RADIOENGINEERING, VOL. 19, NO.

4, DECEMBER 2010 601

Influence of Sea Surface Roughness

on the Electromagnetic Wave Propagation
in the Duct Environment
Xiaofeng ZHAO, Sixun HUANG

Institute of Meteorology, PLA Univ. of Sci. & Tech., Nanjing 211101, China

[email protected], [email protected]

Abstract. This paper deals with a study of the influence of surface [11]. In this paper, we will employ Miller-Brown-
sea surface roughness on the electromagnetic wave propa- Vegh model to study the influence of wind-driven rough-
gation in the duct environment. The problem of electro- ness on the electromagnetic wave propagation in the
magnetic wave propagation is modeled by using the para- evaporation duct environment. A detailed discussion will
bolic equation method. The roughness of the sea surface is be focused on different numerical values for various
computed by modifying the smooth surface Fresnel reflec- important parameters, i.e. sea wind speed, evaporation duct
tion coefficient to account for the reduction in the specular height, operating frequency, and antenna altitude.
reflection due to the roughness resulting from sea wind
The remainder of this paper is organized as follows.
speed. The propagation model is solved by the mixed Fou-
In section 2, a simple description of the electromagnetic
rier split-step algorithm. Numerical experiments indicate
wave propagation model is introduced. Section 3 con-
that wind-driven roughened sea surface has an impact on
structs the relationship between the rough sea surface and
the electromagnetic wave propagation in the duct environ-
the smooth sea surface through the reflection coefficient.
ment, and the strength is intensified along with the incre-
Finally, a discussion of simulations of different parameters
ment of sea wind speeds and/or the operating frequencies.
is presented in section 4.
In a fixed duct environment, however, proper disposition of
the transmitter could reduce these impacts.

2. The Electromagnetic Wave

Propagation Model
Keywords
For many years now PE has been used to model elec-
Electromagnetic wave propagation, parabolic tromagnetic wave propagation in a complex environment.
equation, sea wind speed, evaporation duct. PE can be derived from the wave equation under certain
assumptions, and it has the form [7],

1. Introduction  2 u ( x, z )
z 2
 2ik 0
u ( x, z )
x
 
 k02 n 2 x, z   1 u  x, z   0 (1)
Ducted propagation over the sea surface can seriously
impact shipboard radar and communications, which is where u represents a scalar component of the electric field
extremely important in telecommunications for the Navy. for horizontal polarization or a scalar component of the
The physics of propagation is affected by ever-changing magnetic field for vertical polarization, k0 is the free-space
atmospheric conditions and by complex features on the sea wave number, x is the range axis, z is the height axis, and n
surface [1]. Numerous numerical methods are available for is the index of refraction. Owing to n is very close to unity,
predicting electromagnetic wave propagation such as for environmental inputs, modified refractivity M is used to
geometric optics, normal mode analysis, parabolic equation describe the information of the atmospheric environment.
(PE) method, and combinations of the above [2-5]. The M is defined by
most popular approach of the mentioned above is PE which  z  (2)
could be used to predict electromagnetic wave propagation M ( x, z )   n( x, z )  1   106
in a complex environment. Previously, use of the PE to  ae 
investigate propagation in an inhomogeneous atmosphere
where ae is the radius of the earth.
and over irregular terrain has been published a lot [6-10].
However, the researches of the propagation over the rough The two most popular approaches to numerically
sea surface are very few. In 2003, Guillet et al. employed solving PE are the use of implicit finite differences and the
PE to studied low grazing angle propagation above rough Fourier split-step algorithm. However, the latter is much
602 XIAOFENG ZHAO, SIXUN HUANG, INFLUENCE OF SEA SURFACE ROUGHNESS ON THE ELECTROMAGNETIC WAVE …

more numerically efficient [3], and is adopted in our cal-  R   S  exp[2(2 g ) 2 ]  I 0 [2(2 g ) 2 ] (8)
culations. Generally, split-step PE solutions to electromag-
and
netic problems implemented either Dirichlet or Neumann
 h sin 
boundary conditions corresponding to horizontally or verti- g (9)
cally polarized propagation, respectively, over a perfectly 
conduct surface. In rough sea surface, however, the per- where I0(x) is the modified Bessel function J0(ix), and σh is
fectly conduct surface assumption is not sufficient and use the standard deviation of the sea surface elevation deter-
of a surface impedance is highly desirable. The impedance mined from wind speed, μ (m/s), by
boundary condition is
 h  0.0051 2 . (10)
 z ( x, z ) z 0   u ( x, 0)  0 (3)
and As mentioned above, when modifying reflection coef-
ficients using the Miller-Brown-Vegh model, the resulting
 1  
  ik0 sin    (4) expression for α has a strong dependence on the local graz-
 1   ing angle θ which is, in general, a function of geometry
where Γ is the Fresnel reflection coefficient and θ is the and, more importantly, atmospheric refractivity. Dockery
grazing angle measured from the local surface tangent , i.e. and Kuttler have pointed out that geometric optics (GO)
the complement of the incidence angle. could provide a qualified grazing angle estimate in the
evaporation duct environment [12]. Therefore, GO will be
The mixed Fourier transform introduced as a rigorous adopted in our computation.
method to incorporate surface impedance into split-step
solution of the parabolic equation was described in detail in
[4], and a fast computation algorithm for the mixed Fourier
transform split-step solution of PE was presented in [12].
4. Numerical Experiments
The purpose of this paper is to study the influence of
sea surface roughness to the electromagnetic wave
3. Sea Surface Roughness Modeling propagation in the evaporation duct environment. First,
an evaporation duct model should be selected. For
Using the impedance boundary condition, the reflec- thermally neutral conditions in which the air and sea
tion coefficient Γ should be determined. In terms of the law temperature are equal, the modified refractivity M can be
of reflection, for a giving grazing angle θ, the smooth sur- determined at any height z by the relationship [16]
face reflection coefficient ΓS for horizontal polarization and
vertical polarization could, respectively, be described as M ( z )  M (0)  0.125 z  0.125d ln[( z  z0 ) / z0 ] (11)
[13]
where d is evaporation duct height, ln[·] is the natural loga-
sin   ns2  cos 2  rithm, and z0 is momentum roughness length of 1.5×10-4 m.
 SH  (5)
sin   ns2  cos 2  In the following, two experiments are performed. For
every experiment, a Gaussian antenna pattern is used and
and
horizontal polarization is assumed. The elevation angle is
ns2 sin   ns2  cos 2  0 deg and the beamwidth is 1 deg. The detailed expression
 SV  (6)
ns2 sin   ns2  cos 2  of the Gaussian antenna pattern could refer to [7].
Experiment 1: The evaporation duct height and the
where ns is the complex dielectric constants, defined by antenna altitude are fixed at 30 m and 10 m, respectively.
Then, the propagation characteristics at different sea wind
ns    i 60 . (7)
speeds and operating frequencies are investigated. The sea
Here, ε and σ are the relative permittivity and conductivity wind speeds are changeable, from 0 m/s to 10 m/s.
respectively, which are functions of radio frequency Using the PE/mixed Fourier split-step algorithm, the
recommended by International Radio Consulting Commit- value of one-way propagation loss is computed. Here, the
tee (CCIR), and λ is the electromagnetic wave length. computation domain is fixed on [100 km  200 m], and the
One approach to modeling sea surface roughness in- horizontal increment and vertical increment are 500 m and
volves multiplying the smooth surface Fresnel reflection 1 m, respectively. The coverage diagram (dB) for one-way
coefficients by a roughness parameter, which has been propagation loss is shown in Fig. 1, in which the operating
calculated by Ament [14] for a Gaussian distribution of frequency is 8000 MHz, Fig. 1(a) is for sea wind speed of
heights and corrected by Miller et al. [15]. The rough sur- 0 m/s, Fig. 1(b) is for 10 m/s, and Fig. 1(c) is the absolute
face reflection coefficient, ΓR, is taken as difference of the above two figures.
RADIOENGINEERING, VOL. 19, NO. 4, DECEMBER 2010 603

Fig. 1. The coverage diagram (dB) for one-way propagation loss. (a) is for 0 m/s; (b) is for 10m/s, (c) is the absolute difference of (a) and (b).

Fig. 2. Comparison of propagation loss at antenna altitude (10 m) in a 30 m evaporation duct between smooth sea surface (sea wind speed of
0 m/s, solid line) and rough sea surface (sea wind speed of 10 m/s, dashed line). The operating frequency is 8000 MHz.

Fig. 3. Relative errors at different operating frequencies and different sea wind speeds.
604 XIAOFENG ZHAO, SIXUN HUANG, INFLUENCE OF SEA SURFACE ROUGHNESS ON THE ELECTROMAGNETIC WAVE …

In order to analyze the influence of wind-driven From Fig. 1-3, it is clearly seen that, considering the
roughness on the electromagnetic wave propagation, the sea surface roughness, the electromagnetic propagation
values of propagation loss at the antenna altitude are se- loses more energy than the smooth surface case, and the
lected to compute the relative error of the propagation loss propagation losses are strengthening with the distance. Fig.
between the smooth sea surface and the rough sea surface. 3 shows that with the increment of sea wind speed and/or
The relative error Λ is defined as operating frequency, the relative errors between smooth
N
sea surface and rough sea surface become larger and larger.
PLiu  PLi0

1 In 10000 MHz and 10 m/s case, the relative error could
  100% (12)
N i 1
PLi0 even exceed 8.7%. However, when the sea wind speed is
under 5 m/s, the errors could be limited within 1.2%.
where PLu is the value of propagation loss corresponding to Experiment 2: The operating frequency is fixed at
the wind-driven rough sea surface and PL0 corresponding 8000 MHz. The relationship between evaporation duct
to smooth surface, N is the number of discrete points in the height and antenna altitude will be studied. The evapora-
horizontal direction. The values of propagation loss along tion duct heights are set at 10 m, 20 m, and 30 m. Within
the horizontal distance at the antenna altitude 10 m in the duct, the antenna altitudes are changeable, from 1 m to
Fig. 1(a) and 1(b) are presented in Fig. 2. Using equation the duct height. Fig. 4 shows the relative error between
(12), the relative errors at different operating frequencies smooth surface case (sea wind speed of 0 m/s) and rough
and different sea wind speeds are shown in Fig. 3. surface case (sea wind speed of 10 m/s).

Fig. 4. Relative error between smooth surface case (sea wind speed of 0 m/s) and rough surface case (sea wind speed of 10 m/s) at different
evaporation duct heights and different antenna altitudes.

From Fig. 4, we could see the position of antenna alti- usually exists near the top of the duct. However, if the
tude within the duct has great impact upon the propagation transmitter is setting at the top of the duct, it is not in favor
loss in the rough sea surface case. The stronger the evapo- of increasing radar detecting distance. In our simulations,
ration duct exists, the larger the relative error of propaga- the maximum sea wind speed is selected as 10 m/s. In
tion loss will be generated. In 10 m duct, the maximum practical duct environment, the sea wind speed is usually
error presents at transmitter altitude of 2 m, in 20m duct, below 10 m/s because larger wind speed is destructive to
the maximum error presents at 12 m, and 5 m in 30 m the generation of the evaporation duct. If the sea wind
evaporation duct. When the relative error reaches the speed is not very large (under 5 m/s), we could ignore the
maximum value, it will have a tendency to decreasing with influence of the sea surface roughness to the electromag-
the increment of the antenna altitude. netic wave propagation in the duct environment.
The above two experiments show that, in the evapora-
tion duct environment, the electromagnetic wave propaga-
tion characteristics over the rough sea surface are different 5. Conclusion
from that over the smooth sea surface. In general, the er- This paper has investigated the influence of the sea
rors enlarge with the increment of sea wind speeds, operat- surface roughness on the electromagnetic wave propaga-
ing frequencies and evaporation duct height. At a fixed tion in the evaporation duct environment. Considering the
atmosphere environment, proper disposition of the trans- sea surface roughness, the impedance boundary condition
mitter could reduce these errors. The minimum error value was used, and the Miller-Brown-Vegh model was em-
RADIOENGINEERING, VOL. 19, NO. 4, DECEMBER 2010 605

ployed to compute the wind-driven rough surface reflection [9] OZGUN, O. Recursive two-way parabolic equation approach for
modelling terrain effects in tropospheric propagation. IEEE
coefficient. Two numerical experiments were performed in
Transactions on Antennas and Propagation, 2009, vol. 57, no. 9, p.
terms of different sea wind speeds, operating frequencies, 2706-2714.
evaporation duct heights and antenna altitudes, and present
[10] VALTR, P., PECHAC, P., KVICERA, V., GRABNER, M.
a straightforward understanding of the relationship of these A terrestrial multiple-receiver radio link experiment at 10.7 GHz –
parameters to the roughness influence on propagation comparisons of results with parabolic equation calculations.
characteristics. However, the Miller-Brown-Vegh model Radioengineering, 2010, vol. 19, no. 1, p. 117-121.
doesn't take into account the influence of the surface [11] GUILLET, N., FABBRO, V., BOURLIER, C., COMBES, P. F.
geometry, which will be investigated in our future work. Low grazing angle propagation above rough surface by the
parabolic wave equation. In Geoscience and Remote Sensing
Symposium, 2003.
[12] DOCKERY, G. D., KUTTLER, J. R. An improved impedance-
Acknowledgements boundary algorithm for Fourier split-step solution of the parabolic
wave equation. IEEE Transactions on Antennas and Propagation,
1996, vol. 44, no. 12. p. 1592-1599.
This work has been supported by the National Natural
Science Foundation of China, grant No. 40775023, “Re- [13] YANG, X. Q., ZHAO, J. S., WANG, Y. Electromagnetic Field
and Electromagnetic Wave. Peking: National Defence Industry
fractivity estimation from radar sea clutter by incorporating Press, 2002.
regularization techniques into data assimilation method”.
[14] AMENT, W. S. Toward a theory of reflection by a rough surface.
In Proc. IRE. 1953, vol. 41, p. 142-146.
[15] MILLER, A. R., BROWN, R. M., VEGH, E. New derivation for
References rough surface reflection coefficient and for the distribution of sea-
wave elevations. In IEE Proc, 1984, vol. 131, no. 2, p. 114-116.

[1] YAN, H. J., FU, Y., HONG, Z. J. Introduction to Modern Atmo- [16] HITNEY, H. V., VIETH, R. Statistical assessment of evaporation
spheric Refraction. Shanghai: Science and Educational Press, 2006. duct propagation. IEEE Transactions on Antennas and
Propagation, 1990, vol. 38, no. 6, p. 794-799.
[2] HE, G. Y., LU, C. C., HONG, J. C., DENG, H. Computation and
Measurement of Electromagnetic Scattering. Peking: National
Defence Industry Press, 2006.
[3] VALTR, P., PECHAC, P. Tropospheric refraction modeling using
ray-tracing and parabolic equation. Radioengineering, 2005, vol.
14, no. 4, p. 98-104.
[4] KUTTLER, J. R., DOCKERY, G. D. Theoretical description of the
About Authors ...
parabolic approximation/ Fourier split-step method of representing
Xiaofeng ZHAO was born in Jiangsu, China. He received
electromagnetic propagation in the troposphere. Radio Science,
1991, vol. 26, no. 2, p. 381-393. his M.Sc. degree from PLA University of Science and
Technology in 2009. Now he is working towards his Ph.D.
[5] HITNEY, H. V. Hybrid ray optics and parabolic equation methods
for radar propagation modelling. In Proc. IEE Int. Conf. Radar. at the Institute of Meteorology of the same university. His
Brighton (U.K.), 1992, p. 58-61. research interests include radiowave propagation modeling
[6] BARRIOS, A. E. Parabolic equation modelling in horizontally and atmospheric refractivity inversion from radar sea
inhomogeneous environments. IEEE Transactions on Antennas clutter.
and Propagation, 1992, vol. 40, no. 7, p. 791-797.
Sixun HUANG was born in Shanghai, China. He received
[7] BARRIOS, A. E. A terrain parabolic equation model for
his M.Sc. degree in applied mathematics from Fundan
propagation in the troposphere. IEEE Transactions on Antennas
and Propagation, 1994, vol. 42, no. 1, p. 90-98. University in 1981. He is now a Professor at the Institute of
Meteorology, PLA University of Science and Technology.
[8] GRABNER, M, KVICERA, V. Clear-air propagation modelling
using parabolic equation method. Radioengineering, 2003, vol. 12, His research interests are in the field of radiowave propa-
no. 4, p. 50-54. gation modeling and remote sensing technologies.