Lidar system applied in atmospheric pollution monitoring

Download Lidar system applied in atmospheric pollution monitoring

Post on 03-Oct-2016




0 download

Embed Size (px)


  • Lidar system applied in atmospheric pollutionmonitoring

    K. Fredriksson, B. Galle, K. Nystrom, and S. Svanberg

    A lidar system, incorporating tunable dye lasers and a 25-cm diam Newtonian telescope, has been con-structed and applied in atmospheric pollution monitoring. The system, which is fully controlled by a spe-cially designed microcomputer, has been used in several field tests, where stack effluents as well as the ambi-ent air have been monitored. Results from particle, NO2, and SO2 measurements are discussed.

    IntroductionRemote sensing methods are becoming increasingly

    important for monitoring the state of the earth's at-mosphere. By using laser techniques, atmosphericpollution can be studied in long-path absorption or at-mospheric backscattering experiments. Of the manyreviews that have been written on this topic, we wouldlike especially to mention the 1976 monograph editedby Hinkley.1 The laser radar or lidar approach offersparticular advantages because of its single-ended natureand great flexibility. Pulsed laser light is transmittedinto the atmosphere, and light backscattered fromparticles and aerosols is collected by an optical telescopeand analyzed. Apart from particle content information,knowledge of the 3-D distribution of certain molecularpollutants can be obtained using the so-called dial(differential absorption lidar) technique. Here theparticle backscattering is measured for two wavelengths,where the gas to be investigated absorbs strongly andweakly, respectively. This technique was pioneered bySchotland,2 and early laser work was performed byRothe et al., 3 Grant et al., 4 and others. In 1975 westarted a program of atmospheric laser probing at thisdepartment. In this paper we describe the lidar systemthat we constructed and discuss results from severalfield tests, where particles, NO 2, and SO2 were studied.Our system incorporates tunable dye lasers and uses aNewtonian telescope. A specially designed micro-computer fully controls the measurement procedure.

    The authors are with Chalmers University of Technology, PhysicsDepartment, S-412 96 Goteborg, Sweden.

    Received 6 March 1979.0003-6935/79/172998-06$00.50/0. 1979 Optical Society of America.

    For a lidar system, information from a distance R isobtained at a time delay t = 2R/c after the transmissionof the laser pulse, traveling with the velocity of light c.The power P\(R, AR) received from the interval AR atthe range R is given by the general lidar equation:

    Px(R,AR) = CWAbN(R)R2

    exp-2SR[o(X)n(r) + oaN(r)]dr , (1)where W is the transmitted pulse power, and N(R) isthe concentration of scatterers with the backscatteringcross section b. The exponential describes the at-tenuation of the beam and the backscattered light. Theattenuation is caused by absorption due to moleculesof concentration n(r) and absorption cross section o-(X)and scattering and absorption due to particles of con-centration N(r) and cross section -a*. C is a systemconstant. In particle studies it is sometimes a reason-able approximation to neglect the absorption term. If,then, 0ib is assumed to be a constant, N(R) can be di-rectly mapped in three dimensions. As the value of Ubis not generally known, only relative concentrationdistributions are usually obtained. In dial measure-ments of gas concentrations, use is made of the fact thatwavelength regions can be found where (A) stronglychanges for a small wavelength change, while rb and astay approximately constant. In these conditions, theevaluation of the experimental data is quite simple. Ifmore than one gas absorbs in the wavelength region ofinterest, a summation must be performed in the expo-nential factor.

    Lidar SystemThe lidar system used in these measurements is

    schematically shown in Fig. 1. The system includesthree different pulsed lasers, a Newtonian telescope,and detection, controlling, and processing electronics.

    2998 APPLIED OPTICS / Vol. 18, No. 17 / 1 September 1979

  • Fig. 1. Schematic diagram of the lidar system.

    The choice of laser transmitter is dependent on themeasuring technique employed. A 400-kW nitrogenlaser with 10-nsec pulses at 337 nm is used in measure-ments of particles and aerosols. The repetition rate is10-50 pulses/sec. In NO2 measurements, a dye laserpumped by the nitrogen laser is used. The dyes Cou-marin 45 or Coumarin 47, giving about 0.5 mJ in 5-nsecpulses in the 440-480-nm region, are used to producelight of suitable wavelengths in the NO2 absorptionband. Wavelengths around 300 nm in the UV absorp-tion band of SO2 are obtained with a frequency-doubledflashlamp pumped dye laser as 1-,usec pulses with theenergy 0.4 mJ at a repetition rate of 10-25 Hz. Thewavelength setting of the dye lasers is made by steppermotors.

    The Newtonian telescope has a 25-cm mirror with afocal length of 100 cm and is constructed with a mountfor vertical and horizontal turnings. The turnings areeffected with computer-controlled motors. The laserbeam is transmitted to an output mirror (placed insidethe telescope) by means of first-surface aluminummirrors in such a way that the transmitted laser beamand the optical axis of the telescope are almost coaxialat every positioning of the system. Alternatively, theflashlamp pumped dye laser can be placed on a tablerotating together with the telescope. The field of viewof the telescope normally corresponds to 2 mrad tomatch the divergence of the dye lasers. Interferencefilters, 5-10 nm broad with a transmission of 35-55%,are used to suppress the background skylight. Anothervariation is that the system may be used with a 500-mmBausch & Lomb grating monochromator.

    The spectrally filtered lidar signal is detected by anEMI 9558 QB PMT. A fast transient digitizer, Bio-mation 8100 with 2048 8-bit channels, is used to convertthe signal to time-resolved digital data. The data aretransferred to a microcomputer for signal averaging.

    This unit has a memory consisting of 4096 24-bit wordsthat may be divided into sixteen or fewer parts. Themaximum transient frequency is 400 Hz. The numberof transients to be averaged, the laser wavelength set-ting, the telescope steering,-and the readout on a strip-chart recorder are controlled by the microcomputer,which also can do arithmetic operations. Thus themeasuring cycle can be fully automatized. The mea-surement process is monitored on oscilloscopes viacontinuous readouts of the individual and the averagedlidar signals. In some lidar measurements a boxcaraverager is used instead of the transient digitizer. Inthis case the lidar signals from only one or two distancesare studied at each laser pulse. In some measurementmodes a chopper is used to block every second or ninthshot from the laser. Then rf interference, caused by thenitrogen laser, and the background signal level aresubtracted from the lidar signals. The lidar system isdescribed in more detail in an institute report.5 Inmeasurements over the city of Goteborg, the system hasbeen placed in a suitably located laboratory within thePhysics Department; during field tests it was housed ina small truck.Measurements of Particles

    We have used the lidar system in several studies ofparticle emissions from industrial smoke stacks.Measurements of relative particle distributions are easyto perform using elastically backscattered light andneglecting weak effects of beam attenuation. To obtaingood spatial resolution it is important to use a laser ofshort pulse length. Normally we have used the 10-nsecUV pulses from the N2 laser. In Fig. 2 the result of avertical scan downwind an industrial area is shown.

    1000 1500 m

    800 1000 1200 1400 m

    Fig. 2. Relative particle distribution downwind an industrial area,obtained in a vertical lidar scan employing a nitrogen laser. An ex-perimental curve illustrating the evaluation procedure is included.

    1 September 1979 / Vol. 18, No. 17 / APPLIED OPTICS 2999

  • mg/0,

    1000 ' I-


    I ., ., . . .I.I





    14hO5m 1 4 h3 0 m 15hO0M 15h30 m

    REMOTE SENSINGI , . . . I , ,t | , I . .I I , I .14h05n 1430M 15h 00

    Fig. 3. Comparison between lidar and in-stack measurements ofparticle concentrations. The lidar curve features the amplitude ofthe elastic plume echo as recorded at 220-m distance. The curvedeviations are discussed in the text. The in-stack instrument mea-sured the attenuation of white light across the plume diameter. This

    instrument was calibrated against sampling measurements.

    The lidar system was placed about 1 km away from thearea to be investigated. In the illustration, an exampleof a recorded lidar curve is shown, featuring several echostructures from spreading plumes. The 1/R2 curvefrom the background particle concentration is clearlyseen. In addition, 1/R2 curves corresponding to higherconcentrations are indicated. The particle distributiondiagram given in the illustration was constructed bylocating the intersections between such equidistantcurves and the actual backscattering profile for differentelevation angles. We have used similar techniques tomap the extensions of isolated spreading plumes. Thiscan readily be done even for plumes invisible to thenaked eye. Plume mapping techniques have beendiscussed, e.g., by Uthe and Wilson.6

    If absolute particle loads in stack effluents are to bemeasured, the lidar system should be pointed to theplume as close to the mouth of the stack as possible.This approach avoids both influences due to wind anddue to condensing water droplets. Because of thecomplexity of the Mie scattering theory and the lack ofdetailed information on particle characteristics, it isnormally necessary to provide an in-stack calibration.If the particle concentration is not too high, the ampli-tude of the plume echo can be used in the evaluationprocedure. For dense plumes it is preferable to mea-sure the beam attenuation by the plume, measured inthe light backscattered from the air before and behindthe plume ("plume opacity").7 In Fig. 3 a particlemeasurement is illustrated. The remote sensing curvedisplays the amplitude of the plume echo vs time. Themeasurement was performed using the boxcar averager,

    set to sample the intensity at a delay time correspondingto the light round-trip time to the plume, in the actualcase situated at 220-m distance from the lidar system.Comparison is made with the in-stack curve obtainedwith a calibrated recording instrument. The partslacking on each curve correspond to zero calibrationintervals. A good correlation between the curves wasfound except during the situations when the in-stackinstrument was out of range. From the curves theproper calibration for the particular type of smoke (froman iron-alloy melting furnace) could be established.

    Measurements of NO2In measurements of gaseous pollutants, the dial

    technique is normally employed. As mentioned above,this technique is based on the difference in absorptioncross section of a certain molecular gas at two or moreclose lying wavelengths. At normal atmospheric con-ditions the backscattering and the absorption of parti-cles and aerosols are slowly varying functions of thewavelength. Thus, a fast wavelength dependence in thelidar Eq. (1) will be solely due to absorbing gases.making the exponential factor change. The systemconstant C may also have a fast wavelength dependence,e.g., from Fabry-Perot effects in the transmitting andreceiving optics, but this is automatically taken care ofby the evaluation procedure. The dial equation is ob-tained from Eq. (1) as

    P^(R) = C'* exp{-2 fr X1) - o-(X2)]n(r)dr}j (2)where C' is a new constant, and it is assumed that onlyone atmospheric gas has a differential absorption at thetwo wavelengths XI and X2. If o(X1 ) is larger than U(X 2),it follows from Eq. (2) that the dial signal

    P,\1(R)/PX2(R)will decrease with the distance R in the presence of thestudied gas.

    We have measured NO2 concentrations with the dialtechnique in the 440-460-nm wavelength region. Thecross-section values used in the calculations are thosegiven by Grant et al., 4 with wavelength extensions madein absorption measurements in our laboratory fitted tothese values. Deviations from these, as obtained inlaboratory measurements by Woods and Jolliffe,8 arewithin the uncertainty limits. Possible corrections athigh temperatures in smokestack plumes have not beenconsidered in these measurements.

    Our first measurements on gaseous pollutants con-cerned NO2 concentrations in smokestack plumes of anoil refinery. The lidar system was located at a distanceof 2 km from the stack studied. In Fig. 4 a measure-ment is shown in which the lidar was directed just abovethe top of the smokestack. The backscattered intensityfrom the atmosphere surrounding the plume, measuredfor two different wavelengths, is displayed. The lidarsignal of the plume is far beyond scale and is cutoff inthe figure. This measurement implies an integratedconcentration of 190 i 50 mg/M3 - m, which equals 30i 10 mg/M3 as an average in the plume. The uncer-

    3000 APPLIED OPTICS / Vol. 18, No. 17 / 1 September 1979

  • 30 mg/M3 - m at the top of the smokestack. Measure-ments of exhausts from an aeroengine manufacturerduring tests of jet engines implied very high NO2 con-tents; values up to 900 mg/M3 - m were obtained. TheseNO2 measurements were performed at a range of 1 km.NO2 concentrations in the ambient air have been mea-sured in different types of areas. Tens of parts perbillion have been recorded in the city of Goteborg.When average concentrations over large distances areto be measured, one may use some distant topographictarget instead of the atmosphere for the backscatteringof the laser beam. Such a measurement is shown in Fig.5, where a mountain at a distance of 5.5 km is used as thebackscattering target. This measurement yields anaverage concentration of 2 2-ppb NO2 .

    Fig. 4. Measurement of the NO2 content in the plume from anoil-refinery stack situated 2 km from the remote sensing system. Twolidar curves for two close-lying wavelengths with different absorptioncross sections are shown. A NO2 concentration of 30 + 10 mg/m


    through the 6-m thick plume is inferred.

    N0 20WJI-

    m z.

    L FI- -O W

    Co -




    1 2 3 4 5 km


    Fig. 5. The lidar curve shows topographic-target echo as well asatmospheric backscattering. Measurements at two wavelengths,446.5 nm and 448.5 nm, yielded a mean NO2 concentration of 2 2

    ppb over the 0.5-5.5-km range interval.

    tainty limits of these values are determined by the sta-tistical noise in the measurement curves. The upperlimit of the concentration is further inferred by the in-dividual attenuation in each of the two curves. Theconcentration was found to be much below this valueduring most of the measuring period.

    Plumes from an oil-burning power plant were alsostudied, and the NO2 contents were found to be below

    Measurements of SO2In the dial measurements of SO2 , wavelengths in the

    absorption band around 300 nm are used. The ab-sorption cross-section values used in our evaluations arethose given by Thompson et al. 9 Corrections to thesevalues at high te...


View more >