DOI: 10.1007/s00340-008-3075-6
Appl. Phys. B 92, 295–302 (2008)
Lasers and Optics
Applied Physics B
a. amediek
u
a. fix
m. wirth
g. ehret
DevelopmentofanOPOsystemat1.57 µm
for integrated path DIAL measurement
of atmospheric carbon dioxide
Deutsches Zentrum für Luft- und Raumfahrt (DLR) Oberpfaffenhofen, Institut für Physik der Atmosphäre,
82234 Wessling, Germany
Received: 13 November 2007/Revised version: 19 Mai 2008
Published online: 27 June 2008 • © Springer-Verlag 2008
ABSTRACT Active remote sensing is a promising technique to
close the gaps that exist in global measurement of atmospheric
carbon dioxide sources, sinks and fluxes. Several approaches
are currently under development. Here, an experimental setup
of an integrated path differential absorption lidar (IPDA) is pre-
sented, operating at 1.57 µm using direct detection. An injection
seeded KTP-OPO system pumped by a Nd:YAG laser serves as
the transmitter. The seed laser is actively stabilized by means of
aCO
2
reference cell. The line-narrowed OPO radiation yields
a high spectral purity, which is measured by means of a long
path absorption cell. First measurements of diurnal variations
of the atmospheric CO
2
mixing ratio using a topographic target
were performed and show good agreement compared to simul-
taneously taken measurements of an in situ device. A further
result is that the required power reference measurement of each
laser pulse in combination with the spatial beam quality is a crit-
ical point of this method. The system described can serve as
a testbed for further investigations of special features of the
IPDA technique.
PACS 42.65.Yj; 42.68.Wt; 92.60.hg
1 Introduction
One of the most important topics in earth sys-
tem science today is the investigation of the global carbon
cycle. Identification of sources and sinks of atmospheric car-
bon dioxide is a matter of special importance herein [1, 2].
However, the spatial resolution of the existing global
CO
2
network which is based on about 100 measurement sites
monitoring local
CO
2
concentrations with in situ techniques
is not sufficient for detailed analysis [3]. Remote sensing
from space is a promising approach to close the gaps [4].
Active remote sensing can complement passive sensors
that have disadvantages due to interference of atmospheric
aerosol [5], lack in global coverage, or low sensitivity near
the earth’s surface [6]. Furthermore, pulsed active systems
allow precise determination of the column height. How-
ever, active systems are not yet well developed with regard
to
CO
2
measurements. This can partly be attributed to the
very stringent requirements on measurement accuracy be-
u Fax: +49-8153-281271, E-mail: axel.amediek@dlr.de
low 3–4 ppmv in CO
2
volume mixing ratio [7, 8]. This cor-
responds to a precision of about one order of magnitude
higher compared to existing active systems for measuring
atmospheric constituents, such as water vapor, ozone, or
methane [9].
Current developments of pulsed systems for
CO
2
meas-
urements are described by Koch et al. [10] and Gibert et
al. [11]. They report on test measurements with systems at
a wavelength of
2 µm using heterodyne detection. Ismail et
al. [12] developed a
2-µm system with direct detection. Sys-
tems based on continuous wave lasers are described by Spiers
et al. [13] (
2 µm, heterodyne detection) and Krainak et al. [14]
(
1.57 µm, direct detection). Early approaches are described
by Bufton et al. [15] and Sugimoto et al. [16]. None of the sys-
tems mentioned has demonstrated the required accuracy yet.
Hence the development of new systems using severalmethods
of resolution is necessary.
Against this background we developed and demonstrated
an experimental setup of an IPDA system for column con-
tent measurements of atmospheric carbon dioxide. The idea
was the use of an injection seeded optical parametric oscil-
lator (OPO) system as light source with potassium titanyl
phosphate (KTP) as the non-linear medium. The basic prin-
ciple of using pulsed laser light at
1.57 µm wavelength with
a direct detection system, which was chosen within this study,
has to our knowledge not yet been described. We investi-
gate if such a system is able to achieve the stringent accuracy
requirements.
We assume that our approach of a lidar system is able to
achieve high measurement accuracy due to the fact that the
chosen wavelength range allows for the use of state-of-the-
art detectors that provide high sensitivity and that pulsed light
ensures precise determination of range. Furthermore the dir-
ect detection approach avoids problems with speckle noise in
contrast to heterodyne detection [4].
This paper reports on the setup of the developed system,
its special properties and results of first test measurements that
were performed using a horizontal absorption path with a to-
pographic target (trees) in the distance of
2km.
2 IPDA principle
The IPDA technique also known as hard-target li-
dar is a special case of differential absorption lidar (DIAL)
[9, 13]. Instead of atmospheric backscatter the return of a hard
296 Applied Physics B – Lasers and Optics
target, such as vegetation, buildings, water or clouds, etc.,
is used. The target reflex offers the great advantage of this
method of a very strong backscatter signal which allows high
measurement sensitivity. However, any profile information is
lost.
The received power
P of a hard target reflex (assuming
Lambertian scattering) is given by [17]:
P =
π
E
t
eff
A
R
2
T
opt
T
atm
exp
⎛
⎝
−2
R
0
N
CO
2
(r)σ(λ, r)dr
⎞
⎠
, (1)
with the target reflectance
, the laser pulse energy E,the
effective pulse length of the received laser pulse
t
eff
[7], the re-
ceiver telescope area
A, the distance R between system and
target, the overall optical transmission of the receiver system
T
opt
, the optical transmission due to other atmospheric con-
stituents
T
atm
, the number density of CO
2
molecules N
CO
2
at the position r, and the absorption cross section σ of CO
2
at a given wavelength λ at the position r. The absorption
cross section is calculated on the basis of a Voigt profile [18]
that represents the absorption line shape under atmospheric
conditions.
The relation between the powers of the received online and
offline target reflex pulses
P
on
/P
off
using (1) leads to:
R
0
N
CO
2
(r)∆σ(r)dr =
1
2
ln
P
off
P
on
. (2)
The differential absorption cross section is given by
∆σ =
σ(λ
on
) −σ(λ
off
),atwhichλ
on
and λ
off
correspond to wave-
lengths with molecular absorption (online) and without ab-
sorption (offline), respectively.
The column averaged
CO
2
mixing ratio q
CO
2
is defined
by:
FIGURE 1 Schematic of experimental setup
q
CO
2
=
R
0
N
CO
2
(r)dr
R
0
N
air
(r)dr
, (3)
with the number density of (moist) air
N
air
. The following ex-
pression results from (2) in the case of a horizontal path, at
which
∆σ is constant over r :
q
CO
2
=
1
2N
air
∆σR
ln
P
off
P
on
. (4)
Equation (4) is used for the data interpretation of the measure-
ment results shown below.
3 Experimental set-up
Figure 1 shows the principle of the setup. Special
features are a two-wavelength injection seeded OPO system
providing alternating two narrow-band wavelengths around
1.57 µm, a stabilized online wavelength via reference cell,
a spectral purity control via multipass cell, a spectrum ana-
lyzer, and a direct detection receiver system with power ref-
erence measurement of each outgoing laser pulse. All main
properties are listed in Table 1. The
CO
2
absorption line used
(online:
1572.993 nm,offline:1573.16 nm) provides almost
no interference to atmospheric water vapor and shows low
temperature dependence of the absorption cross section. The
online and offline wavelengths were selected by means of nu-
merical simulations based on the absorption line parameters
of the HITRAN 96 database [19].
The OPO system [20] acts as a tunable frequency con-
verter to generate the desired wavelengths. The fundamen-
tal (
1064 nm) of an injection seeded, Q-switched and flash
lamp-pumped Nd:YAG laser (Continuum NY-61) serves as
the pump radiation for the OPO [21]. The OPO is designed
as a ring cavity to avoid back reflections that may introduce
spectral instabilities or damage to the seed lasers. In combina-
tion with the distributed feedback (DFB) seed lasers that were
AMEDIEK et al. Development of an OPO system at 1.57 µm for integrated path DIAL measurement of atmospheric CO
2
297
Transmitter
Pump source Flash lamp pumped, Q-switched Nd:YAG with integrated injection seeder
Pump wavelength 1064 nm
Repetition rate 10 Hz
Incident pump energy for OPO 70–100 mJ per pulse
Pulse length 8 ns
OPO
Ring cavity length (optical) 92 mm
Non-linear medium KTP, (5×5 ×20) mm
3
, ϑ = 90
◦
, ϕ =0
◦
Phase-matching scheme Non-critical
Output energy 10 mJ per pulse
Output wavelength 1572.1 to 1574.6nm
Pulse length 5 ns
Bandwidth (unseeded) 0.2nm(FWHM)
Threshold 55 mJ unseeded, 40 mJ seeded
Injection seeding
Seed laser DFB telecom diode laser, fiber coupled
Online wavelength 1572.993 nm
Offline wavelength 1573.16 nm
Data rate (online/offline pulse pairs) 5 Hz
Receiver
Telescope Newtonian, 15 cm diameter
Field of view ≈4mrad
Detector InGaAs-PIN (Hamamatsu G8605,
Active area ∅ 1mm)
Data acquisition 12-bit, 400 MHz
TABLE 1 Technical data of the CO
2
-IPDA system
used for the injection seeding it was not necessary to use ad-
ditional protection like a Faraday isolator. KTP was selected
as non-linear medium because of its favorable properties [22].
A non-critical phase-matching scheme is used [23]. The cav-
ity length is fine-tuned via a piezo element. The cavity mirrors
are embedded in a solid aluminum block, leading to a good
passive stability of the OPO. The generated signal radiation
(
1.57 µm) serves as measurement radiation, whereas the idler
radiation at
3.3 µm is not used.
Figure 2 shows the longitudinal mode structure of the
OPO: During this measurement the wavelength of the online
seed laser was continuously tuned. The increase of the OPO
output power during the resonance of the seed laser, at which
the injection seeding is active, represents the OPO modes. The
relative frequency was identified by means of the signal of
a Fabry–P
´
erot interferometer (FPI). The distance between two
FIGURE 2 Measured longitudinal mode structure of
the OPO. The increase of the OPO output power dur-
ing seed laser resonance is displayed. Additionally the
markers of the Fabry–P
´
erot interferometer (FPI) for
relative wavelength calibration are shown
peaks corresponds to twice the cavity length, because a double
circulation in the cavity occurs. It was possible to reduce the
in-between modes by adjustment of the cavity, but a complete
suppression of this occurrence could not be achieved. Fig-
ure 4 depicts the dependence of the OPO output energy on the
pump energy. The conversion efficiency is about
10%.Since
the output energy of
10 mJ per pulse was fully sufficient fur-
ther optimization of the OPO’s efficiency was not undertaken.
Figure 4 indicates that no saturation effects occur. The band-
width of the free running OPO (without injection seeding) is
about
0.2nmFWHM. This is appropriate for the generation
of the two defined wavelengths, online and offline (distance:
0.17 nm), by means of injection seeding, with a single OPO
configuration.
By means of temperature tuning [23, 24] and non-collinear
phase matching [23, 25] (tuning of the angle between pump
298 Applied Physics B – Lasers and Optics
FIGURE 3 Tuning range of the OPO. Four of the possible OPO radiation spectra (unseeded) within the tuning range are displayed. The tuning was per-
formed by means of temperature variation and non-collinear phase matching. Additionally matching CO
2
absorption lines (calculated using HITRAN) are
shown
FIGURE 4 OPO output energy vs. pump energy
beam and signal beam) it was possible to fine-tune the band-
width of the free running OPO and match to the defined online
and offline wavelengths (see Figs. 3 and 5).
The method of injection seeding [26] provides the two de-
fined narrow-band wavelengths that act as online and offline
wavelengths, within the bandwidth of the free running OPO
(see Fig. 5).
Seeding is performed by means of two fiber coupled
commercially available telecom DFB diode lasers (Fujitsu
FLD5F15CA-S9090), one for online and offline, respectively.
Their radiation is alternately coupled to the OPO via a fast
fiber switch (Agiltron “CrystaLatch”) (Fig. 1). Switching is
performed synchronously to the
10 Hz Nd:YAG pump laser
pulses. So the effective measurement data rate is as high
as
5Hz. The DFB laser modules contain an optical isola-
tor (
22 dB) that protects against possibly remaining back
reflections.
The absolute wavelength accuracy needed and the stability
of the online seed laser are achieved by using a 36-m multipass
absorption cell filled with pure
CO
2
(10 hPa pressure) as the
reference [27]. A very high slope of the absorption line’s edge
allows for a precise control loop. It was stabilized to the line
center of the atmospheric pressure-broadened line. For that
purpose a simple edge control was used, since the line max-
imum of the atmospheric line coincides with the edge of the
reference cell’s line, because of the pressure shift of the atmo-
spheric absorption line compared to the reference cell’s line.
A precise knowledge of the pressure shift coefficient [28] is
a precondition to avoid a wavelength bias. The offline seed
laser was not actively stabilized due to its passive stability of
better than
1.3 GHz (monitored by means of the spectrum an-
alyzer), which leads to an error of below
0.1% in the case of
the shown measurements.
Figure 6 shows the achieved stability of
5MHz(rms) of
the online seed laser. In the case of the shown measurements
(horizontal path) an online stability of better than
70 MHz is
needed to achieve an error of below
0.1%. It is assumed that
there is no significant wavelength offset between the radiation
of the seed laser and the radiation emitted by the OPO. An off-
set (for example due to cavity pulling if the cavity is slightly
out-of-tune) can be measured and controlled by a heterodyne
measurement (superposition of seed laser and OPO) [29, 30].
This approach was successfully tested, but not applied during
the shown measurements.
FIGURE 5 Spectra of the OPO radiation measured by means of an optical
spectrum analyzer: narrow-band online and offline radiation during seeded
operation (straight line), free running OPO without seeding (dotted line)and
calculated CO
2
absorption lines (dashed line)
AMEDIEK et al. Development of an OPO system at 1.57 µm for integrated path DIAL measurement of atmospheric CO
2
299
FIGURE 6 Online stabilization. (a) Absorption line scan of CO
2
(36-m multipass cell) at 10 hPa pressure (straight line), calculated line profile (dotted line),
Fabry–P
´
erot interferometer markers for relative wavelength calibration (grey). (b) Time behavior of the stabilized online seeder: the relative frequency is
deduced from the photodiode signal via the calibrated line scan. An accuracy of 5 MHz (rms) is achieved
It is crucial to ensure a sufficient high spectral purity [31]
of the emitted online laser radiation. Otherwise the effect-
ive differential absorption of
CO
2
was undefined, resulting in
a large uncertainty in calculating its mixing ratio. In the case
of the shown measurements a spectral purity of
≥ 99.9% is
needed to achieve a systematic error of
≤0.1%.
The spectral purity was determined by means of a long
pass absorption cell (
210 m) filled with pure CO
2
(35 hPa
pressure). The strongly saturated absorption line acts as
a highly wavelength-stable narrow-band filter (
1.3 GHz
FWHM). Spectral impurity due to OPO side modes or broad-
band emission can accurately be measured by means of this
setup [27].
A spectral purity of the injection seeded OPO online ra-
diation of as high as
99.9% was determined. The OPO cavity
length was manually controlled via a piezo element (Fig. 1) to
maintain a maximum spectral purity.
Figure 7 depicts the configuration that was used for the test
measurements. A Newtonian telescope with a
15 cm diameter
FIGURE 7 Setup of the receiving optics. The light collected by the tele-
scope is homogenized by a scattering disc and focused on the detector. The
power reference measurement is implemented by means of two beamsplitters
that guide a small portion (homogenized by a second scattering disk) of the
outgoing beam to the detector
mirror was used to collect the backscattered light. The usage
of the same detector for power reference and the target reflex
measurements avoids variances in the response that may occur
in the case of two detectors. The implementation of scattering
disks that homogenize the light also plays a key role. It was not
possible to achieve a signal that was sufficiently stable with-
out these components. However, strong signal losses are the
consequence.
A PIN diode was used as detector, although the usage of
an APD is aspired due to the comparatively high sensitiv-
ity. However, APDs currently available have only very small
active areas (about
0.2mm diameter), requiring a more so-
phisticated focusing setup. Hence, in this first setup, the PIN
diode was chosen.
4 Test measurements
4.1 Configuration
The test measurements were performed synchron-
ously with the IPDA system and an in situ device (Carbondio,
Pewatron AG, Switzerland), which detected the
CO
2
mix-
ing ratio of the air outside of the laboratory (
∼10 m above
ground). It is calibrated by means of two gas mixtures with
known
CO
2
mixing ratios (air without CO
2
to set the zero-
point and air containing
520 ppm CO
2
to calibrate the scale).
The IPDA measurement runs along a
2km horizontal path
(about
10 m above ground) from the laboratory to a group of
trees that act as a hard target. The surrounding area is rural and
no notable local
CO
2
sources are known.
Diurnal variations of the atmospheric
CO
2
mixing ratio
that naturally occur near the ground are measured. To get
strong gradients of the mixing ratio the following special situ-
ation was chosen for measurement: the break-up of a ground
inversion that is usually built up in spring and summer under
cloudless conditions at night. In such an inversion layer
the
CO
2
that is mainly produced by vegetation accumulates
near the ground during night and mixing ratios of up to
500 µmol/mol can be observed. When the solar driven atmo-