1932 OPTICS LETTERS / Vol. 27, No.

21 / November 1, 2002

Daylight rejection with a new receiver for potassium resonance

temperature lidars
Cord Fricke-Begemann, Matthias Alpers, and Josef Höffner
Leibniz-Institut für Atmosphärenphysik e.V. an der Universität Rostock, Schloss-Strasse 6, D-18225 Kühlungsborn, Germany

Received May 17, 2002

For daytime operation a new receiver for potassium resonance temperature lidars was developed that includes
a twin Faraday anomalous dispersion optical filter and a high-performance photon-counting avalanche photo-
diode instead of a photomultiplier. The design of a f lat spectral filter shape over the lidar scan range avoids
instrumental error on the temperature results. Initial temperature measurements under daylight conditions
were realized. © 2002 Optical Society of America
OCIS codes: 010.1310, 010.3640, 040.5250, 120.2440, 120.4570, 230.2240.

Laser remote sensing (lidar) is a powerful tool for
ground-based atmospheric research. This technique
allows continuous observations of various basic atmo-
spheric parameters such as temperature, air density,
wind, and composition with high spatial and temporal
resolution. Because of the high solar background
during the day, lidar operation with standard spectral
filtering in the receiver is limited to the dark. A large
number of dynamic processes in the Earth’s atmo-
sphere (e.g., planetary waves, tides) and long-period
changes of atmospheric parameters (e.g., diurnal mean
temperature profiles, day– night variations) can be
explored only if the instrument used allowed daylight
operation with high data quality. This condition
also applies, in general, to measurements during
the polar summer. Above ⬃15-km altitude, for a
lidar receiver with a standard interference filter of
approximately 0.3–1.0-nm spectral bandwidth alone,
the solar background exceeds the lidar backscatter
signal. Therefore, further spectral f iltering with a
much lower bandwidth 共Dl , 10 pm兲 is required.
The common f ilter technology for such narrow spec-
tral bandwidths is the high-resolution Fabry – Perot
etalon. Its disadvantage is the requirement of high
optical quality for the incoming light. A narrow spec-
tral bandwidth combined with a high peak transmis-
sion requires an extremely low beam divergence, i.e.,
large optical apertures, which makes the technical re-
alization of such a filter very expensive (⬃$100,000
U.S.). To tune and lock such a filter on the emitter
wavelength of a lidar, extensive mechanical and ther-
mal stabilization is necessary. Reliable operation of
such a f ilter requires a temperature-stable and vibra-
tion-free laboratory environment. This environment
cannot always be guaranteed, especially for operation
with a mobile lidar under extreme climatological and
operational conditions, as at polar latitudes or on board
a research sea vessel.
A more robust and cost-effective (⬃$15,000
U.S.) technology is an atomic resonance filter, the
so-called Faraday anomalous dispersion optical filter
(FADOF), which plays an important role in free-space
optical communication and remote sensing.1 – 3 The
theory behind the FADOF has been discussed in detail
by Yeh1 and by Yin and Shay.2 A FADOF consists
of an atomic-vapor cell within a permanent magnetic
field between crossed polarizers; the field is oriented
0146-9592/02/211932-03$15.00/0
parallel to the optical axis. The polarization axis of
light close to an atomic resonance wavelength travel-
ing along the optical axis through the atomic vapor is
rotated by the resonant Faraday effect. The degree
of rotation depends on the wavelength, the number of
atoms along the beam path, and the magnetic f ield
strength. By tuning of the magnetic field strength
and vapor-cell temperature, a rotation angle of 90± can
be achieved on a specific atomic resonance line, result-
ing in high theoretical transmission of nearly 100%.
For wavelengths off the resonance line, the rotation
angle deviates from 90±, and the light is (partially)
blocked by analyzer. The blocking efficiency of the
FADOF depends on the optical quality of the polar-
izing optics, whereas its spectral transmission shape
depends mainly on the magnetic f ield strength and
cell temperature. The FADOF does not need highly
collimated light. Only moderate thermal stabilization
of the vapor cell is necessary. The magnetic f ield
can be produced by a permanent rare-earth magnet,
which does not require stabilization. Therefore, the
FADOF has a relatively simple mechanical and optical
structure that also allows operation under extreme
environmental conditions.
The limitation of the FADOF technology to atomic
resonance lines is not a disadvantage for the class of
the so-called metal resonance lidar instruments. In
the altitude range 80–100 km there exist layers of
free metal atoms, e.g., Fe, Ca, Na, and K. Probing the
spectral Doppler broadening of the metal resonance
lines with tunable narrowband laser emitters, espe-
cially Na D2 and K D1 , allows the determination of
atmospheric temperature prof iles at the metal-layer
altitudes.4 – 6 The f irst adaptation of the FADOF
technology to daylight lidar operation was realized for
Na resonance lidar.7,8 So far, a K FADOF has been
developed only for applications (e.g., a solar Doppler
imager3) other than lidar. This work, for what is to
our knowledge the first time, reports on an advanced
daytime FADOF receiver for a K Doppler resonance
lidar. With the FADOF attached to the mobile K
resonance lidar of the Leibniz Institute for Atmo-
spheric Physics (IAP),6 initial daytime measurements
of temperatures in the mesopause region were carried
out on the island of Tenerife (28± N) and during the
summer of 2001 on the arctic island of Spitsbergen
(78± N).
© 2002 Optical Society of America
 November 1, 2002 / Vol. 27, No. 21 / OPTICS LETTERS 1933

For temperature calculations for our K resonance 1-nm spectral width). A further eff icient solar back-
lidar retrievals, the emitter laser is tuned over a suff i- ground reduction can be achieved by minimization of
cient spectral range 共⬃3 pm兲 of the Doppler-broadened the field of view of the lidar. The background radia-
K D1 resonance line (770 nm).6 The spectral trans- tion received from the observed sky angle is dependent
mission shape of any daylight filter must be uniform on the square of its radius. A workable reduction of
over this range, or its transmission must be mea- the field of view of the mobile IAP K resonance lidar
sured with very high accuracy for correction.8 If the was from ⬃790 to 192 mrad. This reduced the solar
spectral filter shape is not uniform, the uncertainty background by a factor of 17.
of its shape determination has a large inf luence on The use of a photon-counting APD instead of a
the accuracy of the temperature results. With a photomultiplier tube (PMT) can significantly enhance
suitable combination of the magnetic field strength
and vapor-cell temperature of the FADOF, the spectral
transmission curve of the K D1 transition can be tuned
to a uniform shape over the lidar scan range, which
avoids an inf luence of the spectral f ilter shape on
the temperature results. Figure 1(a) shows the full
theoretical transmission curve (see also Yeh1) for the
parameters given in Table 1. The curve has a center
peak of 6.7 pm FWHM and two side peaks of ⬃4-pm
width at 10-pm distance. The center peak has a
nearly uniform transmission of more than 99% within
⬃ 6 1.8 pm around its center. The filter integral
of this FADOF f ilter shape is equivalent to that of a
Fabry – Perot etalon with ⬃10 pm FWHM, f inesse of
50, and peak transmission 85%. Figure 1(b) shows a
comparison of the theoretical shape of the center peak
with the measured shape for the experimental conf igu-
ration of the FADOF of the IAP K lidar. The lower
absolute value of the measured peak transmission
(75%) is due to losses in the optical components of the
detection bench (e.g., lenses, vapor-cell windows,
polarizer cubes). Considering the remaining filter
transmission modulation of ⬃1% over the whole scan
range gives a very small correction of 0.5 K to the
atmospheric temperature. Therefore, any uncertain-
ties in filter-shape determination have only negligible Fig. 1. Spectral transmission curve of the K FADOF
inf luence on the temperature results. at the K D1 line (770 nm) with the parameters given in
The operation principle of a FADOF with its two Table 1. (a) Full theoretical shape. (b) Comparison of
the theoretical shape of the center peak with the measured
crossed polarizers requires linearly polarized light. transmission shape for the experimental conf iguration of
Because of the nature of resonance scattering and the IAP K lidar (thick solid curve) and spectral Doppler
because of depolarization in the fiber cable between shape of the K D1 f ine structure line for two different
the receiving telescope and the detection bench, the atmospheric temperatures [in arbitrary units].
light in the detection bench of the resonance lidar is
unpolarized. Therefore, with a single FADOF only Table 1. Instrumental Parameters of the Mobile IAP
half the total backscattered signal can be detected. K D1 Resonance Lidar for Daylight Operation
By using two parallel FADOFs with orthogonal po-
larization axes, one can avoid this signal loss. The Field of view of receiving telescope) 192 mrad
Fiber cable free aperture 365 mm
mobile IAP K resonance lidar receiver is provided with FADOF
such a twin system (Fig. 2). The unpolarized signal Magnetic f ield strengtha 0.18 T
from the receiving telescope is split by a polarizer Vapor-cell temperaturea 125 ±C
cube into two beams with perpendicular polarization. Transmission (measured) 75%
Behind the analyzing polarizers the two beams are Transmission variation over ,1%
realigned by another polarizer cube. scan range
With respect to our K lidar configuration used for Free aperture兾length 25兾25 mm
nighttime measurements, during daytime, a further APDb
background reduction of at least 3 orders of magni- Quantum eff iciencyc 68%
Dark counts 25 counts兾s
tude is necessary to achieve a ratio between the li-
Active area diameterc 170 mm
dar backscatter signal and solar background of greater Imaging lens system for APDd feff 苷 20 mm
than 1. The aim of uniform transmission over a spec- a
tral range of 3 pm def ines a minimum value for the Extracted from the theoretical fit to the measured transmis-
sion curve [Fig. 1(b)].
possible filter bandwidth. The reduction of the solar b
Avalanche photodiode.
background by such a FADOF alone is not sufficient c
Perkin-Elmer SPCM-AQ module.
d
(2 orders of magnitude, compared with an IF filter of Custom-built multilens system.
1934 OPTICS LETTERS / Vol. 27, No. 21 / November 1, 2002
Fig. 2. IAP K resonance lidar detection bench with a twin
FADOF system. B, magnetic f ield.
Fig. 3. Backscatter signal prof ile of the mobile IAP K res-
onance lidar, obtained on August 28, 2001, at 13:00 local
time on the island of Spitsbergen (78±N). The solar eleva-
tion angle was 21.6±. The prof ile is integrated over 4000
laser pulses with an altitude resolution of 200 m.
Fig. 4. Day and nighttime temperatures from 84- to
98-km altitude, measured with the mobile IAP K reso-
nance lidar on August 28 and 29, 2001 on the island of
Spitsbergen (78± N). The maximum solar elevation angle
was 21.6±. LT, local time.
the quantum efficiency (QE) of the receiver. Above
400-nm wavelength, APDs have much higher QEs
than PMTs, with a maximum at 700 nm. Comparing
the best detectors available, we note that the QE of
the APD is 68% at the K D1 resonance wavelength of
770 nm, whereas the PMT reaches only 15%. This
type of low-noise APD (Perkin– Elmer SPCM-AQ)
has an extremely small active area of only ⬃170-mm
diameter. This places great demands on the imaging
optics in front of the APD and their alignment. The
signal enhancement with the APD by a factor of 4
reduces the statistical error of the lidar retrieval and
improves the signal-to-noise ratio by a factor of 2.
With the new receiver design described above, we
achieve signal-to-solar-background ratios of up to 3
in full daylight. This ratio is strongly dependent on
the K number density in the mesospheric metal layer
and on the amount of scatterers in the troposphere.
Haze in the planetary boundary layer and cirrus
clouds can strongly enhance the solar background,
even if their optical depth is negligible. Figure 3
shows a backscatter count rate prof ile, obtained with
the mobile IAP K resonance lidar at 13:00LT on a
clear day on the Norwegian island of Spitsbergen
(78± N). The K peak density was 26 cm23 . The
solar background passing the FADOF produced a
signal of approximately 2 3 105 s21 (1000 counts per
200-m altitude bin and 4000 laser pulses), whereas
the backscatter signal from the K layer was a factor
of 2 higher. This ratio allowed upper-atmosphere
temperature measurements in full daylight. Figure 4
shows the derived atmospheric temperatures from 84
to 98 km above the island of Spitsbergen. The plot
clearly shows wave structures and other temperature
variations, demonstrating the daylight capability of
the advanced K lidar. The altitude range that is
useful for temperature calculations under daylight
conditions has almost the same extension as during
the night.
In conclusion, an upgrade of the mobile IAP K
resonance lidar with a twin FADOF combined with a
reduced field of view of the lidar and a high-quantum-
efficiency avalanche photodiode allows ground-based
lidar measurements of upper-atmosphere tempera-
tures in full daylight. The achievement of a f lat
spectral filter shape at the lidar scan range avoids
inf luences of instrumental error on the atmospheric
temperature results. Initial 24-h temperature data
were obtained on the islands of Tenerife (28± N) and
Spitsbergen (78± N).
We acknowledge useful discussion with B. P.
Williams of Colorado State University concerning the
physical parameter calculations of the FADOF. M.
Alpers’s e-mail address is alpers@iap-kborn.de.
References
1. P. Yeh, Appl. Opt. 21, 2069 (1982).
2. B. Yin and T. M. Shay, Opt. Lett. 16, 1617 (1991).
3. S. Tomczyk, K. Straender, G. Card, D. Elmore, H. Hull,
and A. Cacciani, Sol. Phys. 159, 1 (1995).
4. K.-H. Fricke and U. von Zahn, J. Atmos. Terr. Phys. 47,
499 (1985).
5. C. Y. She, H. Latif i, J. R. Yu, R. J. Alvarez II, R. E. Bills,
and C. S. Gardner, Geophys. Res. Lett. 17, 929 (1990).
6. U. von Zahn and J. Höffner, Geophys. Res. Lett. 23, 141
(1996).
7. H. Chen, C. Y. She, P. Searcy, and E. Korevaar, Opt.
Lett. 18, 1019 (1993).
8. H. Chen, M. A. White, D. A. Krueger, and C. Y. She,
Opt. Lett. 21, 1093 (1996).