Atmospheric Influence on a Laser Beam Observed on the OICETS – ARTEMIS Communication Demonstration Link

Home , OICETS

Atmos. Meas. Tech., 3, 1233–1239, 2010 www.atmos-meas-tech.net/3/1233/2010/ Atmospheric doi:10.5194/amt-3-1233-2010 Measurement © Author(s) 2010. CC Attribution 3.0 License. Techniques

Atmospheric inﬂuence on a laser beam observed on the OICETS – ARTEMIS communication demonstration link

A. Loscher¨ ESA/ESTEC, Keplerlaan 1, 2200 AG Noordwijk, The Netherlands Received: 18 March 2010 – Published in Atmos. Meas. Tech. Discuss.: 10 May 2010 Revised: 26 July 2010 – Accepted: 20 August 2010 – Published: 14 September 2010

Abstract. In 2006 bi-directional optical inter-satellite com- the limits concerning the capacity of X-band downlink sys- munication experiments were conducted between the Japan tems. One possible solution would be the use of optical Aerospace Exploration Agency (JAXA) Optical Inter-orbit terminals, increasing the bandwidth significantly, in a di- Communications Engineering Test Satellite (OICETS) and rect Earth observation satellite-to-ground or Earth observa- the European Space Agency (ESA) multi-purpose telecom- tion satellite-to-relay-satellites-to-ground configuration. The munications and technology demonstration satellite (Ad- links themselves are potentially a source of information con- vanced Relay and Technology MISsion) ARTEMIS. On 5 cerning Earth’s atmosphere depending on the transmission April 2006, an experiment was successfully carried out by geometry. maintaining the inter-satellite link during OICETS’s setting The Optical Inter-orbit Communications Engineering Test behind the Earth limb until the signal was lost. This setup Satellite (OICETS, Japanese name KIRARI) (Jono et al., resembles an occultation observation where the influence 2006) was developed by the Japan Aerospace Exploration of Earth’s atmosphere is evident in the power fluctuations Agency (JAXA) and launched into a circular Low Earth Or- recorded at ARTEMIS’s (and OICETS’s) receiver. These bit (LEO) 23 August 2005 to an altitude of 610 km at an incli- fluctuations do not exist or are at a low level at a link path nation of 97.8◦. The main payload is a laser-based commu- above the atmosphere and steadily increase as OICETS sets nication terminal called the Laser Utilizing Communication behind the horizon until the tracking of the signal is lost. This Equipment (LUCE). It was designed to conduct inter satellite specific experiment was performed only once since atmo- laser communication experiments with the European Space spheric science was not the goal of this demonstration. Nev- Agency’s (ESA) Advanced Relay and Technology MISsion ertheless, this kind of data, if available more frequently in (ARTEMIS) as its counterpart and orbit-to-ground commu- future, can help to study atmospheric turbulence and validate nications (e.g. with the OGS, ESA’s Optical Ground Station). models. The data present here were recorded at ARTEMIS. OICETS is still in orbit and functional but the instrumenta- tion have not been used since the conclusion of the experiments. 1 Introduction ARTEMIS is a multi-purpose telecommunication and technology demonstration satellite. The ARTEMIS mission Free space laser links have been studied in the context began on 12 July 2001 and was subject to an unprecedented of terrestrial wireless communication to establish perfor- rescue that spanned 18 months after a malfunction of the up- mance envelopes and design guidelines (Henninger and Wil- per stage of the Ariane 5 launcher. The spacecraft control fert, 2010). During the last years the capability to employ team managed to raise the satellite, which was stranded in an abnormally low transfer orbit to its intended geostationary this technique for establishing communication links between ◦ satellites and satellite-to-ground was developed in Europe. orbit (GEO) at ∼36 000 km 21.5 East on 31 January 2003 One reason for those developments is the growing amount of using innovative procedures and onboard advanced technol- data generated by Earth Observation Satellites which reached ogy. ARTEMIS is still fully operational serving its purpose in the domains of communication, data relay and navigation. The first ever transmission of an image by laser link from Correspondence to: A. Loscher¨ satellite-to-satellite took place on 30 November 2001 with ([email protected]) the SILEX (Semiconductor Inter-satellite Link Experiment)

Published by Copernicus Publications on behalf of the European Geosciences Union. 1234 A. Loscher:¨ Atmospheric inﬂuence on optical communication links system, which consists of the OPALE (Optical Payload for Inter-Satellite Link Experiment) terminal on ARTEMIS and the PASTEL (PASsenger TELecom) terminal on the French Earth observation satellite SPOT-4 (Systeme` Probatoire pour l’Observation de la Terre), launched on 22 March 1998, in a 832 km sun-synchronous LEO orbit at an inclination of 98.7◦ (Tolker-Nielsen and Oppenhaeuser, 2002). The link experiments between ARTEMIS and OICETS lasted in total ∼8 month. The ﬁrst bi-directional inter-orbit laser communication was established on 9 December 2005, experiments where carried out until the middle of August

2006 resulting in 100 successfully established links (Jono et Figure 1 Observation geometry of the communication experiment between ARTEMIS and OICETS al., 2007). on Fig.5 April 1.2006. Observation geometry of the communication experiment

Usually the link is established well above Earth’s atmo- between ARTEMIS and OICETS on 5 April 2006. The experiment started at 07:21:06 UTC and ended at 07:22:49 UTC, the resulting sphere to ensure an optimal communication environment. geometric link path altitude is shown in Fig. 2. The receiver onboard ARTEMIS lost the signal before160 its counterpart onboard OICETS and fluctuations of the received power For the special experiment conducted on 5 April 2006, the (normalized by the receiver antenna area and given here in nW/m2) are visible a bit earlier link was maintained until the experiment came forcibly to an (~ 5sec,140 which roughly corresponding to 8km) than in the data recorded at OICETS end caused by atmospheric influences. The received power (Takayama et al., 2007). The ARTEMIS120 data record starts at 07:21:36 and ends at 07:22:41. The explanation for started to fluctuate strongly with a decrease of the average re- the earlier disconnection is straightforward: although both beams propagate through the same media100 the signal originating from OPALE on ARTEMIS gets disturbed just before ceived power finally resulting in a disconnection of the link reaching the receiver on OICETS. The beam travels a long distance undisturbed through (Takayama et al., 2007). free space80 before being influenced by Earth’s atmosphere. After emerging from the disturbing volume the distance to the receiver is rather short. The opposite is true for the Altitude [km] The time of the experiment was chosen with respect to beam emitted60 by LUCE on OICETS. The distance to the disturbing volume is rather short the relative positions of ARTEMIS and OICETS so the link but after leaving Earth’s atmosphere the disturbed and deflected beam has to travel a long distance 40which in fact amplifies the power fluctuations recorded at ARTEMIS. The power could be established at an altitude where the residual at- transmitted by LUCE is 100 mW which is the modulated mean value via a telescope of mosphere could be assumed as negligible. Due to the rel- 0.26 m diameter,20 the receiving telescope of OPALE has a diameter of 0.25 m. Figure 3 shows the power detected at ARTEMIS for a 40 second period starting at 0 ative motion of the satellites (basically the propagation of 07:22:00 UTC (the respective altitude range can be seen in Fig. 2) where the red line is the OICETS along its orbital trajectory) a classical occultation mean of 0.1 sec intervals. The link is established above 100 km link altitude and thus no atmospheric influence is to be expected. The procedureTime [UTC] is as follows: Both terminals point event occurs. The laser beams start to propagate through the to each other utilizing an open-loop control based on on-board orbital models. OPALE emits a strong wide-divergence (750 μrad) laser (801 nm) beam scanning a cone of atmosphere at high altitudes reaching gradually denser re- uncertainty.Fig. 2. UponAltitude reception - time LUCE diagram emits its communication of the communication beam (847 nm, experiment 5.5 μrad) gions as OICETS gets closer to the Earth’s rim as seen from between ARTEMIS and OICETS on 5 April 2006. Start and end4 of ARTEMIS. The signal can be tracked as long as the receiver data records are marked in blue for ARTEMIS and red for OICETS. at OICETS can see the transmitter at ARTEMIS (and vice versa, the link is bi-directional), thus, until OICETS sets be- low the horizon and the beams hit the Earth’s surface. In illustrated in Fig. 1, showing ARTEMIS in its geostationary practice, one has to consider ray bending in the atmosphere orbit at ∼36 000 km whereas OICETS has a close to polar in addition to purely geometric considerations to derive the orbit at ∼610 km. At time t1 the bi-directional link acqui- exact ray path. In fact, tracking is lost far above the surface sition is established and the beams propagate undisturbed due to fluctuations in the optical power received caused by trough free space. As OICETS starts to set behind Earth’s atmospheric inhomogenities. atmosphere at t2 as seen from ARTEMIS, fluctuations of the The results presented here are based on the data recorded received power at both detectors start to increase with a de- at ARTEMIS. This experiment focuses on communication crease of the average power received until the link is lost. applications; nevertheless it can provide valuable insights for The data is recorded jointly at OPALE’s detector at 8 kHz the atmospheric science community. and LUCE (LUCE’s sampling rate is 1 kHz). The experiment started at 07:21:06 UTC and ended at 07:22:49 UTC, the resulting geometric link path altitude is 2 Link experiment shown in Fig. 2. The receiver onboard ARTEMIS lost the signal before its counterpart onboard OICETS and fluctua- The main objective of the experiment carried out 5 April tions of the received power (normalized by the receiver an- 2006 was to observe and study the atmospheric influences tenna area and given here in nW/m2) are visible a bit earlier on the optical communication link between ARTEMIS and (∼5 s, which roughly corresponds to 8 km) than in the data OICETS. The timing of the experiment was chosen in a way recorded at OICETS (Takayama et al., 2007). as to establish the bi-directional link between the two satellites in free space and subsequently observe an occultation event when the laser beams penetrated increasingly dense re- gions of Earth’s atmosphere. The observation geometry is

Atmos. Meas. Tech., 3, 1233–1239, 2010 www.atmos-meas-tech.net/3/1233/2010/ A. Loscher:¨ Atmospheric inﬂuence on optical communication links 1235

Figure 3 Optical power detected at OPALE, 07:22:00 UTC was set to 0 in this plot, where the red line is the mean calculated from 0.1 second intervals. Figure 4 Scintillation index computed from the optical power recorded at OPALE for 0.1 second Fig.Figure 3. 4 Opticalshows the corresponding power detected scintillation at OPALE, index Eq. (1) 07:22:00 computed UTCat 0.1 s was intervals, set intervals, 07:22:00 UTC was set to 0 in this plot. Fig. 4. Scintillation index computed from the optical power towhich 0 in steeply this increases plot, where around 30 the seconds. red lineThe same is theis true mean for the calculatedangular fluctuation from of the incoming beam expressed in the angular correction performed by the fine pointing recorded at OPALE for 0.1 s intervals, 07:22:00 UTC was set to 0 in 0.1mechanism s intervals. (FPM) in the x- and y-directions (shown in Fig. 5 and 6). The x- and y- directions are referenced with respect to the FPM’s coordinate system which is defined by this plot. the incoming beam representing the z- axis with x- and y- axes are perpendicular to it. Thus the geometry with respect to the Earth could only be derived knowing the exact attitudeThe (azimuth ARTEMIS and elevation) data of record the CPM. starts Since this at 07:21:36data is not readily and available ends conclusions concerning specific horizontal and vertical effects can not be drawn. It is atevident 07:22:41. in the plotsThe that the explanation fine pointing mechanism for the also earlier gets saturated disconnection after 30 seconds the power received by OPALE, starting from 07:22:00 UTC, and can not compensate for the deflections of the incoming beam any more. The general ispattern straightforward: is similar to that of although power received both by the beams detector. propagate The constant low through level of which is set to 0 in the x-axis of the plot. The red line is the thecorrections same at earlier media times the is most signal likely to originatingbe attributed to combined from micro OPALE vibrations on of mean power calculated at 0.1 s intervals. Starting from ∼7 s ARTEMIS and OICETS; the atmospheric effects lead immediately to strong signal ARTEMISvariations and saturation gets disturbed of the mechanism. just Angle before of arrival reaching errors start theto increase receiver later at which corresponds to a link altitude of ∼66 km first fluctua- onthe OICETS.ARTEMIS side The than at beam the OICETS travels side but a as long mentioned distance the effects undisturbed are immediately tions of the recorded power appear. The fluctuations increase strong (Perlot, 2009). A closer look at the data reveals a small gradual increase of the throughaverage corrections free space applied before (red line in being the plots) influenced but interrupted by by Earth’snumerous saturation atmo- until ∼28 s, which corresponds to a link altitude of ∼36 km, events. This behaviour seems to be caused mainly by wavefront distortions. Thus the FPM sphere.data of ARTEMIS After emerging apparently has from little theuseful disturbing information content volume with therespect dis- to the detector also gets saturated for the first time (received 2 tanceatmospheric to the effects. receiver The situation is ratheris potentially short. a little The bit better opposite on the OICETS is true side. for power >900 nW/m ). Starting from that point in time the Figure 5 Angular fluctuation of the fine pointing mechanism along the x-direction recorded at the beam emitted by LUCE on OICETS. The distance to the powerOPALE, fluctuations07:22:00 UTC was set stay to 0 in at this a plot. high The red level line is saturatingthe average calculated the at detec- 0.1 second intervals of the corrections. disturbing volume is rather short but after leaving Earth’s at- tor frequently. From a brief look at Fig. 3, one might get mosphere the disturbed and deflected beam has to travel a the impression that until loss of lock enough power is re- long distance which, in fact, amplifies the power fluctuations ceived, but this is just a misleading impression caused by the recorded at ARTEMIS. The power transmitted by LUCE is power peaks recorded. The red line in Fig. 3 indicates the 100 mW which is the modulated mean value via a telescope6 0.1 s mean power received. It is evident that around 28 s the of 0.26 m in diameter, the receiving telescope of OPALE has average starts to decline until it reaches very low levels at 7 a diameter of 0.25 m. 40 s corresponding to a link altitude of ∼20 km. The power Figure 3 shows the power detected at ARTEMIS for a peaks stem from the focusing effect of density inhomogeni- 40 s period starting at 07:22:00 UTC (the respective alti- ties inside the atmosphere acting like lenses on the beam. tude range can be seen in Fig. 2) where the red line is Figure 4 shows the corresponding scintillation index the mean of 0.1 s intervals. The link is established above Eq. (1) computed at 0.1 s intervals, which steeply increases 100 km link altitude and, thus, no atmospheric influence is around 30 s. The same is true for the angular fluctuation to be expected. The procedure is as follows: both termi- of the incoming beam expressed in the angular correction nals point to each other utilizing an open-loop control based performed by the fine pointing mechanism (FPM) in the on on-board orbital models. OPALE emits a strong wide- x- and y-directions (shown in Figs. 5 and 6). The x- divergence (750 µrad) laser (801 nm) beam scanning a cone and y-directions are referenced with respect to the FPM’s of uncertainty. Upon reception LUCE emits its communi- coordinate system which is defined by the incoming beam cation beam (847 nm, 5.5 µrad) towards ARTEMIS. When representing the z-axis with x- and y-axes are perpendicular ARTEMIS detects the incoming communication beam from to it. Thus, the geometry with respect to the Earth could only OICETS its own communication beam (819 nm) is switched be derived knowing the exact attitude (azimuth and elevation) on and the beacon is switched off. The pointing accuracy is of the CPM. Since this data is not readily available, conclu- continuously improved by the coarse and fine pointing (CPM sions concerning specific horizontal and vertical effects can and FPM) closed-loop systems and the bi-directional com- not be drawn. It is evident in the plots that the fine pointing munication link is established properly. mechanism also gets saturated after 30 s and can not com- The data is recorded with 8 kHz by OPALE’s receiver on pensate for the deflections of the incoming beam any more. ARTEMIS (received power, x and y direction angular cor- The general pattern is similar to that of power received by rections of the fine pointing mechanism). Figure 3 shows the detector. The constant low level of corrections at earlier www.atmos-meas-tech.net/3/1233/2010/ Atmos. Meas. Tech., 3, 1233–1239, 2010 1236 A. Loscher:¨ Atmospheric influence on optical communication links

Figure 4 Scintillation index computed from the optical power recorded at OPALE for 0.1 second intervals, 07:22:00 UTC was set to 0 in this plot.

Figure 5 Angular fluctuation of the fine pointing mechanism along the x-direction recorded at Figure 6 Angular fluctuation of the fine pointing mechanism along the y-direction recorded at OPALE, 07:22:00 UTC was set to 0 in this plot. The red line is the average calculated at 0.1 second OPALE, 07:22:00 UTC was set to 0 in this plot. The red line is the average calculated at 0.1 second intervalsFig. 5. of theAngular corrections. fluctuation of the fine pointing mechanism along intervalsFig. 6. of theAngular corrections. fluctuation of the fine pointing mechanism along the x-direction recorded at OPALE, 07:22:00 UTC was set to 0 in Thethe scintillationy-direction index recorded is computed at as OPALE, the variance 07:22:00 of the flux normalized UTC was by setthe square to 0 inof this plot. The red line is the average calculated at 0.1 s intervals of thethis average plot. flux: The red line is the average calculated at 0.1 s intervals of the corrections. 2 the corrections. ()X − X SI = (1) ()X 2 The average flux is computed from a sliding mean over 0.1 s intervals comprising 800 7 detector readouts. times is most likely to be attributed to combined micro vi- 3 Link experiment data As mentioned earlier the Fig. 3 and 4 look similar for the power received at OICETS, with brations of ARTEMIS and OICETS; the atmospheric effects the important difference that the pronounced atmospheric effects occur a bit later and lead immediately to strong signal variations and saturation of subsequentlyAnalysing the the link poweris disconnected distribution at a lower ofaltitude. a laser This can beam be related reaching to the observationa detector, geometry which resulting propagated in a shorter through distance between the atmosphere, the LEO satellite resultsOICETS the mechanism. Angle of arrival errors start to increase later and the atmosphere compared to the distance between the GEO satellite ARTEMIS and at the ARTEMIS side than at the OICETS side but, as men- theusually atmosphere. in a log-normalIn our case (data distribution recorded at ARTEMIS) (Churnside, the effects 1987); in terms above of fluctuationsthe atmosphere in optical power a normal reaching distributionthe detector are amplified can be by expected. the distance (the In tioned, the effects are immediately strong (Perlot, 2009). A lateral displacement caused by angular deflections increases with distance). closer look at the data reveals a small gradual increase of the fact, the fluctuations of the optical power observed at the de- average corrections applied (red line in the plots) but inter- Linktector Experiment can be Data attributed to a combination of several effects rupted by numerous saturation events. This behaviour seems Analyzingin our casethe power (Toyoshima distribution of anda laser Araki, beam reaching 2000). a detector, The atmosphericwhich propagated throughinfluence, the atmosphere, beam pointingresults usually errors in a log-normal of the transmitterdistribution (Churnside, and track- 1987); to be caused mainly by wavefront distortions. Thus, the FPM above the atmosphere a normal distribution can be expected. In fact the fluctuations of the opticaling errors power observed of the at receiver the detector (Kiasaleh, can be attributed 1994), to a combination where of the several latter effects is data of ARTEMIS apparently has little useful information in our case (Toyoshima and Araki, 2000). The atmospheric influence, beam pointing content with respect to atmospheric effects. The situation errorsmutually of the transmitter correlated and tracking since errors the of laser the receiver communication (Kiasaleh, 1994), linkwhere islatter a are mutually correlated since the laser communication link is a closed-loop system is potentially a little bit better on the OICETS side. closed-loop system between the two terminals on ARTEMIS and OICETS. As can be seen in Fig. 3 (first 5 s, beam still 8 The scintillation index is computed as the variance of the above atmospheric layers of significant density), the closed- flux normalized by the square of the average flux: loop system reduces the pointing and tracking errors to a minimum, thus, fluctuations are dominated by atmospheric effects. 2 X −X The first 30 s of the time interval displayed in Fig. 3 are SI = (1) (X)2 analysed in more depth, since starting from second 31 the detector is frequently saturated (maximum readout 900 nW/m2) and the respective data is unreliable. The average flux is computed from a sliding mean over The probability density function of the received optical 0.1 s intervals comprising 800 detector readouts. power can be estimated from the data observed at ARTEMIS. As mentioned earlier, Figs. 3 and 4 look similar for the We process the optical power detected to obtain histograms power received at OICETS, with the important difference with respect to the normalized intensity for 1 s periods, in that the pronounced atmospheric effects occur a bit later and total for 30 s (3·10 s). We use a systematic approach to deter- subsequently the link is disconnected at a lower altitude. mine the optimal histogram bin width as proposed in (Scott, This can be related to the observation geometry resulting in 1979, 1992). The bin width w of those histograms is calcu- a shorter distance between the LEO satellite OICETS and lated as follows where n denotes the sample size: the atmosphere compared to the distance between the GEO 1 1 − 1 satellite ARTEMIS and the atmosphere. In our case (data w = 2·3 3 ·π 6 ·σ ·n 3 (2) recorded at ARTEMIS) the effects in terms of fluctuations in optical power reaching the detector are amplified by the dis- which is equal to: tance (the lateral displacement caused by angular deflections − 1 increases with distance). w = 3.49·σ ·n 3 (3)

Atmos. Meas. Tech., 3, 1233–1239, 2010 www.atmos-meas-tech.net/3/1233/2010/

A. Loscher:¨ Atmospheric inﬂuence on optical communication linksFigure 7 The normalized intensity distribution for a period of 10 seconds starting at 07:22:00 UTC 1237 (10 times 1 second).

Figure 8 The normalized intensity distribution for a period of 10 seconds starting at 07:22:10 UTC (10 Figure 7 The normalized intensity distribution for a period of 10 seconds starting at 07:22:00 UTC (10 times 1 second). times 1 second). Fig. 8. The normalized intensity distribution for a period of 10 s Fig. 7. The normalized intensity distribution for a period of 10 s starting at 07:22:10 UTC (10 times 1 s). starting at 07:22:00 UTC (10 times 1 s).

This formulation had been derived assuming a Gaussian distribution, with the standard deviation σ. As already mentioned we can expect a log-normal distribution of the re- 10 ceived intensity as soon as the beam starts to penetrate the atmosphere. This is the part which is of most interest with respect to turbulence. We can account for that fact by intro- ducing the sample Skewness SF and the sample Kurtosis KF factors in calculating the bin width w of the histograms. The

FigureSkewness 8 The normalized is intensity a measure distribution for for thea period asymmetry of 10 seconds starting of theat 07:22:10 probability UTC (10 timesdistribution 1 second). of a real valued random variable and the third

moment of the distribution of a random variable. The Kur- Figure 9 The normalized intensity distribution for a period of 10 seconds starting at 07:22:20 UTC (10 tosis of the probability distribution of a real valued random times 1 second). It Fig.is evident 9. thatThe without normalized atmospheric intensity influence and distribution thus with just for pointing a period uncertainties of 10 s variable, on the other hand, is a measure of the frequency causedstarting by the at relative 07:22:20 motion UTC of the (10satellites times and 1micro-vibrations, s). the received power is quite stable. In Fig. 7 pronounced peaks are visible exhibiting a narrow energy with which extreme deviations appear, it is also denoted as distribution. This tells us that the loop system between ARTEMIS and OICETS works the forth moment. well. Although the distribution still stays narrow as we step forward in time during the first 10 seconds the detected optical power decreases monotonically, which indicates a monotonicallyIt is evident increase in that atmospheric without density. atmospheric influence and, thus, 10 with just pointing uncertainties caused by the relative motion 1 We see a similar picture during the next 10 second interval (Fig. 8) but the distributions = are getting broader and the decrease is not longer monotonic. This means that the beam SF 2 3 (4) of the satellites and micro-vibrations, the received power is 1−0.0060·Skewness+0.27·Skewness +0.0069·Skewness penetrates more turbulent atmospheric regimes resulting in density fluctuations. The densityquite is stable.not monotonically In Fig. increasing 7 pronounced at small scales peaks as arethe beam visible sets exhibit-through increasinglying a narrow lower atmospheric energy layers. distribution. This becomes Thisevident tellslooking us at second that the11 which loop KF = 1−0.2·(1−e−0.7·Kurtosis) (5) actually has the broadest distribution and the lowest power level in Fig. 8. Finally in Fig. 9 similarsystem patterns between are visible ARTEMISwith again broa andder distributions OICETS and works lower power well. levels. Al- Around second 26 atmospheric effects strongly kick in leading to a broad distribution and To account for the real sample distribution Eq. (3) is mul- thusthough quite low the power distribution levels. Although stillquite intense stays power narrow peaks asare werecorded step, the forward average powerin timereceived during is reduced the dramatically. first 10 Th se the shape detected of the probabi opticallity density power function de- tiplied by the Skewness factor SF Eq. (4) if the sample Skew- shifts under the influence of the atmosphere from a close to Gaussian to a more log- ness is >0 and <3 and by the Kurtosis factor KF Eq. (5) if normalcreases distribution, monotonically, as to be expected which (the indicatesclosed-loop tracking a monotonically system minimises in- effectivelycrease pointing in atmospheric & tracking errors). density. The peaks of the distributions move to the left and the kurtosis −3 is >0 and <6. thus to lower received power. Following this procedure 30 histograms have been con- We see a similar picture during the next 10 s interval structed each covering 1 s of observation time, 30 s in total, (Fig.Conclusions 8) but the distributions are getting broader and the de- Thecrease power received is no at longer the OPALE monotonic. detector onboardThis ARTEMIS means duringthat a communication the beam to estimate the probability density functions from the ob- link experiment with the Japanese satellite OICETS has been analyzed. The difference served data. For this analysis the optical power detected at betweenpenetrates this specific more experiment turbulent and other atmospheric optical satellite regimesto satellite communication resulting in links is the fact that the link was maintained until OICETS started to set behind the ARTEMIS was taken starting at 07:22:00 UTC and ending density fluctuations. The density is not monotonically in- 30 s later. To estimate the probability density functions from creasing at small scales as the beam sets through increas-11 the histograms, a data fitting approach was used where for ingly lower atmospheric layers. This becomes evident look- the fitting procedure Gaussian functions have been assumed. ing at second 11 which actually has the broadest distribution and the lowest power level in Fig. 8. Finally in Fig. 9, The intensity I is normalized by the average intensity I0 of the first second analysed. In Figs. 7 to 9 the vertical scaling similar patterns are visible with again broader distributions changes (between max. 0.4 in Fig. 7 to max. 0.04 in Fig. 9), and lower power levels. Around second 26 atmospheric ef- the areas of the underlying histograms were normalized to 1. fects strongly kick in leading to a broad distribution and, thus, quite low power levels. Although quite intense power peaks are recorded, the average power received is reduced

www.atmos-meas-tech.net/3/1233/2010/ Atmos. Meas. Tech., 3, 1233–1239, 2010 1238 A. Loscher:¨ Atmospheric influence on optical communication links dramatically. The shape of the probability density function Scintillations impact many different areas like communi- shifts under the influence of the atmosphere from a close to cation or remote sensing, not only in the optical domain but Gaussian to a more log-normal distribution, as to be expected also in the micro wave (MW) (Gorbunov and Kirchengast, (the closed-loop tracking system minimizes effectively point- 2007). Thus, a better understanding of theses phenomena ing and tracking errors). The peaks of the distributions move will have a positive impact on science and application. to the left and, thus, to lower received power. Acknowledgements. The inter-satellite communication experiments were supported by many people through an international 4 Conclusions cooperation. The author wants to express his thanks to all members of the ARTEMIS and the OICETS teams from ESA and JAXA and The power received at the OPALE detector onboard Z. Sodnik (ESA) for his support. ARTEMIS during a communication link experiment with the Japanese satellite OICETS has been analysed. The dif- Edited by: S. Buehler ference between this specific experiment and other optical satellite-to-satellite communication links is the fact that the link was maintained until OICETS started to set behind the atmosphere in a de facto occultation geometry. Atmospheric References influences caused a disconnection of the link before the tan- gent point could reach the Earth’s surface. Churnside, J. H. and Clifford, S. F.: Log-normal Rician probability- Looking at the data recorded at OICETS in the bi- density function of optical scintillations in the turbulent atmo- directional link one recognizes that atmospheric effects ap- sphere, J. Opt. Soc. Am. A., 4, 10 October, 1923–1930, 1987. pear later and the link is maintained longer than in the data Gorbunov, M. E. and Kirchengast, G.: Fluctuations of radio occultation signals in the X/K band in the presence of anisotropic tur- recorded at ARTEMIS. This can be explained by the distance bulence and differential retrieval performance, Radio Sci., 42(1– between the angular disturbance of the beam and the receiver, 10), RS4025, doi:10.1029/2006RS003544, 2007. which is amplifying the beam displacement. Gurvich, A. S., Kan, V., and Fedorova, V. O.: Refraction angle This effectively works like a “magnifying glass” for the fluctuations in the atmosphere from space observations of stellar fluctuations of the upper atmosphere, and causes earlier link scintillations, Atmos. Ocean. Phys., 31(6), 742–749, 1996. loss. Gurvich, A. S. and Chunchuzov, I. P.: Parameters of the fine density Although peaks appear in the received power as the beam structure in the stratosphere obtained from spacecraft observa- starts to penetrate denser layers of the atmosphere the aver- tions of stellar scintillations, J. Geophys. Res., 108, 4166–4170, age recorded power starts to decline until the tracking is lost. 2003. The slight jitter visible in the x (Fig. 5) and y (Fig. 6) angu- Henninger, H. and Wilfert, O.: An introduction to Free-space Opti- lar corrections at a constant level before atmospheric effects cal Communications, Radioengineering, 19(2), 203–212, 2010. Jono, T., Takayama, Y., Kura, N., Ohinata, K., Koyama, Y., Shi- really kick in can be most likely attributed to micro vibrations ratama, K., Sodnik, Z., Demelenne, B., Bird, A., and Arai, originating onboard the satellites. K.: OICETS on-orbit laser communication experiments, Proc. This kind of data, if available more frequently, could help of SPIE, 6105(610503), 1–11, 2006. to study atmospheric inhomogenities and the related scintil- Jono, T., Takayama, Y., Shiratama, K., Mase, I., Demelenne, B., lation phenomena with the potential to aid in validation of Sodnik, Z., Bird, A., Toyoshima, M., Kunimori, H., Giggen- models. bach, D., Perlot, N., Knapek, M., and Arai K.: Overview of Ongoing work concerning electromagnetic wave propaga- the inter-orbit and the orbit-to-ground lasercom demonstration tion in the Earth’s atmosphere uses different parameterisa- by OICETS, Proc. of SPIE, 6457(645702), 1–10, 2007. tions to take scintillations into account; those formulations Kiasaleh, K.: On the probability density function of signal intensity could be improved. Work has been performed, for exam- in free-space optical communications systems impaired by point- ple, using the fast photometer data from GOMOS (Global ing jitter and turbulence, Opt. Eng., 33(11), 3748–3757, 1994. Perlot, N.: Atmospheric occultation of optical intersatellite links: Ozone Monitoring by Occultation of Stars) (Sofieva et al., coherence loss and related parameters, Appl. Opt. 48, 2290– 2009a, b) for assessment of influence of scintillations on re- 2302, 2009. trieval accuracy from stellar occultation experiments or from Scott, W. D.: On Optimal and Data-Based Histograms, Biometrika, the analysis of measurements performed onboard the former 66(3), 605–610, 1979. Russian space-station MIR (Gurvich et al., 1996). The re- Scott, W. D.: Multivariate Density Estimation, Theory, Practice, sults are used to derive performance envelopes of different and Visualisation, John Wiley & Sons, United States, ISBN:0- kinds of applications affected by scintillation phenomena be- 471-54770-0, 1992. sides using the data to address atmospheric science questions Sofieva, V. F., Kan, V., Dalaudier, F., Kyrol¨ a,¨ E., Tamminen, J., (Gurvich and Chunchuzov, 2003). Bertaux, J.-L., Hauchecorne, A., Fussen, D., and Vanhellemont, F.: Influence of scintillation on quality of ozone monitoring by GOMOS, Atmos. Chem. Phys., 9, 9197–9207, doi:10.5194/acp- 9-9197-2009, 2009.

Atmos. Meas. Tech., 3, 1233–1239, 2010 www.atmos-meas-tech.net/3/1233/2010/ A. Loscher:¨ Atmospheric inﬂuence on optical communication links 1239

Soﬁeva, V. F., Dalaudier, F., Kan, V., and Gurvich, A. S.: Tolker-Nielsen, T. and Oppenhaeuser, G.: In Orbit test result of an Technical note: Scintillations of the double star a Cru ob- Operational Optical Intersatellite Link between ARTEMIS and served by GOMOS/Envisat, Atmos. Chem. Phys., 9, 8967–8973, SPOT4, SILEX, Proc. of SPIE 4635, 1–15, 2002. doi:10.5194/acp-9-8967-2009, 2009. Toyoshima, M. and Araki, K.: Effects of time averaging on opti- Takayama, Y., Jono, T., Koyama, Y., Kura, N., Shiratama, K., cal scintillation in a ground-to- satellite atmospheric propagation, Demelenne, B., Sodnik, Z., Bird, A., and Arai, K.: Observa- Appl. Opt., 39(12), 1911–1919, 2000. tion of atmospheric inﬂuence on OICETS inter-orbit laser communication demonstrations, Proc. of SPIE, 6709, 67091B, 1–9, doi:10.1117/12.736789, 2007.

www.atmos-meas-tech.net/3/1233/2010/ Atmos. Meas. Tech., 3, 1233–1239, 2010