1214 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

CALIPSO Lidar Description and Performance Assessment

WILLIAM H. HUNT
Science Systems and Applications, Inc., Hampton, Virginia

Downloaded from [Link] by guest on 20 June 2020

DAVID M. WINKER, MARK A. VAUGHAN, AND KATHLEEN A. POWELL
National Aeronautics and Space Administration, Hampton, Virginia

PATRICIA L. LUCKER
Science Systems and Applications, Inc., Hampton, Virginia

CARL WEIMER
Ball Aerospace and Technology Corporation, Boulder, Colorado

(Manuscript received 11 September 2008, in final form 10 January 2009)

ABSTRACT

This paper provides background material for a collection of Cloud–Aerosol Lidar with Orthogonal Po-
larization (CALIOP) algorithm papers that are to be published in the Journal of Atmospheric and Oceanic
Technology. It provides a brief description of the design and performance of CALIOP, a three-channel
elastic backscatter lidar on the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO)
satellite. After more than 2 yr of on-orbit operation, CALIOP performance continues to be excellent in the
key areas of laser energy, signal-to-noise ratio, polarization sensitivity, and overall long-term stability, and
the instrument continues to produce high-quality data products. There are, however, some areas where
performance has been less than ideal. These include short-term changes in the calibration coefficients at both
wavelengths as the satellite passes between dark and sunlight, some radiation-induced effects on both the
detectors and the laser when passing through the South Atlantic Anomaly, and slow transient recovery on the
532-nm channels. Although these issues require some special treatment in data analysis, they do not seriously
detract from the overall quality of the level 2 data products.

1. Introduction CALIPSO was launched on 28 April 2006 for a

scheduled 3-yr mission probing the vertical structure
This paper describes the design and performance of
and properties of clouds and aerosols, with the objective
the Cloud–Aerosol Lidar with Orthogonal Polarization
of obtaining a better understanding of how clouds and
(CALIOP), a three-channel elastic backscatter lidar
aerosols form, evolve, and affect weather and climate
that is the prime payload instrument on the Cloud–
(Winker et al. 2007). The satellite orbits in close formation
Aerosol Lidar and Infrared Pathfinder Satellite Obser-
with Aqua, CloudSat, Parasol, and Aura as part of the
vation (CALIPSO) satellite. It provides background
Afternoon Constellation (A-Train) of earth-observing sat-
material for a collection of CALIOP algorithm papers
ellites (Stephens et al. 2002), in a sun-synchronous orbit at
that are to be published in the Journal of Atmospheric
705-km altitude with an ascending node equator crossing
and Oceanic Technology (Winker et al. 2009).
time of 1330 local solar time (LST). The orbit inclination
of 988 provides coverage between 828N and 828S, with
Corresponding author address: William H. Hunt, NASA Lang-
orbit tracks that repeat every 16 days. CALIPSO is the
ley Research Center, M.S. 475, Hampton, VA 23681–2199. product of collaboration between the National Aero-
E-mail: [Link]@[Link] nautics and Space Administration (NASA) and the

DOI: 10.1175/2009JTECHA1223.1

Ó 2009 American Meteorological Society

JULY 2009 HUNT ET AL. 1215

Downloaded from [Link] by guest on 20 June 2020

FIG. 1. CALIOP block diagram, showing simplified optical signal paths to the detectors for all three channels and
signal processing electronics for one channel. Electronics for the other two channels are the same as the one shown.
The different downlink paths and frequencies for Science data and Health and Status data are also shown.

French space agency Centre National d’Etudes Spatiales The transmitter subsystem includes two lasers, each
(CNES). Ball Aerospace and Technologies Corporation with a beam expander, and a beam steering mechanism.
(BATC) was the prime NASA contractor for the pay- The Q-switched diode-pumped Nd:YAG lasers, built by
load, and the PROTEUS spacecraft was built by Alcatel Fibertek under contract to BATC, operate one at a
under contract to CNES. The payload combines an ac- time, and each is designed to operate for the full mission
tive lidar instrument (CALIOP) with passive infrared lifetime. Only the primary laser has been fired since
and visible imagers (Winker et al. 2003). CALIOP is the launch, with the second laser available as a backup.
third lidar launched by NASA to study the earth’s at- The lasers are frequency-doubled to produce simul-
mosphere from space, having been preceded by the Lidar taneous pulses at 1064 and 532 nm. A beam expander on
In-Space Technology Experiment (LITE), launched in each laser reduces the angular divergence of the output
September 1994 on STS-64 (Winker et al. 1996), and the beam, allowing a small receiver field of view (FOV) to
Geoscience Laser Altimeter System (GLAS), launched be used for better solar background rejection. Each la-
in December 2002 on the Ice, Cloud and Land Elevation ser is housed in its own sealed canister filled with dry air
Satellite (ICESAT) (Spinhirne et al. 2005). CALIOP has at standard atmospheric pressure. Energy monitors lo-
already provided far more high-quality global atmo- cated within each canister measure the output pulse en-
spheric data than the two earlier missions, and it con- ergy of each shot at each wavelength, and the energy
tinues to add to that dataset. values are downlinked with both the science data and the
Health and Status (H&S) data. The two lasers are
2. CALIOP description
mounted on a common beam steering system that can
a. Transmitter subsystem adjust their pointing direction for optimizing the boresight
alignment between the transmitter and receiver (Seckar
The lidar consists of a transmitter subsystem that
2005). Key laser specifications are given in Table 1.
sends a laser pulse down through the atmosphere, a
receiver subsystem that collects and measures the light
b. Receiver subsystem
that is backscattered from the laser pulse, a payload
computer that controls the subsystems and does some The receiver subsystem consists of a 1-m-diameter
onboard processing of the signals from the receiver, and all-beryllium telescope, relay optics, detectors, and elec-
the data downlink systems (Fig. 1). tronics mounted on a carbon composite optical bench.
1216 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

TABLE 1. CALIOP transmitter specifications. TABLE 2. CALIOP receiver specifications.

Wavelengths 532 and 1064 nm Telescope diameter 1m

Pulse energy at each wavelength 110 mJ Field of view 130 mrad
(nominal) Optical filter bandwidth (532 nm) 35 pm
Repetition rate 20.16 Hz Optical filter bandwidth (1064 nm) 400 pm
Ground spot spacing 335 m Detector effective quantum efficiency 0.11
Polarization purity at 532 nm .99% (532 nm)
Beam divergence at each wavelength 110 mrad Detector effective quantum efficiency 0.40
(nominal) (1064 nm)
Line width at 532 nm 28 pm Detector dark count rate (532 nm) 2.3 3 103
Detector dark count rate—beginning 2.0 3 107
of mission (1064 nm)
A field stop at the telescope focus limits the lidar FOV Digitizer sample rate 10 MHz
to an angle close to the laser divergence. The telescope Vertical sample spacing 15 m

Downloaded from [Link] by guest on 20 June 2020

entrance pupil is imaged onto the detectors, thereby Electronic bandwidth 2.0 MHz
Vertical resolution as determined by 30 m
eliminating hot spots as a function of viewing angle. A
bandwidth
dichroic beam splitter separates the two wavelengths, Digitizer resolution 14 bits
and a polarization beam splitter separates the 532-nm Maximum dynamic range (merged) 2.5 3 106 (.21 bits)
beam into components parallel to and perpendicular to
the laser polarization. Solar background rejection is
provided by a narrowband etalon in combination with a small ponds would reflect the lidar beam directly back
dielectric interference filter in the 532-nm channels and into the receiver and drive the detector system into
an interference filter alone in the 1064-nm channel. A heavy saturation, an undesirable situation. A pointing
pseudodepolarizer can be moved into the 532-nm beam direction very close to nadir facilitates the identification
on command for measurement of the relative sensitiv- of cirrus clouds with horizontally oriented ice crystals
ities of the two polarization channels. (HOICs) (Platt 1978). In this case the backscatter is
After separation, each beam component goes to its dominated by specular reflections from the ice crystals,
own detector. For the 532-nm channel detectors, pho- making identification of HOICs easy but making the
tomultiplier tubes (PMTs) were chosen because their measurement of cloud optical properties such as depo-
low dark noise more than compensates for their relatively larization and optical depth difficult, if not impossible. If
modest quantum efficiencies. For the 1064-nm channel, the lidar pointing direction is several degrees off nadir,
an avalanche photodiode (APD) was chosen because its then the specular reflections are avoided and cloud
much higher quantum efficiency at that wavelength optical properties can be measured. A small off-nadir
more than compensates for its higher dark noise. angle of 0.38 was chosen at the beginning of the mission,
The outputs of the three detectors go through trans- allowing statistics on HOIC clouds to be obtained. After
impedance amplifiers (TIAs), which convert the detector acquiring more than a year of cloud statistics at 0.38, in
output currents to voltages. To achieve the more than six November 2007 the pointing direction was increased to
orders of magnitude dynamic range required by CALIOP, 3.08 off nadir, where it has remained since that date. The
each TIA output is split and sent in parallel to two am- larger angle avoids most specular reflections from
plifiers with different gains, each of which is followed by a HOICs, thus allowing the physical and optical proper-
14-bit digitizer (designated the high-gain digitizer and the ties of the clouds to be determined more accurately
low-gain digitizer for simplicity). The two digitizer outputs (Hu 2007). The pointing angle offset from nadir was
are later merged during onboard processing, resulting in along track in the forward direction in both cases.
an effective dynamic range of more than 21 bits.
The receiver also includes a built-in test system d. Data types and downlink paths
(BITS) that can be activated on command to test the
Before discussing onboard data processing, it is helpful
performance of the detectors and electronics by illu-
to understand the two primary data categories and their
minating the detectors with known optical profiles from
downlink paths and downlink frequencies. Health and
LEDs. Key receiver specifications are given in Table 2.
Status data include a large number of engineering pa-
rameters along with many parameters derived from the
c. Lidar pointing direction
science profile data. H&S data are downlinked several
The choice of the lidar pointing direction involves times a day via S band and sent to the CALIPSO Mission
balancing several factors. Pointing exactly toward nadir Operations and Control Center (MOCC) in Hampton,
is to be avoided because smooth water surfaces such as Virginia, where their relatively small latency (less than
JULY 2009 HUNT ET AL. 1217

12 h) allows timely evaluation of the payload health and 3) DATA ACQUISITION REGIONS
status and the instrument performance. Science data
The data acquisition timing sequence begins at the
consist of backscatter profiles and ancillary data for all
target altitude and proceeds through the four regions
channels for every shot, along with a second copy of the
shown in boxes in Fig. 2. The operations associated with
H&S data. They are downlinked daily via an X-band
each region are described below.
transmitter on the payload and sent to the Atmospheric
Sciences Data Center (ASDC) at NASA Langley Re- (i) Electronic background measurement region
search Center in Hampton, Virginia, where they are used (112 to 97 km)
for level 1 and level 2 science processing. Because of the
large quantity and longer latency (up to 36 h) of the sci- To avoid the large variations in the digitizer level as
ence data, it is useful to extract any science data infor- the solar background varies during the daytime, each
mation needed for instrument performance evaluation channel’s mean background signal is electronically sub-
tracted. The background signals are averaged over the

Downloaded from [Link] by guest on 20 June 2020

and insert it into the H&S data, as will be discussed below.
97–112-km-altitude region, where the backscatter signal
e. Onboard processing is negligible, and the measured background values are
subtracted electronically from the detector output sig-
Onboard computations of position, altitude, and sun nals and also reported in the H&S data. After the mean
elevation allow automatic reconfiguration and timing background signals are removed, they are replaced with
control. When combined with onboard processing of fixed dc offsets to ensure that the signal levels at the
science data, a large reduction in the downlink data rate digitizers are always positive, even in the presence of
is achieved, allowing a full day’s science data to be negative-going noise variations.
downlinked in a single X-band contact. Onboard pro-
cessing also includes extraction of selected science in- (ii) Offset measurement region (80 to 65 km)
formation for insertion into the H&S data, aiding in the After the signals are digitized, the dc offsets that were
timely monitoring of the health and performance of added before digitization must be subtracted from the
the instrument. The following paragraphs provide more digital data before further processing, leaving only the
details about onboard processing features. backscatter signals. Because the a priori knowledge of
1) RECONFIGURATION FOR NIGHT OR DAY the offsets is insufficient for this purpose, they must be
measured. This is done by averaging the digitized sig-
The optimum receiver gains and offsets are different nals between 65 and 80 km, where the backscatter is
for night and day because of the vastly different back- negligible. The measured offsets for all six digitizers are
ground noise levels. CALIOP changes these param- subtracted from their respective digitized signals, and
eters autonomously based upon the value of the sun– the values are reported in the H&S data, where they are
earth–satellite (SES) angle, which is computed once per used for diagnostic purposes.
second using data from the platform global positioning The standard deviation (RMS noise) is computed for
system (GPS) receiver. Whenever the SES angle is be- all channels for every shot over this same region. Be-
tween 958 and 1658 (equivalent to a solar elevation angle cause there is negligible backscatter at this altitude, this
of less than 258), the lidar is automatically configured for measurement gives the sum of the detector dark noise
night; otherwise, it is configured for day. and the noise due to solar background light. The averaged
2) TIMING CONTROL AND GEOLOCATION values are downlinked in the H&S data, where they are
CALCULATIONS used for diagnostic purposes, and the single-shot values
are downlinked in the science data, where they are used
Before each shot, the footprint location and range to for setting thresholds in the automated layer detection
mean sea level (MSL) for the next shot are computed algorithm during level 1 processing.
from an on-orbit geoid model, utilizing position and After the mean offset and the RMS noise values are
altitude data from the platform GPS receiver and atti- computed, the individual samples from this region are
tude data from the platform star tracker. From these discarded.
values, the range to target altitude and the time to
(iii) Backscatter measurement region
target altitude are computed (Fig. 2) and the data ac-
(40 to 22 km)
quisition timing is set accordingly. Autonomous timing
adjustment not only avoids the collection of unneeded Before downlink, the backscatter data from the low-
data, but it also facilitates the altitude-dependent signal and high-gain digitizers for each channel are combined
averaging discussed later. into a single merged profile. In simplified form, each
1218 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

Downloaded from [Link] by guest on 20 June 2020

FIG. 2. Data acquisition regions shown in terms of their altitudes above MSL. The target altitude and the altitudes
of all the regions below it are fixed. Because the satellite altitude varies with latitude, so does the range to
target altitude, and the time to target altitude must be adjusted on every shot to compensate for the changing range.
The times of all other operations in the data acquisition sequence are kept fixed relative to the target altitude. The
operations defined for each region take place sequentially on every shot, going from the top down.

sample from the high-gain digitizer is examined to see if atmosphere is more homogeneous. Table 3 shows the
it is saturated (full scale). All saturated samples, along amount of horizontal and vertical averaging done on
with their nearest neighbors, are replaced with rescaled board in each of five backscatter altitude ranges. Table 4
values from the low-gain digitizer. The neighbors are gives the spatial resolutions of the downlinked science
included to allow for phase differences between the two data after onboard averaging.
digitizers and for transient effects after the saturated The onboard averaging, along with the merging of data
sample. The rescaling factors are computed for each from the high- and low-gain digitizers, reduces the back-
15-shot frame, and those factors are reported in the scatter data rate by a factor of 17.2 relative to the raw data.
H&S data. Any remaining offset after merging is also In addition to the averaging that is done on the science
subtracted during this process. profile data, much more heavily averaged values are
The profile data rate is further reduced by altitude- extracted from the science data and inserted into the
dependent averaging, both vertically and horizontally. H&S data to aid in performance evaluations in the
The least amount of averaging is done at low altitudes, MOCC. For the H&S data, the profile data from each of
where the natural spatial variability is the greatest, and the five altitude ranges in Table 3 are averaged vertically
the greatest amount is done at high altitudes where the over the entire region and horizontally over 120 shots.

TABLE 3. Onboard averaging of CALIOP data. No 1064-nm channel data are downlinked from the highest altitude region.

Digitizer samples averaged Digitizer samples averaged

Altitude region (km) Shots averaged 532-nm channel 1064-nm channel
30.1 to 40.0 15 20 N/A
20.2 to 30.1 5 12 12
8.2 to 20.2 3 4 4
20.5 to 8.2 1 2 4
22 to 20.5 1 20 20
JULY 2009 HUNT ET AL. 1219

TABLE 4. Spatial resolutions of CALIOP downlinked data resulting from the averaging shown in Table 3. No 1064-nm channel data are
downlinked from the highest altitude region.

Altitude Horizontal Vertical resolution (m) Vertical resolution (m)

region (km) resolution (km) 532-nm channel 1064-nm channel
30.1 to 40.0 5.025 300 N/A
20.2 to 30.1 1.675 180 180
8.2 to 20.2 1.005 60 60
20.5 to 8.2 0.335 30 60
22 to 20.5 0.335 300 300

(iv) Lower offset measurement region enced by the LITE laser, and Afzal (2006) and Afzal

Downloaded from [Link] by guest on 20 June 2020

(211 to 218.5 km) et al. (2007) describe the GLAS laser experience. Sev-
eral factors have contributed to the superior perfor-
Mean values from the lower offset region are mea-
mance of the CALIOP laser (Hovis 2006). Among them
sured and reported in the H&S data but are not other-
are very stringent contamination control, operation at
wise used.
derated energy levels, and careful attention to lessons
4) EXTRACTION OF OTHER INFORMATION FOR learned from previous lasers, including a risk-reduction
THE HEALTH AND STATUS DATA laser similar to the flight design that was operated for
more than two billion shots prior to the CALIPSO
Several other quantities, such as overflow and under-
launch.
flow counts, are computed from the raw low- and high-
gain digitizer data and inserted into the H&S data, thus (ii) Laser canister pressure
preserving information that would otherwise be lost
CALIOP has been experiencing a slow loss of pres-
after onboard processing.
sure in the primary laser canister and will have to switch
to the backup laser in early 2009. Between June 2006
and May 2008, the absolute pressure in the primary laser
3. CALIOP performance
canister dropped from 16 psi to just over 6 psi. The
a. Performance trends backup laser pressure has remained above 17 psi over
the same period (Fig. 4). The loss of pressure has no
Overall, the performance of CALIOP has met or
effect on the laser performance, but it will present a
exceeded nearly all requirements and expectations and
danger of a corona discharge within the canister, and
has shown little degradation over nearly 2 yr of opera-
possibly associated damage to the electronics, if the
tion, as will be illustrated in the following sections. The
amount of adjustment that has been required to main-
tain the performance has been minimal, consisting
mostly of boresight alignments.
1) LASER PERFORMANCE
(i) Laser energy
Based on energy measurements on every shot at both
wavelengths by internal energy monitors, the primary
laser has maintained its performance extremely well,
continuing to produce more than 94% of its initial en-
ergy at both wavelengths (Fig. 3) after almost 2 yr of
near-continuous operation (more than 1.1 billion shots
on orbit). Although the laser energy can be adjusted by FIG. 3. Time history of the energy of the primary laser at both
ground command, no adjustments of any kind have wavelengths, as measured by the internal energy monitors. The
been needed to maintain this high level of performance. loss of energy since the beginning of the mission has been less than
The backup laser has not yet been fired on orbit. 6%, and most of that occurred in several downward steps that are
believed to be due to failures of individual diode bars. The diode
The CALIOP laser has performed much better than bar failure rate and the overall energy degradation rate are similar
the lasers on some previous spaceborne lidar missions. to what was observed during testing of a risk reduction laser prior
Winker et al. (1996) describes some problems experi- to the mission.
1220 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

pressure is allowed to drop too low. A switch to the

backup laser is planned for late February 2009, when the
primary laser pressure is expected to be about a factor of 2
above the corona threshold of 1.9 psi.

(iii) Low-energy laser shots

Although the peak-to-peak energy variation between
shots over a short time is typically less than 1.5%, there
are very occasional shots with much lower energy. As of
1 June 2008, about 0.006% of the total shots (100 per
day out of 1.74 million) had energies less than 45% of
the mean, and that percentage continues to rise. The

Downloaded from [Link] by guest on 20 June 2020

very low-energy shots occur most frequently in the
South Atlantic Anomaly (SAA), a high-radiation re- FIG. 4. Pressure trends for the two CALIOP lasers, showing the
gradual loss of pressure in the primary laser canister (irregular
gion. This implies a probable radiation connection, but solid line) since 1 Jun 2006. Also shown (dashed line) is the pre-
changes in the frequency of occurrence of the low-energy dicted pressure trend for the primary laser, based on a fit to
shots do not appear to be correlated with any changes measurements made soon after launch. The switch to the backup
in the radiation activity within the SAA. The source of laser, which is maintaining its pressure well, will be done when the
this behavior is the subject of an ongoing study, but primary pressure is about a factor of 2 above the corona threshold
of 1.9 psi.
neither the mechanism nor the trend is understood at
this time. The low-energy shots currently do not occur
frequently enough to noticeably impact the science data
quality. this SNR be at least 50 for the specified molecular
backscatter coefficient when averaged over 5 km ver-
2) SIGNAL-TO-NOISE RATIO tically and 1500 km horizontally, averaging distances
The accuracy with which a lidar can measure back- similar to those used for the nighttime calibration of
scatter signals depends on its signal-to-noise ratio the 532-nm parallel channel. To avoid undue influence
(SNR). The SNR of CALIOP is much lower than that from changes in the backscatter coefficient with
from typical ground-based, airborne, or shuttle-based latitude, the SNR was determined by measuring the
lidars because of the large distance between the lidar mean and standard deviation of the backscatter sample
and its targets and the modest power aperture product closest to 30 km over a number of 1500-shot segments
that is dictated by weight and electrical power limita- and then averaging those values to give the SNR at the
tions. Although the SNR can be improved by more downlink resolution. The measured SNR was then
vertical and horizontal averaging, the amount of aver- extrapolated to the specified backscatter coefficient
aging that is acceptable is limited by the spatial scale of value and averaging distances for comparison with the
the target. Early in the CALIOP design phase, day and requirement, assuming that the SNR scales as the
night SNR requirements were established that took into square root of both the signal level and the number of
account the required accuracy and the allowable amount shots.
of averaging for a number of targets and lighting con- The initial on-orbit measurement in June 2006 gave
ditions. Simulations done at various stages of the in- SNR 5 82.6, more than 60% above the requirement
strument development, as well as SNR measurements (Fig. 5). The latest measured value is only 4% below the
using on-orbit data, have verified that the instrument initial value.
exceeds all SNR requirements.
(ii) SNR: 532-nm perpendicular channel at night
(i) SNR: 532-nm parallel channel at night
The SNR requirement for the 532-nm perpendicular
The SNR value that has received the closest scrutiny channel is the same as for the 532-nm parallel channel
is that for the 532-nm parallel channel molecular when measuring the same signal level. Calculations us-
backscatter from 30 km. This measurement is crucial ing data from several polarization calibrations, where
because it is the basis for the calibration of all channels, the two channels are exposed to the same optical signal
day and night. It also is the easiest SNR to verify be- levels, give nearly the same SNR for both channels, thus
cause the signal is free of solar background and aerosol verifying that the perpendicular channel meets its re-
or cloud contamination issues. The requirement is that quirement.
JULY 2009 HUNT ET AL. 1221

FIG. 5. Time history of the nighttime 532-nm parallel channel

Downloaded from [Link] by guest on 20 June 2020

SNR measured values, extrapolated to conditions specified in the
requirements. The most recent value is only about 4% below the
initial value and is more than 60% above the required value of 50. FIG. 6. Time history of the 1064-nm channel dark noise, as
The minimum in early 2008 occurred when both the etalon tuning measured on board and reported in the H&S data. The shape of
and the boresight alignment had drifted from their optimum set- the curve is consistent with a linear increase in the APD bulk dark
tings. Optimizing those adjustments restored much of the lost SNR. current caused by a linear increase in the cumulative radiation
dosage. The APD has been maintained at a constant temperature.

(iii) SNR: 1064-nm channel at night

3) CALIBRATION COEFFICIENT, 532-NM
Although there is no nighttime SNR requirement for PARALLEL CHANNEL AT NIGHT
the 1064-nm channel, it is interesting to examine how
that SNR is affected by the increasing APD dark noise. The calibration coefficient is a good indicator of the
Sun et al. (1997) have shown that APDs in a space en- overall detection sensitivity of the lidar because it defines
vironment experience an increase in bulk dark current the relationship between the backscatter coefficient of
that is proportional to the cumulative radiation expo- the target and the signal measured by the lidar. (Powell
sure, resulting in a gradual increase in dark noise. et al. 2009) In simplified notation, that relationship is
Measurements of the CALIOP 1064-nm channel dark given by Eq. (1):
noise confirm an increase that is consistent with a linear r2 S(z)
increase in the bulk dark current (Fig. 6). C5 , (1)
EGA T 2 (z)b(z)
The effect of the increased dark noise on the 1064-nm
channel nighttime SNR is shown in Fig. 7. The target in where C is the calibration coefficient, r is the range from
this case is a moderate boundary layer aerosol with the lidar to the target, z is the altitude of the target
backscatter coefficient 5 1 3 1023 km21 sr21, averaged above MSL, S(z) is the measured signal from the target,
over 1 km vertically and 25 km horizontally. The most
recent nighttime SNR for the specified target is about
65% of the initial value.

(iv) SNR: Daytime

Even though extensive efforts have been made to keep
the solar background light as low as is reasonable during
daylight, the daylight SNR is nevertheless significantly
lower than the SNR at night. The noise varies with the
scene being viewed and is highest over bright clouds or
snow and ice. A daytime SNR requirement was estab-
lished for each channel, based on specified targets, al-
bedo, and averaging distances. For both 532 channels, the
initial measured daytime SNR was more than 30% above
the requirement and has only decreased approximately FIG. 7. Time history of the nighttime 1064-nm channel SNR
based on a combination of measurements and simulations. Most of
7% since the start of the mission. The daytime SNR of the the 35% decrease in SNR is due to the increase in the APD dark
1064-nm channel has decreased approximately 10% but noise. The time resolution is not sufficient to show short-term
remains about a factor of 3 above its requirement. changes such as the dip in Fig. 5.
1222 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

FIG. 9. Time history of the polarization gain ratio (PGR) based

on data from polarization calibrations. The PGR is the ratio of the

Downloaded from [Link] by guest on 20 June 2020

FIG. 8. Time history of the normalized calibration coefficient for 532-nm perpendicular signal to the 532-nm parallel signal when
the 532-nm parallel channel at night. The calibration coefficient is both detectors are illuminated equally, as is the case during po-
a good indicator of the detection efficiency of the lidar. The slight larization calibration operations. The PGR is used in conjunction
downward trend reflects not only the effects of component deg- with the 532-nm parallel calibration coefficient to calibrate the
radation, but also boresight misalignment and etalon wavelength 532-nm perpendicular channel.
mismatch. Most upward steps result from boresight alignments or
etalon retuning, although there is an unexplained increase of about
2% after changing the off-nadir angle to 38.
4) POLARIZATION GAIN RATIO
The 532-nm perpendicular channel does not have
E is the measured laser energy, GA is the known sufficient SNR to calibrate using a high-altitude mo-
amplifier gain, b(z) is the molecular backscatter coeffi- lecular signal. Instead, this channel’s calibration coeffi-
cient, and T 2(z) is the two-way atmospheric transmis- cient is derived from that of the parallel channel by
sion from the lidar to the target. System characteristics multiplying the 532-nm parallel channel coefficient by
that affect the value of C include the receiver optical the polarization gain ratio (PGR) (Powell et al. 2009).
transmission, the detector gain and quantum efficiency, The PGR is the ratio of the 532-nm perpendicular
the transmitter–receiver wavelength matching, and the channel signal to the 532-nm parallel channel signal
transmitter–receiver boresight alignment. It is not af- when both channels are illuminated by the same light
fected by changes in the laser energy because the ratio levels. It includes the effects of relative detector gains,
S(z)/E should remain constant, independent of E. detector quantum efficiencies, amplifier gains, and op-
Calibration of the 532-nm parallel channel at night is tical efficiencies downstream of the polarization beam
done over an altitude range of 30 to 34 km above MSL, splitter. The PGR is measured during polarization cal-
where all the quantities on the right-hand side of Eq. (1) ibrations, in which a pseudodepolarizer is inserted into
can be determined. Because the atmosphere at this al- the 532-nm channel beam upstream of the polarization
titude is almost purely molecular, the parameters T 2(z) beam splitter, causing the light to be equally divided
and b(z) can be derived from air molecule and ozone between the two 532-nm channels. Under these condi-
number concentrations provided by NASA’s Global tions the PGR is simply the ratio of the signals mea-
Modeling and Assimilation Office (GMAO). With a sured on the two channels. Figure 9 shows the results of
reasonable amount of averaging, the SNR is large all the polarization calibrations that have been per-
enough to get a good measurement of S(z), and all the formed to date. After an initial upward drift, the PGR
other terms on the right are known. has been stable at a value about 6% higher than the
During level 1 processing, multiple calculations of the initial value during the past year. The small amount of
532-nm parallel calibration coefficient are done along initial drift could be due to changes in the detector gains
the night side of each orbit, with each averaged over or any of the other factors listed earlier in this para-
1485 km. The values are relatively constant over the graph. Because it was slow and systematic, it is easily
dark portion of a given orbit, but they do vary somewhat handled in the data processing and is not a cause for
over longer time periods, indicating changes in the lidar concern.
detection efficiency over time. Figure 8 shows that the
5) OTHER CALIBRATION COEFFICIENTS
calibration coefficient had dropped to about 87% of its
initial value in late 2007, but it recovered to more than Only the 532-nm parallel channel at night has suffi-
95% of its initial value after a boresight alignment and cient SNR to be calibrated using high-altitude molecular
retuning of the etalon. backscatter. The calibration coefficients for the 532-nm
JULY 2009 HUNT ET AL. 1223

channels in the daytime and the 1064-nm channel day

and night have to be derived from those of the 532-nm
parallel channel at night (K. A. Powell et al. 2008, un-
published manuscript). Because those calibrations in-
volve several complicating factors, some of which will
be discussed later under Performance Issues, it is not
useful to plot performance trends for them at this time.
6) CLEAR-AIR DEPOLARIZATION RATIO
MEASUREMENTS

The ratio of the calibrated 532-nm perpendicular

channel signal to the calibrated 532-nm parallel channel

Downloaded from [Link] by guest on 20 June 2020

signal is the volume linear depolarization ratio (or depo- FIG. 10. Time history of the measured clear-air depolarization
larization ratio, for short). CALIOP is designed to mea- ratio. The difference between the measured value and the true
sure the depolarization ratio of clouds and aerosols, but is value gives an indication of the total cross talk between the two
not expected to be sensitive enough to make accurate polarization channels. The abrupt jumps in value are believed to
measurements of the depolarization ratio of clear air, es- be due to changes in the etalon polarization resulting from unre-
lieved stress after temperature changes.
timated to be approximately 0.0037 at the bandwidth of the
CALIOP optical filter (Cairo et al. 1999). Nevertheless, it
is informative to look at the values obtained from clear- scatter signal measurements at each position. Based
air measurements because the difference between the upon the relative signals at the four points, a new opti-
measured value and the actual value puts an upper limit mum pointing direction is computed and implemented.
on the amount of crosstalk between the two polarization Boresight alignments are only done at night, since the
channels. Early in the mission the clear-air depolarization SNR is too low to do them in the daylight, and normal
ratio was measured to be 0.006. Later it made an abrupt measurements are disrupted during these operations.
jump to 0.009, and it has made several jumps between Figure 11, which gives he pointing directions after each
approximately those two values since then. The design of the align operations that have been carried out to
objective was to have less than 1% crosstalk between the date, shows that the nighttime alignment has exhibited a
two 532-nm channels (i.e., when there is no depolariza- very slow long-term drift. The most recent optimum
tion by the target, the measured depolarization ratio pointing direction is about 40 mrad away from the initial
should be less than 0.01). Even with the higher of the direction. There is some evidence that the gradual drift
measured values, this design objective has been met. is the result of distortion of the canister of the primary
Figure 10 shows the clear-air depolarization trend up laser as its pressure drifted downward (Fig. 4).
to late May 2008. This plot was generated from the ratio While the alignment is quite stable at night over a
of region averages from H&S data, with approximate short term, it appears that there are significant shifts
corrections for baseline shape and PGR. All the abrupt during the daylight portion of each orbit, as will be
changes in value before 2008 took place upon resuming discussed later in section 3b.
operation after the instrument had been in Safe mode
8) ETALON TUNING
for at least a few hours. The changes in early 2008 are
associated with an etalon scan and subsequent etalon The passband of the 532-nm channel etalon was chosen
retuning. The reason for the abrupt jumps is not known, to be about the same as the line width of the laser (Fig. 12)
but there is reason to believe that they might be asso- to reduce the solar background as much as possible
ciated with unrelieved stress on the etalon after thermal without excessive attenuation of the laser energy. With
changes. This explanation seems to be supported by the the filter passband so close to the laser line width, the two
changes related to the etalon scan and retuning. center wavelengths must be well matched to optimize the
laser throughput. The center wavelength of the etalon is
7) BORESIGHT POINTING DIRECTION
controlled by adjusting its temperature (Zaun 2004). Early
A boresight align operation is carried out relatively in the mission, an etalon temperature scan was carried
infrequently to optimize the boresight alignment be- out in which the etalon temperature was sequentially set
tween the transmitter and the receiver. This operation to a number of different values and the backscatter sig-
consists of moving the laser pointing direction to four nal was measured at each setting. The temperature
points offset from the current position by fixed amounts giving the largest backscatter signal was determined,
in different directions and making high altitude back- and the etalon temperature was set to that value.
1224 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

Downloaded from [Link] by guest on 20 June 2020

FIG. 11. Time history of the pointing directions along two per-
pendicular axes after nighttime boresight alignments. The gradual
drift in the 1y direction is believed to be due to changes in the
laser pointing direction caused by distortion of the laser canister as
its pressure decreased. FIG. 12. An illustration of the close match between the laser line
width and the etalon passband. The laser output and the etalon
transmission are plotted as a function of the wavelength offset
Because both the etalon and the laser are tempera- from the peak.
ture controlled, they can be kept fairly closely matched
for long periods of time without the need for active
used is limited. Such an analysis has shown a surprising
feedback or frequent adjustments. Because of the time
amount of variability in the coefficients when the sat-
involved, scans are only carried out when there are in-
ellite is in sunlight.
dications that they are needed. The original set point
was maintained until February 2008, at which time an- (i) Sunlight effect on 532-nm channel calibration
other scan was done. This scan showed that the center
The 532-nm parallel channel calibration coefficients
wavelengths had become mismatched by about 7 pm,
measured during a night orbit segment with the satellite
causing a signal loss of about 6% (Fig. 13). After this
in darkness show little variability, but as soon as the
scan the etalon temperature was reset to the new opti-
satellite goes into sunlight, the coefficients begin to
mum value, restoring the lost signal.
change noticeably. This effect can be seen in the directly
b) Some performance issues measured calibration coefficients near the end of each
Although the overall performance has been excellent,
allowing for the production of high-quality science pro-
ducts, there are some performance characteristics that
require special care in data processing to obtain the
most accurate results.
1) THERMAL EFFECTS ON CALIBRATION
COEFFICIENTS

As was discussed earlier, all other calibration coeffi-

cients are derived from the 532-nm parallel channel
nighttime calibration coefficients. The process of trans-
ferring those coefficients to the 532-nm channel in the
daytime and to the 1064-nm channel both night and day
has proven to be more difficult than anticipated. Careful
offline analysis of the data, including looking at tropo-
spheric molecular signals after cloud clearing and much
averaging (K. A. Powell et al. 2008, unpublished man- FIG. 13. Results of the etalon temperature scan performed in
uscript), allows the calibration coefficients to be evalu- February 2008 showing the backscatter signal as a function of the
etalon temperature set point. The new optimum temperature
ated much more accurately than can be done during setting corresponds to a center wavelength shift of about 6.9 pm
production level 1 processing, when the processing flow from the previous setting. Shifting to the new setting resulted in a
is more restricted and the quantity of data that can be 6% signal increase.
JULY 2009 HUNT ET AL. 1225

night orbit. Although direct calibration measurements

are not possible during the day portion of the orbit,
offline analysis shows that the variation becomes even
more pronounced as the lidar progresses further into
the sunlight. The initial data release (version 1) did not
take this variability into account when computing the
daytime calibration coefficients. Figure 14 is an exam-
ple of the scattering ratios over the day portion of an
orbit, computed using the version 1 calibration coeffi-
cients (K. A. Powell et al. 2008, unpublished manu-
script). The value should be close to unity at all times,
but instead it shows a systematic dip to about 0.70 ap-

Downloaded from [Link] by guest on 20 June 2020

proximately 2200 s after going into daylight. This gen-
eral behavior is consistent from month to month, but
the size of the dip varies with season, with midsum- FIG. 14. Illustration of the effect of thermal misalignment on the
mer showing the smallest dip. This behavior points to 532-nm channel daytime scattering ratio. The 532-nm channel
scattering ratio, computed using version 1 daytime calibration
a thermally induced misalignment as being the cause of
coefficients and including only clear-air profiles from 8 to 12 km, is
the variation. BATC has done some end-to-end mod- plotted as a function of the time after the start of the daytime
eling that tends to support this theory (Lieber et al. portion of the orbit. The scattering ratio should be near unity at all
2007). times. Deviations of up to 30% from unity are believed to be due
Offline analysis of the first 18 months of data has to daytime thermal misalignments that are not taken into account
in the version 1 calibration. The amount of the deviation is a
made it possible to derive more accurate daytime cali-
function of the time of year, with the largest deviations occurring in
bration coefficients for the past data and to predict more January.
accurate values for future data. These improved values
were incorporated into the version 2 processing soft-
ware, and all of the data have now been reprocessed ibration coefficients, but it also shows a latitudinal
with that software version. Even further refinement is dependence both night and day. Figure 15 illustrates
expected in future data releases. both of these effects. When the satellite is in the earth’s
shadow (blue crosses), the 1064/532 channel ratio rises
(ii) Sunlight effects on 1064-nm channel calibration
in a nearly linear manner as the satellite goes from
Transfer of the 532-nm channel calibration coeffi- north to south. This behavior is typical of the nighttime
cients to the 1064-nm channel has proven to be even behavior at all times of year, although the slope is a
more complicated. The calibration approach that is weak function of the time of year. Once the satellite
currently being used is to obtain the 1064-nm channel hits sunlight (red circles), there is some sign of a mo-
calibration coefficients by multiplying the 532-nm channel mentary increase in the ratio, and then it drops as the
calibration coefficients by the relative sensitivities of satellite reaches its southernmost point and starts back
the 1064- and 532-nm channels, which we will call the north. It reaches its lowest value at around 208 north,
channel ratio. The channel ratio is obtained by mea- where it is almost 30% lower than for the same latitude
suring the ratio of the backscatter signals at the two in the dark, before it starts to rise again as the satellite
wavelengths from strong high-altitude cirrus clouds, approaches darkness. The amount of difference be-
after correction for extinction above the clouds. By tween night and day is a function of the time of year.
restricting the measurements to strongly scattering The difference is largest in midwinter and is quite small
clouds, the contribution of the molecular backscatter is in midsummer. The difference between the ratio in
negligible, so the measured backscatter is almost en- sunlight and shadow seems to indicate a thermally in-
tirely due to particulate backscatter. Because of the duced misalignment similar to what has been previ-
large particles, the ratio of the backscatter at the two ously discussed for the 532-nm channel in the daytime.
wavelengths was initially assumed to be unity and In this case, not only must there be misalignment, but
constant throughout an orbit. However, further study the misalignment must affect the two wavelengths
of the data has shown that the channel ratio is not differently, causing a change in the channel ratio. The
constant over an orbit. Not only does it show differ- latitudinal dependence of the channel ratio at night is
ences between day (satellite in sunlight) and night harder to explain. Several possibilities, including a
(satellite in shadow), with characteristics somewhat small change in the nighttime alignment, are being
similar to those discussed for the 532-nm channel cal- considered, but no conclusions have been reached.
1226 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

Downloaded from [Link] by guest on 20 June 2020

FIG. 15. Scatterplot of the normalized ratio of the 1064-nm channel raw signal to the 532-nm channel
raw signal (channel ratio), plotted as a function of latitude and color-coded according to satellite lighting
conditions. Arrows show the direction of satellite motion for each lighting condition. There is a large
difference between the shadow and sunlit values in January, but very little difference in July. Both months
show a distinct latitudinal dependence. The data values for these plots are taken from H&S data using
only those samples that closely approximate the conditions used for 1064-nm channel calibration. The
currently implemented 1064-nm calibration algorithm is based on the assumption that the channel ratio is
constant, independent of latitude. These plots clearly do not support that assumption.

Although some changes were made in the 1064-nm calibration measurements are bypassed and historical
channel calibration in the version 2 software, inclusion calibration coefficient values are used if the noise level
of the effects described above will have to wait until a becomes too high.
still later release.
3) SLOW TRANSIENT RECOVERY
2) RADIATION-INDUCED NOISE
Measurements of weak signals immediately after a
The CALIOP 532-nm channel detectors occasion- strong signal can give misleading results on the 532-nm
ally produce radiation-induced current spikes that are channels because the PMTs exhibit a decaying noise tail
as much as two orders of magnitude larger than the after being exposed to a large signal. The 1064-nm
pulses produced by single photoelectrons. Such pulses channel APD does not exhibit this effect. This is illus-
are fairly rare outside the SAA but are much more trated in Fig. 17, which shows the backscatter from the
frequent when passing through the SAA. Individual surface of Antarctica for both wavelengths. The browse
spikes can adversely affect signal averages in low-signal images for both channels show a strong (probably sat-
regions, and multiple pulses can increase the dark noise urated) return from the ice surface, but the 532-nm
level, with a corresponding decrease in the SNR. For channel image on the left shows a rainbow effect from
example, the 532-nm channel RMS dark noise in the the decaying false signal from below the surface, whereas
middle of the SAA is more than 30 times higher than the 1064-nm channel image on the right does not show
that outside the SAA, leading to a decrease in the this effect. The noise tail is present on the 532-nm
nighttime high-altitude SNR by more than a factor of 5. channels after any signal, but it is most easily seen and
The effect on the SNR from clouds and aerosols at characterized after a surface return, where the back-
lower altitudes is much less because the backscatter scatter signal goes quickly from a small value to a very
signal is so much larger. Figure 16 shows the geographic large value and then quickly back to zero. More quan-
distribution of PMT dark noise, clearly showing regions titatively, Fig. 18 shows the relative intensity of the noise
where there is increased noise due to radiation. tail from the 532-nm parallel channel detector as a
Some special processing steps are taken to minimize function of time after a surface return. Although the data
the effects of the radiation-induced noise spikes on the shown are for a nearly saturated signal, this plot is a good
532-nm parallel channel calibration (Powell et al. 2009). representation of the relative response after an impulse-
Isolated noise spikes are simply removed from the data like signal of any amplitude, so long as the peak signal is
whenever they can be identified. More extreme mea- not saturated. It even applies to a moderately saturated
sures are needed in the vicinity of the SAA, where signal if the noise tail is taken as relative to the true peak
JULY 2009 HUNT ET AL. 1227

Downloaded from [Link] by guest on 20 June 2020

FIG. 16. Map showing increased 532-nm parallel channel dark noise in regions with high cosmic ra-
diation densities. All orbit tracks for a 16-day period from late September and early October 2006 are
plotted, with each point color coded by the 532-nm parallel channel RMS dark noise. The South Atlantic
Anomaly stands out distinctly in bright colors just to the left of center. Also visible are curved regions
with increased dark noise near both poles, approximately along the auroral ovals. The straight light blue
bands at around 6658 latitude are not dark noise; rather, they are the result of light scattered into the
telescope when the sun is just above the horizon.

rather than the measured peak. The impact of the slow at 532 nm. It utilizes onboard processing to increase its
transient recovery on science products is still being in- autonomy, decrease its downlink data rate, increase its
vestigated, as are potential correction algorithms. dynamic range, and extract science data for inclusion in
the Health and Status data. Its on-orbit performance has
continued to exceed nearly all requirements after more
4. Summary
than 2 yr of operation, producing high-quality science
CALIOP is a three-channel elastic backscatter lidar, data products. The energy of the primary laser has de-
making measurements at 1064 nm and two polarizations creased less than 10%, although a slow pressure leak

FIG. 17. Browse images of the 532- and 1064-nm backscatter signals from the ice-covered surface of
Antarctica illustrating the slow transient recovery of the 532-nm channel PMT. The 532-nm channel
signal shows a rainbow of colors below the surface as the PMT output goes slowly to zero after the surface
return. The 1064-nm channel signal goes immediately to zero after the surface return. The location of the
surface, as given by a digital surface elevation model, is shown by a red line in both images.
1228 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 26

Cairo, F., G. Di Donfrancesco, A. Adriani, L. Pulvirenti, and

F. Fierli, 1999: Comparison of various linear depolariza-
tion parameters measured by lidar. Appl. Opt., 38,
4425–4432.
Hovis, F., 2006: Qualification of the laser transmitter for the
CALIPSO aerosol lidar mission. Solid State Lasers XV:
Technology and Devices, H. J. Hoffman and R. K. Shori, Eds.,
International Society for Optical Engineering (SPIE Pro-
ceedings, Vol. 6100), 516–525.
Hu, Y., 2007: Depolarization ratio–effective lidar ratio rela-
tion: Theoretical basis for space lidar cloud phase dis-
crimination. Geophys. Res. Lett., 34, L11812, doi:10.1029/
2007GL029584.
Lieber, M., C. Weimer, M. Stephens, and R. Demara, 2007: De-
velopment of a validated end-to-end model for space-based

Downloaded from [Link] by guest on 20 June 2020

lidar systems. Lidar Remote Sensing for Environmental
FIG. 18. Quantitative illustration of the slow transient recovery Monitoring VIII, U. N. Singh, Ed., International Society for
of the 532-nm channel PMT. The peak at t 5 0 is the normalized Optical Engineering (SPIE Proceedings, Vol. 6681), 66810F,
signal from a nearly saturated surface return, broadened by the doi:10.1117/12.731277.
analog filter response and onboard averaging. The optical back- Platt, C. M. R., 1978: Lidar backscatter from horizontal ice crystal
scatter signal goes to zero immediately after the surface return, but plates. J. Appl. Meteor., 17, 482–488.
the detector output decays slowly over many microseconds, after Powell, K. A., and Coauthors, 2009: CALIPSO lidar calibration
an initial drop to about 0.02 times the peak signal. The amplitude algorithms. Part I: Nighttime 532-nm parallel channel and
of the noise tail relative to the peak signal is largely independent of 532-nm perpendicular channel. J. Atmos. Oceanic Technol.,
the peak amplitude for an impulse-like signal. in press.
Seckar, C., L. Guy, A. DiFronzo, and C. Weimer, 2005: Per-
formance testing of an active boresight mechanism for use
will necessitate a switch to the backup laser in early in the CALIPSO Spaceborne Lidar Mission. Optomechanics
2009. The signal-to-noise ratios for all channels remain 2005, A. E. Hatheway, Ed., International Society for Optical
comfortably above their requirements for most cases Engineering (SPIE Proceedings, Vol. 5877), 319–330.
and have been decreasing at a very modest rate. The ex- Spinhirne, J. D., S. P. Palm, W. D. Hart, D. L. Hlavka, and
E. J. Welton, 2005: Cloud and aerosol measurements from
ception is in the South Atlantic Anomaly, where the high
GLAS: Overview and initial results. Geophys. Res. Lett., 32,
radiation level generates excessive noise from the 532-nm L22S03, doi:10.1029/2005GL023507.
channel detectors, causing the SNR to drop below re- Stephens, G. L., and Coauthors, 2002: The CloudSat mission and
quirements and necessitating special processing treat- the A-Train: A new dimension of space-based observations of
ment to deal with the noise pulses. Depolarization mea- clouds and precipitation. Bull. Amer. Meteor. Soc., 83,
1771–1790.
surement capabilities have exceeded expectations, with
Sun, X., D. Reusser, H. Dautet, and J. Abshire, 1997: Measure-
very low cross talk between channels. Boresight stability, ment of proton radiation damage to Si Avalanche photodi-
as well as the resulting calibration constant stability, has odes. IEEE Trans. Electron Devices, 44, 2160–2166.
been quite good at night but less so in the daytime, so Winker, D. M., R. H. Couch, and M. P. McCormick, 1996: An
accurate determination of daytime calibration coeffi- overview of LITE: NASA’s lidar in-space technology exper-
iment. Proc. IEEE, 84, 164–180.
cients continues to be a work in progress, especially for
——, J. Pelon, and M. P. McCormick, 2003: The CALIPSO
the 1064-nm channel. Slow transient recovery of the mission: Spaceborne lidar for observation of aerosols and
532-nm channel causes some error in measurements clouds. Lidar Remote Sensing for Industry and Environment
immediately after a strong backscatter signal. Its impact Monitoring III, U. N. Singh et al., Eds., International So-
on science data products has not yet been fully evalu- ciety for Optical Engineering (SPIE Proceedings, Vol.
4893), 1–11.
ated, but it is thought to be relatively small.
——, W. H. Hunt, and M. J. McGill, 2007: Initial performance
assessment of CALIOP. Geophys. Res. Lett., 34, L19803,
REFERENCES doi:10.1029/2007GL030135.
——, M. Vaughan, A. Omar, Y. Hu, K. Powell, Z. Liu, W. Hunt,
Afzal, R. S., 2006: Performance of the GLAS laser transmitter. and S. A. Young, 2009: Overview of the CALIPSO mission
Solid State Lasers XV: Technology and Devices, H. J. Hoffman and CALIOP data processing algorithms. J. Atmos. Oceanic
and R. K. Shori, Eds., International Society for Optical En- Technol., in press.
gineering (SPIE Proceedings, Vol. 6100), 539–547. Zaun, N. H., C. Weimer, Y. Sidorin, and D. Lunt, 2004: Solid-
——, A. W. Yu, J. L. Dallas, A. Melak, A. T. L. Lukemire, etalon for the CALIPSO lidar receiver. Earth Observing
L. Ramos-Izqueirdo, and W. Mamakos, 2007: The Geoscience Systems IX, W. L. Barnes and J. J. Butler, Eds., International
Laser Altimeter System (GLAS) laser transmitter. IEEE J. Society for Optical Engineering (SPIE Proceedings, Vol.
Sel. Top. Quantum Electron., 13, 511–536. 5542), 141–145.