The atmospheric lidar remote sensing groups of NOAA Environmental Technology Laboratory, NASA Marshall Space Flight Center, and Jet Propulsion Laboratory have developed and flown a scanning, 1 Joule per pulse, CO2 coherent Doppler lidar capable of mapping a three-dimensional volume of atmospheric winds and aerosol backscatter in the planetary boundary layer, free troposphere, and lower stratosphere. Applications include the study of severe and non-severe atmospheric flows, intercomparisons with other sensors, and the simulation of prospective satellite Doppler lidar wind profilers. Examples of wind measurements are given for the marine boundary layer and near the coastline of the western United States.
© Optical Society of America
Many atmospheric flows occur in the absence of precipitation and optically-dense cloud. Therefore these flows are amenable to study by active optical remote sensing techniques utilizing backscattered signals from naturally-occurring aerosols that passively trace the flow. Flows accompanied by precipitation and thick cloud, which can be probed using other technologies such as radar, often contain important aspects that are amenable to lidar remote sensing as well, such as the environment of a severe thunderstorm. Over the past two decades coherent Doppler lidar has been demonstrated to be an effective tool for atmospheric research under a variety of meteorological conditions. The versatility of deployment of coherent Doppler lidar is enhanced significantly when it is placed on an airborne platform. One can not only study flows and other atmospheric features that are inaccessible or inadequately sampled by more conventional sensors, but also study properties of surface targets. It is also possible to simulate aspects of prospective satellite Doppler wind lidar that cannot be addressed using ground-based measurements.
The atmospheric lidar remote sensing groups of NASA Marshall Space Flight Center (MSFC), NOAA Environmental Technology Laboratory (ETL), and the Jet Propulsion Laboratory (JPL) together have developed an airborne Doppler lidar system capable of mapping the wind and aerosol backscatter distribution over large volumes of the planetary boundary layer (PBL), free troposphere and lower stratosphere in regions of adequate backscatter. This instrument, the Multi-center Airborne Coherent Atmospheric Wind Sensor (MACAWS), was developed during 1992-5 in large part by using hardware resources and expertise acquired in the course of previous atmospheric research programs. The transmitter/receiver is the CO2 coherent Doppler lidar from the highly successful NOAA Windvan, which is capable of emitting nearly 1 J/pulse at 10.6 μm and which can achieve measurement coverage of as much as 30 km . This approach permitted a more-straightforward development and considerable savings, while resulting in the most sophisticated and versatile airborne coherent Doppler lidar to date. Although other airborne coherent Doppler lidars have been developed [2,3], or are being developed , to our knowledge MACAWS is the only system specifically designed to measure high-resolution fields of horizontal wind velocities over a three-dimensional volume.
The concept of wind field measurements with airborne Doppler lidar was demonstrated in 1981 using a 14 mJ/pulse coherent lidar capable of achieving 6–10 km coverage in the lower troposphere . A pseudo-dual Doppler technique was employed wherein the lidar beam was scanned sequentially to infer the two-dimensional horizontal wind field relative to the aircraft. The scanning capability was subsequently enhanced to permit measurement of wind fields at several vertical levels . The same measurement technique is used for MACAWS, but with substantially greater coverage╍and scientific utility╍which is made possible by a significantly more-powerful transmitter. The balance of this article describes the MACAWS instrument, examples from field programs in 1995-6, and potential research applications. A MACAWS World Wide Web page has been established which contains additional details and examples .
2. Instrument description
The MACAWS hardware system consists principally of the laser transmitter, receiver, telescope, optical table, scanner, inertial measurement unit, and computer (Fig. 1). The primary operating characteristics are summarized in Table 1; the optical layout is summarized in Ref. . The transmitter is a frequency-stable, transverse-excited atmospheric pressure (TEA) CO2 laser ; several modifications were required to ensure reliability, safety, and compatibility with the aircraft environment . The receiver consists of a cryogenically-cooled infrared detector and supporting optics and electronics configured for coherent signal detection [1,9]. The folded telescope consists of a 0.3 m diameter off-axis paraboloidal primary mirror and secondary mirror shared by the transmitter and receiver in a monostatic configuration. The table assembly consists of a ruggedized optical table, separable into two sections to facilitate integration; the table itself is upheld by a three-point support structure which is fastened to the aircraft seat tracks. A large portion of the optical table was developed previously by JPL for a program to survey the global aerosol backscatter distribution . The scanner is composed of two computer-controlled, independently-rotating germanium wedges which refract the transmitted beam in the desired direction . The scanner was developed for atmospheric research programs by NASA MSFC for the first airborne coherent Doppler lidar system (ADLS) . The original scanner had the capability to refract the beam anywhere within a full cone angle of ~40 deg. In the present configuration the wedges have been replaced with elements of greater angular thickness which permit scanning anywhere within a full cone angle of ~64 deg. A dedicated inertial measurement unit (IMU), mounted beneath the scanner, senses aircraft attitude and speed at a rate of 20 s-1 . The IMU and telescope were developed for the first ADLS as well. The computer consists of a Unix-based operations control system (OCS) which orchestrates the functioning of each subsystem. The OCS also processes, displays, and stores raw lidar data (in limited quantities) and processed data, along with scanner settings, IMU data, location and ground speed derived from the Global Positioning System (GPS), and aircraft housekeeping data. MACAWS is presently configured for the NASA DC-8 research aircraft, which has a service ceiling of 12.5 km and a range of over 9400 km.
During flight laser pulses are transmitted to the atmosphere through the scanner, which is mounted within the left side of the aircraft ahead of the wing. Aerosols, clouds, or the surface scatter a small portion of the incident radiation backward along the line-of-sight (LOS) to the receiver. In order to maintain precise beam pointing, IMU measurements are input to the OCS, which in turn issues commands to the receiver and scanner to compensate for aircraft attitude and speed changes. Using the same IMU measurements during signal processing, the OCS and receiver calculate and subtract the frequency contribution to the Doppler-shifted signal representing the component of aircraft motion along the line of sight. The resulting, range-resolved LOS velocities represent the component of wind motion with respect to the earth. Measurement coverage varies with lidar system settings (laser output energy, pulse repetition frequency, LOS resolution, number of pulses averaged) and atmospheric conditions (aerosol backscatter distribution, and attenuation by water vapor, carbon dioxide, and clouds). Atmospheric signal processing is done in real time; a poly-pulse-pair velocity estimation algorithm, implemented digitally as a matched-filter frequency domain estimator , is used to calculate LOS velocities . For each range gate, the fast Fourier transform (FFT) of the truncated autocorrelation function is calculated from digitized, complex samples of the time-varying output of the signal detector. The peak of the frequency spectrum is then found with high resolution by fitting a quadratic curve to the three points nearest the peak of the FFT. The peak of the fitted function corresponds to the LOS velocity estimate. Pulses with excessive frequency variation due to system anomalies, such as transient optical misalignments due to turbulence, are flagged, are excluded from signal processing and wind estimation, and are termed “bad” pulses. On-board displays of LOS velocity, two-dimensional wind fields, and backscattered signal intensity provide inflight mission guidance as well as a means to assess subsystem performance and overall data quality.
Before research flights commence, the IMU is physically aligned with the aircraft so that measurements of roll, pitch, and pointing direction agree with those of the aircraft inertial navigation system (INS). Scanner pointing is then calibrated relative to the IMU-indicated aircraft orientation . Finally, the intensity response of the lidar transmitter/receiver is calibrated by comparing backscattered signals from a target of known reflectance with expected signal intensity calculated from the lidar signal-to-noise (SNR) equation using measured system parameters . The resulting calibration factor permits conversion of relative signal intensity (dB) to units of absolute backscatter (m-1 sr-1).
The manner in which the atmosphere is sampled with the lidar beam is determined by the science objective(s), three-dimensional distribution of the feature or process of interest, aircraft altitude, aerosol backscatter distribution, attenuation, and range to target. Fig. 2(a-c) illustrates the possible sampling geometries. In the vertical profiling mode, Fig. 2(a), the beam is maximally refracted to 32 deg up or down relative to the aircraft. A quasi-vertical cross section of LOS velocity and aerosol backscatter may then be obtained by maintaining a constant flight heading . Alternatively, the aircraft may fly one or more orbits; each orbit permits the lidar to sample the atmosphere in a pattern resembling the familiar velocity-azimuth-display (VAD) common to ground-based radar and lidar. Two principal applications of this approach are possible. First, a single horizontal wind profile above (clockwise orbit) or below (counterclockwise) the aircraft may be calculated using techniques originally developed for ground-based radar, e.g., . Second, by conducting the orbits at different roll angles, the angular dependence of surface scattering may be measured.
The second, and unique, capability of MACAWS is that of remotely sensing two-dimensional (2-D) wind fields. Fig. 2(b) illustrates a plan view of the sampling pattern of lidar beams. Each beam may be composed of three or more pulses that are combined during signal processing to improve LOS velocity accuracy and coverage. During scanning the beam is alternately directed ~20 deg forward and aft of normal relative to the aircraft heading. At each “point” of intersection, a 2-D velocity can be calculated due to the angular separation between perspectives. The velocity field thus represents the component of wind within the scan plane. Without rapid and precise compensation for aircraft motions, turbulence experienced by the aircraft could cause the scanner to misdirect one or more beams outside of the measurement plane. Using the IMU measurements of aircraft pitch, roll, and track angle, the OCS attempts to compensate for turbulent motion by issuing appropriate commands to the scanner. The angular thickness of the germanium wedges limits the 2-D measurement capability to ±25 deg in the vertical, beyond which there is insufficient angular separation between fore and aft beams to calculate 2-D velocity accurately.
An important extension of the 2-D scanning capability is three-dimensional (3-D) coverage that can be achieved by generating multiple scan planes, Fig. 2(c). The present OCS configuration permits up to five scan planes with arbitrary vertical angular spacing. In general the extent over which measurable signals may be obtained is a function of: 1) aircraft altitude, subject to air traffic control restrictions and aircraft service ceiling; 2) angular separation between uppermost and lowermost scan planes, subject to the refractive limit of the scanner; 3) aerosol backscatter distribution, a function of the aerosol physical, chemical, and optical properties; and 4) attenuation of the incident and scattered laser radiation, depending on concentrations of water vapor, CO2, aerosols, and optical thickness of cloud (if present). Coverage is generally more-extensive in the PBL, where higher concentrations of larger aerosols are responsible for larger aerosol backscatter values hence stronger signal-to-noise ratios, e.g., .
Resolution along the flight track Δx in each scan plane may be approximated by:
where d1 is the delay in repositioning the scanner wedges to an adjacent elevation angle in the fore or aft direction (~0.1 s), d2 is the scanner delay between the fore and aft pointing directions (~0.6 s), na is the number of scan planes, ng is the number of pulses averaged during signal processing, nb is the number of pulses rejected during signal processing, P is the laser pulse repetition frequency (s-1), and Vg is the aircraft ground speed. In practice turbulence slightly increases the time required to reposition the scanner wedges. Turbulence can also cause brief periods of laser mode degradation or frequency jitter; the on-board pulse quality discriminator rejects these pulses. Both effects degrade both the along-track and cross-track resolution. For the case of measurements in the PBL assuming Vg = 125 m s-1, P = 20 s-1, na = 5, nb = 2, and ng = 10, Eq. (1) yields Δx ≅ 1.0 km.
During scanning the measured LOS velocity with respect to the aircraft is dominated by aircraft motion. Therefore, this large velocity component must be characterized accurately in order to determine the residual, ground-relative wind motion. This requires accurate knowledge of the scanner settings, aircraft attitude and speed, and frequency distribution of the outgoing pulse . Analysis of the ground calibration measurements yielded a 0.1 deg root-mean-square (rms) uncertainty in scanner pointing angle, which is equivalent to a spatial uncertainty of 17 m at 10 km range. This uncertainty was confirmed by analysis of ground strikes. Corresponding velocity errors were less than 0.4 m s-1 for ground speeds of 232 m s-1 (450 kt). The largest source of velocity uncertainty is attributed to the IMU estimates of ground speed, which can amount to 0.5 – 4 m s-1. This source of uncertainty varied from flight to flight, but can be reduced in-flight or during postprocessing by using ground speed data from the aircraft INS or GPS. Wind velocities derived from the aircraft INS may differ from lidar winds at close range by ~1 m s-1 or less when a more-accurate source for ground speed is used. Measurement of aerosol backscatter coefficients is carried out by inverting the equation for the coherent lidar signal-to-noise ratio after computing the SNR [1,16]. This estimate requires knowledge of several parameters including system range response, optical efficiency, atmospheric extinction, shot noise level and pulse energy. For MACAWS, system optical efficiency parameters are measured while the aircraft is stationary; these values are then applied to the flight data. Given an accurate measurement of the system parameters, backscatter coefficients are measurable to within ~3 dB if the key system parameters do not change significantly between calibrations. By comparison, atmospheric aerosol backscatter at 10.6 μm wavelength may vary over five orders of magnitude or more, e.g., . More details on measurement uncertainties are found in Ref. .
Laboratory integration and ground tests were conducted at JPL, Pasadena, California, during March 1 - July 19, 1995; the first atmospheric returns were obtained on May 18, 1995. The first flight tests were conducted during 13–26 September 1995 over the western US and eastern Pacific Ocean; subsequent flight experiments were conducted 31 May - 2 July 1996 over the western and central US and eastern Pacific Ocean. Between flight programs, modifications were made to improve performance, especially under turbulent flow conditions. This section contains examples from the 1995 and 1996 flight experiments.
Accurate mapping of the marine PBL wind structure is important to understanding the interaction between the atmosphere and ocean. During the 1995 flights, MACAWS was used to characterize the PBL wind regime near the coast of Oregon as part of a field experiment to study oceanic internal waves. Fig. 3 illustrates the 3-D wind distribution observed as the aircraft flew just above the top of the PBL. Weaker wind speeds and a shift in wind direction were measured by the aircraft INS; this feature is characteristic of weak wind shear across the top of the PBL associated with the flow transition to the free troposphere. At the lowest elevation angle the wind field was measured down to the ocean surface, below which the lidar beam is extinguished. “Clustering” of wind vectors, and variations in along-track and LOS resolution, are due to signal processor settings and the effect of turbulence on scanner and laser performance (Eq. (1)). Clustering arises when LOS resolution is considerably finer than along-track resolution, in this case 300 m and ~730 m, respectively.
Surface winds off the California coast are northwesterly to northerly in summer, resulting from flow around the east side of the subtropical high-pressure system over the Pacific Ocean. This high is accompanied by subsidence and a strong marine temperature inversion usually a few hundred meters deep. The coastal mountain-range topography interacts with the northerly flow in this marine-inversion layer to produce a variety of interesting flow phenomena. For example, when this flow passes one of the many capes and points that protrude into the wind along the California coast, structures referred to as “hydraulic expansion fans” have been found . Such features are marked by strong variation along the vertical and cross-shore directions. To study this variability the aircraft flew sets of parallel line segments just offshore past Point Arena on 30 June 1996 during 1950–2110 UTC at an altitude of 0.49 km. Fig. 4 shows the wind distribution observed within five scan planes. Figs. 3 and 4 also serve to illustrate the wind vector displays that are produced in real time. The data from all elevation angles in Fig. 4 were reanalyzed along constant-height levels as shown in Fig. 5. Evident in the marine PBL is the northerly flow, the strong variability in the cross-shore direction (especially at 150 m ASL), and the structural changes in the vertical. Grid spacing in the east-west and north-south directions was 300 m, and 25 m in the vertical. The National Center for Atmospheric Research’s (NCAR) software package ‘Custom Editing and Display of Reduced Information in Cartesian space’ (CEDRIC) was used to determine the 2-D wind velocities using the two-equation solution . Interpolation of the raw wind fields to a Cartesian grid is an essential step toward incorporation of MACAWS data into atmospheric numerical forecast models.
4. Summary and conclusions
In 1992 the MACAWS team set out to develop what has become perhaps the most powerful and sophisticated airborne coherent Doppler lidar system in the scientific history of atmospheric remote sensing. We have successfully achieved that goal, and can now concentrate our attention on the many scientific applications of which the instrument is capable. With spatial resolution of order 1 km, ~1 m s-1 velocity accuracy, a Joule-class laser transmitter for extended propagation in optically-clear air, and the range and duration afforded by a large multi-engine aircraft, MACAWS provides a unique means to study a variety of atmospheric processes and features that may be poorly sampled by existing or planned sensors such as radar wind profiles or Doppler radars.
Numerous research applications for MACAWS have been identified , including but not limited to the following. New measurements are needed within tropical cyclones in order to improve forecasts of intensification and tracking. Numerical modeling techniques can now identify regions of the hurricane where more observations are needed in order to reduce forecast uncertainties . Plans are underway to employ MACAWS in such a manner during the 1998 Atlantic hurricane season. In particular, MACAWS has the potential to measure uniquely the winds within the optically-clear eye in the free troposphere; conventional airborne weather radars currently require the addition of reflective material (chaff) in order to visualize the flow. Observations in the eye, at low levels near the rain bands, and elsewhere in the cyclone where observations are not precluded by optically-dense cloud, hold the potential to guide forecast model improvements.
Research into coastal processes is hampered by a lack of detailed observations. Interaction of marine boundary layer flows with coastal topography may strongly influence coastal meteorological conditions, as illustrated in Figs. 4 and 5. Another example is the “southerly surge” phenomenon which can affect the meteorology of the southern California coast, in extreme cases affecting flight operations at the Los Angeles International Airport, e.g., . Data from a weak southerly surge case were obtained during the 1996 flights and are currently being analyzed. The strengths of the MACAWS platform in investigating coastline meteorological systems make it ideal for probing the structure of these surges, and they are expected to be an important target for future research flights.
In addition to atmospheric research, MACAWS has applications to the design, performance simulation, and validation of existing or planned satellite sensors. The concept of direct measurement of global tropospheric winds from space with Doppler lidar has been studied for some time, e.g., [21–23]. Such measurements would fundamentally improve our understanding of global and climate change, as well as global and regional-scale hydrological cycles . In the absence of a heritage of satellite Doppler lidar wind measurements, performance simulations with measured╍rather than simulated╍data are highly desirable to reduce uncertainties in lidar simulation models and to begin to develop the necessary interpretive skills. Some issues can only be addressed from the airborne perspective, such as utilization of the frequency distribution and backscattered intensity of ground returns from land or ocean surface during signal processing. MACAWS has the capability to simulate a number of scanning strategies, as well as to assess satellite Doppler lidar in the presence of clouds and organized atmospheric flow structures.
Our experience has demonstrated that the technology necessary for the reliable operation of high-power, frequency-stable CO2 lidar is mature and presents no technological risks. Moreover, use of a large, high-energy CO2 laser poses no special integration problems. Finally, the sharing of existing hardware, software, and expertise, the minimization of new hardware development, and the identification of mutually-compatible research interests, has resulted in both a substantial cost savings and the successful development and application of a world-class airborne coherent Doppler lidar system.
We gratefully acknowledge Dr. Ramesh K. Kakar, Atmospheric Dynamics and Remote Sensing Program, Office of Mission to Planet Earth, NASA Headquarters, and the Atmospheric Lidar Division, Environmental Technology Laboratory, NOAA Environmental Research Laboratories, without whose support this program would not be possible. We acknowledge the assistance of the DC-8 pilots and ground support team of the NASA Ames Research Center. We also acknowledge Diane Samuelson, NASA Marshall Space Flight Center, who helped prepare the QuickTime animations.
References and links
03. Richmond, R., and D. Jewell, “U.S. Air Force ballistic winds program,” Preprints 9th Conf. Coherent Laser Radar, Linköping, Sweden, (Swedish Defence Research Establishment, Stockholm, 1997), pp. 304–307.
04. Werner, C., P. Flamant, G. Ancellet, A. Dolfi-Bouteyre, F. Köpp, H. Herrmann, C. Loth, and J. Wildenauer, “WIND: An advanced wind infrared Doppler lidar for mesoscale meteorological studies,” Proc. 5th Conf. Coherent Laser Radar, Munich, (Deutsche Forschungsanstalt fur Luft- und Raumfahrt, Munich, 1989), pp. 35–38.
05. Bilbro, J. W., G. H. Fichtl, D. E. Fitzjarrald, and M. Krause, “Airborne Doppler lidar wind field measurements,” Bull. Amer. Meteorol. Soc. , 65, 348–359 (1984). [CrossRef]
07. Rothermel and J., MACAWS World Wide Web page, http://wwwghcc.msfc.nasa.gov/macaws.html.
08. Howell, J. N., R. M. Hardesty, J. Rothermel, and R. T. Menzies, “Overview of the first Multicenter Airborne Coherent Atmospheric Wind Sensor (MACAWS) experiment,” Proc. SPIE , 2833, 116–127 (1996). [CrossRef]
11. Rye, B. J., and R. M. Hardesty, “Spectral matched filters in coherent laser radar,” J. Mod. Opt. , 41, 2131–2144 (1994). [CrossRef]
12. Lee, R. W., and K. A. Lee, “A poly-pulse-pair signal processor for coherent Doppler lidar,” Coherent Laser Radar for the Atmosphere, OSA Technical Digest Series, (Optical Society of America, Washington, DC, 1980), WA2, 1–4.
13. Rothermel, J., D. R. Cutten, R. M. Hardesty, J. N. Howell, S. C. Johnson, D. M. Tratt, L. D. Olivier, and R. M. Banta, “The Multi-center Airborne Coherent Atmospheric Wind Sensor,” Bull. Amer. Meteorol. Soc. , accepted (1998). [CrossRef]
14. Browning, K. A., and R. Wexler, “The determination of kinematic properties of a wind field using Doppler radar,” J. Appl. Meteorol. , 7, 105–113 (1961). [CrossRef]
17. Winant, C. D., C. E. Dorman, C. A. Friehe, and R. C. Beardsley, “The marine layer off northern California: An example of supercritical channel flow,” J. Atmos. Sci. , 45, 3588–3605 (1988). [CrossRef]
18. Mohr, C. G., and L. J. Miller, “CEDRIC - A software package for Cartesian space editing, synthesis, and display of radar fields under interactive control,” Preprints 21st Radar Meteorological Conference, Edmonton, Alta., Canada, (American Meteorological Society, Boston, 1983), pp. 569–574.
19. Zhang, Z., and T.N. Krishnamurti, “Ensemble forecasting of hurricane tracks,” Bull. Amer. Meteorol. Soc. , 78, 2785–2796 (1997). [CrossRef]
20. Emanuel and K. B., et al., “Report of the first prospectus development team of the U.S. Weather Research Program to NOAA and the NSF,” Bull. Amer. Meteorol. Soc. , 76, 1194–1208 (1995).
21. Huffaker, R. M., M. J. Post, J. T. Priestley, F. F. Hall Jr., Richter R. A., and R. J. Keller, “Feasibility studies for a global wind measuring satellite system (WINDSAT): Analysis of simulated performance,” Appl. Opt. , 22, 1655–1665 (1984).
22. M. J. Kavaya, G. D. Spiers, E. S. Lobl, J. Rothermel, and V. W. Keller, “Direct global measurements of tropospheric winds employing a simplified coherent laser radar using fully scalable technology and technique,” Proc. SPIE , 2214, 237–249 (1994). [CrossRef]
23. W. E. Baker, et al., “Lidar-measured winds from space: a key component for weather and climate prediction,” Bull. Amer. Meteorol. Soc. , 76, 869–888 (1995). [CrossRef]