Airborne forward-pointing UV Rayleigh lidar
for remote clear air turbulence detection:
system design and performance
PATRICK VRANCKEN,
1,
*MARTIN WIRTH,
1
GERHARD EHRET,
1
HERVÉ BARNY,
2
PHILIPPE RONDEAU,
2
AND HENK VEERMAN
3
1
Deutsches Zentrum für Luft- und Raumfahrt (DLR), Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany
2
THALES Avionics, 25 Rue Jules Védrines, 26027 Valence Cedex, France
3
Netherlands Aerospace Centre (NLR), Anthony Fokkerweg 2, 1059CM Amsterdam, The Netherlands
*Corresponding author: patrick.vrancken@dlr.de
Received 20 July 2016; revised 30 September 2016; accepted 13 October 2016; posted 14 October 2016 (Doc. ID 270899);
published 10 November 2016
A high-performance airborne UV Rayleigh lidar system was developed within the European project DELICAT.
With its forward-pointing architecture, it aims at demonstrating a novel detection scheme for clear air turbulence
(CAT) for an aeronautics safety application. Due to its occurrence in clear and clean air at high altitudes (aviation
cruise flight level), this type of turbulence evades microwave radar techniques and in most cases coherent Doppler
lidar techniques. The present lidar detection technique relies on air density fluctuation measurement and is thus
independent of backscatter from hydrometeors and aerosol particles. The subtle air density fluctuations caused by
the turbulent air flow demand exceptionally high stability of the setup and in particular of the detection system.
This paper describes an airborne test system for the purpose of demonstrating this technology and turbulence
detection method: a high-p ower UV Rayleigh lidar system is installed on a research aircraft in a forward-looking
configuration for use in cruise flight altitudes. Flight test measurements demonstrate this unique lidar system
being able to resolve air density fluctuations occurring in light-to-moderate CAT at 5 km or moderate CAT at
10 km distance. A scaling of the determined stability and noise characteristics shows that such performance is
adequate for an application in commercial air transport.
© 2016 Optical Society of America
OCIS codes: (010.1330) Atmospheric turbulence; (010.7060) Turbulence; (280.7060) Turbulence; (010.3640) Lidar; (280.3640) Lidar;
(290.5870) Scattering, Rayleigh.
http://dx.doi.org/10.1364/AO.55.009314
1. INTRODUCTION
In commercial aviation, clear air turbulence (CAT) encounter is a
leading cause of injuries to cabin crew and passengers and results in
M$ or M€ damage per year to airlines. Nonfatal aircraft accidents
and incidents are concentrated en route where, in turn, turbulence
encounter is the main cause for such injuries [1]. CAT encounter
further yields important fatigue to aircraft structures. Thus impact
mitigation is of high interest to the aeronautics sector .
CAT is a phenomenon that is difficult to forecast with mere
provision of probability of occurrence charts over vast areas that
typically cannot be fully avoided by aircraft (e.g., in the vicinity
of jet streams or in the lee of mountain ridges). An in-flight
forward turbulence detection of a turbulent zone ahead would
allow for a warning, such as a fasten seat belt sign given by the
flight deck. Further mitigation may consist in a slight adjust-
ment of the flight state within the aircraft’s envelope (e.g.,
deceleration) or evasion maneuvers.
Turbulence in clear air, though, defies detection by aeronau-
tics weather radar because it relies on the backscatter of radio
frequency waves on hydrometeors. Here active optical sensing
with lidar appears as being the only possible and/or useful
means for remotely detecting CAT [2].
In principle, airborne turbulence detection by lidar is con-
ceivable by different methods: wind speeds and variations along
the flight path (i.e., along the lidar line-of-sight) may readily be
detected by Doppler wind lidar (DWL). Though, despite its
apparent advantages, coherent DWL relies on backscatter from
aerosols that are not sufficiently present at cruise flight altitudes
for a thoroughly reliable and, in particular, long-range detection
of CAT. Direct-detection Doppler techniques working on the
molecular Rayleigh backscatter (with interferometric evaluation
of the Doppler-shifted spectrum) necessitate a high backscat-
tered photon number for fringe analysis. A short-range imple-
mentation, for the quantification of wind speeds of gusts just
9314
Vol. 55, No. 32 / November 10 2016 / Applied Optics
Research Article
1559-128X/16/329314-15 Journal © 2016 Optical Society of America
ahead of an aircraft, has been demonstrated within the
European AWIATOR project [3]. A long-range application
as discussed here, however, imposes excessive requirements
with respect to laser power or signal averaging.
Another, more photon-efficient detection principle [4] relies
on tiny fluctuations of air tem perature (and thus density) when
air parcels undergo the up- and downwelling motion within the
turbulent airflow (see Section 2). Thus vertical wind speeds
may be derived from air density (hence molecular backscatter)
fluctuations. Vertical wind speed is also the most important
parameter to know because it modifies the angle of attack of
the airflow; thus it directly acts on the instantaneous lift and
creates the known “bumps” and “air holes.”
This principle of lidar turbulence detection by air density
fluctuation has already been tested from ground-based lidar
with some success [5].
In this paper, we report on the development, flight tests, and
metrological performance of an airborne Rayleigh lidar system
to exploit the air density fluctuation method. This activity was
performed within the European-funded FP7 DELICAT project
for validating the appropriability of the method [4] and dem-
onstrating the functional characteristics of the lidar.
The paper is organized as follows: In the next section, the air
density lidar approach is presented. Section 3 details and illus-
trates the configuration and technical details of the airborne
demonstrator lidar system and its layout in the cabin of the
test aircraft. Section 4 exemplifies the constraints of the lidar
measurements with respect to meteorological conditions. It
then gives a detailed evaluation of the lidar performance based
on reference measurements at cruise flight altitudes. It is shown
that density fluctuations as occurring in light-to-moderate CAT
(or stronger) may, in principle, be detected by the present lidar
at distances of 5–10 km. A scaling of the determined perfor-
mance illustrates how the present lidar may extend its detection
range to 25 km, which would put it, from the requirements
point of view, in the position of a real-world aeronautics safety
application.
2. CAT DETECTION WITH RAYLEIGH LIDAR
Clear air turbulence results from a gravity wave (GW) field and is
generated by the saturation and breaking of this field [5]. This, in
reality, complex GW field may be engendered by orography,
such as mountain waves, by convection as above thunderstorms
or even shallow convection, by instabilities in stratified shear
layers (Kelvin–Helmholtz instabilities in the vicinity of jet stream
borders), and commonly by a combination of these.
For the matter of lidar detection, a relationship between the
vertical wind speed w and the air temperature T and thus den-
sity ρ may be derived, as shown in [4]: From potential and
actual temperature gradients (lapse rate) and static stability
N (Brunt–Väisälä frequency), the following expression may
be der ived for the temperature (thus density) of an air parcel,
vertically displaced by Δ z:
Δρ
ρ
−
ΔT
T
Δz ·
N
2
g
: (1)
In order to relate the relative change in density Δρ∕ρ
to the vertical wind speed w, we consider the critical
Richardson number Ri
c
0.25, below which turbulence sets
in. With Ri N
2
∕S
2
and for the shear S du∕dz ≈ 2w∕Δz,
it follows w N · Δz [4]. One may thus deduce
Δρ
ρ
−
ΔT
T
w ·
N
g
. (2)
With Eq. (2) we obtained a simple relationship between
density fluctuations and vertical velocity related to turbulent
events. With typical values for N of 0.01 and 0.02 rad/s
for the troposphere and stratosphere, respectively, and some
5–30 m/s vertical gust peak speed ˆw, these air density variations
are subtle, i.e., on the percent level. Considering pure molecu-
lar backscatter, they will appear as variations of the lidar signal,
superimposed to the “standard” variance arising from photon
(and other) noise. For resolving the turbulence variance at use-
ful ranges (i.e., some 10 km in front of an aircraft traveling at
Ma ≥ 0.8) a high synthetic signal-to-noise ratio of, say
SNR
av
100, has to be achieved by substantial averaging
of individual lidar signals. Here these air density fluctuations
can be considered “frozen” over the considered detection/
averaging time span (of some seconds to tens of seconds) com-
pared with the characteristic rotation times of the turbulent
vortices given by above values of N (i.e., some 5 to 10 min).
In order to most efficiently exploit the molecular backscatter
β
mol
(in order to measure air density), a short laser wavelength is
favored here because the backscatter cross section scales with
the fourth power the frequency (β
mol
∝ υ
4
). As will be expli-
cated in Section 3.A, ultraviolet radiation has additional advan-
tages, such as better appropriateness for eye-safety norms and a
more favorable aerosol to molecular backscatter ratio. Indeed,
for aircraft en route altitudes, the typical UV lidar backscatter
ratio between aerosol and molecules R
b
− 1 β
aer
∕β
mol
amounts to less than 8 × 10
−3
most of the time [6,7], which
is worse with factors of 10–40 for NIR wavelengths. Then
we may assume for UV wavelengths:
Δβ
mol
β
mol
≅ w ·
N
g
: (3)
However, the above given statistical/climatological value of R
b
may not be expected to occur during all portions of a cruise
flight. Higher aerosol loads may in fact mask the desired molecu-
lar backscatter fluctuations (refer also to Section 4.B). Therefore,
for use in an aeronautics application, a high-resolution spectral
(HSR) filter should be employed in order to select the spectrally
narrow aerosol return and use only the molecular wings of the
broad Rayleigh–Brillouin spectrum.
As mentioned above, the concept of CAT detection by air
density fluctuation has been tested with ground-based lidar in-
strumentation from Observatoire de la Haute Provence within
the French precursor project MMEDTAC in 2008/2009 [5],
even though ground-based detection (and longer-term averag-
ing) suffers from the advection of the turbulent patch by the
horizontal wind prevailing at these altitudes. The European
Commission (EC) funded Sixth Framework (FP6) project
FLYSAFE studied this technological alternative for use in aero-
nautics [8,4]. Within the FP7 project DELICAT (2009–2013),
such an airborne Rayleigh lidar system has been developed and
flown for demonstration of its performance and proof of con-
cept. The central requirement formulated for the DELICAT
Research Article
Vol. 55, No. 32 / November 10 2016 / Applied Optics 9315
project was to measure a 1% air density fluctuation at 5 km
distance in cruise flight altitude, corresponding to moder-
ate CAT.
Within the project, the lidar system was integrated in a re-
search aircraft in a forward-pointing arrangement. In this con-
figuration, the aircraft passes through the farthest lidar-probed
zones (recorded detection range: 15 km) after less than 1:30–
2:00 min (depending on altitude/flight speed), provided the
absence of horizontal wind shear during the respective flight
portion. With the assumption of frozen turbulence, the aircraft
itself then acts as a “truth” sensor for the turbulence. An inertial
reference system (IRS) delivers high-resolution acceleration for
all six axes. This macroscopic aerodynamic turbulence sensor,
though, is rather elaborate to exploit in terms of turbulent ver-
tical velocity, necessitating complex aircraft and also turbulence
models. Thus, the aircraft was also equipped with a fast total air
temperature probe also delivering a low-pass filtered turbulence
“truth” according to Eq. (2). The respective prevailing larger-
scale Brunt–Väisälä frequency N may be interpolated from
temperature measurements between ascents and descents next
to a turbulence encounter or taken from numerical weather
analysis.
3. LIDAR SYSTEM
A synopsis of the developed airborne lidar system is depicted in
Fig. 1. A high power laser transmitter sends short pulses on a
beam steering device. This device compensates the attitude and
movements of the ai rcraft and ensures the horizontal projection
of the laser beam on the ahead-lying flightpath. The backscat-
tered light takes the same direction, is collected by a telescope,
filtered, and projected onto a set of detectors. These elements,
together with their performance, are described in Sections A
through F.
The lidar system is mounted within a stiff rack structure,
which is adapted to the NLR research aircraft PH-LAB, a
modified Cessna Citation 2 aircraft. To allow the horizontal
projection of the laser beam onto the flight path, it is equipped
with a special aerodynamically optimized fairing under which is
located the forward bending mirror. Figure 2 shows photo-
graphs of this carbon-fiber reinforced plastic structure on the
starboard side of the aircraft and a view of the lidar system in-
side the cabin.
A. Transmitter
For a given realistically achievable laser power, this application
calls for high pulse energies with low repetition rate rather than
low pulse energy with high rate since the averaged SNR
av
scales
linearly with the former and by the square-root with the latter.
In the same logic, regarding the choice of an appropriate
wavelength, one may consider maximizing the following simple
figure of merit:
F:o:M: η
c
ν
· R
β
λ
· Q:E:; (4)
where η
c
ν
is the laser frequency conversion efficiency, R
β
λ
the
ratio of Rayleigh backscatter coefficients at different wave-
lengths, and Q.E. the quantum efficiency of a detector at a
considered wavelength. Table 1 shows typical achievable values
of this F.o.M. for the three fundamentals of the well-proven
solid-state Nd:YAG laser for two different detector types,
photomultiplier tubes (PMT) and avalanche photodiodes
(APD).
This exercise shows the principal equivalency of using
the green second harmonic and the UV third harmonic.
However, because the visible harmonic imposes more serious
eye-safety considerations than the invisible third harmonic
(see Section 3.G), a UV laser source based on nonlinear fre-
quency conversion is favored.
As a source, the several times flight-proven pump laser of the
DLR WALES lidar was chosen. It is thoroughly described in
[9]; some details are given in the following (cf. Fig. 3). The
laser is of the MOPA (master oscillator, power amplifier) de-
sign, with a monolithic intrinsically single-mode running
Nd:YAG master resonator. This diode-pumped NPRO (non-
planar ring oscillator) laser emits about 150 mW of infrared
laser pulses with a length (FWHM) of 8 ns at a rate of
4 kHz. Its frequency may be tuned and modulated both by
temperature and mechanical stress. These techniques are used
for locking it to a molecular reference. The NPRO oscillator is
stress-modulated with a sine wave that is in phase with the
reference signal (see below) for the subsequent power ampli-
fiers. A small part of the IR radiation is directly frequency
Fig. 1. Synopsis of the DELICAT lidar system.
Fig. 2. Left: view on the outside starboard fuselage with the fairing
for laser beam transmission and reception. Right: complete lidar and
beam steering system with operator interface.
9316 Vol. 55, No. 32 / November 10 2016 / Applied Optics
Research Article
doubled (SHG) and fed through an iodine vapor absorption
cell generating a signal on a photodetector (PD). This is us ed
in a lock-in technique to stabilize the laser to the absorption
line center yielding an absolute frequency stability below
1 MHz and 300 kHz on short time scales (<1min). Note
that this frequency locking is not required in the current setup
but was used nonetheless. The laser is thus apt for use in a pos-
sible future setup, including a high spectral resolution lidar
setup (as addressed in Section 2).
The absolute repetition rate reference (at 100 Hz) of the
MOPA setup is imposed by the driver current cycle of the power
amplifier (PA) stages. The residual timing jitter of the passively
Q-switch generated pulses is less than 0.5 μs(at1σ) leading to a
low pulse-to-pulse power variation (see below). The amplifier
setup is composed of a small-signal double-pass amplifier and
two single-pass main amplifiers. With a combined gain of
40 dB, the laser thus delivers pulse energies of up to 400 mJ.
Measurements of the fundamental at this power level yield a
beam quality of M
2
1.5. A photodiode (PD) registers the out-
going pulses used for triggering of the data acquisition (see
Section 3.D).
The infrared radiation is then fed into a KTP crystal for
second-harmonic generation (SHG) in type II configuration
(oe → o). The phase-matching is achieved coarsely by angle
tuning and finely by temperature tuning. For this purpose,
the KTP crystal is heated to approximately 80°C, and the whole
setup is accommodated in a heat-insulated compartment.
Thus, at full laser power, an SHG conversion efficiency of
up to 55% is achieved. Upon SHG, both the fundamental
as well as the second harmonic feature a certain elliptic polari-
zation. Therefore, a set of two-wavelength zero-order wave-
plates with λ∕4 and λ∕ 2 delay are employed to adjust the
two harmonic’s polarization states to the linear states. Then
the beams are fed into a BBO crystal for sum frequency gen-
eration. Here, the phase matching is achieved by angular tuning
within a Piezo-driven two-axes mount. Together, a third-
harmonic generation (THG) efficiency of up to 30% is
achieved. Figure 4 shows the THG efficiency dependency over
the incoming laser power and the respective attained power
at 355 nm.
The UV radiation is then led over a first highly dichroic
mirror; the transmitted infrared and green portions are fed into
a beam dump. The UV part is directed through a motorized
zero-order λ∕2 waveplate in order to rotate its polarization
for optimizing the transmission losses over the skew arrange-
ment of many (eight) mirrors to follow in the subsequent op-
tical setup. Some leakage of the UV light behind a second
dichroic mirror is focused on a fast PIN photodetector; the gen-
erated signal is fed through a sample and hold circuit and digi-
tized. This monitoring of the laser pulse energy of every emitted
pulse is stored alongside the lidar information (see Section 3.D).
Figure 5 shows a comparison of this internal laser pulse energy
Fig. 3. Schematic optical layout of the transmitter. See text and [9]
for details and acronyms. Only main elements are shown here.
Fig. 4. Third-harmonic generation efficiency and output power.
Fig. 5. UV laser pulse energy and comparative power measurement
with external reference. The line shows a linear regression with mean
residuals of 60 mW. These originate mainly in the pyroelectric detec-
tor, which features a dispersion of around 40 mW (vertical error bars).
Research Article
Vol. 55, No. 32 / November 10 2016 / Applied Optics 9317
measurement with an external (continuous) pyro-electric
sensor.
The measurement reveals the linearity of this pulse energy
measurement process and the low pulse-to-pulse energy jitter of
less than 0.5%, which have been measured quite equally inter-
nally (dispersion 0.1–0.4 mJ) and by the reference (dispersion
40 mW). For the internal pulse measurement, this includes the
pulse-to-pulse energy fluctuation originating in the passive
Q-switch pulse timing jitter, subsequent fluctuation in laser
amplification, which is then further deteriorated by the non-
linear efficiency of the THG process. This implies an excellent
precision of this digital pulse energy measurement.
The main high-power UV laser beam is expanded by an-
other Galilean telescope to a diameter of 13 mm. Further,
its divergence is adjusted to a value Θ 150 μrad. The beam
quality factor in the UV was determined to M
2
4.3 (see also
Fig. 6), which is in good agreement with theory on beam
quality degradation with nonlinear frequency conversion:
1 ≤ M
2
3ν
∕M
2
1ν
≤ 3, depending on conversion efficiency [10].
The beam is then guided through a shutter device that al-
lows the lidar operator in the aircraft cabin and the flight deck
to remotely block the laser output without interrupting laser
operation. Another dichroic mirror directs the UV beam into
the subsequent beam feed.
Opto-mechanically, the whole laser and harmonic genera-
tion assemblies are based on standard laboratory holders and
some custom parts, both from aluminum and steel alloys. A
flight-proven compact rugged vibration- and ambient condi-
tion resilient design is followed. Figure 7 shows a photo of
the all-self-contained transmitter. The upper part (visible in
the image) contains all the optical elements referred to above
(as in Fig. 3), with the master oscillator to the far right rear, and
the three greenish amplifiers in the red compartment and the
THG in the magenta module. The THG module may be ex-
changed to any other optical parametric oscillator of the
WALES lidar series [9,11].
The lower half of the transmitter body contains all electron-
ics, in particular the DC/DC converters and diode drivers of
the laser power amplifier stages. The intermediate level of
the IR pump laser features a water-cooling circuit directly feed-
ing the pump chambers with de-ionized water. The overall
power consumption at full optical output power is 800 W.
The chief part of this power is evacuated as heat by the water
cooling circuit. The heat is then exchanged to another cooling
circuit based on some aeronautics coolant oil. This circuit
transports the heat charge to an external cooling plate, which
is installed in a cabin windows aperture. The transmitter mea-
sures 935 mm⨯412 mm⨯257 mm, including the THG com-
partment with a total mass of 106 kg. It is integrated in a stiff
rack (see Section 3.F) with the other main optical subsystems
such as receiver and beam guidance device.
B. Laser Beam Guidance System
The lidar system is arranged in a classical monostatic arrange-
ment, with the laser beam being transmitted over a mirror super-
posed to the secondary mirror of the receiver telescope (receiver
see next section). To this end, the laser beam is guided through a
tubular system from the transmitter exit to this transmit mirror.
From here on, the optical transmit and receive paths are
common. The tubular beam feed comprises a piezo-electrically
motorized mirror. This serves to fine-tune the transmit beam
into the receiver field-of-view (FOV), which is defined by the
telescope optics and field stop.
The common transmit/receive path is guided over a com-
plex arrangement of mirrors and windows out of the aircraft
cabin where it is folded forward onto the flight path.
During steady flight (high-altitude cruise) where the CAT
measurements should take place, the aircraft is subjected to fol-
lowing movements: Change of aerodynamic angle of attack due
to different altitudes, speeds, and change of mass and center of
gravity (due to fuel consumption); low frequency residual air-
craft dynamic modes such as a phugoid and Dutch roll that
may not fully be damped by autopilot or other means; and
movements due to the light background turbulence. In order
to ensure probing the same air volume at distances of 5 to
Fig. 6. Far-field beam pattern of the third harmonic of the laser.
Fig. 7. Photo of the (opened) transmitter. See text for details.
9318 Vol. 55, No. 32 / November 10 2016 / Applied Optics
Research Article
![Fig. 12. (a) Backscatter ratio at 355 nm derived from [6,7], actually giving a “median of median values” from 80 flights in six different areas and seasons of the Atlantic Ocean region, shown at the flight altitudes relevant for DELICAT. The graph also gives the 25% and 10% quantiles for the unfavorable case. (b) Determined maximum permissible aerosol mass concentration for the flight tests.](/figures/fig-12-a-backscatter-ratio-at-355-nm-derived-from-67-2x7ppt29.webp)
![Table 5. Turbulence Levels According to [32]](/figures/table-5-turbulence-levels-according-to-32-1zjoxfjc.webp)
