Characterization of atmospheric turbulence effects over 149 km propagation path using multi-wavelength laser beacons

Transcription

1 Characterization of atospheric turbulence effects over 149 k propagation path using ulti-wavelength laser beacons Mikhail A. Vorontsov School of Engineering, University of Dayton, Dayton, Ohio 45469, USA Gary W. Carhart Coputational and nforation Sciences Directorate, U.S. Ary Research Laboratory, Adelphi, Maryland 0783, USA V. S. Rao Gudietla Directed Energy Directorate, Air Force Research Laboratory, Kihei, Maui, Hawaii 96753, USA Thoas Weyrauch School of Engineering, University of Dayton, Dayton, Ohio 45469, USA Eric Stevenson Boeing LTS, Kihei, Maui, Hawaii 96753, USA Svetlana L. Lachinova nstitute for Systes Research, University of Maryland, College Park, Maryland 074, USA Leonid A. Beresnev Coputational and nforation Sciences Directorate, U.S. Ary Research Laboratory, Adelphi, Maryland 0783, USA Jiang Liu Coputational and nforation Sciences Directorate, U.S. Ary Research Laboratory, Adelphi, Maryland 0783, USA Karl Rehder Boeing LTS, Kihei, Maui, Hawaii 96753, USA and Ji F. Riker Space Vehicles Directorate, Air Force Research Laboratory, Kirtland AFB, ew Mexico 87117, USA ABSTRACT We describe preliinary results of a set of laser bea propagation experients perfored over a long (149 k) near-horizontal propagation path between Mauna Loa (Hawaii sland) and Haleakala (sland of Maui) ountains in February 010. The distinctive feature of the experiental capaign referred to here as the Coherent Multi-Bea Atospheric Transceiver (COMBAT) experients is that the easureents of the atospheric-turbulence induced laser bea intensity scintillations at the receiver telescope aperture were obtained siultaneously using three laser sources (laser beacons) with different wavelengths (λ 1 = 0.53 μ, λ = 1.06 μ, and λ 3 = 1.55 μ). The presented experiental results on intensity scintillation characteristics reveal coplexity of the observed phenoena that cannot be fully explained based on the existing atospheric turbulence odels. 1. TRODUCTO There is a growing interest in developent of optical systes capable of operation over long atospheric distances (path lengths up to and over 100 k) in various atospheric conditions and engageent scenarios. Aong these systes are long-range laser counications, reote sensing, active and passive iaging, target tracking and designation, and laser bea projection (directed energy) systes. Propagation of optical waves over a long distance through volue atospheric turbulence can result in significant changes in characteristics of received or transitted waves and thus draatically ipact perforance of the optical systes.

2 The turbulence effects over atospheric propagation paths are coonly associated with fluctuations of intensity (scintillations), foration of wavefront phase topological singularities (branch points), wander and widening of laser bea, spatial non-unifority of the turbulence-induced iage degradation, etc. [1 4]. Currently, analysis of atospheric turbulence-induced effects is based on theoretical and nuerical odels that are derived fro the classical Kologorov-Obukhov turbulence theory [5 8], developed in 1940's 1960's. This theory was extensively validated through a nuber of various atospheric experients. The ajority of these experients were perfored over relatively short distances (coonly with path lengths of less than several kiloeters [9 1]) and only a few experients over distances exceeding 100 k [13 15]. evertheless the odels derived fro the classical Kologorov-Obukhov turbulence theory are coonly used for perforance assessent of long-range optical systes. The existing gap in experiental validation of turbulence odels over long-range propagation distances ay result in significant errors, leading to iscalculations of optical syste capabilities and conceptual istakes in their design. The risk related with unconscious extension of the classic Kologorov-Obukhov turbulence odel for analysis of atospheric turbulence effect on optical wave propagation over long-range distances has been recognized and experiental evidence of noticeable deviations fro this odel prediction have already been observed in a few long-range experients [13 16]. Several physical factors can contribute to such unanticipated (non-classical) behavior of optical wave characteristics that are not accounted for in the existing theories, including dynaics of large-scale air flows leading to strongly non-stationary refraction effects, ipact of ocean-land interface, nonunifority of air pressure, teperature, and wind profiles along the propagation path. Unfortunately, due to coplexity (both logistical and technical) and cost which are coonly associated with long-range optical wave propagation experients, the collected experiental data are not sufficiently representative to deterine if soe iportant adjustents can be ade to the existing theory to account for extended turbulence effects, or if fundaentally new approaches are required. n this paper we intend to partially fill out the existing gap in experiental assessent of extended turbulence effects on laser bea propagation over long-range paths by providing a large aount of intensity scintillation data obtained over a 149 k long propagation path using light sources with different wavelength. The intensity scintillations are coonly characterized by the noralized variance of optical wave intensity fluctuations easured using a point photo-detector located at a point r = {x,y} of receiver aperture, ( ( r) ( r) ) σ ( r) =, (1) ( r) where (r) is the short-exposure intensity at the photo-detector point and angular brackets denote averaging over the enseble of statistically independent realizations of refractive index perturbations of the propagation ediu (atospheric turbulence). t is assued that rando fluctuations of intensity are hoogeneous and isotropic and σ r in Eq. (1) is independent of coordinate vector r, that is, σ () r = σ. hence the noralized variance ( ) n accordance with the classical and ore recent atospheric turbulence theories, the noralized intensity scintillation variance σ can be represented as 7/6 σ = λ F ( Cn, l0, L0, z, h), () where λ is the wavelength of the propagating quasi-onochroatic wave and F( Cn, l0, L0, z, h ) is a factor represented here as a function of the propagation path characteristics (distance z and elevation h) and the ajor turbulence paraeters: structure constant C n, inner scale l 0, and outer scale L 0. The particular for of function F( C, l, L, z, h ) depends on theoretical odels and approxiations used to obtain expression () [17, 18]. n 0 0 n this respect, in principle, validity of one or another theoretical odel can be tested through easureents of the intensity scintillation variance σ ( r ) and independent calculation of the factor F( Cn, l0, L0, z, h ) in Eq. (). evertheless, such an approach requires knowledge of the propagation path and turbulence characteristics the function F( Cn, l0, L0, z, h ) depends on. n practice, these paraeters cannot be easured independently and directly.

3 This proble however can be overcoe using siultaneous easureents of intensity scintillations corresponding to optical waves at different wavelengths that are originating fro a single laser beacon and hence propagating along identical atospheric path. As follows fro Eq. (), the ratio of the noralized variances σ ( λ 1) and σ ( λ ) corresponding to the laser beacon wavelengths λ 1 and λ represents a value independent of the factor 7 6 F( C, l, L, z, h ): σ ( λ ) σ ( λ ) = ( λ λ ). n The idea of atospheric turbulence characterization using ulti-wavelength laser beacons with siultaneous easureents of intensity scintillations at different wavelengths has been ipleented in the COMBAT experients. n these experients intensity scintillations were siultaneously easured using three closely located identical laser beacons that generated truncated Gaussian beas at the following wavelengths: λ 1 = 0.53 μ, λ = 1.06 μ, and λ 3 = 1.55 μ. To siplify notations, the corresponding beacons are referred to as green (G), infrared (R), and laser counications (COM) beacons. n accordance with Eq. (), the ratios of the corresponding T T T intensity scintillation variances expected fro the atospheric turbulence theory, R GR, R GCOM, and R, are given by T σ (0.53μ) T σ (0.53μ) T σ (1.06μ) RGR = =.4; R GCOM = = 3.50; R RCOM = = (3) σ (1.06 μ) σ (1.55μ) σ (1.55μ) E E E n the COMBAT experients, the experiental values of the ratios R GR, R GCOM, and R corresponding to Eq. (3) were directly obtained using siultaneous easureents of statistically representative ensebles of shortexposure intensity patterns (, r λ1 ), (, r λ ), and (, r λ3 ) inside the receiver telescope aperture of diaeter 90 c located a distance 149 k fro the corresponding laser beacons. With the understanding that over long propagation paths the assuptions of stationary and isotropic turbulence with hoogeneous statistics the basis for the Kologorov-Obukhov turbulence theory are not always satisfied, we expected soe deviation in values between T T T E E the ratios R GR, R GCOM, and R in Eq. (3), as predicted by the theory, and the values R GR, R GCOM, and E R obtained experientally. The real surprise was the degree of disconnect between predictions based on hoogeneous turbulence theory and easureents as illustrated by the data obtained and discussed below.. COMBAT EXPERMETAL SETTG Laser bea propagation experients were perfored over the propagation path of length L 149. k between the laser beacon platfor located on the Mauna Loa ountain (Hawaii island) at elevation h ML = 3397 (11,140 ft), and the receiver telescope on the top of the Haleakala ountain (Maui island) at elevation h H = 3058 (see Fig. 1). Due to the Earth's curvature, the iniu height along the propagation path h in 775 corresponds to the distance 60 k fro the receiver site. The laser beacon assebly at the Mauna Loa Observatory coprised three single-ode fiber colliators (each with a clear aperture of 6 ) and the corresponding lasers: a fiber-coupled laser diode at the wavelength of λ 1 = 0.53 µ and two fiber lasers with wavelengths of λ = µ and λ 3 = 1.55 µ. The fiber colliators were Fig. 1. Elevation profile along the propagation path fro the Mauna Loa OAA observatory to the AEOS telescope on Haleakala.

4 Fig.. Conceptual scheatic of the COMBAT experiental syste. ounted together with an aiing telescope into a gibal syste (see Fig. ), which allowed for pointing with a sallest angular step size of 1.75 µrad, corresponding to a lateral bea deflection of 6 c at the receiver telescope over 149 k distance. Each fiber colliator had eans for anual horizontal, vertical, and axial alignent of the fiber tip with respect to the colliating lens to adjust bea pointing and focus. For co-alignent of the fiber colliators' optical axes with the aiing telescope, a corner-cube retro-reflector was ounted in front of the beacon assebly in such way that soe part of the light fro two colliators at a tie was directed into the aiing telescope and focused onto the caera surface located in its focal plane. Using the 1064 n colliator as reference, the focal spots of all three beas were brought to coincide in the telescope focus, which in turn co-aligned the colliated beas. After the alignent procedure, the retro-reflector was reoved and the beacon assebly was pointed toward the receiver telescope [3.67 Air Force Advanced Electro-Optical Syste (AEOS telescope)] using the beas' forer spot location in the caera's field-of-view as aipoint. A searchlight installed in front of the AEOS telescope helped to achieve and aintain correct aiing after sunset, when the telescope was no longer visible with the aiing telescope. Another search light installed near the beacon assebly at the Mauna Loa Observatory allowed for alignent of the AEOS telescope toward the beacons. After propagation to Maui site, the optical waves fro the beacons were received by the AEOS telescope and sent through the coudé path to the deforable irror, which is part of the AEOS adaptive optics syste operating in the visible wavelength band. The adaptive optics syste was off and the deforable irror was flattened during the experients. Colliated light fro the adaptive optics bench (bea diaeter 10 c) was sent to one of the AEOS coudé roos where the bea was subdivided into six separate subapertures, each having a usable diaeter of 5 (Fig. ). Each subaperture corresponds to an area with a diaeter of 91.7 c at the telescope pupil considering the de-agnification of the receiver area fro 3.67 diaeter at the telescope's priary irror to 10 c in the coudé roo. Three of the subapertures were used to record pupil-plane and focal-plane irradiance distributions of the received beas (Fig. ). arrow bandpass filters were used to dedicate each subaperture to a specific wave fro a single beacon. n this paper we discuss only easureents of the pupil-plane intensity distributions, which were recorded using three Sensors Unliited (SU640SDWHvis) 14-bit caeras. The caeras were operating in a windowed ode with pixels' window, providing an actual sensor area of with a pixel size of 5 5 μ and a 100% fill factor. A four-to-one bea reducer was installed in the optical train of each subaperture to atch the bea (pupil) size with the active window size.

5 3. MEASUREMETS AD RESULTS 3.1 Data Recording and Processing The COMBAT easureents were perfored fro early evening to iddle night fro Feb. 1 to Feb. 1, 010. Data collection consisted of a set of experiental trials (typically fro 10 to 30) between 7:00 p.. and 11:30 p.. during each night of easureents. During each trial a set of nearly fraes of short-exposure intensity patterns in the for of 14-bit digital arrays { n (r,l,λ 1 )}, { n (r,l,λ )}, and { n (r,l,λ 3 )} were recorded siultaneously using three identical caeras located in the iage planes of the telescope subapertures that are dedicated to each beacon. Here n is the frae nuber in the experiental trial (n = 1,, 9980), r,l is the transversal coordinate vector in the plane of photo-array (,l = 1,, 56), and λ q is the received light wavelength (q = 1,, 3). The tie of the first frae capture by all caeras within each trial was triggered fro a PC coputer. The capture rate (between 00 and 350 fraes per second dependent on the trial) and the integration tie (between 1 s and s) were identical for all caeras. The averaged deviation between capturing of intensity patterns n (r,l,λ 1 ), n (r,l,λ ), and n (r,l,λ 3 ) by different caeras did not exceed 1.0 s. The physical size of the caera's pixel 5 5 μ corresponds to the telescope receiver aperture size of approxiately Thus the easureents obtained at each pixel of the photo-array correspond to the received light intensity averaged over 3.6- square aperture in the AEOS telescope pupil plane. According to the caera specification, variation in sensitivity fro pixel to pixel did not exceed 1% and therefore was neglected. To avoid aperture boundary effects related with sall aplitude slow displaceent of the received bea centroid during the tie of easureents, the intensity scintillation data were analyzed inside the circle area corresponding to the sub-aperture boundary reduced by 10 pixels of the photoarray. To exclude the data recorded by photo-array pixels that are not properly functioning (broken or blocked by presence of dust on optics or sensor), the following procedure was applied. For each easureent trial we calculated variance of the intensity fluctuations at each pixel (, ) (, ) σ r pix l = n l r n, (4) 1 where n = ( 1 n r, l) is the averaged value of pixel intensity corresponding of the set of short-exposure n= iages belonging to a single easureents trial. The pixels were excluded in the further analysis if the corresponding values of the intensity fluctuation variance σ pix ( r, l ) for these pixels were less than 1% fro the aperture-average value. n addition, we perfored the caera offset correction by capturing a "dark ap" for each of the caeras at the beginning of each data collection trial. The values of the "dark ap" iages were then subtracted fro all subsequent data frae captures. The dark aps were recorded by blocking all external light fro entering the optical syste at the experient roo's coudé port. However, this dark ap capture procedure does not take into account any external background light such as the light fro the sky, which should also be reoved. The ipact of this external background light was reoved by calculating a ap of iniu intensities detected during each trial for each pixel. This "in ap" data was subtracted fro all the fraes of the trial. ote that the difference in the intensity variance (4) brought about by using this additional in ap correction or not only produces a change of a few percent in its value and the true value ust lie in the range between the two possibilities. t is believed that ost of the in ap detected intensity is in fact sky background illuination and therefore the data presented akes use of the in ap correction. 3. Received Power Fluctuations ntensity scintillations observed during ost of experients were surprisingly strong. The short-exposure intensity patterns show the presence of speckles of very different spatial scales: fro few centieters or less to large speckles of approxiately the size of the receiver aperture (~0.9 ). These large-size speckles were seen as spikes and deep fadings in the received optical power signal. n ost experients the intensity scintillation patterns did not show coherent (directional) otion but rather rando appearance and disappearance. As rare exceptions in windy conditions at the receiver site, we observed occasional coherent otion of sall-scale speckles. Consider first the received light power fluctuations data siultaneously recorded for all three beacons. For each short-exposure intensity pattern (iage frae), the optical power averaged over the aperture is defined as M n n ( ) P = r, (5)

6 Fig. 3. Teporal dynaics of received optical power fluctuations P ˆ = Pn ax n Pn for laser beacons with wavelengths λ 1 = 0.53 μ (top, green), λ = 1.06 μ (iddle, red), and λ 3 = 1.55 μ (botto, blue) corresponding to the experiental trials labeled as "3beacon19" (left colun) and "3 beacon_c" (right colun). The date and recording tie are shown on the top of each colun. During each trial, three sequences of short-exposure fraes with s integration tie were captured siultaneously. Several selected bright and dark spikes in the received power for the trial "3beacon19" are shown by BS and DS arks. The aplitudes of ax ax the biggest spikes in the data set "3beacon19" are: Pλ = 14.8P 1 λ (green beacon), P 1 λ = 40.0P λ (R beacon) ax and Pλ = 8.7P 3 λ (COM beacon). Here P 3 λ, P 1 λ, and P λ are the averaged received power values for each 3 wavelength. where M = To siplify notation, in Eq. (5) and below we use a single index to denote the caera pixel coordinate vector. The dependencies of the noralized power Pn ˆ( ) on the consequent frae nuber n or physical tie t are shown in Fig. 3 for two representative easureents trials. n this figure, we can clearly see the presence of large spikes on the received power which indicate the occasional appearance of large-size bright and dark speckles at the receiver aperture for all three wavelengths. Bright speckles can result in up to ten-fold increase in received power. ote that for the "3beacon19" trial with λ = 1.06 μ (red), the aplitude of spikes appeared to be drastically higher. The appearance of large-aplitude bright or dark spikes in received power can be explained by the existence of slowly changing/oving large-scale coherent structures of refractive index that play a role of large aperture lenses with several kiloeter-long focal lengths. As can be seen fro Fig. 3, there sees to be no correlation between the appearance of bright and dark speckles for laser beacons of different wavelengths. The characteristic snapshots of the received light pupil-plane intensity distributions for beacons with different wavelengths are illustrated in Fig. 4. The snapshots are chosen fro the vicinities of bright spikes BS, BS (1) BS 3 indicated in Fig. 3. The iages in Fig. 4 show that the spatial scale of intensity scintillations is strongly dependent on the wavelength, i.e., the larger the wavelength the larger the speckles. This observation is consistent (1) 1 (1), and

7 Fig. 4. Snapshots of pupil-plane intensity distributions of received light, n( r ), fro beacons with three different wavelengths, λ 1 = 0.53 μ, λ = 1.06 μ, and λ 3 = 1.55 μ, for "3beacon19" easureent trial. First three coluns represent experiental data; frae nubers n are denoted above the intensity iages. Last colun contains typical snapshots of intensity distributions within a single subaperture obtained nuerically. with the theoretical findings and was also confired by our nuerical siulations (see the ost right colun in Fig. 4). 3.3 ntensity Scintillations Spatial non-unifority of intensity in each short-exposure intensity pattern (frae) of a trial can be characterized by the set of values (one per a single frae) referred to here as spatial variances M ( ) ( ) S n = n n n S σ r, (6) 1 where n = M ( ) = 1 n r is the aperture-average intensity for the nth frae. The index S in Eq. (6) denotes averaging over spatial coordinate (over pixels). The sequence of σ S ( n ) values characterizes teporal changes of spatial variance occurring during the tie of easureents trial. Correspondingly, by averaging σ S ( n ) over the set of saples, which corresponds to tie averaging, we obtain statistical characteristic σ = σ ( n) that can be S used for evaluation of spatial non-unifority of intensity for the entire trial. Dependencies σ S ( n ) for three different wavelengths and two easureent trials are shown in Fig. 5. The averaged values σ S are given inside the corresponding plots. Another and ore coonly used approach for intensity scintillations analysis is based on coputation of the noralized variance of intensity fluctuations at a single photo-array pixel, where ( ) 1 ( ) ( ) = ( ) ( ) ( ) n S σ r r r r, (7) r = n = 1 n r. The averaging over the set of fraes in Eq. (7) is equivalent to tie averaging over nearly 50 seconds. The noralized variance (7) is referred to as the scintillation index. By averaging σ ( r ) over all pixels inside the receiver aperture (with exception of "bad" pixels described above) we obtain the aperture-average scintillation index σ = σ ( r ). S S

8 Fig. 5. Spatial variances of short-exposure intensity non-unifority inside receiver subaperture, σ S, as functions of frae nuber n for "3beacon19" (left) and "3 beacon_c" (right) easureent trials and three wavelengths, λ 1 = 0.53 μ (top, green), λ = 1.06 μ (iddle, red), and λ 3 = 1.55 μ (botto, blue). n order to evaluate whether the observed intensity scintillations can be considered as a stationary rando process, consider analogues of the scintillation index obtained using either increasing nuber n of subsequent fraes σ ns ( n ) or "oving" (window) averaging of n sequential fraes σ WS ( nn, W ), where n W is a nuber of fraes in the selected averaging window. Both σ ns ( n ) and σ WS ( n ) dependencies are shown in Fig. 6; the values σ S are given inside the corresponding plots. The gray-scale iages shown at the botto of Fig. 6 illustrate spatial non-unifority of the scintillation index σ ( r ) obtained by averaging different nuber of fraes in the trial for different beacons. The indication of spatial hoogeneity of intensity scintillation process is spatial unifority of the iages apping scintillation index σ ( r ). As it can be seen fro σ ( r )-iages in Fig. 6, averaging over fraes results in quite unifor σ ( r) -iages for the "3 beacon_c" easureent trial, while the corresponding σ ( r) -iages for the "3beacon19" trials are still highly non-unifor. This indicates statistical non-stationary character of the intensity scintillation process during this easureent trial. The sae conclusion regarding non-stationary character of the scintillation process can be also derived fro the analysis of dependencies of σ ns ( n ) and σ WS ( nn, W ) in Fig. 6. This highly non-stationary behavior of the scintillations can be associated with the presence of the large-scale intensity scintillations coponents (large-scale bright and dark speckles). Assuing that the ergodic hypothesis can be applied, the obtained values σ S and σ S should be approxiately equal since they are obtained by changing the order of spatial and teporal averaging [15]. However, when coparing the corresponding values σ S and σ S calculated for the two selected data sets (trials) in Figs. 5 and 6, we see that these values are quite different and so application of the ergodic hypothesis is questionable. Bar diagras suarizing the results for σ S and σ S obtained for a nuber of easureent trials during the COMBAT experiental capaign along with the wave-optics siulation results are shown in Fig. 7. Bars' heights

9 Fig. 6. Evolution of the aperture-average scintillation indices σ ns ( n ) (solid lines) and σ WS ( nn, W ) for n W = 000 (dashed lines) for three different wavelengths, λ 1 = 0.53 μ (top, green), λ = 1.06 μ (iddle, red), and λ 3 = 1.55 μ (botto, blue), and for "3beacon19" (left), and "3 beacon_c" (right). Corresponding intensity variance iages, σ ( r ), are shown at the botto. indicate spatial variance σ S (top) and scintillation index σ S (botto) values, while color of the bars denote the wavelength of the laser beacons used in experients, i.e., λ 1 = 0.53 μ (green), λ = 1.06 μ (red), and λ 3 = 1.55 μ (blue) beacons, respectively. Each set of adjacent green, red, and blue bars corresponds to a different easureent trial whose nae is indicated below the set; nubers at the tops of each set denote local tie of the experient. Groups of several sets of bars correspond to different dates of the experients, specified below the diagra. uerical siulation results obtained for the HV-5/7 altitude profile odel of refractive index structure paraeter (see, e.g., [19]) are shown by the first group of bars. Several conclusions can be drawn fro the coparison of the experiental with the nuerical siulation data. First, the experients showed significantly ( 4 ties) larger scintillation index values. Second, contrary to the predictions indicated by the wave optics siulation results, the scintillation index for 1.06 μ laser bea (red bars) was noticeably larger than the one for the shorter wavelength beacon (0.53 μ, green bars) in alost all experients. n addition, using the nuerical siulation and experiental data presented in Fig. 7, we coputed the ratios of E E E intensity scintillation variances (indices) R GR, R GCOM, and R as defined by Eq (3). n Table 1 the obtained E E E fro two easureent trials ratios R GR, R GCOM, and R are copared with the corresponding theoretical values T T R GR, GCOM T R, and R obtained based on the expression () that describes the wavelength dependence

10 Fig. 7. Bar diagra suarizing experiental results for a large set of experiental trials obtained during three days (nights) of easureents during COMBAT experiental capaign for intensity scintillation spatial variances σ S (top) and intensity scintillation indices σ S (botto) for three different wavelengths, λ 1 = 0.53 μ (green), λ = 1.06 μ (red), and λ 3 = 1.55 μ (blue). of the noralized intensity scintillation variance in accordance with the classical and ore recent atospheric turbulence theories. As can be seen fro Table 1, the ratios of the intensity scintillation variances for different wavelengths [cf. Eq. (3)] coputed using both nuerical siulation and experiental data do not atch those expected fro theoretical calculations. Table 1. Ratios of scintillation indices for different wavelengths. R GR R GCOM R Theory [Eq. (3)] "nuerics" "3beacon19" "3 beacon_c"

11 4. SUMMARY Results of the COMBAT experients ake questionable the assuption that stationary and isotropic turbulence with hoogeneous statistics is valid for long-range propagation paths. Statistical characteristics of the observed intensity scintillation patterns (spatial and teporal) are highly diverse and dependent on local weather conditions and global coherent structures of teperature gradient dynaics along the propagation path. n ost experients the intensity scintillation and spatial variances easured exceeded by 4 ties the corresponding values obtained using both analytics and nuerical siulations based on the classical Kologorov-Obukhov turbulence odel. The observed intensity scintillation patterns for 149-k near-horizontal propagation are characterized by the increased appearance of both large (~1 ) and sall (< 10 5 ) speckles. Their appearance can be associated with the presence of strong turbulent layers near the transitter (large-scale speckles) and receiver (sall-scale speckles). The frequent (0.3 s 1.0 s) appearance of large-aplitude spikes in received power indicates the existence of slowly changing/oving large-scale coherent structures with sharp refractive index changes at their boundaries. n all experients the intensity scintillation patterns do not show coherent (directional) otion but rather rando appearance and disappearance. As rare exceptions in windy conditions at the receiver site, we observed occasional coherent otion of sall-scale speckles. These results challenge the assuption of "frozen" turbulence (Taylor hypothesis) coonly used in analyses. n addition, the existence of nuerous well-developed sall-scale speckles akes the possibility of active/adaptive copensation questionable. For the three-wavelength beacons (0.53 μ, 1.06 μ and 1.55 μ), the wavelength dependence of the intensity scintillation variance does not follow existing assuptions that the longer the wavelength the less the turbulence effect on the laser bea. Aong other "anoalies" observed in the experients are the isatches between the predicted and the easured aperture-averaging factor and spatial autocorrelation functions (not discussed in this paper). 5. ACKOWLEDGEMETS This project was supported through the HEL Joint Technology Office contract and the Cooperative Agreeents between the US Ary Research Laboratory and both the University of Dayton and University of Maryland. The authors also want to thank Dr. John Barnes (Mauna Loa Observatory/OAA), Ms. Laura Ulibarri (AFRL Maui Branch Chief), Dr. Skip Willias (AFRL Maui Deputy Branch Chief), Bob Lytle, Kevin Moore, Ed Walker, Sarah Loney, Eddie Kanai (all Boeing, Maui) for support for the experiental capaigns. 6. REFERECES [1] V.. Tatarskii, The Effects of the Turbulence Atosphere on Wave Propagation. Jerusale: srael Progra for Scientific Translations, [] V. E. Zuev, Propagation of visible and infrared radiation in the atosphere. ew York: John Wiley & Sons, [3] A. shiaru, Wave Propagation and Scattering in Rando Media. ew York: Acadeic Press, [4] D. L. Fried and J. L. Vaughn, Branch cuts in the phase function, Appl. Optics, vol. 31, no. 15, pp , 199. [5] A.. Kologorov, The local structure of turbulence in incopressible viscous fluid for very large Reynolds nubers, Dokl. Akad. auk SSSR, vol. 30, no. 4, pp , 1941, [English translation in Turbulence: Classic Papers on Statistical Theory (ed. S. K. Friedlander and L. Topper), pp nterscience, ew York, 1961]. [6] A. M. Obukhov, On the distribution of energy in the spectru of turbulent flow, Dokl. Akad. auk SSSR, vol. 3, no. 1, pp. 4, [7] V.. Tatarskii, Wave Propagation in a Turbulent Mediu, ser. McGraw-Hill Series in Electrical Engineering. ew York: McGraw-Hill, [8] S. M. Rytov, Y. A. Kravtsov, and V.. Tatarskii, Principles of Statistical Radiophysics V: Wave Propagation Through Rando Media. Berlin: Springer-Verlag, 1989.

12 [9] M. E. Gracheva and A. S. Gurvich, Strong fluctuations in the intensity of light propagated through the atosphere close to the Earth, zv. Vyssh. Uchebn. Zaved. Radiofiz., vol. 8, no. 4, pp , 1965, [Radiophys. Quantu Electron., vol. 8, no. 4, (1965)]. [10] W. A. Coles and R. G. Frehlich, Siultaneous easureents of angular scattering and intensity scintillation in the atosphere, J. Opt. Soc. A., vol. 7, no. 8, pp , 198. [11] R. L. Phillips and L. C. Andrews, Measured statistics of laser-light scattering in atospheric turbulence, J. Opt. Soc. A., vol. 71, no. 1, pp , [1] A. Consortini, F. Cochetti, J. H. Churnside, and R. J. Hill, nner-scale effect on irradiance variance easured for weak-to-strong atospheric scintillation, J. Opt. Soc. A. A, vol. 10, no. 11, pp , [13] W. A. Bernard, B. M. Welsh, M. C. Roggeann, and R. J. Feldann, Atospheric turbulence characterization of a low-altitude long horizontal path, in age Propagation through the Atosphere, Proc. SPE vol. 88, , C. Dainty and L. R. Bissonnette, Eds. Bellingha, Wash.: SPE, [14]. Perlot, D. Giggenbach, H. Henniger, J. Horwath, M. Knapek, and K. Zettl, Measureents of the beawave fluctuations over a 14 k atospheric path, in Free-Space Laser Counications V, Proc. SPE vol. 6304, 63041O, A. K. Majudar and C. C. Davis, Eds. Bellingha, Wash.: SPE, 006. [15] W. M. Hughes and R. B. Holes, Pupil-plane iager for scintilloetry over long horizontal paths, Appl. Optics, vol. 46, no. 9, pp , 007. [16] M. S. Belen'kii, E. Cuellar, K. A. Hughes, and V. A. Rye, Preliinary experiental evidence of anisotropy of turbulence at Maui Space Surveillance Site, in Proceedings of the 006 AMOS Conference, S. Ryan, Ed., 006, pp [17] R. S. Lawrence and J. W. Strohbehn, A survey of clear-air propagation effects relevant to optical counications, Proc. EEE, vol. 58, no. 10, pp , [18] R. J. Hill and R. G. Frehlich, Onset of strong scintillation with application to reote sensing of turbulence inner scale, Appl. Optics, vol. 35, no. 6, pp , [19] J. C. Ricklin, S. M. Hael, F. D. Eaton, and S. L. Lachinova, Atospheric channel effects on free-space laser counication, J. Opt. Fiber. Coun. Rep., vol. 3, no., pp , 006.