You are on page 1of 13

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1417

Theory of optical scintillation


L. C. Andrews
Department of Mathematics and Center for Research and Education in Optics and Lasers, University of Central Florida, Orlando, Florida 32816

R. L. Phillips
Director, Florida Space Institute, Department of Electrical and Computer Engineering, University of Central Florida, Orlando, Florida 32816

C. Y. Hopen
Department of Mathematics and Florida Space Institute, University of Central Florida, Orlando, Florida 32816

M. A. Al-Habash
Department of Mathematics and Center for Research and Education in Optics and Lasers, University of Central Florida, Orlando, Florida 32816 Received August 11, 1998; revised manuscript received January 5, 1999; accepted January 19, 1999 A heuristic model of irradiance uctuations for a propagating optical wave in a weakly inhomogeneous medium is developed under the assumption that small-scale irradiance uctuations are modulated by large-scale irradiance uctuations of the wave. The upper bound for small turbulent cells is dened by the smallest cell size between the Fresnel zone and the transverse spatial coherence radius of the optical wave. A lower bound for large turbulent cells is dened by the largest cell size between the Fresnel zone and the scattering disk. In moderate-to-strong irradiance uctuations, cell sizes between those dened by the spatial coherence radius and the scattering disk are eliminated through spatial-frequency ltering as a consequence of the propagation 2 2 2 2 2 process. The resulting scintillation index from this theory has the form 2 I x y x y , where x de2 notes large-scale scintillation and y denotes small-scale scintillation. By means of a modication of the Rytov method that incorporates an amplitude spatial-frequency lter function under strong-uctuation conditions, tractable expressions are developed for the scintillation index of a plane wave and a spherical wave that are valid under moderate-to-strong irradiance uctuations. In many cases the models also compare well with conventional results in weak-uctuation regimes. Inner-scale effects are taken into account by use of a modied atmospheric spectrum that exhibits a bump at large spatial frequencies. Quantitative values predicted by these models agree well with experimental and simulation data previously published. In addition to the scintillation index, expressions are also developed for the irradiance covariance function of a plane wave and a spherical wave, both of which have the form B I ( ) B x ( ) B y ( ) B x ( ) B y ( ), where B x ( ) is the covariance function associated with large-scale uctuations and B y ( ) is the covariance function associated with small-scale uctuations. In strong turbulence the derived covariance shows the characteristic two-scale behavior, in which the correlation length is determined by the spatial coherence radius of the eld and the width of the long residual correlation tail is determined by the scattering disk. 1999 Optical Society of America [S0740-3232(99)01606-3] OCIS codes: 010.0010, 010.1300, 030.7060, 030.0030.

1. INTRODUCTION
An optical wave propagating through a random medium such as the atmosphere will experience irradiance uctuations (called scintillation), even over relatively short propagation paths. Scintillation is caused almost exclusively by small temperature variations in the random medium, resulting in index-of-refraction uctuations (i.e., optical turbulence). Theoretical and experimental studies of irradiance uctuations generally center on the scintillation index

semble average or, equivalently, a long-time average. In weak-uctuation regimes [dened as those regimes for which the scintillation index (1) is less than unity], derived expressions for the scintillation index show that it is proportional to the Rytov variance
2 1 1.23C n 2 k 7/6L 11/6,

(2)

2 I

I 2 I2

1,

(1)

where the quantity I denotes irradiance (intensity) of the optical wave and the angle brackets denote an en0740-3232/99/061417-13$15.00

where C n 2 is the index-of-refraction structure parameter, k is the optical wave number, and L is the propagation path length between transmitter and receiver. The Rytov variance represents the scintillation index of an unbounded plane wave in weak uctuations but is otherwise considered a measure of optical turbulence strength when extended to strong-uctuation regimes by increasing either C n 2 or the path length L, or both. It is known that
1999 Optical Society of America

1418

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

the scintillation index increases with increasing values of the Rytov variance (2) until it reaches a maximum value greater than unity in the regime characterized by random focusing, so called because the focusing caused by largescale inhomogeneities achieves its strongest effect. With increasing path length or inhomogeneity strength, the focusing effect is weakened by multiple scattering, and the uctuations slowly begin to decrease, saturating at a level for which the scintillation index approaches unity from above. Qualitatively, saturation occurs because multiple scattering causes the optical wave to become increasingly less coherent as it propagates, eventually appearing like extended multiple sources, each scintillating with a distinct random phase. The process of optical wave propagation through random media has been studied for more than 35 years. By a random medium, we mean a turbulent medium or one for which the index of refraction of the medium exhibits random spatial variations that are large with respect to optical wavelength. As yet there is no tractable solution to the problem of irradiance uctuations from rst principles of electromagnetic wave propagation that applies to all conditions of optical turbulence. Early investigations concerning the propagation of unbounded plane waves and spherical waves through random media led to the classical monographs published in the early 1960s by Tatarskii1 and by Chernov,2 but their scintillation results are limited to weak uctuations. The saturation effect of the optical wave was rst observed experimentally by Gracheva and Gurvich3 in 1965. This work attracted much attention and stimulated a number of theoretical and experimental studies devoted to irradiance uctuations under conditions of strong turbulence. Based on weak-uctuation theory, Tatarskii1,4 predicted that the correlation length of the irradiance uctuations is on the order of the rst Fresnel zone L / k . However, measurements5,6 of the irradiance covariance function under strong-uctuation conditions revealed that the correlation length decreases with increasing values of the 2 Rytov variance 1 and that a large residual correlation tail emerges at large separation distances. That is, in the strong-uctuation regime the correlation length of irradiance uctuations is determined by the spatial coherence radius 0 of the optical wave, and the width of the residual tail is characterized by the scattering disk L/k0 . Several qualitative models describing the underlying physics associated with amplitude or irradiance uctuations were developed in the mid-1970s. Yura7 generalized Tatarskiis geometrical-optics model to include diffraction effects and the loss of spatial coherence of the wave as it propagates into the strong-uctuation regime. His results are primarily an order-of-magnitude estimate rather than a rigorous quantitative derivation. Clifford et al.8 extended Tatarskiis theory to the log-amplitude variance under strong uctuations and showed why the smallest scales of irradiance uctuations persist into the saturation regime. This latter model was subsequently modied by Hill and Clifford9 and called the heuristic theory. Although quantitative predictions from Yuras physical model and the heuristic theory do not fully agree with known results,10,11 the basic qualitative arguments

presented in these models are still valid. The rst widely accepted asymptotic theory for the saturation regime under the assumption of negligible inner scale was published in 1974 by Gochelashvili and Shishov.12 Innerscale models were later introduced by Fante13 for the plane wave and by Frehlich14 for the spherical wave. Unfortunately, numerical results from the asymptotic theory underpredict measured15 18 and simulation19 23 data of the scintillation index in the saturation regime. Kolmogorov theory assumes that turbulent eddies range in size from a macroscale to a microscale, forming a continuum of decreasing eddy sizes. The largest cell size smaller than those at which turbulent energy is injected into a region denes an effective outer scale of turbulence L 0 , which near the ground is roughly comparable with the height of the observation point above ground. An effective inner scale of turbulence l 0 is associated with the smallest cell size before energy is dissipated into heat. Each of these cells is actually a randomly changing volume of high or low refractive index. The energy distribution of these cells is described by spectral models such as those below, which in reality are only locally homogeneous.24 Numerical results deduced from theoretical models of scintillation depend strongly on the assumed model for the spatial power spectrum of refractive-index uctuations. If we ignore outer-scale effects, which are usually not important in scintillation studies, the commonly used spectral models are all specializations of n 0.033C n 2 11/3f l 0 , (3)

where is the magnitude of the spatial wave number, l 0 is the inner scale, and f ( l 0 ) is a factor that describes inner-scale modications of the basic power-law form. For example, the Kolmogorov spectrum is characterized by f ( l 0 ) 1, whereas f ( l 0 ) exp(l0 /5.92) 2 in the case of the Tatarskii spectrum,4 the latter sometimes called the traditional spectrum. A more accurate model for scintillation studies is provided by the Hill spectrum,21 23,25 but, because it is described in terms of a second-order differential equation that must be solved numerically, the Hill spectrum cannot be used in analytic developments. An analytic approximation to the Hill spectrum is given by the modied atmospheric spectrum26,27 in which f l 0 exp 2 / l 2 1 1.802 / l 0.254 / l 7/6 ,

l 3.3/l 0 .

(4)

Numerical comparison of results based on this latter spectrum and the Hill spectrum reveals differences no larger than 6% but generally within 1% 2% of each other.27,28 In this paper we use a modication of the Rytov method1,27 to develop a relatively simple model for irradiance uctuations that is applicable in moderate-to-stronguctuation regimes. To do so, we make the following basic observations and/or assumptions: 1. Atmospheric turbulence as it pertains to a propagating wave is statistically nonhomogeneous. 2. The received irradiance of an optical wave can be modeled as a modulation process in which small-scale

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1419

(diffractive) uctuations are multiplicatively modulated by large-scale (refractive) uctuations. 3. Small-scale processes and large-scale processes are statistically independent. 4. The Rytov method for optical scintillation is valid even into the saturation regime with the introduction of a spatial-frequency lter to account properly for the loss of spatial coherence of the optical wave in strong-uctuation regimes. 5. The geometrical-optics method can be applied to large-scale irradiance uctuations. These observations and/or assumptions are based on recognizing that the distribution of refractive power among the turbulent eddy cells of a random medium is described by an inverse power of the physical size of the cell, with the smallest cells having the weakest refractive power and the largest cells having the strongest. As a coherent wave begins to propagate into a random medium, the wave is scattered by the smallest of the turbulent cells (on the order of millimeters) by diffraction. The large turbulent cells act as refractive lenses with focal lengths typically on the order of hundreds of meters or more. Thus, although they act as weak-focusing (or defocusing) lenses, they can increase (or decrease) the amplitude of the wave by a signicant amount for even short propagation distances. Diffractive scattering spreads the wave as it propagates, so the amplitude of the wave illuminating the subsequent small cells is more uniformly distributed. Refractive and diffractive scattering processes are compound mechanisms, and the total scattering process acts like a modulation of small-scale uctuations by large-scale uctuations. Small-scale contributions to scintillation are associated with turbulent cells smaller than the Fresnel zone L / k or the coherence radius 0 , whichever is smaller. Largescale uctuations in the irradiance are generated by turbulent cells larger than that of the rst Fresnel zone or the scattering disk L / k 0 , whichever is larger, and can be described by the method of geometrical optics. Under strong-uctuation conditions spatial cells having size between those of the coherence radius and the scattering disk contribute little to scintillation.8,9 That is, because of the loss of spatial coherence, only the very largest cells nearer the transmitter have any focusing effect on the illumination of small diffractive cells nearer the receiver, and eventually even these large cells cannot focus or defocus. When this loss of coherence happens, the illumination of the small cells is (statistically) evenly distributed and the uctuations of the propagating wave are due just to random interference of a large number of diffraction scatterings of the small cells. Thereby the compound effect subsides, and the optical eld uctuations approach a complex Gaussian distribution for which the corresponding irradiance probability density function (PDF) approaches a negative exponential distribution.

is the identication of the PDF for the amplitude or the irradiance of the wave. Over the years many irradiance PDF models have been proposed with varying degrees of success. The most widely accepted models have all evolved from an assumed modulation process in which the irradiance can be expressed as a product I xy . 29 36 Regardless of the particular PDF model, if we assume that the irradiance can be expressed as I xy , where x arises from large-scale turbulent eddies and y arises from statistically independent small-scale eddies, then the second moment of irradiance is
2 I 2 x 2 y 2 1 2 x 1 y ,

(5)

and are normalized variances of x and y, rewhere spectively. For convenience we have taken I 1. Based on Eq. (5), the implied scintillation index is
2 2 2 I 1 x 1 y 1 2 2 2 2 x y xy.

2 x

2 y

(6)

A. Modied Rytov Theory In weak-uctuation regimes the conventional method of analysis is called the Rytov theory. Under the Rytov approximation, the eld of an optical wave propagating at distance L from the source is represented by1,4,27 U r, L U 0 r, L exp r, L , (7)

where r is the observation point in the transverse plane at propagation distance L, U 0 ( r, L ) is the optical eld in the absence of turbulence, and ( r, L ) is a complex phase perturbation caused by random inhomogeneities along the propagation path. The validity of the Rytov method as described by Eq. (7) is generally limited to singlescattering regimes because it does not take into account the role of the decreasing transverse spatial coherence radius of the propagating wave, viz., that only those turbulent cells smaller than the coherence radius and those larger than the scattering disk contribute to irradiance uctuations in multiple-scattering regimes. To modify the Rytov approximation in multiple-scattering regimes, we assume that Eq. (7) can be expressed in the form U r, L U 0 r, L exp x r, L y r, L , (8)

where x ( r, L ) and y ( r, L ) are statistically independent complex phase perturbations that are due only to largescale and small-scale uctuations, respectively. Note that additivity in the argument of the exponential function in Eq. (8) is equivalent to a modulation process of the small-scale amplitude uctuations by large-scale amplitude uctuations. B. Spatial Filter Functions The modulation process discussed above addresses the inherent problem of nonhomogeneous statistics of atmospheric turbulence when applied to optical wave propagation. In addition, we treat the atmosphere like a linear lter that induces an amplitude spatial lter function that accounts for the loss of spatial coherence of the propagating wave. The lter function is formally introduced by replacing the refractive-index spectrum model (3) with

2. MODULATION PROCESS AND MODIFIED RYTOV THEORY


One of the goals in studying the scintillation phenomenon of an optical wave propagating through optical turbulence

1420

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

n 0.033C n 2 11/3G , l 0 , where G , l 0 G x , l 0 G y

(9)

2 11/3 f l 0 exp 2 . (10) x 2 y 2 11/6


In Eq. (10) the parameter x is a large-scale (or refractive) spatial-frequency cutoff much like an inner-scale parameter, and y is a small-scale (or diffractive) spatialfrequency cutoff similar to an outer-scale parameter. In this fashion G ( , l 0 ) acts like an amplitude (or irradiance) spatial lter function that permits only low-pass spatial frequencies x or high-pass spatial frequencies y at a given propagation distance. The highpass lter G y ( ) is similar in shape to the lter introduced by Yura [Eq. (41) in Ref. 7], although its functional form is different. The functional forms of these lter functions are chosen primarily on the basis of mathematical convenience. For the same reason, inner-scale effects in the model (10) are represented only in the large-scale lter function under all uctuation conditions.

uctuation conditions are characterized by the emergence of these two quantities as the dominant cell sizes, viz., the scattering disk forms the lower bound of the large refractive cells, and the coherence radius forms the upper bound of the small diffractive cells. Thus we deduce that the cutoff spatial frequencies introduced in the spatial lter function (10) satisfy the asymptotic behavior

x y

2 3 2 1

weak uctuations , strong uctuations weak uctuations . strong uctuations

(13)

(14)

In the following analysis, it is useful to introduce the nondimensional quantity L 2 / k , which is simply the ratio of the Fresnel zone area to the cell size area. In terms of this ratio of areas, the wave numbers in relations (12) correspond to L 12 k L k0 , 2 L 22 k 1, L 32 k k 02 L . (15) The relative cell sizes dened in relations (15) are shown in Fig. 1 for the limiting cases of weak and strong scintillations. The weak-scintillation regime is dened by L / k 0 2 1, in which the refractive and diffractive scales correspond roughly to 0 1 and 1, respectively [see Fig. 1(a)]. Strong uctuations are characterized by L / k 0 2 1, and hence the refractive cells correspond to 0 k 0 2 / L and the diffractive cells correspond to L / k 0 2 [see Fig. 1(b)]. Cell sizes in which k 0 2 / L L / k 0 2 do not contribute signicantly to irradiance uctuations in strong-uctuation regimes because of the loss of coherence of the impinging wave. Based on the above discussion, it is reasonable to assume that at arbitrary distance L into the random medium there exists an effective scattering disk L / kl x and an effective correlation length l y that identify the spatial frequencies x and y . These scale sizes are dened, respectively, by

3. SCINTILLATION INDEX FOR ZERO-INNER-SCALE MODEL


In the absence of both inner-scale and outer-scale effects, the irradiance is mainly affected by cell sizes described by l1 0 l2
spatial coherence radius , Fresnel zone size , scattering disk ,

L / k

l3 L/k0

(11)

which in turn correspond to the wave numbers (or spatial frequencies)

k L

1/2

k0 L

(12)

For constant C n 2 the plane-wave coherence radius is 0 (1.46C n 2 k 2 L ) 3/5 in both weak- and strong-uctuation regimes,1,4,27 whereas the Fresnel zone L / k denes the correlation length in only weak irradiance uctuations. A scattering disk is dened by the refractive cell size l at which the focusing angle F l / L is equal to the average diffraction angle, which, under weak uctuations, is D 1/k L / k 1/kL and, in strong uctuations, is D 1/k 0 . Hence, in weak uctuations, the Fresnel scale l 2 characterizes both the correlation length of the irradiance uctuations and the scattering disk. Cell sizes smaller than the Fresnel zone cause diffractive distortions of the optical wave, whereas those larger than the Fresnel zone cause refractive distortions such as focus and tilt. At the onset of strong uctuations, the coherence radius approaches the size of the Fresnel zone, and all three cell sizes (11) are roughly equal (i.e., l 1 l 2 l 3 ); this happens in the vicinity of the focusing regime. For yet stronger-uctuation conditions, the correlation length is dened by the spatial coherence radius 0 , which is now smaller than the Fresnel zone, and the scattering disk L / k 0 is larger (i.e., l 1 l 2 l 3 ). Strong-


L kl x 1

x2

c 1L k c3 L/k

c2


L k0 ,

(16)

l y2

y2

c4

02

(17)

Fig. 1. Relative positions of scale sizes in (a) weak and (b) strong turbulence.

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1421

where the scaling constants c 1 , c 2 , c 3 , and c 4 will be chosen in Subsection 3.A on the basis of known asymptotic behavior of the scintillation index. A. Plane Wave To begin, let us consider the scintillation index of an unbounded plane wave that has propagated a distance L through optical turbulence. Under weak-uctuation theory and the Rytov method, the scintillation index can be expressed in the form1,4,27

moderate-to-strong uctuations for which x 1. Along similar lines the second term of the lter function (10) leads to the small-scale log-irradiance scintillation dened by
2 2 ln y 1.06 1

1 0

y 11/6 1 cos d d

2 1.272 1 y 5/6, 2

y 1,

(25)

2 I
2 ln I

2 exp ln I

2 ln I,

2 1

1,

(18)

is the log-irradiance variance dened under where the Rytov approximation by


2 2 2 ln I 8 k

where y L y / k and we have assumed that y is sufciently large that the approximation in Eq. (25) is valid. From Eqs. (16) and (17), we see that the nondimensional quantities in Eqs. (24) and (25) are

2 1.06 1


L 0 1 0

n 1 cos


2z
k

d dz

1 c 1 c 2L / k 02

1/c 1 , k 0 / c 2L ,
2

L / k 02 1 , L / k 02 1 (26) L / k 02 1 . L / k 02 1 (27)

11/6 1 cos d d .

(19)

y c 3 c 4L / k 02

c3 , c 4L / k 0 ,
2

In the last step, we have assumed a conventional Kolmogorov spectrum and introduced the nondimensional quantities z / L and L 2 / k . Performing the integration in Eq. (19), we obtain the well-known result1
2 2 I 1 0.847


L k 02

5/6

2 1 1,

(20)

where we use Eq. (18) and the fact that L / k 0 2 2 6/5 1.22( 1 ) . At the other extreme, the asymptotic behavior of the scintillation index in the saturation regime is described by12

2 I 1

0.86
4/5 1

1 0.919


k 02 L

1/3

In weak uctuations the sum of Eqs. (24) and (25) should 2 be approximately 1 . Although it is somewhat ad hoc, we note that the choice x y 3 will give this result (not necessarily an optimum choice). Thus, in Eqs. (26) and (27) it follows that c 1 1/3 and c 3 3. In strong uctuations we expect the large-scale scintillation to die out eventually, leading to the conclusion that the smallscale scintillation 2 y must necessarily approach the asymptotic limit of unity. For this to happen, we require 2 2 2 ln y ln 2, and thus y 1.7L / k 0 , L / k 0 1, or c 4 1.7. The scintillation index (23) in the saturation regime therefore reduces to
2 2 I 2 exp ln x 1

2 1 1.

(21) 1

0.24 1

If we invoke the Rytov theory for the small-scale and large-scale scintillations, then
2 2 x exp ln x 1, 2 2 y exp ln y 1,

4/5 c2 1

3,

7/6

2 1 1,

(28)

(22)

2 2 where ln x and ln y are large-scale and small-scale logirradiance scintillations, respectively. It follows, therefore, that the scintillation index (6) can also be expressed 2 2 directly in terms of ln x and ln y according to 2 2 2 I exp ln x ln y 1.

where we have also used the asymptotic form x k 02/ c 2L . By equating relation (28) with the asymptotic result (21), we see that c 2 1/3. Hence the parameters (26) and (27) for the plane-wave case under general uctuation conditions become

(23)

3 1 L / k 02

When only the large-scale component of the lter function (10) with f ( l 0 ) 1 is used, the resulting logirradiance scintillation is
2 2 2 ln x 8 k

L / k 02 1
2

3k0 /L,

L / k 02 1

, (29)

L 0

n G x 1 cos
1

2 1.06 1


L k
7/6


2z
k

y 3 1.7L / k 0 2

3, 1.7L / k 0 ,
2

L / k 02 1 L / k 02 1

d dz

(30) 2 6/5 With L / k 0 2 1.22( 1 ) , it follows from Eqs. (24) and (25) that
2 ln x 2 0.54 1 12/5 7/6 1 1.22 1

4/3 exp 2 / x 2 d d
(24)
2 ln y

2 0.15 1 x 7/6,

2 0.54 1 , 4/5 , 0.43/ 1

2 1 1 2 1 1

, (31)

where x L x 2 / k and we have invoked the geometricaloptics approximation. The geometrical-optics approximation is valid here, since the lter function eliminates high-spatial-frequency contributions, particularly in

2 0.509 1

12/5 5/6 0.69 1

2 0.509 1 ,

ln 2,

2 1 2 1

1 1

, (32)

1422

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

and consequently, the scintillation index (23) for a plane wave in the absence of inner scale is given by

2 I exp

2 0.54 1 12/5 7/6 1 1.22 1 2 . 0 1

2 0.509 1 12/5 5/6 1 0.69 1

1,

(33)

B. Spherical Wave In the case of a spherical wave, the scintillation index based on f ( l 0 ) 1 and conventional Rytov theory leads to1,4
2 2 2 I exp ln I 1 ln I , 2 1 1,

(34)

where
2 ln I

8 k

2 2

L 0 1 0

n 1 cos

2
k

z 1 z/L

d dz
2 1.06 1

Fig. 2. The solid curves depict the scintillation index for the plane-wave and spherical-wave models in the absence of innerscale effects. The dashed curves correspond to the asymptotic theory valid for the saturation regime, and the dotted curve is the small-scale scintillation for the spherical-wave model.

11/6 1 cos 1 d d ,
2 1 1. (35)

The choice of scaling constant in Eqs. (39) and (40) for L / k 0 2 1 is based on knowledge that the sum of Eqs. 2 . With L / k 0 2 (37) and (38) is approximately 0.4 1 2 6/5 1.22( 1 ) , Eqs. (37) and (38) lead to
2 0.17 1 12/5 7/6 1 0.167 1

Results deduced from Eq. (35) and the corresponding asymptotic theory provide expressions for the scintillation index of a spherical wave in weak- and strong-uctuation regimes given by4,37

2 ln x

2 0.17 1 , 4/5 , 1.37/ 1

2 1 1

2 I

, 2 1 1 (41) . (42)

2 0.4 1 ,

2 1 1

2.73
4/5 1

2 1

(36)

2 ln y

2 0.225 1 12/5 5/6 1 0.259 1

2 0.225 1 ,

2 1 1 2 1 1

ln 2,

Relying on these asymptotic results, we develop the largescale and small-scale log-irradiance scintillation in a manner analogous to that for a plane wave. For example,
2 2 ln x 1.06 1

Hence the scintillation index for the spherical-wave model in the absence of inner-scale effects takes the form

1 0

2 I exp

11/6

2 0.17 1 12/5 7/6 1 0.167 1 2 . 0 1

2 0.225 1 12/5 5/6 1 0.259 1

1,

(43)

exp / x 1 cos 1 d d
2 0.015 1 x 7/6, 2 2 ln y 1.06 1

(37)
y 11/6

1 0

1 cos 1 d d
2 1.272 1 y 5/6,

(38)

In Fig. 2 we plot the predicted scintillation index of the plane wave [Eq. (33)] and the spherical wave [Eq. (43)] as a function of the strength-of-turbulence parameter 1 . Peak scintillation occurs in the vicinity of 1 2 in the plane-wave case and 1 4 in the spherical-wave case. The dotted curve is the small-scale scintillation of a spherical wave, which approaches unity from below. Since published experimental data contain inner-scale effects, we cannot compare them directly with our results.

where in this case we select the nondimensional quantities x and y according to

8 1 0.137L / k 0 2

8, 58.4k 0 2 / L ,

y 8 1.7L / k 0 2

L / k 02 1 , L / k 02 1 (39) L / k 02 1 . L / k 02 1 (40)

4. SCINTILLATION INDEX FOR NONZERO-INNER-SCALE MODEL


When inner-scale effects become important, previous studies21 23 have shown that the atmospheric power spectrum is best described by a modied spectrum with highwave-number rise.25,26 That is, the traditional Tatarskii spectrum with inner scale typically underestimates the scintillation index.

8, 1.7L / k 0 ,
2

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1423

Under weak irradiance uctuations in the case of an unbounded plane wave, the scintillation index based on the modied spectrum is described by27,28

The large-scale log-irradiance scintillation in the planewave case is dened by


2 2 ln x 1.06 1

2 I L

2 3.86 1

1 1/Q l

2 11/12

1.507
1 Ql
2 1/4

sin

sin

5 4

tan1 Q l


sin

11 6

tan

Ql

4 3

tan1 Q l

2 1, 1

1 0

11/6 exp / Q l / x

1 1.802 / Q l 1/2 0.254 / Q l 7/12 1 cos d d


2 0.15 1 x 7/6 l 0 ,

0.273
1 Q l 2 7/24

3.50Q l 5/6 ,

(47)

(44) and for a spherical wave the comparable expression is


2 2 11/12 2 sin I L 3.86 1 0.40 1 9/Q l

where x ( l 0 ) is the inner-scale counterpart of x dened by

27,28

2.610
9 Ql
2 1/4

sin

sin

5 4

tan1

Ql 3

4 3

tan1


Ql 3

11 6

tan1

Ql 3

(45)

x l 0

xQ l

x Ql

0.252


1 1.753

x x Ql
.

1/2

7/12 6/7

x Ql

(48)

0.518
9 Q l 2 7/24
2 1, 1

3.50Q l 5/6 ,

Note that x ( l 0 ) x as l 0 0. The same expression (47), multiplied by 1/10, arises in the spherical-wave case. Under moderate-to-strong irradiance uctuations, we can approximate the nondimensional quantity x in Eq. (48) by an expression similar to that used in the zero-innerscale case; that is,

where Q l 10.89L / kl 0 2 is a nondimensional inner-scale parameter. Asymptotic expressions for the scintillation index in the saturation regime, based on the modied atmospheric spectrum, have not previously been published. Such expressions are (see Appendix A)

x,p

3 1 0.49L / k 0
2

3
2 1 0.50 1 Q l 1/6

plane wave , (49)

2 I

1 1

2.39
2 Q l 7/6 1/6 1

2 1 Q l 7/6 100

plane wave

7.65
2 Q l 7/6 1/6 1

. ,
2 1 Q l 7/6

(46)

100

spherical wave

Under general conditions the size of the inner scale l 0 relative to the Fresnel zone L / k is an important consideration. For example, in weak irradiance uctuations associated with short propagation paths, the inner scale may be of similar size to or larger than the Fresnel zone; hence there will be little contribution to scintillation from eddy cells smaller than the inner scale. On the other hand, over longer propagation path lengths the inner scale can be much smaller than the Fresnel zone. In this latter situation, cells the size of the inner scale and smaller contribute most to small-scale scintillation, and large-scale scintillation is dominated by cells larger than L / kl 0 . Eventually, however, the coherence radius becomes smaller than the inner scale, and small-scale scintillation depends less and less on cells the size of the inner scale in the saturation regime. Large-scale scintillation, which continues to depend on the inner scale, begins to diminish in the saturation regime, since only those cells larger than the scattering disk are strong enough to still cause focusing effects.

x,s

8 1 0.068L / k 0
2

8
2 1 0.069 1 Q l 1/6

spherical wave ,

(50)

2 where L / k 0 2 1.02 1 Q l 1/6 in the presence of inner 38 scale. These two expressions lead to the large-scale log-irradiance scintillations dened, respectively, by

2 2 ln x , p l 0 0.15 1

1 1.753 0.252

x,pQ l x,p Q l

7/6

x,p x,p Q l

x,p x,p Q l


7/12

1/2

plane wave ,

(51)

1424

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

Fig. 3. Scintillation index of a plane wave with inner-scale effects for 0.488 m and C n 2 5 1013 m2/3.

2 2 ln x , s l 0 0.015 1

1 1.753 0.252

x,sQ l x,s Q l

7/6

x,s x,s Q l

x,s x,s Q l


7/12

1/2

5 mm. In both gures we set 2 / k 0.488 m and C n 2 5 1013 m2/3 and allowed the propagation distance L to vary. We illustrate the effect of C n 2 variations for a xed inner scale l 0 and 0 in Fig. 5 for l 0 1 mm and 8 mm and for two values of C n 2 in each case. Although predicted scintillation for a given 0 value and small inner scale does not change much with C n 2 , the change is quite signicant for large inner scales. Nonetheless, the results are consistent with experimental data3,18 and with simulation data,21,22 but the former inherently contain considerable scatter, making accurate comparisons somewhat difcult. To compare our results directly with simulation data,21 we show in Fig. 6 the predicted scintillation index as a function of inner scale for 1 5.48 in the plane-wave case and for 0 5.48 in the spherical-wave case. The curves in both cases are similar to those of the empirical formulas 2 I 1.74 3.02 0.35 0 0.092 1 0.6( kl 0 2 / L ) 1/2 and 2 I 5.56( kl 0 2 / L ) 1/2 taken from Ref. 21 for plane waves and spherical waves, respectively. The maximum difference for the plane-wave case is 18% (for l 0 8 mm) above that of the simulation data, and it is 21% (for l 0 2 mm) for the spherical-wave case.

spherical wave .

(52)

When the coherence radius 0 is smaller than the inner scale l 0 , small-scale log-irradiance scintillation can be described by Eq. (25), even for nonzero inner scale. Note that the condition 0 l 0 corresponds to 2 I 1 in all experimental data reported by Consortini et al.18 Thus, in strong turbulence we assume that small-scale log2 irradiance scintillations ln y are adequately described under moderate-to-strong irradiance uctuations by Eqs. (32) and (42) for the plane-wave and spherical-wave cases, respectively. Based on Eq. (23), the scintillation index for a plane wave and that for a spherical wave in the presence of nite inner scale are, respectively,
2 2 I exp ln x , p l 0

2 0.509 1 12/5 5/6 1 0.69 1

Fig. 4.

Same as Fig. 3, but for a spherical wave.

1 (53)

2 2 I exp ln x , s l 0

plane wave ,
2 0.225 1 12/5 5/6 1 0.259 1

spherical wave . (54)

In Figs. 3 and 4, we illustrate the scintillation index for plane waves [Eq. (53)] and spherical waves [Eq. (54)] with several values of inner scale. For the spherical wave, we 2 plot the curves as a function of 0 , where 0 2 7/6 11/6 0.5C n k L is the Rytov variance for a spherical wave. Because weak-uctuation inner-scale effects are ignored in the small-scale scintillation model, the portion of the curves for 1 1 and 0 1 tends to be above that predicted by the weak-scintillation results of relations (44) and (45), particularly in cases when l 0

Fig. 5. Scintillation index of a spherical wave with two values of inner scale, 0.488 m, and a 1-order-of-magnitude change in C n2.

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1425

2 B ln y 1.06 1

1 0

y 11/6J 0 k / L

1 cos d d
2 1.265 1


k2 Ly

5/12

K 5/6


k 2 y L
1/2

(58)

where y is dened by Eq. (30) and K ( ) is a modied Bessel function. For the case of a spherical wave, the large-scale and small-scale log-irradiance covariances are, respectively,
2 B ln x 1.06 1

1 0

11/6 exp / x J 0 k / L

1 cos 1 d d
Fig. 6. Scintillation index as a function of inner scale with 0 5.48 for the spherical-wave case and 1 5.48 for the planewave case. The dashed curves represent simulation data taken from Ref. 21, and the solid curves are based on the theory presented here.
2 0.015 1 x 7/6

3F 3
2 B ln y 1.06 1

5. IRRADIANCE COVARIANCE FUNCTIONS FOR ZERO-INNER-SCALE MODEL


In this section we develop expressions for the large-scale and small-scale irradiance covariance functions for the case when l 0 0. Our representation for the total covariance function is the same as that discussed by Rickett et al.39 for scintillation in the interstellar medium. Specically, based on the assumed modulation process, the irradiance covariance function can be expressed as a sum: B I B x B y B x B y , (55)

1 0

7 3 7 k 2 x , , , 2; , 3, 1; 6 2 2 4L

(59)

y 11/6J 0 k / L

1 cos 1 d d
2 1.272 1 y 5/6 1 F 2

1 1 3 k 2 y ; , ; 2 6 2 4L

2 0.889 1 k 2 / L 5/6 1 F 2

4 11 7 k 2 y , ; , ; 3 6 3 4L (60)

where denotes separation distance between two points on the wave front. Here B x ( ) is the covariance of the large-scale process, and B y ( ) is the covariance of the small-scale process. The covariance of irradiance (55) is related to the covariance of log irradiance in a manner similar to that of the variance, viz., B I exp B ln I 1 exp B ln x B ln y 1, (56)

where x is dened by Eq. (39) and 3 F 3 and 1 F 2 are generalized hypergeometric functions.40 These expressions lead to the covariance function for the plane wave given by
2 B I exp ln x 1F 1

2 0.994 ln y


7 6 k 2 y L

; 1;

k 2 x 4L

5/12

K 5/6


k 2 y L
1/2

1, (61)

where B ln x() and B ln y() are the large-scale and smallscale log-irradiance covariances, respectively. In the particular case of a plane wave, the log-irradiance covariance of the large-scale process is given by
2 B ln x 1.06 1

and for the spherical wave the covariance function becomes


2 B I exp ln x 3F 3

1 0

11/6 exp / x J 0 k / L

2 ln y 1F 2

1 cos d d
2 0.15 1 x 7/6 1 F 1

7 6 ; 1;

k 2 x 4L

(57)

2 0.70 ln y

where x is dened by Eq. (29) and 1 F 1 denotes a conuent hypergeometric function.40 The corresponding logirradiance covariance of small-scale uctuations is approximated by

1F 2


k 2 y L

7 3 7 k 2 x , , 2; , 3, 1; 6 2 2 4L

1 1 3 k 2 y ; , ; 2 6 2 4L
5/6

4 11 7 k 2 y ; , ; 3 6 3 4L

1,

(62)

where the log-irradiance variances are dened in Section 3.

1426

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

In the saturation regime, the asymptotic theory predicts that the covariance can be expressed as a sum of three terms from a series expansion, i.e.,12,14,37
hf lf hf B I C 0 C1 C1 hf C0

where D ( ) is the wave structure function. Once again, for the case of a plane wave, the corresponding highspatial-frequency component deduced from Eq. (58) in the saturation regime takes the form B y exp 0.86

A
4/5 1

b 1 / l 3 b 2 / 0 ,

2 1 1,

5/6

K 5/6

1.30

(63) where superscripts lf and hf refer, respectively, to low spatial frequency and high spatial frequency. The constant A 0.43 in the plane-wave case, A 1.37 in the spherical-wave case, and l 3 L / k 0 is the scattering disk. To compare Eq. (55) directly with the form of relation (63), we rst make the identications
hf C0 B y , lf C1 B x B ln x , hf C1 B x B y B ln x B y ,

exp 2 / 0 5/3 exp 2 / 0 5/3 1 exp 2 / 0 5/3 ,


2 0 , 1 . (69)

(64)

Thus, for sufciently small separation distances, we have agreement between Eqs. (68) and (69). In general, howhf ( ). ever, B y ( ) decreases to zero more slowly than C 0 hf ( ) 1/e 2 at separation distance 0 and That is, C 0 approaches zero near 2 0 , whereas B y ( ) 1/e at 0 and does not approach zero until separation distance 4 0 . It is customary to equate the irradiance correlation length with either the 1/e or the 1/e 2 value of the normalized covariance function b I B x B y B x B y
2 2 2 2 x y xy

where we use the fact that the large-scale log-irradiance covariance function decreases toward zero in the satura2 1. From these tion regime, so that B x ( ) B ln x(), 1 relations and the fact that x 3 k 0 2 / L 3 0 / l 3 and y 1.70L / k 0 2 for a plane wave in the saturation regime, we deduce from expressions (57) and (58) that b 1 / l 3 1F 1

(70)

b 2 / 0 1F 1


32 ; 1; 2 , 6 4l3 7 3 /0 ; 1; 6 4 l3 /02 7
2

(65)

For the purpose of making comparisons with conventional perturbation theory, we note that in the weak-uctuation regime the normalized covariance as predicted here in both the plane-wave case ( x y 3) and the spherical-wave case ( x y 8) has the asymptotic behavior

b I

exp 0.86

5/6

K 5/6

1.30

1 .

(66)

Note that whereas b 2 ( / 0 ) depends on the transverse spatial coherence radius 0 and on the scattering disk l 3 , the function b 1 ( / l 3 ) depends only on the scattering disk. The representation for b 2 ( / 0 ) is quite complicated based on the asymptotic theory, so direct comparisons are difcult to make, but b 1 ( / l 3 ) from the asymptotic theory is12,37,41 b 1 / l 3 0.736

(71) By conventional perturbation theory, the corresponding expressions are42

1 2.37 k 2 / L 5/6 1.82 k 2 / L , l0 l0

L / k L / k

plane wave spherical wave

1 2.22 k 2 / L 5/6 1.97 k 2 / L ,

1 2.36 k 2 / L 5/6 1.71 k 2 / L , l0 l0

b I

L / k
2

plane wave
5/6

1 2.2 k / L

L / k

1.71 k 2 / L ,
spherical wave

u 4/3J 0 0.658 u / l 3
5 8 du d .

exp u 5/3 5/3 1

(67)

Numerical values arising from the function b 1 ( / l 3 ) in Eq. (65) are remarkably close to those from Eq. (67). In particular, both expressions are unity when 0 and cross the axis at / l 3 1.9. The low-frequency (large-scale) terms eventually go to zero, and the irradiance covariance is dominated by the high-frequency (small-scale) covariance function. According to the asymptotic theory, the high-frequency leading-order term is described by
hf C0 exp D

exp 2 / 0 5/3 ,

2 1 ,

(68)

(72) Clearly, the difference between Eqs. (71) and (72) for small separation distances is minor. In Fig. 7 we show the normalized plane-wave covariance function (70) for various values of the Rytov variance 2 depicting weak-turbulence conditions ( 1 1), the focus2 4), and strong-turbulence conditions ing regime ( 1 2 50). This behavior agrees qualitatively with (1 known results. Namely, in weak scintillations the correlation length of irradiance uctuations is determined by the Fresnel zone L / k , whereas in strong-uctuation regimes it is dened by the spatial coherence radius 0 . In Fig. 8 the normalized covariance function (70) for a 2 1), spherical wave is shown for weak uctuations ( 0 2 the focusing regime ( 0 4), and strong uctuations 2 36). In both gures the long correlation tail for (0 strong turbulence is characterized by the scattering disk L / k 0 . The emergence of two scales under strong uc-

Andrews et al.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1427

6. SUMMARY
Within the framework of small-scale scintillation being modulated by large-scale scintillation, we have developed tractable expressions for the scintillation index and the irradiance covariance functions under moderate-to-strong irradiance uctuations. The method of approach adopted here is largely heuristic, based for the most part on plausible qualitative arguments that have been put forth by other researchers7 9,30 37,41 and also on observations made about irradiance measurements.5,6,17,39 The twoscale process associated with scintillation is described in detail in the classic paper by Prokorov et al.37 Also, Strohbehn41 discusses the two-scale process and provides a summary of some of the early heuristic approaches to scintillation. Relying on the modied atmospheric spectrum with high-wave-number bump, we assume that inner scale effects in strong uctuations arise only in the large-scale scintillationsmall-scale scintillation under saturation conditions depends more on the size of the spatial coherence radius than on the inner scale. Thus, in the attempt to achieve the desired level of simplicity and tractability of the scintillation models, the weak-irradianceuctuation results may not always agree exactly with well-known results based on weak-uctuation theories. However, the quantitative discrepancy in the weakuctuation regime is often minor, particularly for small inner scales. Extension of the method to include Gaussian-beam waves is currently under way and will be presented in a future paper.

Fig. 7. Normalized plane-wave covariance for various strengths of turbulence.

APPENDIX A
Fig. 8. Normalized spherical-wave covariance for various strengths of turbulence.

In this appendix we derive the asymptotic formulas (46) for the plane-wave and spherical-wave cases, based on the modied atmospheric spectrum. In the saturation regime, the asymptotic theory predicts that the scintillation index can be expressed in the form12 14
2 2 2 I L 1 32 k L

exp

1 0

n sin2

L k

h , d d d ,

L2 2k

h ,

(A1)

where the exponential function acts like a low-pass spatial lter dened by the plane-wave structure function of phase D ( ). The function h ( , ) is dened by h ,

Fig. 9. Spatial scales of irradiance uctuations normalized by the Fresnel zone L / k and plotted as a function of the strength of turbulence. The emergence of two scales occurs at the onset of strong uctuations.

1 ,
1 ,

(A2)

tuations is illustrated in Fig. 9 for both the plane wave and the spherical wave, where the small-scale correlation length and the large-scale scattering disk are each scaled by the Fresnel zone L / k .

where the parameter 0 for a plane wave and 1 for a spherical wave. 2 Q l 7/6 100 and the coherence radius of the When 1 wave 0 is much less than the inner scale of turbulence l 0 , the phase structure function for a plane wave based on the modied atmospheric spectrum can be approximated by38 D ( ) 1.87C n 2 k 2 Ll 0 1/3 2 . Hence, for the planewave case, we nd that

1428

J. Opt. Soc. Am. A / Vol. 16, No. 6 / June 1999

Andrews et al.

L k

2 Q l 1/6 h , d 1.02 1


L2 k

2 1

2 3

,
(A3)

ACKNOWLEDGMENT
Funding for this work was partially provided by the Ballistic Missile Defense Organizations Innovative Science and Technology Directorate and administered by the Space and Naval Warfare Systems Center, San Diego, California, under contract N66001-97-C-6008. Address correspondence to L. C. Andrews at the location on the title page or by phone, 407-823-2418; fax, 407823-6253; or e-mail, landrews@pegasus.cc.ucf.edu.

and Eq. (A1) leads to the expression


2 2 7/6 I ,pl L 1 0.982 1 Q l


1 0

2
1
2 1.02 1 Q l 7/6 2 1
2 7/6 3

1.802 5/3 7/6

2
1
2 1.02 1 Q l 7/6 2 1

2 5/3 3

REFERENCES
1. 2. V. I. Tatarskii, Wave Propagation in a Turbulent Medium, translated from Russian by R. A. Silverman (McGraw-Hill, New York, 1961). L. A. Chernov, Wave Propagation in a Random Medium, translated from Russian by R. A. Silverman (McGraw-Hill, New York, 1960). M. E. Gracheva and A. S. Gurvich, Strong uctuations in the intensity of light propagated through the atmosphere close to the earth, Izv. Vyssh. Uchebn. Zaved. Radioz. 8, 717 724 (1965). V. I. Tatarskii, The Effects of the Turbulent Atmosphere on Wave Propagation (Keter, Jerusalem, 1971). J. Dunphy and J. Kerr, Scintillation measurements for large integrated-path turbulence, J. Opt. Soc. Am. 63, 981 986 (1973). M. Gracheva, A. S. Gurvich, S. S. Kashkarov, and V. V. Pokasov, Similarity relations for strong uctuations of light in a turbulent medium, Sov. Phys. JETP 40, 1011 1016 (1974). H. T. Yura, Physical model for strong optical-amplitude uctuations in a turbulent medium, J. Opt. Soc. Am. 64, 59 67 (1974). S. F. Clifford, G. R. Ochs, and R. S. Lawrence, Saturation of optical scintillation by strong turbulence, J. Opt. Soc. Am. 64, 148 154 (1974). R. J. Hill and S. F. Clifford, Theory of saturation of optical scintillation by strong turbulence for arbitrary refractiveindex spectra, J. Opt. Soc. Am. 71, 675 686 (1981). R. G. Frehlich, S. M. Wandzura, and R. J. Hill, Logamplitude covariance for waves propagating through very strong turbulence, J. Opt. Soc. Am. A 4, 2158 2161 (1987). R. J. Hill and R. G. Frehlich, Onset of strong scintillation with application to remote sensing of turbulence inner scale, Appl. Opt. 35, 986 997 (1996). K. S. Gochelashvili and V. I. Shishov, Saturated uctuations in the laser radiation intensity in a turbulent medium, Sov. Phys. JETP 39, 605 609 (1974). R. L. Fante, Inner-scale size effect on the scintillations of light in the turbulent atmosphere, J. Opt. Soc. Am. 73, 277 281 (1983). R. G. Frehlich, Intensity covariance of a point source in a random medium with a Kolmogorov spectrum and an inner scale of turbulence, J. Opt. Soc. Am. A 4, 360 366 (1987); erratum, 4, 1324 (1987). G. Parry, Measurements of atmospheric turbulenceinduced intensity uctuations in a laser beam, Opt. Acta 28, 715 728 (1981). R. L. Phillips and L. C. Andrews, Measured statistics of laser-light scattering in atmospheric turbulence, J. Opt. Soc. Am. 71, 1440 1445 (1981). W. R. Coles and R. G. Frehlich, Simultaneous measurements of angular scattering and intensity scintillation in the atmosphere, J. Opt. Soc. Am. 72, 1042 1048 (1982). A. Consortini, R. Cochetti, J. H. Churnside, and R. J. Hill, Inner-scale effect on irradiance variance measured for weak-to-strong atmospheric scintillation, J. Opt. Soc. Am. A 10, 2354 2362 (1993).

0.254 7/4 7/6


1

2
2 Q l 7/6 2 1 1 1.02 1
2 7/4 3

d . (A4)
4. 5. 6.

3.

The evaluation of these last integrals in closed form is not 2 Q l 7/6 known. However, if we impose the condition 1 100, a good approximation to the sum of the three integrals is given by
2 I ,pl L 1

2.39
2 Q l 7/6 1/6 1

2 1 Q l 7/6 100. (A5)

7. 8.

In the spherical-wave case, we rst note that

L k

h , d

2 0.34 1 Q l 1/6


L2 k

1 ,
2 2

9. 10.

(A6)

which leads to
2 I ,sph L


1 0

2 0.982 1 Q l 7/6 1

2 1 2
2 Q l 7/6 2 1 2 7/6 1 0.34 1

11.

d
12. 13.

1.802 5/3 7/6

2 1 2
2 Q l 7/6 2 1 2 5/3 1 0.34 1

14.

0.254 7/4 7/6


1

15.

2 1 2
1
2 0.34 1 Q l 7/6 2 1

2 7/4

d . (A7)

16. 17. 18.

2 Q l 7/6 100, Eq. (A7) can be approximated by For 1 2 I ,sph L

7.65
2 Q l 7/6 1/6 1

2 1 Q l 7/6

100.

(A8)

Andrews et al. 19. 20. 21. , Intensity images and staJ. M. Martin and S. M. Flatte tistics from numerical simulation of wave propagation in 3-D random media, Appl. Opt. 27, 2111 2126 (1988). , Simulation of point-source J. M. Martin and S. M. Flatte scintillation through three-dimensional random media, J. Opt. Soc. Am. A 7, 838 847 (1990). , G. Wang, and J. Martin, Irradiance variance S. M. Flatte of optical waves through atmospheric turbulence by numerical simulation and comparison with experiment, J. Opt. Soc. Am. A 10, 2363 2370 (1993). , C. Bracher, and G.-Y. Wang, ProbabilityS. M. Flatte density functions of irradiance for waves in atmospheric turbulence calculated by numerical simulations, J. Opt. Soc. Am. A 11, 2080 2092 (1994). R. J. Hill and R. G. Frehlich, Probability distribution of irradiance for the onset of strong scintillation, J. Opt. Soc. Am. A 14, 1530 1540 (1997). V. I. Tatarskii and V. U. Zavorotnyi, Wave propagation in random media with uctuating turbulent parameters, J. Opt. Soc. Am. A 2, 2069 2076 (1985). R. J. Hill and S. F. Clifford, Modied spectrum of atmospheric temperature uctuations and its application to optical propagation, J. Opt. Soc. Am. 68, 892 899 (1978). L. C. Andrews, An analytical model for the refractive index power spectrum and its application to optical scintillations in the atmosphere, J. Mod. Opt. 39, 1849 1853 (1992). L. C. Andrews and R. L. Phillips, Laser Propagation through Random Media (Society of Photo-Optical Instrumentation Engineers, Bellingham, Wash., 1998). W. B. Miller, J. C. Ricklin, and L. C. Andrews, Effects of the refractive index spectral model on the irradiance variance of a Gaussian beam, J. Opt. Soc. Am. A 11, 2719 2726 (1994). E. Jakeman and P. N. Pusey, A model for non-Rayleigh sea echo, IEEE Trans. Antennas Propag. AP-24, 806 814 (1976). E. Jakeman and P. N. Pusey, The signicance of K-distributions in scattering experiments, Phys. Rev. Lett. 40, 546 550 (1978). 31. 32.

Vol. 16, No. 6 / June 1999 / J. Opt. Soc. Am. A

1429

33.

34. 35.

22.

23. 24. 25. 26. 27. 28.

36.

37. 38. 39. 40.

41.

29. 30.

42.

E. Jakeman, On the statistics of K-distributed noise, J. Phys. A 13, 31 48 (1980). L. C. Andrews and R. L. Phillips, I K distribution as a universal propagation model of laser beams in atmospheric turbulence, J. Opt. Soc. Am. A 2, 160 163 (1985). L. C. Andrews and R. L. Phillips, Mathematical genesis of the I K distribution for random optical elds, J. Opt. Soc. Am. A 3, 1912 1919 (1986). J. H. Churnside and R. J. Hill, Probability density of irradiance scintillations for strong path-integrated refractive turbulence, J. Opt. Soc. Am. A 4, 727 733 (1987). J. H. Churnside and S. F. Clifford, Log-normal Rician probability-density function of optical scintillations in the turbulent atmosphere, J. Opt. Soc. Am. A 4, 1923 1930 (1987). J. H. Churnside and R. G. Frehlich, Experimental evaluation of log-normally modulated Rician and IK models of optical scintillation in the atmosphere, J. Opt. Soc. Am. A 6, 1760 1766 (1989). A. M. Prokhorov, F. V. Bunkin, K. S. Goshelashvili, and V. I. Shishov, Laser irradiance in turbulent media, Proc. IEEE 63, 790 809 (1975). C. Y. Young and L. C. Andrews, Effects of a modied spectral model on the spatial coherence of a laser beam, Waves Random Media 4, 385 397 (1994). B. J. Rickett, W. A. Coles, and G. Bourgois, Slow scintillation in the interstellar medium, Astron. Astrophys. 134, 390 395 (1984). L. C. Andrews, Special Functions of Mathematics for Engineers, 2nd ed. (Society of Photo-Optical Instrumentation Engineers, Bellingham, Wash., 1998) [formerly published as 2nd ed (McGraw-Hill, New York, 1992)]. J. W. Strohbehn, Modern theories in the propagation of optical waves in a turbulent medium, in Laser Beam Propagation in the Atmosphere, J. W. Strohbehn, ed. (Springer, New York, 1978), Chap. 3. R. S. Lawrence and J. W. Strohbehn, A survey of clear-air propagation effects relevant to optical communications, Proc. IEEE 58, 1523 1545 (1970).

You might also like