Practical considerations for the design of CO_2 lidar systems
Post on 02-Oct-2016
Embed Size (px)
Practical considerations for the design of CO2 lidar systems
Jay A. Fox, Cynthia R. Gautier, and Jeffrey L. Ah
A 10.6-,um single laser lidar system has been utilized to monitor the amplitude, standard deviation, andcorrelation of returns from foliage, hillside, and man-made targets as a function of the lidar system divergenceand mode shape, the receiver field of view and receiver/transmitter alignment tolerance, the repetition rate,and the sampling time. Studies of the dependence of the system sensitivity on signal averaging and signalcorrelation demonstrate performance comparable with that achieved with reported dual laser lidar systems.
1. IntroductionIt is well known that differential absorption lidar
(DIAL) is extremely useful for the remote sensing andquantitative measurement of gaseous constituents ofthe atmosphere. Most of the recently reported directdetection studies have dealt, either directly or indi-rectly, with improvements of the accuracy of DIALsystems.1 -7 The experimental work has generally in-volved the analysis of large data sets obtained overcomparatively long intervals of time (10-20 min) andhas utilized multiple lasers operating at relatively lowpulse repetition rates (10 Hz). These experimentshave yielded statistically useful results, but little infor-mation was obtained that would be useful for the de-sign of more compact lidar systems with optimizedsensitivity.
This paper attempts to address this need by present-ing data bearing on the basic design parameters of adirect detection DIAL lidar system. Consideration isgiven to the optimum useful magnitude of both thetransmitter divergence and receiver field of view. Inaddition, the relative importance of transmitter/re-ceiver alignment is explored as well as the possibility ofutilizing multitransverse-mode beams to enhance sys-tem performance. The effect of varying the pulserepetition rate as well as the data collection interval isalso presented. Finally, the possibility of using a sin-gle tunable laser source is discussed. It will be shownthat for applications requiring minimum size and
Jeffrey Ahl is with Science Applications International Corpora-tion, McLean, Virginia 22102; the other authors are with U.S. ArmyCenter for Night Vision and Electro-Optics, Fort Belvoir, Virginia22060.
Received 18 July 1987.
weight and operating from a stationary platform, itmay be beneficial to use such a system. In fact, datawill be presented that suggest that signal averagingwith a single-laser system may be superior to thatpreviously reported with multilaser lidars.
11. Experimental ApparatusThe CO2 laser lidar system is shown schematically in
Fig. 1 and consists of a CO2 TEA laser, a variablemagnification beam expander, a beam steering mirror(not shown), an 18-cm (7-in.) aperture Dall-Kirkhamtelescope, HgCdTe detectors for monitoring the trans-mitted and received beam intensities, a pyroelectricarray for monitoring far-field beam patterns, a dual-channel 200-MHz digitizer, and a computerized dataacquisition system. This system has monitored back-scatter returns from foliage and hillside targets atranges exceeding 3.5 km, with carrier-to-noise ratiosexceeding 10, dependent on atmospheric absorptivity.
The system utilized a grating tuned Laser Science,Inc. model 150G TEA laser yielding 120-mJ pulses in amultitransverse-mode beam. An intracavity apertureallowed single-transverse-mode operation with -60-mJ output energy. Laser pulses exhibited a 150-nsduration gain switched spike and a 1-2-gs tail. Thesingle-mode beam had a measured divergence of 3mrad and the multimode beam exhibited a divergenceof 5 mrad. Virtually no triggering jitter was observedfrom the thyratron triggered power supply. The laserwavelength was monitored with a spectrum analyzer,and the laser was typically operated on the 10P(22)transition to minimize the effects of absorbing atmo-spheric species.
The output of the laser was directed through a beamexpander with stepwise selectable magnifications of 1,2, 4, and 6. A portion of the resulting beam was thenfocused through a 2-m focal length lens onto a 128-element pyroelectric array, with an element width of0.1 mm and an element-to-element gap of 0.015 mm.
1 March 1988 / Vol. 27, No. 5 / APPLIED OPTICS 847
Fig. 1. Schematic of the TEA CO2 laser lidar system.
Thus, the far-field beam profile could be observed witha 50-grad pixel resolution. Monitoring revealed thatthe laser exhibited
to the noise variance at frequencies below a few hertz.Larger amplitude changes (exceeding 50%) have beenobserved to occur on a time scale of several minutes.
Throughout these experiments, both the carrier-to-noise ratio and transmitter energy variance were moni-tored. Most of the data analyzed exhibited CNRsexceeding 100:1 at 1 km and 40:1 at 2 km, and thereforedid not represent the dominant noise source in themeasurement. The transmitter standard deviationranged between 1% and 2%, matching values deter-mined with a pyroelectric energy meter. Electronicand digitization noise were measured by injecting areproducible signal into the detector preamp. Thecomputed standard deviation was below 1% and thuswas not a substantial contributor to the observed lidarsignal variance.A. Transmitter/Receiver Alignment
One important consideration in constructing a lidarsystem is maintaining the relative alignment betweenthe transmitted beam and the receiver field of view. Itis obvious that the receiver must image the targetilluminated by the transmitted beam. However, ifthere is any misalignment between the receiver and thetransmitter, there will be a reduction in the receivedsignal, which will eventually give rise to an increase inthe noise level. Of course, a designer would alwayswant to construct a lidar in which the transmitter andreceiver were aligned as well as possible; but it is im-portant to know the effect of misalignment so that areasonable tolerance can be specified. We have per-formed an experiment that is designed to investigatethis issue.
Treetop foliage at a range of 2 km was irradiatedwith a 3-mrad divergence, single-transverse-modebeam and the normalized return signals were mea-sured. The transmitted beam was then moved rela-tive to the receiver by known amounts which weremeasured in the far field with the pyroelectric array.The results can be seen in Fig. 2. Note that the stan-dard deviation of the returns is almost unaffected bytransmitter movements as large as 1.5 mrad eventhough a substantial portion of the beam has beenmoved out of the field of view of the receiver as evi-denced by the fact that the return has decreased toonly -60% of the value at perfect alignment. Thus,although the carrier-to-noise ratio is affected by trans-mitter/receiver misalignment, the sensitivity (signalvariance) of the system has not been substantiallydegraded. Therefore, the receiver field of view cansafely be reduced to a value only slightly larger thanthe transmitter spot size on target.
B. Beam SizeOne of the principal lidar system design consider-
ations is selection of the proper transmitter divergenceand receiver field of view. As discussed above, thereceiver field of view should be selected to be slightlylarger than the transmitted spot size on the target.Selection of the transmitter divergence is subject tothe following considerations. Changes in beam size
NORMALIZED RETURN / 0.10
1.0 - 3 8 - -8 / 0.08
in\ w > / ~ ~~~~~~~~~~~~~~~ 0.06
0.5 - 0.04STANDARD DEVIATION
I l l n inn
-3 -2 -1 0 1 2 3"_
TRANSMITTER OFFSET (mradFig. 2. Dependence of the normalized lidar return signal and stan-dard deviation as a function of the relative transmitter/receiveroffset angle for a foliage target at 2 km and a 3-mrad TEMoo trans-
TRANSMITTER DIVERGENCE (mrad)Fig. 3. Uncorrelated noise level from a foliage target at 2 km as afunction of the transmitter divergence monitored at 40 Hz with afixed 5-mrad receiver field of view. The continuous line represents
a curve fit.
are expected to affect the lidar signal variance in sever-al ways. (1) Increases in the beam size on the targetwill reduce the speckle noise by increasing the numberof independent speckle cells averaged by the receiver.(2) Increasing the area of the irradiated target spot sizewill enhance the averaging of spatial reflectivity varia-tions. This will minimize the effects of target motionor beam steering produced by either atmospheric tur-bulence or laser pointing jitter. (3) Increasing thereceiver field of view or the transmitter divergence willimprove the transmitter/receiver overlap and will min-imize the effect of variations in atmospheric absorptiv-ity.
The effect of varying the irradiated spot size wasmonitored by reducing the divergence of the transmit-ted beam by factors of 1, 2, 4, and 6. Typical resultsare shown in Fig. 3 for a multimode beam with a 5-mrad divergence, illuminating a foliage target 2 kmaway at a repetition rate of 40 Hz. As expected, thesignal variance is smaller for the larger divergencebeam. The theoretical speckle limit of 0.8%, assumingsix uncorrelated modes, is substantially lower than theobserved uncorrelated noise level. This is the case for
1 March 1988 / Vol. 27, No. 5 / APPLIED OPTICS 849
Table . Standard Deviation (%) for Returns from Diffuse TargetsRange
Targets (km) Multimode Single-mode MM/SMTrees 1 2.6 + 0.3 3.2 0.2 0.81 0.11
2 4.4 0.2 4.9 + 0.4 0.75 I 0.07
Plywood 1 3.1 i 0.5 4.1 + 0.5 0.76 + 0.15
both single-mode and multimode data taken at rangesof 0.7-3 km. Autocorrelation was also measured dur-ing this experiment and there was no discernible de-pendence of the values on the beam divergence.C. Transverse Modes
Several recent lidar systems have been designed thatutilize single-transverse-mode TEA lasers, due to aconcern that fluctuations in the profile of a multimodelaser would significantly increase the system variance.These profile fluctuations could be due to either varia-tions in the mixture of the intensities of the individualtransverse modes on a shot-to-shot basis, or to changesoccurring during a single-laser pulse. Measurementsdetailed in this section, however, demonstrate thatsystem variance with a multimode laser is at least aslow as that resulting from a single-mode laser of thesame divergence, and that the additional energy out-put (approximately a factor of 2) available with a mul-timode laser may significantly extend the operationalrange of the lidar system.
As noted previously, the laser spatial mode structurecould be selected in a reproducible manner by means ofa variable diameter intracavity aperture, and theshape of the resulting beam could be monitored bymeans of the pyroelectric array. Measurements wereperformed on both treetop foliage and a 24-m (8-ft)square painted plywood target. The divergences ofthe single-mode and multimode transmitter beamswere measured to be -3 and 5 mrad, respectively.
Typical results comparing single-mode vs multi-mode performance are displayed in Table I. All datahave been corrected for differences in the carrier-to-noise ratios for the single-mode and multimode (MM)beams. It can be seen that use of the MM beamreduces the standard deviations for signals returningfrom a foliage target by as much as 25%, while thereduction for a plywood target is up to 40%.
The reduction in the magnitude of the standarddeviation could be due to two effects: (1) The irradiat-ed target size for the multimode beam is -60% largerthan for the single-mode beam. (2) The number of
Table Ill. Short-Term Fluctuations of the Normalized Return and StandardDeviation
Data collection Standardtime Average deviation(s) return (%)
3 0.87 i 0.01 3.4 0.36 0.86 0.01 3.0 0.1
12 0.86 0.01 3.3 0.424 0.85 0.01 3.2 + 0.2
uncorrelated frequencies of the multimode beam islarger than for the single-mode beam.
As shown in Fig. 3, for a foliage target the standarddeviation decreases by -25% when the divergence of amultimode beam is increased from 3 to 5 mrad. Asthis is the same enhancement achieved by changingfrom a single-mode to a multimode beam, it appearsthat the additional speckle noise reduction due totransverse mode frequency decorrelation is not signifi-cant. This is consistent with the earlier observationthat the uncorrelated noise level is substantially abovethe computed speckle noise limit.
Data taken with a plywood target also show a 25%reduction in standard deviation when switching from asingle-mode to a multimode beam. As the targetdepth is insufficient to produce decorrelation of thetransmitted frequencies, this result is consistent withthe interpretation of the results from the foliage tar-gets.
D. Sampling TimeA possible area of concern to the lidar user might be
the variation with time of return signals reflected fromtopographic targets. Both long- and short-term fluc-tuations could have a deleterious effect on lidar utility.During the months of September and December 1986,data were taken to investigate any long-term changes.The results are given in Table II. A variety of targetswere used at 1- and 2-km ranges. A burst of 160 shotsat a repetition rate of 40 Hz was fired every minute fordurations of 13-17 min. As can be seen from the table,variations in the return signals are quite small (2-3%),while the standard deviations did not change morethan -10%. Similar results were obtained during themonths of May and June.
The effect of short-term variations was also investi-gated. A foliage target at a distance of 1800 m wasirradiated with bursts of 120 shots for times extendingfrom 3 to 24 s. Table III illustrates the results of thisexperiment. Apparently, both the magnitude and the
Table II. Long-Term Fluctuations of the Normalized Return and Standard Deviation
StandardRange Normalized deviation
Target (km) return (%) Autocorrelation NotesTreetop foliage 2 0.84 0.03 2.9 0.2 0.20 0.12 13-min duration, Sept. 1986
Plywood 1 1.05 + 0.02 4.2 + 0.3 0.40 0.11 15-min duration, Sept. 1986
Shrubs 2 1.02 0.02 4.9 0.5 0.12 0.09 17-min duration, Dec. 1986
850 APPLIED OPTICS / Vol. 27, No. 5 / 1 March 1988
Table IV. Repetition Rate Dependence of the Lidar Returns
StandardRepetition Normalized deviation
rate return (%) Autocorrelation40 0.94 0.03 4.35 0.77 0.440 0.05120 0.91 + 0.01 4.56 0.38 0.407 0.08110 0.93 0.02 4.06 0.35 0.174 0.1535 0.91 0.01 4.49 + 0.67 0.149 i 0.044
standard deviation of the reflected return signals areunaffected by any atmospheric or other variations thatmight have occurred during these time intervals. Asabove, this result was typical of others obtained overthe May through December experimental period.E. Repetition Rate
As noted previously, although the laser was capableof operating at a repetition rate of up to 150 Hz, dataacquisition limitations gave a practical upper limit of40 Hz. Nevertheless, this rate was great enough toessentially avoid effects of slow changes in atmospher-ic conditions as noted by others, where it has beensugge...