• Banakh, V. A., , Smalikho I. N. , , Köpp F. , , and Werner C. , 1995: Representativity of wind measurement with a cw Doppler lidar in the atmospheric boundary layer. Appl. Opt, 34 , 20552067.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beniston, M., 1998: From Turbulence to Climate. Springer-Verlag, 328 pp.

  • Betout, P., , Ch. Werner , , and Burridge D. , 1989: Atmospheric Laser Doppler Instrument (ALADIN). ESA Lidar Working Group Rep. SP-1112, 64 pp.

    • Search Google Scholar
    • Export Citation
  • Bilbro, J. W., 1980: Atmospheric laser Doppler velocimetry: An overview. Opt. Eng, 19 , 533542.

  • Bilbro, J. W., , DiMarzio C. , , Fitzjarrald D. , , Johnson S. , , and Jones W. , 1986: Airborne Doppler lidar measurements. Appl. Opt, 25 , 39523960.

  • Chanin, M. L., , Garnier A. , , Hauchecorne A. , , and Porteneuve J. , 1989:: A Doppler lidar for measuring winds in the middle atmosphere. Geophys. Res. Lett, 16 , 12731276.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chanin, M. L., , Cress A. , , and Wergen W. , 2001: Impact of wind profile observations on the German Weather Service's NWP system. Meteor. Z.,. . 10 , 91101.

    • Search Google Scholar
    • Export Citation
  • Curran, R. J., and and Coauthors, 1987: Laser Atmospheric Wind Sounder (LAWS). NASA Instrument Panel Rep. Vol. IIg, 134 pp.

  • Garnier, A., , and Chanin M. L. , 1992: Description of a Doppler Rayleigh lidar for measuring winds in the middle atmosphere. Appl. Phys, B55 , 3540.

    • Search Google Scholar
    • Export Citation
  • Hollingsworth, A., , and Lönnberg P. , 1987: The verification of objective analysis: Diagnostics of analysis system performance. ECMWF Tech. Rep. 142, 223 pp.

    • Search Google Scholar
    • Export Citation
  • Korb, C. L., , Gentry B. M. , , Li S. X. , , and Flesia C. , 1998: Theory of the double-edge technique for Doppler lidar wind measurements. Appl. Opt, 37 , 30973104.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lorenz, E., 1982: Atmospheric predictability experiments with a large numerical model. Tellus, 34 , 505513.

  • McGill, M. J., , Skinner W. R. , , and Irgang T. D. , 1997: Analysis techniques for the recovery of winds and backscatter coefficients from a multiple channel incoherent Doppler lidar. Appl. Opt, 36 , 12531268.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McGill, M. J., , Hart W. D. , , McKay J. A. , , and Spinhirne J. D. , 1999: Modeling the performance of direct-detection Doppler lidar systems including cloud and solar background variability. Appl. Opt, 38 , 63886397.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Post, M. J., , and Cupp R. E. , 1990: Optimization a pulsed Doppler lidar. Appl. Opt, 29 , 41454157.

  • Rees, D., , and McDermid I. S. , 1990: Doppler lidar atmospheric wind sensor: Reevaluation of a 355-nm incoherent Doppler lidar. Appl. Opt, 29 , 41334144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rees, D., , Nelke G. , , Fricke K-H. , , von Zahn U. , , Singer W. , , von Cossert G. , , and Lloyd N. D. , 1996: The Doppler wind and temperature system of the Alomar lidar. J. Atmos. Terr. Phys, 58 , 18271842.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stoffelen, A., , Becker B. , , Eyre J. , , and Roquet H. , 1994: Theoretical studies of the impact of Doppler wind data: Preparation of a database. ESA Rep. CR(P)-3943, 187 pp.

    • Search Google Scholar
    • Export Citation
  • Streicher, J., , Leike I. , , and Werner C. , 1998: ALIENS: Atmospheric lidar end-to-end simulator. Proc. Fifth Int. Symp. on Atmospheric and Ocean Optics, Tomsk, Russia, International Society for Optical Engineering, 380–386.

    • Search Google Scholar
    • Export Citation
  • Vinnichenko, N. K., , Pinus N. Z. , , Shmetter S. M. , , and Shur G. N. , 1973: Turbulence in the Free Atmosphere. Consultants Bureau, 287 pp.

  • Werner, C., , Wildgruber G. , , and Streicher J. , 1991: Representativity of wind measurements from space. European Space Agency Contract 8664/90/HGE-1, 89 pp.

    • Search Google Scholar
    • Export Citation
  • Werner, C., and and Coauthors, 2001: Wind Instrument. Opt. Eng., in press.

  • Winzer, P., , Leeb W. , , Leike I. , , Streicher J. , , and Werner Ch , 1997: Coherent detection at low photon number per measurement interval (DELPHI). ESA/ESTEC Rep. Contract 11733/95/NL/CN, 157 pp.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Block diagram of the virtual instrument simulation program ALIENS for coherent detection (left column) and IALIENS for direct detection (right column)

  • View in gallery

    Front panel of signal.vi, computing the return signal. There are different boxes for the selection of the lidar parameters from the left: atmosphere, instrument position, transmitter, lidar parameters, the selected pulse profile, and the received optical signal in the right corner. The distance from the satellite to the ground for the Nadir angle 30° is 561 km; the optical signal is visible in the lowest 3 km

  • View in gallery

    Front panel of ialiens.vi. There are different boxes for the selection of the lidar parameters from the left: atmosphere, transmitter, and lidar parameters. The detection chain is selectable, as are the filter parameters (see Fig. 4). The wind LOS profile is shown on the right with the true wind and the detected photons from the two channels

  • View in gallery

    Steps for direct detection Doppler lidar (molecular return). First, action of the medium resolution filter; second, action of the high-resolution, dual-channel Fabry–Perot etalon; and third, action of the high resolution filter on a random spectrum

  • View in gallery

    Altitude vs transmission (clear atmosphere for the wavelength 0.35 μm)

  • View in gallery

    Altitude vs backscatter coefficients (clear atmosphere for the wavelength 0.35 μm). Black line corresponds to molecules, gray line to aerosols

  • View in gallery

    LOS components as a result of the simulation of a 15-J CO2 laser. Black line corresponds to mean true data from 57 shots, gray line in left figure corresponds to simulated LOS wind speed, gray line in right figure is one realization of turbulent wind speed

  • View in gallery

    Comparison of WIND profiles measured with the airborne Doppler lidar WIND with the simulated data using the DWD LM

  • View in gallery

    Measured LOS components as color scale vs scan angle and altitude: (a) measurement and (b) simulated using wind field given by the DWD and with the WIND parameters (0.1-J output energy and 1-MHz chirp). (c) Same as (b), but with 0.3-J and 0.2-MHz chirp

  • View in gallery

    Simulated wind LOS components vs scan angle in 8-km altitude for (a) an improved system (0.3-J and 0.2-MHz chirp) and (b) WIND Doppler lidar (0.1-J and 1-MHz chirp)

  • View in gallery

    Profiles of LOS wind components vs height and distance between 30.0°N, 117.3°W and 51.63°N, 75.0°W for 1200 UTC 20 Jan. (a) True wind LOS from the global model of the DWD, (b) CO2-Doppler lidar (see Table 1), (c) 2-μm Doppler lidar (Table 1), and (d) UV direct detection Doppler lidar (Table 1)

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 55 55 3
PDF Downloads 14 14 3

Virtual Doppler Lidar Instrument

View More View Less
  • 1 Institute of Atmospheric Physics, DLR, Wessling, Germany
  • | 2 Institute of Atmospheric Optics, Russian Academy of Sciences, Tomsk, Russia
  • | 3 German Weather Service, Offenbach, Germany
© Get Permissions
Full access

Abstract

Doppler lidars measure the range-resolved line-of-sight wind component by extracting the Doppler shift of radiation backscattered from atmospheric aerosols and molecules. A virtual instrument was developed to simulate wind measurements by flying virtually over the atmosphere. The atmosphere contains all components that influence the lidar, that is, wind, turbulence, aerosols, clouds, etc. For a selected time period, a dataset of the atmospheric conditions from the global model and the local model was provided by the German Weather Service. Three different Doppler lidar systems were simulated for this report: a coherent airborne conical scanning 10-μm Doppler lidar, a 10-μm and a 2-μm spaceborne system, and a spaceborne incoherent ultraviolet Doppler lidar.

Corresponding author address: Dr. Christian Werner, DLR-Institute of Atmospheric Physics, P.O. Box 1116, D-82230 Wessling, Germany. Email: christian.werner@dlr.de

Abstract

Doppler lidars measure the range-resolved line-of-sight wind component by extracting the Doppler shift of radiation backscattered from atmospheric aerosols and molecules. A virtual instrument was developed to simulate wind measurements by flying virtually over the atmosphere. The atmosphere contains all components that influence the lidar, that is, wind, turbulence, aerosols, clouds, etc. For a selected time period, a dataset of the atmospheric conditions from the global model and the local model was provided by the German Weather Service. Three different Doppler lidar systems were simulated for this report: a coherent airborne conical scanning 10-μm Doppler lidar, a 10-μm and a 2-μm spaceborne system, and a spaceborne incoherent ultraviolet Doppler lidar.

Corresponding author address: Dr. Christian Werner, DLR-Institute of Atmospheric Physics, P.O. Box 1116, D-82230 Wessling, Germany. Email: christian.werner@dlr.de

1. Introduction

One of the major deficiencies of the current meteorological network is the lack of global wind data. New approaches will be necessary, among which the use of spaceborne instrumentation is notable. One candidate has such a sensor is the Doppler lidar (Curran et al. 1987; Betout et al. 1989), Doppler lidars measure the range-resolved line-of-sight (LOS) wind component by extracting the Doppler shift of radiation backscattered from atmospheric aerosols (Bilbro 1980; Bilbro et al. 1986) or from molecules (Garnier and Chanin 1992). Range resolution is in the order of a few hundred meters.

Several studies were performed to show the impact of Doppler lidar wind data from space on numerical weather prediction. At European Centre for Medium-range Weather Forecasts a database of simulated observations (Stoffelen et al. 1994) has been created. Different observing system simulation experiments showed the usefulness of a Doppler lidar in space (Lorenz 1982; Hollingworth and Lönneberg 1987).

A virtual instrument was developed to produce LOS wind components. This instrument is based upon the Delphi study of the European Space Agency (ESA; Winzer et al. 1997) and the improved Atmospheric Lidar End-to-End Simulator (ALIENS; Streicher et al. 1998). By adding a scanning procedure and platform movement, one gets a new, dynamic version of the virtual instrument. This requires a 3D model of the atmosphere. In this paper a short description of the virtual instrument is given (section 2), and information on the airborne (section 3) and spaceborne (section 4) versions follow.

2. Virtual instruments

a. General

Virtual instruments provide a powerful tool to test and investigate sensor performances under various environmental and instrumental conditions without hardware development. Also, sets of simulated data can be used for further tests and investigations, such as signal processor development or (as in this case) impact of data on weather forecasting. Our simulation tool is written in LabVIEW (National Instruments). Figure 1 shows a block diagram of the main modules.

Parameters of the Doppler lidar instrument (e.g., laser wavelength, pulse shape and power, transceiver characteristics, etc.) can be chosen either for heterodyne detection or for direct detection. The pulse shape is either Gaussian (for a solid-state laser) or similar to a gain-switched spike (for a CO2 laser). One can choose platform such as an aircraft, satellite, or ground-based station together with its parameters (see, e.g., Fig. 2).

The atmosphere is divided into slices of height intervals (1.5 m as lowest value), where the optical beam is extincted and scattered. Clouds strongly influence the extinction and scattering. The frequency is shifted due to the LOS component of the prevailing wind velocity through the laser focus volume (Post and Cupp 1990).

The optical signal is specular. Speckles result from destructive and constructive interference of waves, scattered by randomly distributed particles. Due to the stochastic simulation, temporal and spatial speckles appear in the signal. Different shots into the same atmosphere therefore lead to different return signals based on the random distribution of the scatterers.

For a coherent instrument, the signal-field generation is followed by optical mixing with the local oscillator field on the detector. This was performed with the additive Gaussian noise approximation (AGNA) module, which was developed by the Vienna University of Technology (Winzer et al. 1997). Digitization and signal processing are simulated after that. A number of different estimators like pulse pair (PP) or maximum likelihood can be selected. The result of the simulation is the comparison of calculated wind profile with the input wind field. The direct-detection virtual instrument works with two techniques: the double-edge technique (Chanin et al. 1989) for the detection of the molecular signal, and multichannel Fizeau receiver for the aerosol signal. A detector chain [charge coupled device (CCD) array] with the relevant errors follows, and after that, accumulation and wind speed estimation. A dynamic version of ALIENS produces signals for a given flight path or satellite track.

b. Heterodyne detection

Figure 2 shows the front panel with the input and output parameters grouped together in boxes of common relation. The two graphs on the right show the pulse monitor (top), which is the signal of the laser source, and the atmospheric return signal (bottom)—both are in units of optical power versus time or range, and both are modulated in the intermediate frequency band.

In this example the pulse monitor shows a typical gain-switched CO2 laser pulse with initial spike, whereas the atmospheric return, here for a spaceborne system, reflects the behavior of the atmosphere. The last 5-km of the measurement path contains the highest amount of scatterers (dust particles of the lower atmosphere); therefore, the amplitude increases.

The tool further includes

  • satellite/airborne/ground geometry calculations (position and scan pattern),
  • arbitrary atmospheric conditions (wind field, backscatter profiles, and clouds), and
  • arbitrary laser source (pulse shape and frequency chirp).

The signal generation is followed by a sophisticated modeling of the heterodyne receiver front end. An important feature of the model is its validity for an extremely low signal-to-noise ratio, that is, for a few signal photons per measurement interval. It consists of a balanced receiver with two photodetectors, an electronic circuit subtracting the two currents yielding the beating term, and an electronic preamplifier.

This signal now includes all noise contributions in our example (heterodyne detection), mainly consisting of shot noise (tiny quantum fluctuations in the electromagnetic field of the local oscillator plus the atmospheric return radiation).

This module further takes into account

  • nonideal beam splitter and photo mixer,
  • degree of polarization mismatch,
  • wave front distortion due to nonideal optical components,
  • heterodyne efficiency (amplitude and phase mismatch between signal and local oscillator),
  • local oscillator amplitude and phase noise,
  • detector dark currents, and
  • shot and excess noise.

The processing unit includes a virtual analogue-to-digital (A/D) converter and the frequency analysis subroutines.

The process of A/D conversion is virtually simulated—the data values exist inside the computer as digital information only—but the program pretends to handle the signal coming from the AGNA detector as if it was an electrical current converted directly into the digital domain, including all errors induced by that process. These errors are due to

  • the finite sampling rate (Nyquist theorem),
  • amplitude steps (bit noise), and
  • bit fluctuations (swapping bits).

The next step in the digital domain is the demodulation of the atmospheric return signal using the laser source frequency, which may differ from shot to shot (jitter), plus the local oscillator offset frequency. This is done by a complex multiplication in the time domain and low-pass filtering of the signal. Finally a frequency analysis is applied to the signal to find the Doppler shift in a LOS direction in respect to the laser beam.

There are several frequency estimators implemented, working either in the time domain or in the frequency domain:

  • Fourier transform (peak finder),
  • PP and polypulse pair,
  • adaptive notch filter, and
  • maximum likelihood.

Accumulation of subsequent shots can be included optionally.

c. Direct detection

Figure 3 shows the front panel of the direct detection virtual instrument with all relevant parameters. The lidar equation is used to calculate the mean number of received aerosol and molecular photons from one slice. It is assumed that the photon number obeys Poisson statistics. The mean number of photons is then varied according to the Poisson distribution. There were a lot of different approaches for the direct detection (Chanin et al. 1989; McGill et al. 1999; Rees and McDermid 1990; McGill et al. 1997; Rees et al. 1996); we selected a kind of double-edge technique for the simulation (Korb et al. 1998). Each photon carries a random wavelength according to the aerosol/molecular spectrum. The aerosol spectrum is assumed to be Gaussian with full-width half-maximum (FWHM) equal to the laser FWHM linewidth. The molecular spectrum is also Gaussian (Brillouin scattering is neglectable for a wavelength of 355 nm). Its width is given by the prevailing temperature. The photon wavelength is shifted according to the wind speed.

The Doppler shift occurs either on molecules or on aerosols. For the UV-wavelength, the molecular backscatter dominates. Caused by the Brownian motion, the spectrum of the molecular signal is much broader than the spectrum of the aerosol signal. After a diplexer, the received signal is split into two channels, one aerosol and one molecular.

1) Aerosol channel

Polarization, low resolution, and medium resolution filters influence photons only via the transmission factor. A photon is transmitted with probability τ, where τ is the transmission factor.

The aerosol receiver is a Fizeau high-resolution multichannel etalon. Its transmission is assumed to have a repeated Lorenzian shape. The spectral range imaged onto the detector (CCD array) is assumed to be less than 1 order of interference, such that only one Lorenzian transmission curve has to be taken into account. Furthermore it is assumed that the peak is imaged onto the center of the CCD array for zero wind speed. A signal photon is transmitted into channel i with probability τi(λ), where τi(λ) is the wavelength-dependent transmission factor. Background photons reaching the aerosol receiver are taken into account via their equivalent bandwidth. This includes the repeated Lorenzian structure of the filters. The electrons are collected by the CCD structure for 50 subsequent shots. Due to detector dark current, readout noise, preamplifier noise, electronic noise of the digitalization chain, and digitalization noise, additional electrons are added on each channel. The number of electrons in a channel are added for 14 series of 50 shots to yield a return of 700 shots altogether.

The resulting pattern is analyzed and if a useful maximum is detected, the energetic centroid of the seven channels around the signal peak is computed. The result is a measure of the wind velocity.

Finally, additional sources of measurement errors (high-resolution etalon temperature and optomechanical stability, calibration errors, alignment errors, and Rayleigh background curvature) are taken into account.

2) Molecular channel

Figure 3 shows the action of optical components on the backscattered signal for the molecular channel. The medium resolution filter reduces the number of signal photons (molecules and aerosols) due to the wavelength-dependent reflection coefficient. A photon is reflected with probability 1 − τ, where τ is the transmission factor (Fig. 4a). The molecular receiver is a Fabry–Perot high-resolution, dual-channel etalon (Fig. 4b). The center wavelengths of the two channels are assumed to be situated symmetrically with respect to the laser frequency. The number of photons reaching the receiver depends on the wavelength-dependent transmission of the Fabry–Perot etalon. A photon is transmitted with probability τ(λ), where τ(λ) is the wavelength-dependent transmission factor.

The transmission curve is assumed to have a Lorenzian shape. Each transmitted photon produces an electron (with some probability, given by the detector quantum efficiency). The photons are distributed with probability 1/2 to either of the two channels.

The electrons are collected by the CCD structure for 50 subsequent shots. Due to detector dark current, readout noise, preamplifier noise, electronic noise of the digitalization chain, and digitalization noise, additional electrons are added on each channel. This procedure is repeated 14 times and the electrons on either channel are added to yield the return for altogether 700 shots. Due to the symmetrical arrangement of the etalons, the number of photons reaching the receiver in either channel is equal to zero wind speed (in absence of noise). The shift of the spectrum according to the wind speed results in a different number of electrons registered by the two channels.

A calibration curve (detector response) has been computed beforehand using smooth (not random) Rayleigh spectra for different temperatures and LOS wind speeds. To find the actual wind speed, the temperature of a range gate has to be determined. In this simulation, the mean temperature of the range gate is assumed to be known. Then due to the detector response curve, an LOS wind speed is selected. Finally, additional sources of measurement errors (e.g., medium-resolution filter stability, calibration errors, alignment errors) are taken into account.

d. Atmospheric model

The atmosphere is divided into small slices of frozen atmosphere (wind speed and temperature are constant for each slice) of variable depth (here 15 m is sufficient). Real atmospheric values for wind speed, temperature, pressure, transport coefficients, and coverage coefficients provided by the German Weather Service (DWD) for a period of 10 days (19–30 January 1998) around the globe are taken into account. The data are organized on a net of points, separated by 1.125° longitude direction and 1.121° latitude. The vertical spacing of the data in 20 layers lies between 0 and 30 km and is not equidistant, but closer near the earth's surface. Wind speed fluctuations (turbulence) according to real atmospheric data (calculated from current temperature, pressure, and transport coefficients) are taken into account. Fluctuations of subsequent shots are correlated. Backscatter coefficients and transmission are taken from the ESA Atmosphere Dynamics Mission (ADM) data (at this time only a median atmospheric model).

Clouds are also taken into account: absorption and transmission of various cloud types according to the ESA ADM data, and cloud coverage due to real atmospheric data. Figures 5 and 6 illustrate the atmospheric components for clear atmosphere for the UV wavelength of the direct-detection Doppler lidar.

e. Wind turbulence

A 2D turbulent wind field in a vertical plane oriented in the direction of the azimuth angle θ is simulated on the base of the Karman model of turbulence with parameters depending on height (Banakh et al. 1995). The Karman model parameters, namely, the variance of wind velocity fluctuations and the outer scale of turbulence, are specified differently for the surface layer, for the troposphere up to the height of 10 km, and for a free atmosphere above 10 km. In the surface layer the calculation of these parameters is based on the knowledge of the dissipation rate of turbulent kinetic energy. This dissipation rate is estimated in accordance with the turbulent kinetic energy (TKE) parameterization of the boundary layer of the atmosphere (Beniston 1998) using DWD data of the (transfer coefficients, temperature, and wind velocity) for the first three layers. For the free atmosphere above 10 km the relative wind velocity variance and the outer scale have constant values as shown by empirical data (Vinnichenko et al. 1973). They are given by corresponding empirical constants. For the troposphere between the surface layer and 10 km, the necessary parameters are approximated by the available models in such a way that the velocity variance and outer scale at the boundaries of the tropospheric range take the values calculated for the surface layer and free atmosphere.

A measurement under such turbulent conditions with a resolution of 1000 m would yield many different frequency contributions and therefore produce a mean frequency output with a broadened peak. Accumulation of shots is necessary. The turbulence simulation leads to a modification of the DWD wind profile, as shown in Fig. 7.

3. Airborne version of the virtual instrument: Comparison with experimental data

For the application of the virtual instrument, the parameters of the airborne Wind Infrared Doppler lidar (WIND) were used (Werner et al. 2001). WIND was developed within a common project by CNRS/CNES/DLR and is an airborne coherent infrared Doppler lidar for wind velocity measurement. The system is based on pulsed CO2 laser technology, heterodyne detection, and a conical scanning system. The validation was performed using the Windprofiler of the Meteorological Observatory Lindenberg and with the forecast model (LM) of the German Weather Service. The flight was performed at 1330 UTC on 12 October 1999. The local model provides data for all the levels in the atmosphere with the same information as for the global model within a finer grid.

The virtual instrument is attached to the atmospheric data (LM) for 1330 UTC and provides simulated LOS wind components by virtually flying over the local model atmosphere. One can compare simulated signals with the measured signals.

Figure 8 shows the result. The data are profiles of the horizontal wind accumulated over five conical scans (i.e., over 100-s measurement time or 10 × 50 km grid close to the ground). For the flight path a very strong wind appears in the flight level. The figure shows that measured and simulated data are in good accordance.

This comparison with the virtual instrument can be used to improve the hardware. The measurement values in Fig. 8 are placed near the ground and at an altitude between 8 and 10 km. This was caused by the limited performance of the system. With the available laser output energy, a good signal-to-noise ratio can only be achieved at cloud level or in the boundary layer. This can be simulated using the virtual instrument. Figure 9 shows three conical scans—one from the experiment and two from the virtual instrument with similar parameters to the experiment and with slightly improved parameters.

One can see the simulation (Fig. 9b) meets the measurement (Fig. 9a) and the improvement (Fig. 9c) of the system could fill the gaps in Fig. 7. In one level at 8 km (Figs. 9b,c) a comparison was made to show the scanning and the velocity azimuth display procedure. Figure 10 shows two simulations: one for the improved system (Fig. 10a) and one for the normal WIND system (Fig. 10b).

Figure 10 includes the sine-wave fitting to get the wind speed and wind direction for one level at an 8-km height (selectable using the cursor shown in Fig. 9). From the measurements one gets 37 m s−1 from 330° (Fig. 8). The simulation (Fig. 9) also shows the LOS values, which are much noisier for the WIND parameters (Fig. 9a), compared to the simulated system parameters (Fig. 9b). For Fig. 8, all data points with an error larger than 2 m s−1 were rejected.

4. Spaceborne Doppler lidar virtual instrument

Again, a dynamical version of the Doppler lidar simulation tool is used to provide sets of simulated LOS wind measurements from space. The sensor is based on the ALIENS virtual instrument with an addition—that the sensor can fly over a selected region. DWD global model atmospheric data (winds, cloud coverage, temperature, and transport coefficients) of the time period (1200 UTC 19 January–1200 UTC 30 January 1998) were used as an input. A target region over North America was selected and all measurements falling into that region were simulated with three different Doppler lidar systems, two heterodyne detection systems, and one direct detection system in the UV-wavelength region. Table 1 shows the sensor parameters. The orbit height was 400 km for all the lidar systems.

A nonscanning system concept was used. Werner et al. (1991) showed that LOS wind components are useful per se. By using one single LOS component, one needs the knowledge of the wind direction from the models. The so-called meteorological first guess of the wind velocity is used to get the horizontal wind vector out of the single LOS component. This was tested during the impact study and is published (Cress and Wergen 2001).

The true LOS component value was compared with the calculated one. The results for the three simulated Doppler lidars are shown in Figs. 11a–d. Wind moving array from the instrument appears in red and wind moving toward the instrument is blue.

Figure 11a gives the true LOS values for one track over North America performed on 1200 UTC 20 January 1998. One can identify the information losses for the CO2 system (Fig. 11b) below cloud layers and above 20 km. The 2-μm Doppler lidar (Fig. 11c) gets most of the information in the cloud layer and in the boundary layer. The UV system (Fig. 11d) is similar to the large CO2 system but gets data in the boundary layer or below clouds, which may be affected by multiple scattering.

In order to estimate the potential benefit of the additional LOS lidar wind profiles on the numerical weather forecasts, an impact study was initiated. It consists of two parts. In the first study, the importance of wind profile data in general was assessed. In the second part, the impact of the additional lidar winds will be estimated.

The basic prognostic variables of current numerical weather prediction (NWP) models are geopotential, humidity, and wind. In the extratropics, there exists a strong coupling between the geopotential and the wind field—the so-called geostrophic relation. It allows for the derivation of approximate wind components from geopotential gradients. It is therefore important to find out whether these geostrophic winds are sufficient or whether observed wind profiles result in a more realistic definition of the initial state for NWP.

To answer this question, a parallel data assimilation with the DWD global model was performed, in which the wind profile observations from radiosondes and aircraft over the United States and Canada were excluded. By comparing the forecast quality between the runs using or not using the wind data or using lidar wind data instead, an estimate of their importance can be given. By tracing back the differences in the forecasts to the initial difference, sensitive areas can be identified (Cress and Wergen 2001).

5. Summary

The new software package provides a powerful and convenient tool for Doppler lidar simulation. All the instrument parameters can be set individually. This software tool can be used to find the optimal lidar parameters for the envisaged application or to interpret lidar results. The single units of the software package, such as the processing unit, can even be used separately (e.g., to process measured Doppler lidar data). The dynamical versions of the sensor simulation are designed to process input data of a real changing atmosphere. Sets of simulated data have been used for investigations of the impact of lidar wind measurements on numerical weather prediction.

The model assumptions of the simulation have been validated comparing real Doppler lidar data from the airborne WIND system, ground observations, and local model of the German Weather Service.

A demo version can be found online at http://www.op.dlr.de/ipa/LIDAR/WIND/ALIENS/ for either the Macintosh PowerPC (aliens.sit.bin) or Windows 95/NT (aliens.zip) operating system. See also http://www.op.dlr.de/ipa/LIDAR/WIND/impact.html and http://www.op.dlr.de/ipa/LIDAR/WIND/localmodel.html.

Acknowledgments

Some of the modules have been originated from the ESA study of Coherent Detection at Low Photon Number Per Measurement Interval (DELPHI) Contract 11733/95/NL/CN. The heterodyne receiver front end was developed by Peter Winzer and Walter Leeb from Technische Universitaet Wien, Institut fuer Nachrichtentechnik und Hochfrequenztechnik. This project was supported by the German Agency for Affairs of Space Travel under Contract EP9624 and the Federal Ministry for Education, Science, Research, and Technology represented by the Project for Biology, Energy, and Environment under Contract 03EE9624. Cooperation with the Institute of Atmospheric Optics in Tomsk was supported by the International Bureau Ru-138 and by the Russion Foundation of Basic Research (Grants 98-05-03131 and 00-05-64033). The direct detection Doppler lidar virtual instrument was developed by ESA Contract CCN 5 to ESTEC Contract 13018/98/NL/GD.

REFERENCES

  • Banakh, V. A., , Smalikho I. N. , , Köpp F. , , and Werner C. , 1995: Representativity of wind measurement with a cw Doppler lidar in the atmospheric boundary layer. Appl. Opt, 34 , 20552067.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beniston, M., 1998: From Turbulence to Climate. Springer-Verlag, 328 pp.

  • Betout, P., , Ch. Werner , , and Burridge D. , 1989: Atmospheric Laser Doppler Instrument (ALADIN). ESA Lidar Working Group Rep. SP-1112, 64 pp.

    • Search Google Scholar
    • Export Citation
  • Bilbro, J. W., 1980: Atmospheric laser Doppler velocimetry: An overview. Opt. Eng, 19 , 533542.

  • Bilbro, J. W., , DiMarzio C. , , Fitzjarrald D. , , Johnson S. , , and Jones W. , 1986: Airborne Doppler lidar measurements. Appl. Opt, 25 , 39523960.

  • Chanin, M. L., , Garnier A. , , Hauchecorne A. , , and Porteneuve J. , 1989:: A Doppler lidar for measuring winds in the middle atmosphere. Geophys. Res. Lett, 16 , 12731276.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chanin, M. L., , Cress A. , , and Wergen W. , 2001: Impact of wind profile observations on the German Weather Service's NWP system. Meteor. Z.,. . 10 , 91101.

    • Search Google Scholar
    • Export Citation
  • Curran, R. J., and and Coauthors, 1987: Laser Atmospheric Wind Sounder (LAWS). NASA Instrument Panel Rep. Vol. IIg, 134 pp.

  • Garnier, A., , and Chanin M. L. , 1992: Description of a Doppler Rayleigh lidar for measuring winds in the middle atmosphere. Appl. Phys, B55 , 3540.

    • Search Google Scholar
    • Export Citation
  • Hollingsworth, A., , and Lönnberg P. , 1987: The verification of objective analysis: Diagnostics of analysis system performance. ECMWF Tech. Rep. 142, 223 pp.

    • Search Google Scholar
    • Export Citation
  • Korb, C. L., , Gentry B. M. , , Li S. X. , , and Flesia C. , 1998: Theory of the double-edge technique for Doppler lidar wind measurements. Appl. Opt, 37 , 30973104.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lorenz, E., 1982: Atmospheric predictability experiments with a large numerical model. Tellus, 34 , 505513.

  • McGill, M. J., , Skinner W. R. , , and Irgang T. D. , 1997: Analysis techniques for the recovery of winds and backscatter coefficients from a multiple channel incoherent Doppler lidar. Appl. Opt, 36 , 12531268.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McGill, M. J., , Hart W. D. , , McKay J. A. , , and Spinhirne J. D. , 1999: Modeling the performance of direct-detection Doppler lidar systems including cloud and solar background variability. Appl. Opt, 38 , 63886397.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Post, M. J., , and Cupp R. E. , 1990: Optimization a pulsed Doppler lidar. Appl. Opt, 29 , 41454157.

  • Rees, D., , and McDermid I. S. , 1990: Doppler lidar atmospheric wind sensor: Reevaluation of a 355-nm incoherent Doppler lidar. Appl. Opt, 29 , 41334144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rees, D., , Nelke G. , , Fricke K-H. , , von Zahn U. , , Singer W. , , von Cossert G. , , and Lloyd N. D. , 1996: The Doppler wind and temperature system of the Alomar lidar. J. Atmos. Terr. Phys, 58 , 18271842.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stoffelen, A., , Becker B. , , Eyre J. , , and Roquet H. , 1994: Theoretical studies of the impact of Doppler wind data: Preparation of a database. ESA Rep. CR(P)-3943, 187 pp.

    • Search Google Scholar
    • Export Citation
  • Streicher, J., , Leike I. , , and Werner C. , 1998: ALIENS: Atmospheric lidar end-to-end simulator. Proc. Fifth Int. Symp. on Atmospheric and Ocean Optics, Tomsk, Russia, International Society for Optical Engineering, 380–386.

    • Search Google Scholar
    • Export Citation
  • Vinnichenko, N. K., , Pinus N. Z. , , Shmetter S. M. , , and Shur G. N. , 1973: Turbulence in the Free Atmosphere. Consultants Bureau, 287 pp.

  • Werner, C., , Wildgruber G. , , and Streicher J. , 1991: Representativity of wind measurements from space. European Space Agency Contract 8664/90/HGE-1, 89 pp.

    • Search Google Scholar
    • Export Citation
  • Werner, C., and and Coauthors, 2001: Wind Instrument. Opt. Eng., in press.

  • Winzer, P., , Leeb W. , , Leike I. , , Streicher J. , , and Werner Ch , 1997: Coherent detection at low photon number per measurement interval (DELPHI). ESA/ESTEC Rep. Contract 11733/95/NL/CN, 157 pp.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Block diagram of the virtual instrument simulation program ALIENS for coherent detection (left column) and IALIENS for direct detection (right column)

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 2.
Fig. 2.

Front panel of signal.vi, computing the return signal. There are different boxes for the selection of the lidar parameters from the left: atmosphere, instrument position, transmitter, lidar parameters, the selected pulse profile, and the received optical signal in the right corner. The distance from the satellite to the ground for the Nadir angle 30° is 561 km; the optical signal is visible in the lowest 3 km

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 3.
Fig. 3.

Front panel of ialiens.vi. There are different boxes for the selection of the lidar parameters from the left: atmosphere, transmitter, and lidar parameters. The detection chain is selectable, as are the filter parameters (see Fig. 4). The wind LOS profile is shown on the right with the true wind and the detected photons from the two channels

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 4.
Fig. 4.

Steps for direct detection Doppler lidar (molecular return). First, action of the medium resolution filter; second, action of the high-resolution, dual-channel Fabry–Perot etalon; and third, action of the high resolution filter on a random spectrum

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 5.
Fig. 5.

Altitude vs transmission (clear atmosphere for the wavelength 0.35 μm)

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 6.
Fig. 6.

Altitude vs backscatter coefficients (clear atmosphere for the wavelength 0.35 μm). Black line corresponds to molecules, gray line to aerosols

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 7.
Fig. 7.

LOS components as a result of the simulation of a 15-J CO2 laser. Black line corresponds to mean true data from 57 shots, gray line in left figure corresponds to simulated LOS wind speed, gray line in right figure is one realization of turbulent wind speed

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 8.
Fig. 8.

Comparison of WIND profiles measured with the airborne Doppler lidar WIND with the simulated data using the DWD LM

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 9.
Fig. 9.

Measured LOS components as color scale vs scan angle and altitude: (a) measurement and (b) simulated using wind field given by the DWD and with the WIND parameters (0.1-J output energy and 1-MHz chirp). (c) Same as (b), but with 0.3-J and 0.2-MHz chirp

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 10.
Fig. 10.

Simulated wind LOS components vs scan angle in 8-km altitude for (a) an improved system (0.3-J and 0.2-MHz chirp) and (b) WIND Doppler lidar (0.1-J and 1-MHz chirp)

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Fig. 11.
Fig. 11.

Profiles of LOS wind components vs height and distance between 30.0°N, 117.3°W and 51.63°N, 75.0°W for 1200 UTC 20 Jan. (a) True wind LOS from the global model of the DWD, (b) CO2-Doppler lidar (see Table 1), (c) 2-μm Doppler lidar (Table 1), and (d) UV direct detection Doppler lidar (Table 1)

Citation: Journal of Atmospheric and Oceanic Technology 18, 9; 10.1175/1520-0426(2001)018<1447:VDLI>2.0.CO;2

Table 1.

Sensor parameters for virtual spaceborne lidars

Table 1.
Save