.3 ) reduces to the finite linear system:
4.6
4.6 where
is a column vector with elements
and
is a matrix with elements
with
The kernel of the integral operator in (Equation4.15
4.2
4.2 ) is smooth, and therefore the inversion is severely ill-posed. The linearized Equation (Equation4.16
4.3
4.3 ) inherits the ill-posedness. In order to deal with this difficulty, we use the Tikhonov regularization to solve (Equation4.16
4.3
4.3 ) with the corresponding regularization parameter
. With regularization, the Equation (Equation4.19
4.6
4.6 ) leads to:
4.7
4.7
Remark 4.2
In (Equation4.204.7
4.7 ),
is a complex matrix and
is a complex vector. We use
and
instead of
and
in order to obtain a real vector
in (Equation4.20
4.7
4.7 ).
Before introducing the iterative algorithm based on the nonlinear integral equation, we define the relative error . Let
denote the reconstructed rough surface. For a fixed wave number
and the measured data
corresponding to the exact rough surface
, we denote the corresponding scattered field on the receiving points by
and define the relative error
for the reconstructed rough surface
by
For a fixed wave number
, the iterative algorithm to solve the integral schemes (Equation4.14
4.1
4.1 ) and (Equation4.15
4.2
4.2 ) is as follows. Assume that using Algorithm 4.1 we can obtain the reconstruction of the rough surface
for the fixed
and the corresponding relative error
. With the iterative algorithm for a fixed wave number, we can give the iterative algorithm for multiple frequency near-field data as follows. Given the tolerance
, the initial guess
(in this paper let
be a constant function) and the constant
, we propose the nonlinear integral equation iteration algorithm:
5 Numerical examples
In the final section, we present some numerical experiments to exhibit the effectiveness of the nonlinear integral equation method described in Section 4.
In all examples, we set the regularization parameter in (Equation4.20
4.7
4.7 ). For the relative noise level
, the perturbed data are constructed in the following way
where
with
normally distributed. In all figures, we denote the exact rough surface by solid (blue) lines, the initial guess by dash-dot (black) lines and the reconstructed rough surface by dashed (red) lines.
Example 1
For the first example, we consider the reconstruction of a rough surface which is approximately local and parameterized by5.1
5.1 with different width and different amplitude. We set
,
and
in (Equation4.17
4.4
4.4 ) and (Equation4.18
4.5
4.5 ). We implement the experiments with
for incident plane waves and
for incident point sources. Due to the limitation of space, for plane wave incidence, we only show the reconstructions results from exact data and from
noisy data when
in Figures and , while for incident point sources we only show the results for
in Figures and . We can see the reconstructions for incident plane waves and point sources are both very accurate even with
noise added.
Figure 3. Reconstruction of (Equation5.215.1
5.1 ) from exact data for incident plane wave with
,
and
.
![Figure 3. Reconstruction of (Equation5.215.1 f(t)=0.6+0.2exp(-0.16t2)-0.4exp(-0.25(t+8)2)+0.5exp(-0.49(t-10)2)5.1 ) from exact data for incident plane wave with ε=0.10, ρ=0.90 and k=1,9,15.](/cms/asset/7efa7371-d114-4a95-82ff-4ca4b1f32cf8/gipe_a_922077_f0003_oc.gif)
Figure 4. Reconstruction of (Equation5.215.1
5.1 ) from 10% noisy data for incident plane wave with
,
and
.
![Figure 4. Reconstruction of (Equation5.215.1 f(t)=0.6+0.2exp(-0.16t2)-0.4exp(-0.25(t+8)2)+0.5exp(-0.49(t-10)2)5.1 ) from 10% noisy data for incident plane wave with ε=0.15, ρ=0.90 and k=1,9,15.](/cms/asset/7e9810b1-3feb-4f97-8654-8a6d28f03f34/gipe_a_922077_f0004_oc.gif)
Figure 5. Reconstructions of (Equation5.215.1
5.1 ) from exact data for incident point sources with
,
and
.
![Figure 5. Reconstructions of (Equation5.215.1 f(t)=0.6+0.2exp(-0.16t2)-0.4exp(-0.25(t+8)2)+0.5exp(-0.49(t-10)2)5.1 ) from exact data for incident point sources with ε=0.15, ρ=0.90 and k=1,3,7.](/cms/asset/9aa103e3-2e79-48c0-96e7-5982dda3e53c/gipe_a_922077_f0005_oc.gif)
Figure 6. Reconstruction of (Equation5.215.1
5.1 ) from 10% noisy data for incident point sources with
,
and
.
![Figure 6. Reconstruction of (Equation5.215.1 f(t)=0.6+0.2exp(-0.16t2)-0.4exp(-0.25(t+8)2)+0.5exp(-0.49(t-10)2)5.1 ) from 10% noisy data for incident point sources with ε=0.20, ρ=0.90 and k=1,3,7.](/cms/asset/92c687d7-7e1d-4101-ad8a-e82d7fef0181/gipe_a_922077_f0006_oc.gif)
Figure 7. Reconstruction of (Equation5.225.2
5.2 ) from exact data for incident plane wave with
,
and
.
![Figure 7. Reconstruction of (Equation5.225.2 f(t)=0.6+0.2sin(2t)+0.3sin(6t).5.2 ) from exact data for incident plane wave with ε=0.40, ρ=0.90 and k=3,5,11.](/cms/asset/10e93134-4529-4360-8bad-d7d2c52a1f9d/gipe_a_922077_f0007_oc.gif)
Figure 8. Reconstruction of (Equation5.225.2
5.2 ) from 10% noisy data for incident plane wave with
,
and
.
![Figure 8. Reconstruction of (Equation5.225.2 f(t)=0.6+0.2sin(2t)+0.3sin(6t).5.2 ) from 10% noisy data for incident plane wave with ε=0.40, ρ=0.95 and k=3,5,11.](/cms/asset/cae45f65-c9af-4e14-889c-8117272bbe61/gipe_a_922077_f0008_oc.gif)
Figure 9. Reconstruction of (Equation5.225.2
5.2 ) from exact data for incident point sources with
,
and
.
![Figure 9. Reconstruction of (Equation5.225.2 f(t)=0.6+0.2sin(2t)+0.3sin(6t).5.2 ) from exact data for incident point sources with ε=0.20, ρ=0.95 and k=3,7,11.](/cms/asset/2d7651ad-0002-4d6f-94e0-3e639b26246f/gipe_a_922077_f0009_oc.gif)
Figure 10. Reconstruction of (Equation5.225.2
5.2 ) from 10% noisy data for incident point sources with
,
and
.
![Figure 10. Reconstruction of (Equation5.225.2 f(t)=0.6+0.2sin(2t)+0.3sin(6t).5.2 ) from 10% noisy data for incident point sources with ε=0.20, ρ=0.95 and k=3,7,11.](/cms/asset/bed13da7-d59d-4aac-92d9-ef9c0a479070/gipe_a_922077_f0010_oc.gif)
Figure 11. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident plane wave with
,
and
.
![Figure 11. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident plane wave with ε=0.60, ρ=0.90 and k=3,7,11.](/cms/asset/57038cdb-f699-43e2-8cb0-affe663dc72a/gipe_a_922077_f0011_oc.gif)
Figure 12. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident plane wave with
,
and
when the measurements were taken at
.
![Figure 12. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident plane wave with ε=0.60, ρ=0.90 and k=3,7,11 when the measurements were taken at x2=2.50.](/cms/asset/4c1ab47b-1317-4417-8f4d-98dea296ae9c/gipe_a_922077_f0012_oc.gif)
Figure 13. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident plane wave with
,
and
when the measurements were taken at
.
![Figure 13. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident plane wave with ε=0.60, ρ=0.90 and k=3,7,11 when the measurements were taken at x2=3.50.](/cms/asset/79f771f1-58cb-4bc7-9484-ff352f6afc30/gipe_a_922077_f0013_oc.gif)
Figure 14. Reconstruction of (Equation5.235.3
5.3 ) from 10% noisy data for incident plane wave with
,
and
.
![Figure 14. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from 10% noisy data for incident plane wave with ε=0.60, ρ=0.90 and k=3,7,11.](/cms/asset/11e3f3a9-884f-465e-bd43-1ab722b2af24/gipe_a_922077_f0014_oc.gif)
Figure 15. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident point sources with
,
and
.
![Figure 15. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident point sources with ε=0.30, ρ=0.90 and k=3,7,11.](/cms/asset/2add9aa4-7faa-459a-ad5d-21cc81c881dc/gipe_a_922077_f0015_oc.gif)
Figure 16. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident point sources with
,
and
when the measurements were taken at
.
![Figure 16. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident point sources with ε=0.30, ρ=0.90 and k=3,7,11 when the measurements were taken at x2=2.50.](/cms/asset/d8247ddf-6dc1-4f53-9793-ea07e1f9fada/gipe_a_922077_f0016_oc.gif)
Figure 17. Reconstruction of (Equation5.235.3
5.3 ) from exact data for incident point sources with
,
and
when the measurements were taken at
.
![Figure 17. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from exact data for incident point sources with ε=0.30, ρ=0.90 and k=3,7,11 when the measurements were taken at x2=3.50.](/cms/asset/4f7f16d2-b40a-4b2b-ad93-5fe08a070054/gipe_a_922077_f0017_oc.gif)
Figure 18. Reconstruction of (Equation5.235.3
5.3 ) from 10% noisy data for incident point sources with
,
and
.
![Figure 18. Reconstruction of (Equation5.235.3 f(t)=1-0.2cos(0.01t2)exp(-sin(t)).5.3 ) from 10% noisy data for incident point sources with ε=0.30, ρ=0.90 and k=3,7,11.](/cms/asset/ff9efe50-ff52-426f-a600-3312ba3e39db/gipe_a_922077_f0018_oc.gif)
Example 2
In the second example, we attempt to reconstruct a nonlocally rough surface which is periodic with the parameterization5.2
5.2 We let
,
and
in (Equation4.17
4.4
4.4 ) and (Equation4.18
4.5
4.5 ). We implement the experiments at
. Because of the limited space, we only present the results from exact data and from
noisy data for incident plane waves for
in Figures and . The reconstruction results for point sources are presented in Figures and when
. From the figures in this example, it can be seen that the algorithm works well for nonlocally rough surfaces when the incoming wave is a plane wave, but the reconstruction is not good enough when the incoming wave is point sources. We attempt to increase the wave numbers in Figures and , but the reconstruction is not improved very much.
Example 3
In the last example, we test our algorithm on a rough surface which is nonlocal and nonperiodic, and formulated by5.3
5.3 We let
,
and
in (Equation4.17
4.4
4.4 ) and (Equation4.18
4.5
4.5 ) and implement the experiments with
. Due to the limitation of space, we only show a part of results. For plane wave incidence, we show the reconstruction results from the exact measurement data and from
noisy data at
and
in Figures and . In order to see the effect of our algorithm at different measurement heights, we also present the reconstruction results by using the exact scattering data measured at different heights
and
in Figures and . For incident point sources, the results from the exact data and from
noisy data with
and
are presented in Figures and . The reconstruction results from the exact data at the heights
and
are given in Figures and . As we can see from the figures, the reconstructions are very accurate, especially when
. Even
noise is added to the measurements, the reconstructed surfaces still agree well with the exact one. Furthermore, when the measurement height is within a reasonable range, smaller measurement heights (that is, the measurement place is closer to the exact surface) give better results in both cases of incident plane waves and point sources with a single wave number. However, the effect of measurement height is weak when using multiple frequency data.
Through the examples in this section we can see that the nonlinear integral equation method works well for the problem of inverse scattering by sound-soft rough surfaces. For locally rough surface and gently rough surface, we obtain satisfactory results whether the incident field is plane waves or point sources. However, for the nonlocally rough surfaces with considerable oscillations, the reconstruction results for incident plane waves are better than that for incident point sources.
Acknowledgements
The authors would like to thank Prof. Bo Zhang, Dr. Jiaqing Yang and Dr. Haiwen Zhang for the valuable discussions. The authors also thank the referees for their constructive comments and suggestions which helped improve the paper.
Additional information
Funding
References
- Chandler-Wilde SN, Zhang B. Electromagnetic scattering by an inhomogeneous conducting or dielectric layer on a perfectly conducting plate. Proc. R. Soc. Lond. A. 1998;454:519–542.
- Chandler-Wilde SN, Ross CR. Uniqueness results for direct and inverse scattering by infinite surfaces in a lossy medium. Inverse Prob. 1995;11:1063–1067.
- Chandler-Wilde SN, Ross CR. Scattering by rough surfaces: the Dirichlet problem for the Helmholtz equation in a non-locally perturbed half-plane. Math. Methods Appl. Sci. 1996;19:959–976.
- Chandler-Wilde SN, Ross CR, Zhang B. Scattering by infinite one-dimensional rough surfaces. Proc. R. Soc. Lond. A. 1999;455:3767–3787.
- Chandler-Wilde SN, Zhang B. A uniqueness result for scattering by infinite rough surfaces. SIAM J. Appl. Math. 1998;58:1774–1790.
- Zhang B, Chandler-Wilde SN. Integral equation methods for scattering by infinite rough surfaces. Math. Methods Appl. Sci. 2003;26:463–488.
- Chandler-Wilde SN, Zhang B. On the solvability of a class of second kind integral equations on unbounded domains. J. Math. Anal. Appl. 1997;214:482–502.
- Chandler-Wilde SN, Zhang B, Ross CR. On the solvability of second kind integral equations on the real line. J. Math. Anal. Appl. 2000;245:28–51.
- Arens T, Chandler-Wilde SN, Haseloh KO. Solvability and spectral properties of integral equations on the real line: ∐. Lp-spaces and applications. J. Integral Equ. Appl. 2003;15:1–35.
- Meier A, Arens T, Chandler-Wilde SN, Kirsch A. A Nyström method for a class of integral equations on the real line with applications to scattering by diffraction gratings and rough surfaces. J. Integral Equ. Appl. 2000;12:281–321.
- Burkard C, Potthast R. A multi-section approach for rough surface reconstruction via the Kirsch-Kress scheme. Inverse Prob. 2010;26:45007–45029.
- Burkard C, Potthast R. A time-domain probe method for three-dimensional rough surface reconstructions. Inverse Prob. Imaging. 2009;3:259–274.
- Lines C. Inverse scattering by unbounded rough surfaces [PhD thesis]. London: Department of Mathematics, Brunel University; 2003.
- Bao G, Lin J. Imaging of local surface displacement on an infinite ground plane: the multiple frequency case. SIAM J. Appl. Math. 2011;71:1733–1752.
- Zhang H, Zhang B. A novel integral equation for scattering by locally rough surfaces and application to the inverse problem. SIAM J. Appl. Math. 2013;73:1811–1829.
- Bao G, Li P. Inverse medium scattering problems for electromagnetic waves. SIAM J. Appl. Math. 2005;65:2049–2066.
- Bao G, Li P. Near-field imaging of infinite rough surfaces. SIAM J. Appl. Math. 2013;73:2162–2187.
- Bao G, Lin J, Triki F. A multi-frequency inverse source problem. J. Differ. Equ. 2010;249:3443–3465.
- Chandler-Wilde SN, Potthast R. The domain derivative in rough-surface scattering and rigorous estimates for first-order perturbation theory. Proc. R. Soc. Lond. A. 2002;458:2967–3001.
- Kress R, Rundell W. Nonlinear integral equations and the iterative solution for an inverse boundary value problem. Inverse Prob. 2005;21:1207–1223.
- Ivanyshyn O, Johansson T. Nonlinear integral equation methods for the reconstruction of an acoustically sound-soft obstacle. J. Integral Equ. Appl. 2007;19:289–308.
- Ivanyshyn O, Kress R. Inverse scattering for surface impedance from phase-less far field data. J. Comput. Phys. 2011;230:3443–3452.
- Ivanyshyn O, Kress R. Nonlinear integral equations for solving inverse boundary value problems for inclusions and cracks. J. Integral Equ. Appl. 2006;18:13–38.
- Eckel H, Kress R. Nonlinear integral equations for the inverse electrical impedance problem. Inverse Prob. 2007;23:475–491.
- Meier A. Numerical treatment of integral equations on the real line with application to acoustic scattering by unbounded rough surfaces [PhD thesis]. London: Department of Mathematical Sciences, Brunel University; 2001.
- Colton D, Kress R. Inverse acoustic and electromagnetic scattering theory. 2nd ed. Berlin: Springer; 1998.
- Kress R. Linear integral equations. 2nd ed. Berlin: Springer Verlag; 1999.
- Lebedev NN. Special functions and their applications. New York (NY): Dover; 1972.
- Potthast R. Fréchet differentiability of boundary integral operators in inverse acoustic scattering. Inverse Prob. 1994;10:431–447.
Appendix 1.
In this appendix, we give the formulations of and
in (Equation4.16
4.3
4.3 ).
The Fréchet derivative of
at
with respect to the direction
can be obtained by differentiating the kernel with respect to
(see [Citation29]). Therefore, the derivative is given by:
where