The world’s scientific community agrees on the factthat human activities have an impact on the climate.Global warming and the induced climate change areamong the most important environmental issuesnowadays.1 The enhanced greenhouse effect that isdue to human activities emitting carbon dioxide,methane, nitrous oxide, chlorofluorocarbons, or ozoneprecursors is investigated in great detail today. De-spite the great efforts made during the past decade toevaluate the radiative effects of aerosols and watervapor, the level of scientific understanding of thisproblem has been assessed as very low in the lastreport of the International Panel for ClimateChange.1 It is considered as the largest uncertainty
G. Larcheveque, I. Balin, R. Nessler, P. Quaglia, V. Simeonov, H.van den Bergh, and B. Calpini �[email protected]� are withthe Air Pollution Laboratory, Swiss Federal Institute of Technol-ogy, CH-1015, Lausanne, Switzerland.
Received 24 October 2001; revised manuscript received 25 Feb-ruary 2002.
in the understanding of the Earth’s radiation bud-get.2 Water vapor is a primary greenhouse gas, and,despite its small amount in the free troposphere andstratosphere, recent studies have shown that it mayplay an essential role in the Earth’s climate.3–6
Aerosols affect the heat balance of the Earth bothdirectly by reflecting and absorbing solar radiationand by absorbing and emitting some terrestrial in-frared radiation and indirectly by influencing theproperties and processes of clouds and possibly bychanging the heterogeneous chemistry of reactivegreenhouse gases.2,3
For global atmosphere model predictions, thechanges arising from the aerosol scattering and ab-sorption of radiation, referred to as the direct radia-tive forcing, can be accurately calculated at least inprinciple once the optical constants, size distribution,and atmospheric concentration of the aerosols areknown. Hence it is a primary necessity to create aquantitative data set describing on a large scale theaerosol vertical, horizontal, and temporal distribu-tion, including its variability. To create such data-base on a European scale, the European AerosolResearch LIdar NETwork �EARLINET� was set up in1999.7,8 The network comprises 21 lidar stations
distributed over most of Europe. The lidar measure-ments are supported by a suite of more conventionalobservations. Most of the stations employ alreadyexisting instruments, the prevailing part of which aremultiwavelength systems. More than half of thestations have the ability to perform aerosol Ramanmeasurements.
This paper focuses on describing a combined mul-tiwavelength Raman elastic-backscatter lidar systemspecially built for measurements in the EARLINETnetwork and on some of the results obtained with thelidar. The system was installed in 1999 at the Jung-fraujoch Research Station �latitude, 46.55°N; longi-tude, 7.98°E; 3580 m above sea level �ASL��.Because of its location in the Swiss Alpine region, thestation is above the planetary boundary layer most ofthe time, and the aerosol load and the humidity levelsare extremely low. These are unique conditions forhigh-altitude lidar observations in central Europe,giving researchers the possibility to perform mea-surements of atmospheric extinction and water-vapor content in the upper troposphere, avoiding theperturbation caused by the planetary boundary layer.Furthermore, the Jungfraujoch Observatory is aworld-known atmospheric monitoring station per-forming observations since the 1950’s.9–13 It is anessential part of different international networksfor monitoring of the atmosphere, and many differ-ent instruments, providing well-established qualitydata, are installed there. The availability of insitu measurements gives an additional opportunityto compare and link the extinction and backscatterinformation obtained by the lidar to the in situ data.
The Jungfraujoch lidar station was designed as amultifunctional system allowing multiwavelengthaerosol measurements by elastic and Raman meth-ods, depolarization measurements at 532 nm, andwater-vapor mixing-ratio measurements by the Ra-man technique. The specifications of the lidar sys-tem are given in Table 1. Future implementation ofpure rotational Raman temperature measurements
for the lower altitudes, completed with Rayleigh tem-perature measurements for the higher altitudes anda differential-absorption-lidar tropospheric andstratospheric ozone channel, will expand the capabil-ity of the instrument. The results will be comparedwith the data from instruments for column and insitu measurements of similar atmospheric parame-ters already installed at the station.
2. Methodology
The advantages and drawbacks of aerosol lidars forderiving optical aerosol parameters have been widelydiscussed in the literature.14,15 There are two simi-lar approaches, the Fernald and the Klett inversionalgorithms,16,17 for solving the ill-posed problem ofdetermining an aerosol extinction coefficient from asingle wavelength lidar. Another approach is to useRaman scattering from atmospheric nitrogen as anadditional signal with a well-known backscatter co-efficient.18,19 The disadvantage of this method is thelow efficiency of the spontaneous vibrational Ramanscattering that results in a low signal level, thussuitable predominantly for nighttime measurements.In our data treatment we have used Fernald andRaman techniques to retrieve the aerosol extinctionand backscatter coefficients. By use of the sponta-neous vibrational Raman signals from atmosphericnitrogen and water vapor, the range-resolved water-vapor mixing ratio is retrieved; this technique hasbecome a standard procedure for nighttime water-vapor measurement.20
If the atmosphere is probed by a polarized laserbeam, the backscatter light will maintain the polar-ization within a limit of 2%–3% if the scattering isdue to molecules �Rayleigh scattering� or homoge-neous spherical particles �Mie scattering� in thesingle-scattering approximation, whereas nonspheri-cal particles or multiple scattering will induce somedegree of depolarization. The degree of depolariza-tion in lidar applications is measured usually by thedepolarization ratio, which is the ratio between the
Table 1. Specifications of the Jungfraujoch Lidar Systema
Backscattering 355 355 extinction profile at 355 nm 30 100–200 30 min nightBackscattering 532 532 extinction profile at 532 nm 30 100–200 30 min nightBackscattering 355 355 lidar ratio at 355 nm 30 100–200 30 min night–dayBackscattering 532 532 lidar ratio at 532 nm 30 100–200 30 min night–dayBackscattering 1064 1064 lidar ratio at 1064 nm 30 100–200 30 min night–dayDepolarization 532 532 parallel
532 perpendicular30 100–200 30 min night–day
H2O mixing ratio 408 � 387 calibration value 12 100–200 2 h night
intensities of the returned cross-polarized andparallel-polarized signals. By analysis of the depo-larization ratio, information about particle shape andphase or presence of multiple scattering can be de-rived.21,22
As already mentioned, the aerosol size distributionis a key parameter. Its retrieval is possible if inde-pendent measurements of the aerosol extinction andbackscatter coefficients are available at only a fewlidar wavelengths.23,24 The Jungfraujoch lidar per-mits the determination of three backscatter coeffi-cients and two extinction coefficients. Table 1summarizes the different wavelengths and measuredparameters that are currently performed at the Jung-fraujoch together with their respective limit of detec-tion.
3. System Description
The Jungfraujoch lidar is based on a two-telescopeconfiguration in order to cover a maximum operatingrange. The system was mounted in the astronomi-cal dome of the Jungfraujoch Station. The lidar re-ceiving optics and electronics are placed on thesecond floor of the dome, and the transmitting optics,the computers for controlling the laser source, andthe acquisition system together with the laser coolingunit are installed on the first floor as shown in Fig. 1.Figure 2 is a picture of the Jungfaujoch lidar in op-eration.
The laser transmitter is based on a tripled Nd:YAGlaser �Coherent, Infinity�. The laser produces 3.5-nspulses with energies of as much as 400 mJ at 1064nm. Owing to the original scheme of the amplifierthat employs a phase-conjugated mirror for thermallensing correction, the laser repetition rate can betuned continuously from 0.1 to 100 Hz. The laserparameters are specified in Table 2. Two �-bariumborate crystals perform the frequency doubling andtripling of the fundamental frequency. Beam split-ters separate the three output wavelengths, 1064,532, and 355 nm. The corresponding beams are ex-
panded by three 5X beam expanders and directedonto three dielectric steering mirrors. These mir-rors are controlled independently by piezoelectric-driven x–y mounts to align the laser beams within thefield of view of the telescopes. To control preciselythe polarization state of the transmitted 532-nm ra-diation, an air-spaced Glan–Thompson prism is in-serted into the optical path of the 532-nm beam.
In the receiving part a dual-telescope configurationis employed to reduce the dynamic range of the sig-nals. A Newtonian, 20-cm telescope with an aper-ture ratio of f�4 covers the lower range from 400- to12000-m altitude above ground level. The telescopeis pointed vertically and can be tilted at 5° from thevertical. The astronomical telescope originally in-stalled into the dome and used earlier for stellar pho-tometry will provide stratospheric and mesosphericmeasurements. It is a Cassegrain type with a mainmirror diameter of 760 mm and an aperture ratio off�14. It has German equatorial mounting with slow-motion drives on both �polar and elevation� axes thatwe use to align the telescope versus the laser beams.Even though preliminary tests have been performedwith this telescope and lidar signals from altitudes toas high as 60-km above ground level were detected,regular measurements are still not being carried out.
At the present moment the received wavelengthsfor each telescope include three elastically scatteredwavelengths and two spontaneous Raman signalsfrom nitrogen and water vapor. The optical signalsreceived by each of the telescopes are separated spec-trally by two filter polychromators. They are builtwith an identical optical layout for both telescopesbut differ by physical size and the way of opticalcoupling to the respective telescope. The filter poly-chromators and coupling optics are designed and op-timized by use of a ray-tracing analysis �OSLO, version6, software�. The optical layout of the short-rangereceiver without the receiving telescope is schemati-
Fig. 1. Schematic of the Jungfraujoch lidar installation in theastronomical cupola.
Fig. 2. Nighttime picture of the Ecole Polytechnique Federal Lau-sanne lidar system at Jungfraujoch.
cally depicted in Fig. 3. The different wavelengthsare separated by dichroic beam splitters �BS� andthen filtered by sets of broadband �typically, 4 nm�
and narrow-band �typically, 0.5 nm� interference fil-ters �F� specific for each wavelength. Neutral-density �ND� filters are employed to adapt the light
Channel 607 nmBeam-splitter efficiency 77%Filter transmission 42%Out-of-band filter transmission 10�9 �relative to the peak transmission�Detector PMT, Thorn EMI, B9202
Channel 106 nmBeam-splitter efficiency 73%Filter transmission 58%Out-of-band filter transmission 10�8 �relative to the peak transmission�Detector APD, EG&G, C30954�5E
Data acquisition systemType Seven input-channels in analog and photon-counting modes with two triggersModel LicelMaximum count rate �photon-counting mode� 200 MHzMaximum voltage �analog mode� 20, 100, 500 mVMinimum time-bin width 50 nsNumber of time bins 16000
intensity of the signals to the corresponding photo-multiplier tube �PMT� sensitivity. The importantoptical parameters of the filter polychromators to-gether with the photodetectors’ efficiencies are sum-marized in Table 2.
Two different polarization states with planes ofpolarization parallel and perpendicular to the polar-ization plane of the transmitted beam are separatedfrom the 532-nm signal by a Wollaston prism �WP� inorder to define the depolarization ratio. A 2X beamcompressor adapts the size of the incoming beam tothe prism aperture. The Wollaston prism, the beamcompressor, and the two detecting PMTs aremounted on a holder that can be rotated preciselyaround the common axis of the optical elements.This design allows precise alignment and an easyway of calibrating the depolarization block by revers-ing the position of the parallel and perpendicularPMTs.
PMTs perform the detection of the optical signalsfor the UV and visible signals and by Si avalanchephotodiodes �APD� �EG&G, C30954�5E� for 1064-nmsignals. Elastically backscattered signals at 355and 532 nm �with parallel and perpendicular polar-izations� are detected by photosensor modules�Hamamatsu, 6780-06�. The new, metal-channeldynode configuration of the PMT used in this moduleensures short pulse duration and good pulse-heightdistribution, which in turn make possible both theanalog and the photon-counting modes of detectingthe signals. Despite these excellent features, thePMT shows nonnegligible spatial nonuniformity.An improvement of the PMT uniformity leading tovariations of less than 2% in the sensitivity of the
central part of the photocathode was achieved by ad-dition of a lens and a diffuser before the PMT, asdescribed in Ref. 25. The Raman wavelengths aredetected by two types of PMT for 387 and 408 nm�Thorn-EMI, QA9829A� and for 607 nm �Thorn EMI,B9202A� signals. The PMTs used in the Ramanchannels were specially selected to work in a photon-counting mode.
The acquisition of the signals is performed by seventransient recorders �Licel GmbH�. Each transientrecorder combines a 250-MHz photon-counting unitwith a 12-bit 20-MHz analog-to-digital converter. Alow repetition rate alternating storing of the data ispossible for each of the transient recorders by use oftwo sets of memories consisting of a 24-bit RAM anda 16-bit RAM for the analog and for the photon-counting data, respectively.
The data stream is directed to one of these memorysets depending on the states of two external triggers.This feature of the device allows us to record the datafrom the Newtonian and the Cassegranian telescopesby turns by use of a single acquisition system. Thecontrol over the transient recorders, as well as thepreliminary and final treatments of the data, is per-formed on a personal computer by use of LABVIEW andMATLAB programs.
4. Results
In the framework of the EARLINET project the lidarhas been operating every other week since May 2000.Examples of data recorded with the tropospheric tele-scope during this period will be discussed here insome detail.
Range-corrected signals �RCS�, which are lidar sig-nals multiplied by the square of the distance, at thethree elastic-backscatter wavelengths and the Ra-man wavelength at 387 nm are shown in Fig. 4.RCS signals in analog mode recorded on 28 August2000 between 22h00 and 22h30 local time are shownin the left panel, whereas the right panel shows thesame set of data but acquired in photon-counting
Fig. 3. Schematic of the lidar detection setup �lens, L; beam split-ter, BS; filter, F; Wollaston prism, WP; photomultiplier tube, PMT;avalanche photodiode, APD; neutral-density filter, ND; diffuser, D.
Fig. 4. �left panel� Range-corrected signals at 355, 387, 532, and1064 nm in analog mode and �right panel� photon-counting mode.Arrows mark the break in slope of the Raman channel that is dueto the aerosol extinction in the cloud.
mode. Note that the logarithmic scale used for theRCS intensity is on the horizontal scale. In this li-dar configuration, the three laser beams are emittedoff axis with respect to the telescopes. At short dis-tance, in which a full overlap of the laser light withthe telescope field of view is not achieved, the mea-surements are not used. The data shown in Fig. 4 arevalid only from a lower altitude of approximately 4km ASL �or 420 m above the lidar site�. A strongaerosol layer between approximately 7 and 11 km�possibly a subvisible or thin cirrus cloud� is welloutlined on the RCS at the elastic wavelengths. Thesame layer is seen in the Raman channel as a changeof the slope of the RCS owing to the increased opticalthickness. Arrows in the panels point to the breakin slope of the RCS at the bottom of the layer. Thefact that this relatively strong layer does not give riseto the Raman signal is a proof for the valid rejectionof the elastic-backscatter light in the Raman channelof the system.
The vertical profiles of the aerosol backscatter co-efficients at 355, 532, and 1064 nm calculated by theFernald inversion method from the already presentedRCS are shown in Fig. 5. The backscatter coeffi-cients for each of the 355- and 532-nm wavelengthsby use of both analog and photon-counting signalsshow extremely good agreement. In the Fernald for-malism the weak point is that a lidar ratio must bepostulated at each wavelength a priori; thus retriev-ing these coefficients provides values that are morerelative rather than absolute. On the basis of in situobservation at Jungfraujoch26 and also in accordancewith Ref. 27, a value of 15 sr for the lidar ratio wasselected for each wavelength, a value that was keptconstant at any altitude considered in these exam-ples. As will be seen later, this value is also in goodaccordance with the one retrieved in the case of theRaman formalism applied to the 355-nm elastic lidarsignal for which the nitrogen Raman channel at 387nm was used as an additional channel, thus avoidingany hypothesis about the lidar ratio. The data was
smoothed by a variable gliding average �from 37.5 to150 m in analog mode and 75 to 225 m in photon-counting mode� prior to inversion. The integrationtime was 30 min. In the profiles shown in Fig. 5, tworegions with different light-scattering properties canbe distinguished. In the region between 8 and 11km ASL, the values of the aerosol backscatter coeffi-cients for the three wavelengths are equal and of theorder of 1.5 � 10�5 m�1 sr�1. This indicates that theaerosol particles are large compared with any of theemitted wavelengths. In this case the scattering ef-ficiency is almost independent of the wavelength, andthe aerosol contribution is larger than the molecularcontribution. The second type of aerosol light scat-tering is seen at an altitude region between 4 and 7km ASL in which the laser wavelengths are of thesame order of magnitude as the aerosol mean size.Here the ratio between the particle size of the parti-cles and the laser wavelength is the key parametergoverning this particle-scattering effect. In thiscase the three aerosol backscatter coefficients differsignificantly. For example, at an altitude of 5 kmASL, the backscatter coefficient is equal to 2 � 10�6
m�1 sr�1, 1 � 10�6 m�1 sr�1, and 0.5 � 10�6 m�1
sr�1, respectively, at 355, 532, and 1064 nm. In thatcase the molecular scattering is prevailing over theaerosol scattering.
Figure 6 shows the comparison between the aerosolextinction and the backscatter coefficients obtainedwith in one case the Fernald inversion at 355 nm witha constant lidar ratio of 15 sr �average value from theRaman lidar ratio� and in the other case the inversionalgorithm by use of the Raman nitrogen signal at 387nm combined with the elastic lidar signal at 355 nm.Owing to the weaker Raman signals, the averagingtime is 1 h, a larger variable gliding average �300 to900 m� is used, and the comparison is shown hereonly in photon-counting mode because the Ramannitrogen signal intensity is more than 3 orders of
Fig. 5. Aerosol backscatter coefficients in analog �A� and photon-counting �PC� mode retrieved from the data in Fig. 4 when theFernald inversion formalism was applied and a lidar ratio of 15 srwas assumed.
Fig. 6. Aerosol extinction and backscatter coefficients at 355 nmobtained with the Fernald formalism and by combinations of Ra-man and elastic signals, retrieved from the data in Fig. 4. Thecorresponding lidar ratio of the Raman profiles is given in the rightpanel, and a constant value of the lidar ratio of 15 sr is used for theFernald formalism.
magnitude lower than the elastic one. The profilesare compared in the cloud region between 7.5 and10.5 km ASL. An angstrom parameter equal to zero�no wavelength dependence� is taken, and the molec-ular atmosphere is assumed at an altitude of 14 kmASL to retrieve the backscatter coefficient. The den-sity profile is calculated with a U.S. standard atmo-sphere28 fitted to the temperature and pressuremeasured at the Jungfraujoch Station. The extinc-tion and backscatter coefficients obtained are in goodagreement. Between 8.5 and 9.5 km, the retrievedcoefficients are also similar to the values proposed inRef. 27 for cirrus conditions with an extinction coef-ficient of 1.4 � 10�5 m�1, a backscattering coefficientof 2 � 10�4 m�1 sr�1, and a lidar ratio of 13.5 sr.
The temporal evolution of the aerosol backscattercoefficients on 28 August 2000 between 16h00 and24h00 local time calculated by the Fernald inver-sion method at 355 and 1064 nm is shown in Figs.7 and 8, respectively. The time resolution is halfan hour, and the lidar ratio is chosen to equal 15 sr.The blank parts on the plot are due to strong ex-tinction inside the cloud. The top of the aerosollayer is around the tropopause while its bottomdecreases in time until the layer reaches an altitudeof 2 km above the Jungfraujoch lidar station at mid-night. During these measurements, a Saharan dustevent was recorded simultaneously over Europe bythe EARLINET network. In particular, the aerosolbackscatter values extending from 4 to 7 km ASLwere shown to be linked to the presence of Saharandust particles. Peak values of the order of 1.5–5.5 �10�6 m�1 sr�1 were observed by several lidar systemswithin EARLINET operating at 355 nm.29 Similarvalues were detected from our system as shown inFig. 5. Note that in this case at 1064 nm, the pres-ence of these particles exhibits a weaker contributionto the aerosol backscatter coefficient. This is an-other example of the strong wavelength dependenceof the backscatter coefficient.
The sources of uncertainties in the estimate of theaerosol backscatter coefficient by the Fernald inver-sion with a constant lidar ratio of 15 sr are investi-gated in Fig. 9. Here we consider only the errorsources that are associated with the statistical errorthat is due to the signal detection,30 the systematicerrors linked with the estimate of the lidar ratio,31
the total backscatter coefficient at the reference alti-tude, and the molecular backscatter coefficient.32
They are taken into account according to Ref. 33 andreported in the figure as the ratio of the uncertaintydivided by the aerosol backscatter coefficient. Addi-tional error sources that are due to multiple scatter-ing34,35 or misalignment of the lidar transmitter arenot considered here. The relative error on the back-scatter profiles is calculated from the data samplesshown in Fig. 5 assuming a Poisson distribution36 forthe statistical error that is due to the signal detection
Fig. 7. Temporal evolution of the aerosol backscatter coefficientat 355 nm on 28 August 2000 between 16h00 and 24h00 local time.
Fig. 8. Temporal evolution of the aerosol backscatter coefficientat 1064 nm on 28 August 2000 between 16h00 and 24h00 localtime.
Fig. 9. Relative error on the aerosol backscatter coefficients inanalog and photon-counting modes calculated from the data in Fig.5, taking into account the statistical error that is due to the signaldetection, the systematic errors that are due to the estimate of thelidar ratio, the total backscatter coefficient at the reference alti-tude, and the molecular backscatter coefficient.
and uncertainties of 3% on the molecular coefficientand 10% on the lidar ratio. The relative error on theaerosol backscatter ratio remains below 8% to as highas altitude of 11 km ASL �below the cirrus cloudlayer�, whereas at a higher altitude the statisticalerror on the signal detection increases owing to a poorsignal-to-noise ratio.
The depolarization ratio brings additional informa-tion on the aerosol shape �spherical versus nonspheri-cal particles�. In Fig. 10 a typical measurement ofthe depolarization ratio is shown. It exhibits valuesbelow the cloud layer of the order of 2% that is indic-ative of an aerosol-free atmosphere. This low base-line value of the depolarization ratio is obtainedbecause of the special design of the detection unit byuse of a Wollaston prism as polarizer. In Fig. 10 acloud layer is seen between 6.5 and 8.5 km ASL witha maximum depolarization ratio around 7.8 km. Afirst break in the vertical profile of the depolarizationratio around 6.5 km ASL indicates that part of thewater vapor was condensed and frozen in the form ofsmall crystallites, thus causing this increase of thedepolarization ratio. The latter has a value that isclose to its detection limit �2%� at 6.5 km ASL andincreases almost linearly to 9% at 7.2 km ASL. It isassumed to be a region of the cloud characterized bya mixture of water droplets and small ice particles. At7.2 km ASL a second break in the profile is observed,followed by a maximum depolarization ratio of theorder of 40% at 7.8 km ASL. It is indicative of acloud layer containing only frozen particles �crystal-lites� in this case. The aerosol backscatter coeffi-cients that were simultaneously recorded �only the�532nm vertical profile is shown in Fig. 10 for clarity�had values of the order of 4 � 10�7, 2 � 10�7, and 1 �10�8 m�1 sr�1 at 355, 532, and 1064 nm below thecloud �6.5 km ASL�. Increasing scattering towardshorter wavelengths is a general characteristic ofnonabsorbing particles that are small compared withthe wavelength of the laser light.37 At 7.8 km ASL,these coefficients reached maximum values of the
order of 3 � 10�6, 2 � 10�6, and 7 � 10�7 m�1 sr�1.In this case the different backscattering values aredepicting a wavelength dependence that is totallydifferent from the one presented in Fig. 5.
Since August 2000, nighttime measurements of thewater-vapor mixing ratio were performed on a regu-lar basis in the middle and upper troposphere.38
The water vapor is measured by Raman lidar takingadvantage of the spontaneous vibrational Ramanscattering of an incident laser beam by atmosphericN2 and H2O molecules. The third harmonic of theNd:YAG laser at 355 nm is used, with the Raman-shifted wavelengths at, respectively, 387 nm from N2and 408 nm from H2O. In a first approximation thewater-vapor mixing ratio H2O�z� defined in units ofg�kg of dry air is proportional to the ratio of the twoRaman-backscattered signals assuming a constantmixing ratio for N2. The lidar system efficiency isnot known a priori, and a reference value must beused as a calibration point for retrieving an absolutewater-vapor profile by the Raman lidar.
Figure 11 shows the water vapor measured by Ra-man lidar from Jungfraujoch for a total integrationtime of 2 h and a vertical resolution of 75 m. Thelidar profile is compared with parallel balloon mea-surements by use of a chilled cooled mirror hygrom-eter �i.e., Snow White� launched from the SwissMeteorological Station in Payerne �located approxi-mately 80 km west from the Jungfraujoch Station�.At the altitude of 4.75 km ASL, the absolute water-vapor content measured by balloon is used as a ref-erence value for the lidar profile. Both profiles arein good agreement, especially if one recalls here thatthe lidar measurements were performed with theshort-range telescope �primary diameter of only 20cm�. Error sources in the water-vapor lidar mea-
Fig. 10. Depolarization ratio at 532 nm in analog and photon-counting modes. The backscatter coefficient at 532 nm when theFernald inversion formalism is used and a lidar ratio of 15 sr isassumed is shown for direct comparison.
Fig. 11. Water-vapor mixing ratio measured by Raman lidar andby balloon �Snow White sonde, Swiss Meteorological Station inPayerne�. The calibration point for the Raman lidar data is usedat an altitude range of 4.75 km ASL.
surements were presented in detail in Ref. 38 with anoverall uncertainty of below 5% for clear-sky condi-tions but that may reach 20% in the case of hazyconditions.
5. Conclusion
The multiwavelength aerosol lidar constructed at theJungfaujoch Alpine Station in Switzerland was pre-sented with the aim of providing extinction and back-scatter profiles at 355 and 532 nm, backscatterprofiles at 1064 nm, and depolarization measure-ments at 532 nm. Even if the experimental condi-tions on site are difficult with some extremeatmospheric conditions �temperature ranging from�50 °C to �10 °C, maximum wind speed to as greatas 200 km�h, two thirds of the atmospheric pressureat sea level�, the system has shown reliable resultswith measurements on a regular basis. The dataacquisition is performed both in analog and inphoton-counting modes, thus allowing the aerosol re-trieval over a higher vertical range. The depolariza-tion that is a good indicator of the aerosol shape in theatmosphere was demonstrated with an excellent re-jection ratio �essentially no optical cross talks in thedetection box between the parallel and the perpen-dicular channels�, thanks to the use of a Wollastonprism as a polarizer. Range-resolved water-vapormeasurements were demonstrated as well with anachievable range of approximately 10 km ASL by useof the short-range telescope only. Regular measure-ments will soon be performed with both telescopesimultaneously �see Fig. 1�, thus significantly in-creasing the range of observation for both the aerosoland the water-vapor content in the higher atmo-sphere.
The authors kindly thank the Foundation Jung-fraujoch Gornergrat for the different research facili-ties at Jungfraujoch, the Swiss MeterologicalInstitute �Payerne Station� for the radio-soundingdata and the use of meteorological data from Jung-fraujoch, and the Swiss National Foundation and theSwiss Federal Office for Education and Science forfinancial support of the lidar system and its imple-mentation in the European EARLINET lidar net-work.
References1. International Panel for Climate Change, “Climate change
2001—the scientific basis,” available at www.ipcc.ch.2. R. J. Charlson and J. Heintzenberg, ed., Aerosol Forcing of
Climate �Wiley, New York, 1995�.3. J. Houghton, Global Warming: The Complete Briefing �Cam-
bridge U. Press, Cambridge, England, 1997�.4. M. T. Chahine, “The hydrological cycle and its influence on
climate,” Nature 359, 373–380 �1992�.5. D. O. Starr and S. H. Melfi, ed., The Role of Water Vapor in
Climate: A Strategic Research Plan for the Proposed GEWEX,Water Vapor Project �GVaP�, NASA Conf. Publ. CP-3120�NASA, Hanover, Md., 1991�.
6. D. Rind, “Just add water vapor,” Nature 281, 1152–1153�1998�.
7. J. Bosenberg, A. Ansmann, J. Baldasano, D. Balis, C. Bock-mann, B. Calpini, A. Chaikovsky, P. Flamant, A. Hagard, V.
Mitev, A. Papayannis, J. Pelon, D. Resendes, J. Schneider, N.Spinelli, T. Trickl, G. Vaughan, G. Visconti, and M. Wiegner,“EARLINET: a European aerosol research lidar network,” inAdvances in Laser Remote Sensing, J. Pelon, C. Loth, and A.Dabas, eds. �Editions de l’Ecole polytechnique, Palaiseau,France, 2001�, pp. 155–158.
8. J. Bosenberg, ed., “EARLINET scientific report, February 2000to January 2001,” available at http:��lidarb.dkrz.de�earlinet�scirepl.pdf.
9. U. Baltensperger, H. W. Gaggeler, D. T. Jost, M. Lugauer, M.Schwikowski, E. Weingartner, and P. Seibert, “Aerosol clima-tology at high-alpine site Jungfraujoch, Switzerland,” J. Geo-phys. Res. D102, 19707–19715 �1997�.
10. L. Delbouille and G. Roland, “High-resolution solar and atmo-spheric spectroscopy from the Jungfraujoch high-altitude sta-tion,” Opt. Eng. 34, 2736–2739 �1995�.
11. M. de Maziere, M. van Roozendael, C. Hermans, P. C. Simon,P. Demoulin, and G. Roland, “Quantitative evaluation of thepost-Pinatubo NO2 reduction and recovery, based on 10 yearsof FTIR and UV-visible spectroscopic measurements at theJungfraujoch,” J. Geophys. Res. 103, 10849–10858 �1998�.
12. A. Heimo, R. Philipona, C. Frohlich, C. Marty, and A. Ohmura,“The Swiss atmospheric radiation monitoring networkCHARM,” in Proceedings of the WMO Technical Conference onMeteorological and Environmental Instruments and Methodsof Observation �World Meteorological Organization, Geneva,Switzerland, 1998�.
13. D. Maier, N. Kampfer, W. Amacher, M. Wuthrich, J. de la Noe,P. Ricaud, P. Baron, G. Beaudin, C. Viguerie, J.-R. Pardo, J. D.Gallego, A. Barcia, J. Cernicharo, B. Ellison, R. Siddans, D.Matheson, K. Kunzi, U. Klein, B. Franke, J. Louhi, J. Mallat,M. Gustafsson, A. Raisanen, and A. Karpov, “EMCOR radiom-eter: calibration and first tests,” in Microwave Remote Sensingof the Atmosphere and Environment, T. Hayasaka, D. L. Wu,Y.-Q. Jin, J.-S. Jiang, eds., Proc. SPIE 3503, pp. 362–374.
14. W. B. Grant, E. Browell, R. T. Menzies, K. Sassen, and C. Y.She, ed., Selected Papers on Laser Applications in RemoteSensing �SPIE, Bellingham, Wash. 1997�.
15. R. M. Measures, Laser Remote Sensing: Fundamentals andApplications �Krieger, New York, 1992�.
16. F. G. Fernald, B. M. Herman, and J. A. Reagan, “Determina-tion of aerosol height distributions by lidar,” J. Appl. Meteorol.11, 482–489 �1972�.
17. F. G. Fernald, “Analysis of atmospheric lidar observations:some comments,” Appl. Opt. 23, 652–653 �1984�.
18. A. Ansmann, M. Riebesell, and C. Weitkamp, “Measurement ofatmospheric aerosol extinction profiles with a Raman lidar,”Opt. Lett. 15, 746–748 �1990�.
19. A. Ansmann, M. Riebesell, U. Wandinger, C. Weitkamp, E.Voss, W. Lahmann, and W. Michaelis, “Combined Ramanelastic-backscatter lidar for vertical profiling of moisture, aero-sol extinction, backscatter, and lidar ratio,” Appl. Phys. 55,18–28 �1992�.
20. D. N. Whiteman, S. H. Melfi, and R. A. Ferrare, “Raman lidarsystem for the measurement of water vapor and aerosols in theEarth’s atmosphere,” Appl. Opt. 31, 3068–3082 �1992�.
21. S. R. Pal and A. I. Carswell, “Polarization properties of lidarbackscattering from clouds,” Appl. Opt. 12, 1530–1535 �1973�.
22. W. R. McNeil and A. I. Carswell, “Lidar polarization studies ofthe troposphere,” Appl. Opt. 14, 2158–2168 �1975�.
23. C. Bockmann and J. Wauer, “Algorithms for the inversion oflight scattering data from uniform and non-uniform particles,”Aerosol Sci. 32, 49–61 �2001�.
24. C. Bockmann, “Hybrid regularization method for the ill-posedinversion of multiwavelength lidar data in the retrieval ofaerosol size distribution,” Appl. Opt. 40, 1329–1342 �2001�.
25. V. Simeonov, G. Larcheveque, P. Quaglia, H. Van den Bergh,
and B. Calpini, “The influence of the photomultiplier spatialuniformity on lidar signals,” Appl. Opt. 38, 5186–5190 �1999�.
26. S. Nyeki, U. Baltensperger, I. Colbeck, D. T. Jost, E. Wein-gartner, and H. W. Gaggeler, “The Jungfraujoch high-alpineresearch station �3454 m� as a background clean continentalsite for the measurement of aerosol parameters,” J. Geophys.Res. 103, 6097–6107 �1998�.
27. J. M. Vaughan, N. J. Geddes, P. H. Flamant, and C. Flesia,Establishment of a Backscatter Coefficient and AtmosphericDatabase �Defence Evaluation and Research Agency UK,Farnborough, UK, 1998�.
28. United States Committee on Extension to the Standard Atmo-sphere, U.S. Standard Atmosphere, �National Oceanic and At-mospheric Administration, Washington, D.C., 1976�.
29. A. Papayannis, V. Amoiridis, J. Baldasano, J. Balin, D. Balis,A. Boselli, A. Chaikovsky, B. Chatenet, G. Chourdakis, V.Freudenthaler, M. Frioux, J. Herman, M. Iarlori, S. Kreipl, G.Larcheveque, R. Matthey, I. Mattis, D. Muller, M. Pandolfi, G.Pappalardo, J. Pelon, M. R. Perrone, V. Rizi, A. Rodriguez, L.Sauvage, P. Sobolewski, N. Spinelli, F. de Tomasi, T. Trickl,and M. Wiegner, “Continental-scale vertical profile measure-ments of free tropospheric Saharan dust particles performedby a coordinated ground-based European lidar network �EAR-LINET project�,” submitted to J. Geophys. Res.
30. F. Theopold and J. Bosenberg, “Evaluation of DIAL measure-ment in presence of signal noise,” in Proceedings of the 14th
International Laser Radar Conference �Istituto di Ricerca sulleOnde Elettromagnetiche, Consiglio Nazionale delle Ricerche,Florence, Italy, 1988�.
31. S. Godin, A. I. Carswell, D. P. Donovan, H. Claude, W. Stein-brecht, I. S. McDermid, T. J. McGee, M. R. Gross, H. Nakane,
D. P. J. Swart, H. B. Bergwerff, O. Uchino, P. von der Gathen,and R. Neuber, “Ozone differential absorption lidar algorithmintercomparison,” Appl. Opt. 38, 6225–6236 �1999�.
32. P. B. Russell, T. J. Swissler, and M. P. McCormick, “Method-ology for error analysis and simulation of lidar aerosol mea-surements,” Appl. Opt. 18, 3783–3797 �1979�.
33. J. Taylor, An Introduction to Error Analysis: the Study ofUncertainties in Physical Measurements �University Science,Sausalito, Calif. 1997�.
34. A. Ansmann, U. Wandinger, M. Riebesell, C. Weitkamp, andW. Michaelis, “Independent measurement of extinction andbackscatter profiles in cirrus clouds by using a combined Ra-man elastic-backscatter lidar,” Appl. Opt. 31, 7113–7131�1992�.
35. D. N. Whiteman, K. D. Evans, B. Demoz, D. O. C. Starr, D. C.Tobin, W. E. Feltz, G. J. Jedlovec, S. I. Gutman, G. K. Schwem-mer, M. Cadirola, S. H. Melfi, and F. J. Schmidlin, “Ramanlidar measurements of water vapor and cirrus clouds duringthe passage of hurricane Bonnie,” J. Geophys. Res. 106, 5211–5225 �2000�.
36. J. Bosenberg, “Ground-based differential absorption lidar forwater-vapor and temperature profiling: methodology,” Appl.Opt. 37, 3845–3860 �1998�.
37. E. J. McCartney, Optics of the Atmosphere �Wiley, New York,1976�.
38. I. Balin, G. Larcheveque, P. Quaglia, V. Simeonov, H. vanden Bergh, and B. Calpini, “Water vapor profile by Ramanlidar in the free troposphere from the Jungfraujoch AlpineStation,” in Advances in Global Change Research, M. Benis-ton, ed., �Kluwer Academic, Dordrecht, 2001�, Vol. 10, pp.123–138.
