1

The microphysical properties of wave PSCs retrieved from lidar
measurements during SOLVE–THESEO 2000
Rong-Ming Hu, Kenneth S. Carslaw
School of the Environment, University of Leeds, Leeds, U.K.
Chris Hostetler, Lamont R. Poole
NASA Langley Research Center, Hampton, VA, USA.
Beiping Luo, Thomas Peter, Stefan Füglistaler
Laboratory for Atmospheric Physics, ETH, Zurich, Switzerland.
Thomas J. McGee, John F. Burris
Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, MD, USA.

2

Short title: OPTICAL AND MICROPHYSICAL PROPERTIES OF WAVE PSCS

3

Abstract.

Mountain wave-induced ice clouds in the stratosphere are known to cause the

formation of nitric acid-hydrate particles downwind. Understanding the microphysical properties of
these solid particles is important because they may contribute to a general background of solid PSCs
that is observed, but whose origin is not understood. Based on the limited set of observations of
PSCs directly attributable to mountain waves, it has not been possible to determine their general
microphysical properties. Here we analyze lidar observations from the SOLVE/THESEO-2000
campaign. Between December 1999 and March 2000 seven of the twelve flights of the DC-8 aircraft
showed clear signs of mountain wave-induced clouds, with nitric acid hydrate clouds often extending
many hundreds of kilometers downwind of mountains. Based on T-matrix calculations, we have
developed a technique to estimate the microphysical properties of spherical and non-spherical particles
from multi-wavelength backscatter and depolarization data from the GSFC/LaRC Aerosol lidar. The
technique allows particle radius, condensed mass, and number densities of ice, nitric acid trihydrate and
liquid PSCs to be estimated. Ice clouds were found to contain approximately 1 to 3 ppmv of condensed
water with a number density from 1 to 10 cm−3 , and a narrow size distribution width with mode radii
from 1 to 1.5 µm. Nitric acid hydrate particles downwind of the ice clouds were consistent with 1 to 5
ppbv condensed HNO3 , a number density from 0.5 to 2 cm−3 , and mode radius around 0.5 µm. These
hydrate clouds are characterised by high aerosol backscatter and depolarization and are distinct from
type 1a clouds that are observed away from mountains, which have low aerosol backscatter.

4

1. Introduction
Between December 1999 and March 2000 the NASA DC-8 made 12 flights in the Arctic vortex as
part of the SAGE III Ozone Loss and Validation Experiment and the Third European Stratospheric
Experiment on Ozone (SOLVE/THESEO-2000). Polar stratospheric clouds (PSCs) were observed by
lidar on numerous occasions. Several flights were designed to probe wave-cloud PSCs that had been
forecast to occur over the East Greenland and Scandinavian mountains. Several flights were also made
by the European Falcon aircraft employing the OLEX lidar system. These flights provide a rich dataset
with which to study the formation regions of different PSCs. The wave cloud flights were made possible
by operational forecasts of stratospheric mountain waves over the whole Arctic region, as described by
Eckermann et al. [2001]. In this paper, we use data from the Goddard Space Flight Center/Langley
Research Center (GSFC/LaRC) lidar system, providing observations of aerosol backscatter ratio at
three wavelengths (354, 532 and 1064 nm) and depolarization at one wavelength (532 nm).
Polar stratospheric wave clouds form in regions of strong adiabatic temperature perturbation. The
most common cause of such localized cooling in the Arctic stratosphere is gravity waves produced by
surface topography [Peter et al., 1997; Carslaw et al., 1998; Wirth et al., 1999; Eckermann and Preusse,
1999; Riviere et al., 2000; Dörnbrack et al., 2001a,b], although adiabatic cooling may also occur in the
vicinity of jet streams [Wu and Waters, 1996].
Observations of Arctic wave clouds using airborne lidar have shown that the strong adiabatic
cooling and warming over mountains can lead to the formation of a wide range of different PSCs in
close proximity [Carslaw et al., 1998; Wirth et al., 1999]. These studies also suggest that the wave
clouds may have an effect on PSCs distant from the mountains. For example, PSCs containing solid
particles (most likely nitric acid hydrates) have been observed to extend several hundred kilometers
downwind of the Scandinavian mountains [Carslaw et al., 1998; Riviere et al., 2000] . Carslaw et al.
[1999] used a simple parameterization of mountain wave cooling to show that the net effect of many
such events in the Arctic would be to increase the fraction of the polar stratosphere containing nitric
acid hydrate PSCs.

5

The formation mechanisms of nitric acid hydrate clouds are still not clear. Recent observations
[Biele et al., 2001] suggest that as much as 50% of PSCs contain solid hydrate particles, as revealed
by extensive lidar observations from the Ny Ålesund ground-based station, which is representative of
the deep polar vortex. Analysis of the 1989 DC-8 lidar observations in the Arctic [Toon et al., 2000]
suggests that at least 70% of the PSCs contained solid particles.
The contribution of mountain wave events to this general background of hydrate clouds needs to
be considered. To establish what contribution mountain waves make to solid PSCs in the Arctic vortex
it is necessary to understand both the climatological occurrence of the wave clouds [Carslaw et al.,
1999] as well as the microphysical properties of the clouds that form. Previous studies have provided
a highly detailed analysis of the microphysical properties of several clouds over the Scandinavian
mountains [Carslaw et al., 1998; Wirth et al., 1999; Riviere et al., 2000]. These studies showed a clear
transition from liquid PSC on the upwind side of the mountains to nitric acid hydrates downwind,
with ice clouds in between and provided the first evidence for nitric acid hydrate formation in the
clouds. The most important conclusion of these detailed studies, in the context of solid PSCs on the
synoptic scale, was that the clouds downwind of the mountains contain > 0.2 hydrate particles cm−3 ,
and often much higher number densities were found. This number concentration is greater than occurs
in the more widely observed type 1a clouds. (Two optical modelling studies have recently estimated
the number concentration of hydrate particles in type 1a clouds. Tsias et al. [1999] suggest an upper
limit of about 0.01 hydrate particles cm−3 , while Toon et al. [2000] estimate an upper limit of 1% of
the available nuclei, or typically less than 0.1 cm−3 .) Here, we extend previous studies of mountain
wave clouds by examining the microphysical properties of several clouds observed by the DC-8 aircraft
during SOLVE/THESEO-2000 in December 1999 and January 2000.

2. Lidar measurement technique
The GSFC/LaRC aerosol lidar system measures profiles of aerosol and cloud backscatter at 355,
532 and 1064 nm and aerosol/cloud depolarization at 532 nm.

6

The LaRC Aerosol Lidar is a piggy-back instrument on the GSFC Airborne Raman Ozone and
Temperature Lidar (AROTEL) lidar [McGee et al., 1995; and Burris, 1996; Burris et al., 1998]. The
light source for the aerosol measurements is a Continuum 9050 Nd:YAG laser operating at 50 shots per
second. The laser transmits approximately 600 mJ at 1064 nm, 250 mJ at 532 nm, and 350 mJ at
355 nm. AROTEL also employs an excimer laser transmitting at 308 nm and uses the molecular and
Raman backscatter from the 355 and 308 beams to measure ozone and temperature. Backscattered
light at all wavelengths is collected by a 16-inch diameter Newtonian telescope with a selectable field
stop.
The signals measured by the instrument are composed of backscatter from both air molecules and
aerosol/cloud particles. The aerosol/cloud component of the signal is estimated using density profiles
derived from Data Assimilation Office (DAO) assimilation model results.
The archived products are produced over an altitude range extending from approximately 5 km
to 20 km above the aircraft. Data are acquired at nominal resolutions of 15 m vertical and 500 m
horizontal (2 s integration).
In terms of the backscatter coefficient, the total backscatter ratio and the aerosol depolarization
are defined as
S =

δ =

βaerosol
βair

(1)

βaerosol,⊥
βaerosol,k

(2)

where βaerosol is the backscatter coefficient of aerosol particles and its components βaerosol,⊥ and
βaerosol,k are for perpendicular and parallel polarized radiation. βair is the backscatter coefficient of the
air molecules.

3. Retrieval technique
Multi-wavelength lidar measurements provide an approximate measure of particle size through the
wavelength dependence of the backscatter ratio. In addition, the depolarization of a plane polarized

7

lidar beam can be used as a sensitive indicator of the presence of non-spherical (solid) particles.[e.g.,
Poole and McCormick, 1988; Browell et al., 1990; Toon et al., 1990; McCormick et al., 1990; David et
al., 1998; Gobbi et al., 1998; Carslaw et al., 1998; Tsias et al., 1999; Toon et al., 2000].
Most investigations of PSCs using lidar focus on understanding the occurrence of different PSC
‘types’; that is, whether the clouds are composed of non-depolarizing particles (indicative of liquid
aerosols), depolarizing particles with moderate backscatter ratio (indicative of nitric acid hydrate
particles), or depolarizing particles with high backscatter (indicative of ice particles). The aim of these
studies is often to understand what controls the phase of the particles.
Recently, a few studies have attempted to quantify the microphysical properties of clouds based on
lidar observations. We have previously used aircraft lidar observations of mountain wave-induced clouds
together with microphysical and optical models to derive particle sizes along air parcel trajectories
through the waves [Carslaw et al., 1998; Wirth et al., 1999]. In those studies, good agreement between
the multi-wavelength lidar observations and the modeled optical properties of liquid, nitric acid
hydrate, and ice clouds was obtained, allowing particle size distributions to be determined from the
microphysical model. The technique of combined microphysical/optical modeling of wave clouds was
particularly successful because the microphysical model provided a tight physical constraint on the size
distribution of the particles at any point in the cloud. Tsias et al. [1999] used a similar technique to
constrain the microphysical properties of a nitric acid hydrate cloud observed over the North Atlantic.
Again, a microphysical model was used to constrain what microphysical properties a cloud could have,
based on various assumptions.
A free retrieval of cloud microphysical properties is more difficult. A free retrieval means attempting
to derive the microphysical properties of a cloud based on lidar data without the additional constraint
of a microphysical model. Toon et al. [2000] used the calculated scattering properties of spherical and
non-spherical PSC particles to infer likely sizes and number concentrations of the particles in clouds
observed from the DC-8 in 1989 [Browell et al., 1990]. More recently, Luo et al. [2001] used a look-up
table of scattering properties to derive typical microphysical properties of clouds observed during the

8

SOLVE campaign. They were able to identify several quasi-distinct PSC types and estimate the sizes
and number densities of the particles. Li and Mishchenko [2001] have also used the T-matrix technique
to determine possible particle sizes in PSCs in a similar way, and also examined the effect of assuming
cylindrical particles.
Here, we attempt to derive the microphysical properties of the clouds from the lidar observations
alone. To retrieve aerosol properties from multi-wavelength scattered light intensities is a notoriously
under-determined problem [Twomey, 1977a,b]. The ill-posedness and the associated solution ambiguity
have hampered the inversion problem. Many kinds of inversion methods have been developed to try
to overcome these difficulties, such as the constrained linear inversion method [King, 1982], statistical
methods [Rodgers, 1976; 1990], relaxation methods [Chahine, 1972] or the neural networks approach
[Ishimaru et al., 1990]. Moreover, the positive smoothing constraint regularization combined with the
least square fitting procedure has been developed for improving the constrained linear inversion method
[Liu et al., 1999; Amato et al., 1996].
Aside from these inversion methods, the look-up table (LUT) approach, which we use here, is
an attractive alternative [Tanre et al., 1996, Higurashi and Nakajima, 1999; Luo et al., 2001; Li and
Mishchenko, 2001]. In the LUT method, a precalculated dataset of many values of aerosol scattering
properties (such as backscatter ratio, depolarization, etc.) are compared with measurements until the
best fit is obtained. The scattering properties in the LUT must be calculated for all desired parameters
of the aerosol population that are to be retrieved.

3.1. Method
The T-Matrix light scattering algorithm [Mishchenko, 1993] has been used for calculating scattering
by non-spherical stratospheric particles, with some success [Carslaw et al., 1998; Wirth et al., 1999;
Luo et al., 2001; Li and Mishchenko, 2001]. In this technique, the particles are assumed to have shapes
that allow an exact calculation of scattering efficiency. In the context of analyzing lidar measurements,
spheroidal, cylindrical, and Chebyshev particles have been assumed [e.g., Mishchenko and Sassen, 1998;

9

Carslaw et al., 1998; Li and Mishchenko, 2001].
Here, we assume the PSC particles to be spheroidal. Oblate-like particles are described by an
aspect ratio ε > 1 and prolate particles by ε < 1. The good agreement between the observed and
modeled PSC optical properties in our previous studies gives us confidence that the representation
of non-spherical particles as spheroidal scatterers (and the use of T-matrix codes to calculate the
backscattering ratio) is realistic. Here, we use look-up tables for the particle backscatter ratio at 354,
532, and 1064 nm and the aerosol depolarization at 532 nm for radii between 0.01 and 5 µm (∆r = 0.05
µm) and spheroidal aspect ratios between 0.5 and 1.5 (∆ = 0.05) [Trautmann et al., 2001; Luo et al.,
2001].
The aerosol and cloud particle populations are assumed to be composed of lognormal modes,
described by


dN (r)
N
ln2 (r/rm )
=√
exp −
,
dr
2 ln2 σ
2πr ln σ

(3)

for which the parameters to be retrieved from the lidar data are the particle number concentration
(N ), the (spherical-equivalent) mode radius (rm ), and the width of the lognormal mode (σ), with the
refractive index defined for different types of PSC. The aspect ratio (shape) of the spheroidal particle
(ε) must either be defined or taken as a further unknown. Here, we assume it to be a fourth unknown
to be determined from the retrieval. The mass concentration m (particle mass per mass of air) depends
on N , rm , and σ according to
m=

2
9
4
3
πρrm
N e 2 ln σ ,
3

(4)

where ρ is the mass density of the particle.
There are therefore 4 unknown parameters of the aerosol distribution to be derived for each particle
type (N , rm , σ, and ) and 4 scattering properties on which to base the retrieval (we use here 354, 532
and 1064 nm total backscatter ratio and 532 nm depolarization). The PSC type (nitric acid hydrate,
liquid, or ice) is also, strictly speaking, an unknown. However, solid nitric acid hydrate clouds and
liquid clouds are readily distinguished simply in terms of the aerosol depolarization (essentially zero for
liquids), while ice and hydrate particles are distinguishable due to the normally very high backscatter

10

resulting from the much greater condensed mass of ice.
PSCs almost always contain several particle types within the same cloud element (pixel of a lidar
image). For example, liquid HNO3 /H2 SO4 /H2 O droplets may be present along with nitric acid hydrate
particles, and both will contribute to the aerosol backscatter. Clearly, it is not possible to retrieve
the full complexity of such mixed aerosol populations with just 4 scattering parameters. However, as
we show in section 3.2 the retrieved microphysical properties are reasonably robust for the particular
types of cloud that are found in mountain waves. Liquid (non depolarizing) clouds are likely to be
well retrieved since it can be assumed that they are composed of a single particle type, although it is
necessary to assume that the droplet size distribution is mono-modal. Ice clouds are also likely to be
well retrieved since the scattering efficiency of the ice particles will dominate that of any interstitial
hydrate or liquid particles. The problem arises with mixed hydrate/liquid clouds in which the total
scattering efficiency of the droplets and crystals may be comparable. We discuss the problem of
externally mixed particle populations in section 3.2.
The following steps were used to convert lidar images of scattering properties into equivalent
images of microphysical properties.
Step 1: Firstly, we spatially averaged the lidar data to a resolution of 300 m vertical × 3 minutes
horizontal (approximately 40 km) in order to produce a manageable number of pixels for the retrieval.
For the 532 nm and 1064 nm channels, approximately 12 pixels of the original data were averaged to
produce a single pixel, amounting to about 3000 averaged pixels in each of the images shown here. For
the 354 nm channel, which has a lower resolution, the averaging is over 6 original pixels.
The mean value and variance of backscatter ratio and depolarization for n pixels are

Sλ =

σλS

=

s

Pn

i=1

Pn

Sλ

n

− Sλ )2
,
n−1

i=1 (Sλ

(5)

(6)

11

δ =

σδ =

s

Pn

i=1

δ

(7)

n

Pn

− δ)2
n−1

i=1 (δ

(8)

where Sλ and σλS are the mean value and variance of backscatter ratio at a wavelength λ; δ and σ δ are
the mean value and variance of the aerosol depolarization.
Step 2:

The second step is to distinguish the three different PSC types (liquid, nitric acid

hydrate, and ice) according to their optical properties. This identification is necessary since different
refractive indices are used for the different particle types in the retrieval. Table 1 shows the criteria
that were used to distinguish these PSC types as well as the refractive indices used in the retrieval.
Liquid and ice clouds were retrieved as single size modes. Nitric acid hydrate clouds were assumed
to be composed of nitric acid trihydrate (NAT) and to be externally mixed with liquid aerosols having
10 particles cm−3 and a mode radius of 0.1 µm. As we show in section 3.2 the retrieved properties
of the NAT particles in externally mixed liquid/NAT clouds are relatively insensitive to the assumed
properties of the liquid aerosol.
Step 3:

The lookup tables of S354 , S532 , S1064 and δ532 were searched using all possible

combinations of particle number concentration (N ), condensed mass (m), width of lognormal
distribution (σ) and spheroidal aspect ratio (). Combinations of these four particle properties were
identified such that the calculated optical properties were within the limits

Sλ − σλS < Sλmodel < Sλ + σλS

(9)

δ − σ δ < δ model < δ + σ δ

(10)

where Sλmodel and δ model are the values in the look-up table. Physically reasonable upper and lower
limits were put on N , m, σ and  as follows:

12

0.01 <

N

/cm−3 < 20

0.01 <

m

/ppbv HNO3 < 15

0.1 <

m

/ppmv H2 O < 10

1.05 <

σ

< 1.8

0.5 <



< 1.5

(11)

The particle mode radius was then determined from N , m, and σ. The range of aspect ratios used in
the retrievals is defined by the range that can be calculated with the T-matrix model. Smaller and
larger values cause convergence problems in the numerical solution of the equations. There are no
observations that allow  for stratospheric particles to be constrained.
Note that we do not attempt to accurately retrieve the properties of the clouds with very low
backscatter and moderate depolarization (type 1a clouds), which we expect to have hydrate particle
number densities less than 0.01 cm−3 . These clouds have been studied in detail by Luo et al. [2001].
Here, we focus on the highly backscattering solid and liquid wave clouds.
The uncertainty in the aerosol depolarization becomes large for values of depolarization less than
about 0.1. In such cases, the perpendicular backscatter ratio was used in the retrieval instead.
The look-up table searching procedure results in one or more sets of N , rm , σ and  that satisfy
equations 9 and 10 for each pixel. In the case of multiple solutions, we selected the most likely solution
by finding the minimum of the quantity

2

χ

=

model
Sλ − Sλ,j

X

σλS

λ

+

δ − δjmodel
σδ

!2
!2

(12)

for the j possible solutions satisfying 9 and 10. A similar method has also been used by Martonchik et
al. [1998] in the retrieval of aerosol properties from satellite. In many cases, we found that the solutions

13

were clustered around a single combination of N , rm , σ and , suggesting that a robust determination
of the particle properties had been obtained. In other cases, the solution clustered around two different
sets of N , rm , σ and . In such cases, while it is possible to find the solution that is the minimum of
equation 12, the retrieved particle properties are likely to be genuinely ambiguous. In the following
retrieved lidar images we have not attempted to extract such highly probable alternate solutions, but
plot only the solution that produced the minimum χ2 . In such cases, the existence of alternate solutions
is revealed by spatially patchy retrieved microphysical properties in what might be expected to be a
rather uniform cloud.

3.2. Test of the method
Several factors contribute to the uncertainty in the retrieved microphysical properties, such as the
assumption that the particles are spheroids and the possible presence of more than one size mode. It
is therefore important to test the retrieval method. Unfortunately there were no colocated ER-2 and
DC-8 flights inside the clouds that we are examining here, so a comparison of in situ measurements and
our retrieved size distributions is not possible. However, if we assume that the particles are spheroidal,
we can test the reliability of the retrieval on a synthetic size distribution, which may be composed of a
single particle type or be an external mixture of two types.
We have constructed a bimodal size distribution comprised of liquid aerosols with a fixed mode
radius of either 0.05 or 0.15 µm and various larger modes composed either of ice or NAT. The lidar
quantities S532 , S1064 , and δ532nm were then computed and these values used to retrieve the particle
distribution using the procedure outlined above. The uncertainty of the input scattering properties was
assumed to be ±10%, which is typical of the standard deviation of grouped pixels used in the actual
retrievals (see Step 1 above). For solid particles, the spheroidal aspect ratio was assumed to be 0.9 for
the calculation of the lidar properties but was a free parameter in the retrieval.
Figure 1a shows the retrieval result for pure liquid aerosols with N = 10 cm−3 and rm = 0.05
or 0.15 µm. The retrieved aerosol microphysical properties are in close agreement with the synthetic

14

input size distribution. For the synthetic size distribution with rm = 0.05 µm our retrieval produces
N = 9.2 cm−3 and rm = 0.06 µm. With rm = 0.15 µm the retrieval produces N = 11.9 cm−3 and rm =
0.13 µm. The width of the synthetic lognormal size distribution used here was 1.4. The first retrieval
suggests σ = 1.38 and the second σ = 1.42.
Figure 1b shows the results for a mixed liquid and NAT particle population assuming a NAT
mode with three different mode radii, but in each case with the same total condensed HNO3 of 1 ppbv.
In this calculation, the liquid aerosol distribution was assumed to have the properties N = 6 cm−3
and rm = 0.11 µm. The unknown parameters in the retrieval are the NAT number density, mode
radius, width of the mode, and the spheroidal aspect ratio. Three wavelengths (354, 532, and 1064
nm) and one depolarization (532 nm) were used. The retrieved NAT particle size distributions are in
rather close agreement with the input synthetic distributions. The input/output mode radii of the
NAT particles are as follows: 0.5/0.45 µm, 1.5/1.46 µm, and 3.0/2.73 µm, and the input/output NAT
number densities are 0.43/0.63 cm−3 , 0.016/0.017 cm−3 , and 0.002/0.003 cm−3 . The input asphericity
was 0.9, and the retrieved value was 0.85. The retrieved condensed masses of NAT are within 15% of
the input values. Much of the uncertainty in the retrieved particle properties arises from the unknown
particle asphericity.
The results for ice particles are better than for NAT particles because the light scattering by the
liquid aerosols becomes insignificant, and therefore the retrieval becomes less ambiguous. We assumed
1 ppmv condensed water, a number density 10 cm−3 , and an ice particle mode radius of 1.5 µm for the
synthetic size distribution. The retrieved results were m = 0.98 ppmv, N = 9.7 cm−3 and rm = 1.45
µm.
We have also examined the sensitivity of the retrieved NAT particle properties to varying amounts
of liquid aerosol. As stated previously, the liquid aerosol is likely to be present in most air masses in
which NAT particles exist, but with an unknown and probably highly variable mass. Figure 1c shows
the effect of varying the amount of liquid that is assumed to be present when doing the NAT particle
retrieval. Two liquid distributions were used, one with a mode radius of 0.05 µm and one with 0.15

15

µm. The effect on the retrieved properties of the NAT mode is small. This result suggests that realistic
uncertainties in the amount of liquid aerosol that accompanies the NAT particles is unlikely to have a
significant effect on the properties of the NAT particles that are retrieved.

4. Retrieved Microphysical Properties
In this section we show retrieved microphysical properties of four mountain wave-induced clouds:
January 14, 2000 (a small wave cloud over E. Greenland during the transfer flight); January 25, 2000
(ice, hydrate, and liquid clouds over the Scandinavian mountains); January 23, 2000 (extensive ice
and hydrate clouds over E. Greenland and N. Atlantic); December 10, 1999 (a small cloud without
ice). Plate 1 shows the flight paths and the latitude and longitude versus time so that the geographical
location of the various clouds can be determined from the flight time.

4.1. January 14, 2000
On January 14, 2000, the DC-8 flew a transit flight from Dryden Airbase in the United States to
Kiruna, Sweden, across the Atlantic Ocean and Greenland. A distinct wave-structured PSC is apparent
over the east coast of Greenland, as shown in Plate 2. A strong backscattering feature between 22 and
24 km from 7.38 to 7.41 UT, with a backscatter ratio S > 100 at 1064 nm and strong depolarization
> 0.4 at 532 nm, is indicative of water ice particles [Toon et al., 1990; Carslaw et al., 1998]. The
existence of the ice cloud indicates strong localized cooling at this altitude. A non-depolarizing cloud is
apparent upwind of the ice, which is most likely composed of HNO3 /H2 SO4 /H2 O droplets. Downwind
of the ice is a tail of depolarizing particles with much lower backscatter (typically S < 20 at 1064 nm),
which is indicative of a nitric acid hydrate cloud, as seen in close colocation with ice clouds in several
other similar observations [Carslaw et al., 1998; Wirth et al., 1999].
Using the lidar data, we have attempted to retrieve the microphysical properties of this cloud using
the backscatter ratio of three wavelengths and depolarization at one wavelength. The microphysical
results are shown in Plate 3 and the different particle types referred to can be identified in Plate 3e.

16

The liquid aerosol cloud (7-7.3 UT) upwind of the ice cloud has a peak mass density equivalent to
3 ppbv of HNO3 condensation into the H2 SO4 aerosols, with a number density of approximately 10
cm−3 , a small mode radius of 0.15 µm and width, σ, between 1.4 and 1.8. These retrieved properties
are entirely consistent with expectations for a HNO3 /H2 SO4 /H2 O aerosol population.
The condensed mass of water ice in the middle of the cloud varies from 1.2 to 1.37 ppmv. The size
distribution of ice particles is represented by a lognormal distribution with a total of 4-6 particles cm−3 ,
a narrow mode width, σ, of 1.3, and a mode radius of 1.5-2 µm. The shape of particles is represented
by slightly non-spherical value of aspect ratio ε. The number density of ice particles is a significant
fraction of the available liquid aerosols upwind, implying efficient freezing nucleation in the wave cloud,
as found in earlier studies [Carslaw et al., 1998; Wirth et al., 1999].
The peak mass density of NAT particles is equivalent to 3 ppbv HNO3 condensation as NAT and
is rather uniform through the NAT cloud. The retrieved size and number density of the particles, in
contrast, are more variable. Given that all elements of the NAT cloud are likely to have undergone very
similar thermal histories, it is unlikely that this variability is real. Rather, it is likely to be a result of
ambiguous retrieval. Some patches of the NAT cloud contain approximately 0.4 particles cm−3 , while
others contain approximately 5 particles cm−3 . These variations are reflected also in the retrieved mode
radius, which varies between about 0.2 and 0.7 µm. In all cases, a narrow width of the NAT mode,
σ ∼ 1.1, is indicated (not shown). Although the retrieved properties of the NAT cloud are uncertain,
the results do indicate that the number density is rather high compared with typical type 1a clouds.

4.2. January 25, 2000
On January 25, 2000, the DC-8 flew from Norway to Spitzbergen, then returned near the coast of
Norway, and finally flew two short legs across the Scandinavia mountains. One aim of this flight was to
investigate PSC processes through mountain waves that had been forecast (see Eckermann et al., 2001
for a summary of the mesoscale forecasts for this flight).
Plate 4 shows the lidar observations during the two consecutive flight legs over the Scandinavian

17

mountains (note the reversal of flight direction, which can be seen in Plate 1). The two optically thick
ice clouds were present directly over the mountains, with the hydrate clouds occurring downwind (east
of the mountains). There is a very clear association between the ice clouds and the hydrate clouds,
which strongly suggests that the hydrate particles were formed heterogeneously on ice.
Our retrieval results (Plate 5) are consistent with an ice cloud containing about 2 ppmv water, with
a number density of 5-10 cm−3 and a mode radius of approximately 1.6 µm. The hydrate PSCs are
consistent with an average 3 ppbv condensed HNO3 . Again, the optical properties of the hydrate cloud
are consistent with a NAT particle number densities of either 0.3 cm−3 or about 2 cm−3 . The lognormal
width, σ, is about 1.2. Such a narrow width is consistent with the microphysical model simulations of
NAT particles evolving downwind of a mountain wave ice cloud [Tsias et al., 1999]. The liquid cloud is
composed of small spherical particles with 1-3 ppbv HNO3 condensed as HNO3 /H2 SO4 /H2 O droplets
with a number density from 10-15 cm−3 and a mode radius < 0.3 µm. The other derived microphysical
properties are summarized in Table 2.
As in previous studies [Carslaw et al., 1998; Wirth et al., 1999] we find that the number density
of hydrate particles is less than that of the ice particles. Thus, not all ice particles are effective as
nucleation sites for nitric acid hydrates, the reason for which was explored by Wirth et al. [1999].

4.3. January 23, 2000
On January 23, the DC-8 first flew over the Scandinavian mountains to eastern Greenland, and
finally back to Kiruna, Sweden. One aim of this flight was to probe the wave clouds over the Greenland
mountains that had been forecast by the MWFM mesoscale forecast model [Eckermann et al., 2001].
Plate 6 shows the lidar observations for a short period of the flight during which DC-8 flew
eastwards from the east Greenland coast over the Atlantic Ocean. Both hydrate and liquid PSCs are
apparent in this image. We do not show the optically very dense ice cloud which occurred just to the
west and immediately upwind of this cloud. Again, the three principal PSC types ice, NAT and liquid
were present in close proximity. There are large amplitude wave-like structures apparent in the aerosol

18

and cloud layers.
Our retrieval results (Plate 7) are consistent with a PSC containing NAT particles with 1-5 ppbv
HNO3 , a number density from 0.5-5 particles cm−3 , a very narrow width of log-normal size distribution,
and mode radius 0.5-1 µm. Again, the range of retrieved properties is due to solution ambiguity rather
than real variations within the cloud and, as before, the retrieved quantities cluster about two main
values. The liquid cloud is composed of small droplets with 1-3 ppbv HNO3 in the HNO3 /H2 SO4 /H2 O
solutions, with the number density from 10-15 particles cm−3 , and a small mode radius less than 0.3
µm.

4.4. December 10, 1999
Each of the wave clouds described so far was associated with ice cloud formation and identifiable
mountain wave-driven localized cooling. A second type of solid nitric acid wave cloud was also apparent
during a flight over Spitzbergen on December 5, 1999 and over Franz Josef Land [82 ◦ N, 60 ◦ E] on
December 10, 1999. On December 10, 1999, the DC-8 flew from Kiruna, Sweden towards eastern
Greenland, passed Spitzbergen, then returned to Kiruna. A PSC was observed between 13.5 and 14.5
UT, poleward of about 82 ◦ N, between 18 to 22 km altitude (Plate 8).
The cloud shown in Plate 8 has a backscatter ratio at 1064 nm approximately a factor 10 lower
than the clouds on January 23 and 25, and a depolarization at 532 nm approximately a factor of 5
lower. The temperature on the 50 hPa surface in this area was greater than 190.5 K. There is no
evidence for ice formation as in the previous clouds, although neither can an ice-mediated formation
mechanism for the solid nitric acid particles be excluded from the data. The cloud has a clear wave-like
structure, which may indicate a localized gravity wave-driven cooling and formation of nitric acid
hydrate particles.
Our retrieval results (Plate 9) indicate that this PSC is consistent with a condensed HNO3 amount
of 0.5 ppbv, but with a rather poorly defined number density and size of the NAT particles. As was
the case for the January 14 PSC, the NAT number density appears to be either around 0.03 cm−3 or

19

around 0.4 cm−3 , with a mode radius of either 1-1.5 µm or 0.4 µm, respectively.

5. Summary and discussion
Lidar observations during SOLVE have been used to examine the microphysical properties of
PSCs formed by adiabatic cooling in mountain-induced gravity waves. Patterns of cloud occurrence are
similar to those found in previous aircraft observations of mountain waves; namely, liquid PSCs upwind
of the mountains, ice clouds directly over the mountains, and predominantly nitric acid hydrate clouds
downwind of the mountains. The wave clouds observed during the January deployment of the DC-8
were less clear in their structure than earlier observations [Carslaw et al., 1998; Wirth et al., 1999],
making it difficult to construct air parcel trajectories based on visible wave features. Our approach has
therefore been to attempt a retrieval of cloud microphysical properties using the optical information
alone.
Our technique of retrieving PSC microphysical properties from multi-wavelength lidar data (see
also Luo et al. [2001]) has been tested and shown to be effective for the types of clouds that form
in mountain waves. The wave clouds contain relatively high number densities of particles with sizes
comparable to the lidar wavelength. However, as shown by Luo et al. [2001] the retrieval technique can
only be used to define particle size limits in cases where the particles are very much larger than the
lidar wavelength. Even for the wave clouds, we have shown that the retrieved microphysical properties
can be quite variable throughout a cloud that appears to have rather uniform aerosol backscatter. Such
solution ambiguity is apparent as the speckled regions of the clouds in Plates 3, 5, 7 and 9. We plan to
undertake a similar study using additional lidar wavelengths to see if a the retrieval can be improved.
Comparisons of our retrieved particle sizes with in situ particle size distribution measurements from
balloons or aircraft would also be useful.
A clear picture of the types of clouds produced by mountain wave cooling is now emerging. The
optical properties of the different wave-induced clouds observed during SOLVE were similar to previous
observations of clouds that were attributed to mountain waves [Carslaw et al., 1998; Wirth et al.,

20

1998]. Ice clouds were observed in the coldest parts of the gravity waves and contain high number
densities (typically 5-10 cm−3 ) of ice particles (see Table 2), which is consistent with rapid homogeneous
nucleation of ice from liquid aerosols at low temperatures [Koop et al., 2000]. The nitric acid hydrate
clouds downwind of the ice, which we assume to be composed of NAT particles, have a comparatively
high backscatter ratio (S = 2 at 532 nm) and high aerosol depolarization (δ = 0.1 − 0.2 at 532 nm),
which contrasts markedly with the very low backscatter ratio of the more ubiquitous type 1a hydrate
clouds [Toon et al., 1990; Biele et al., 2001] that are not obviously associated with mountain-wave
effects. The optical properties of the wave-induced hydrate clouds are consistent with a high number
density (0.3-5 cm−3 ) of small (radii < 1 µm) hydrate particles (see Table 2), compared with less than
0.01 cm−3 and radii > 3 µm that are likely to be present in type 1a clouds [Tsias et al., 1999]. Earlier
studies [Carslaw et al., 1998; Wirth et al., 1999] have shown that the high number densities of hydrate
particles is consistent with heterogeneous nucleation on ice in the waves, although we have not been
able to confirm this mechanism here.
It has been suggested that the action of many mountain waves in the Arctic polar vortex could
be a significant source of NAT particles [Carslaw et al., 1999]. An important question is whether
these particles could contribute to a general background of large particles [Fahey et al., 2001] that are
responsible for denitrification. Particles with a radius of 0.5-1.0 µm (the mode radius of our retrieved
distribution) have a fall velocity of approximately 0.1 mm s−1 , which is too low to be of importance:
such particles fall less than 10 m in a day. We see no evidence for particles much larger than 0.5-1.0
µm in the wave clouds we have analyzed. The only way that mountain wave-generated hydrate clouds
could produce much larger particles is if the air below the cloud were supersaturated with respect to
the hydrate. If this were the case, particles could sediment and then grow at a much increased rate
since the gas phase HNO3 would not reduced by the presence of a high number density of hydrate
particles. We explore this possibility in Füglistaler et al. [2002].

Acknowledgments.
We thank M. Kurylo at NASA for enabling our participation in the SOLVE Science Team. This work

21

was funded by grant GST/02/2441 from the U.K. Natural Environment Research Council as part of the Upper
Troposphere–Lower Stratosphere Thematic Programme.

22

References
Amato, U., D. DiBello, F. Esposito, C. Serio, G. Pavese, and F. Romano, Intercomparing the Twomey method
with a multimodel lognormal approach to retrieve the aerosol size distribution. J. Geophys. Res., 101,
19267-19275, 1996.
Biele, J., A. Tsias, K.S. Carslaw, T. Peter, R. Neuber, Coexistence of solid and liquid particles in Arctic
stratospheric clouds studied by lidar in winters 94/95 to 96/97, J. Geophys. Res., in press, 2001.
Browell, E.V., C.F. Butler, S. Iamail, P.A. Robinette, A.F. Carter, N.S. Higdon, O.B. Toon, M.R. Schoeberl,
A.F. Tuck, Airbone lidar observations in the wintertime Arctic stratosphere: Polar stratospheric clouds,
Geophys. Res. Lett., 17, 385-388, 1990.
Burris, J., W. Heaps, B. Gary, W. Hoegy, L. Lait, T. McGee, and M. Gross, Lidar temperature measurements
during the TOTE/VOTE Mission, J. Geophys. Res., 103, 3505-3510, 1998.
Carslaw, K.S., M. Wirth, A. Tsias, B.P. Luo, A. Dornbrack, M. Leutbecher, H. Volkert, W. Renger, J.T.
Bacmeister, T. Peter, Particle microphysics and chemistry in remotely observed mountain polar
stratospheric clouds, J.Geophys.Res., 103, 5785-5796, 1998.
Chahine, M. T., A general relaxation method for inverse solution of full radiative transfer equation, J. Atmos.
Sci., 29, 741-747, 1972.
David, C., C., S. Bekki, S. Godin, and G. Megie, Polar stratospheric clouds climatology over Dumont d’Urville
between 1989 and 1993 and the influence of volcanic aerosols on their formation. J. Geophys. Res., 103,
22163-22180, 1998.
Dörnbrack, A., M. Leutbecher, J. Reichardt, A. Behrendt, K.P. Muller, and G. Baumgarten, Relevance of
mountain wave cooling for the formation of polar stratospheric clouds over Scandinavia: Mesoscale
dynamics and observations for January 1997, J. Geophys. Res, 106 1569-1581, 2001a.
Dörnbrack, A., and M. Leutbecher, Relevance of mountain waves for the formation of polar stratospheric clouds
over Scandinavia: A 20 year climatology, J. Geophys. Res, 106 1583-1593, 2001b.
Eckermann, S.D., and P. Preusse, Global measurements of stratospheric mountain waves from space, Science,
286, 1534-1537, 1999.
Eckermann, S.D., A. Dörnbrack, S. Vosper, and K.S. Carslaw, Mesoscale forecasting of the stratosphere during
SOLVE-THESEO 2000, submitted to J. Geophys. Res., 2001.

23

Fahey, D.W., et al., The detection of large HNO3 -containing particles in the winter Arctic stratosphere Science,
291, 1026-1031, 2001
Fu, Q., and K.N. Liou, Parameterization of the radiative properties of cirrus clouds, J.Atmos.Sci., 50, 2008-2025,
1993.
Gobbi, G.P., G. Di Donfrancesco, and A. Adriani, Physical properties of stratospheric clouds observed during
the Antarctic late winter of 1992, J. Geophys. Res., 103, 10859, 1998.
Heaps, W., and J. Burris, Airborne Raman Lidar, Appl. Opt., 35, 7128, 1996.
Higurashi, A., and T. Nakajima, Development of a two-channel aerosol retrieval algorithm on a global scale
using NOAA AVHRR, J. Atmos. Sci., 924-941, 1999.
Ishimaru, A., R.J. MarksII, L. Tsang, C.M. Lam, and D.C. Park, Particle-size distribution determination using
optical sensing and neural networks, Opt. Lett., 15, 1221-1223., 1990.
King, M. D., Sensitivity of constrained linear inversions to the selection of the Lagrange multiplier, J. Atmos.
Sci., 10341-10356, 1994.
Koop, T., B.P. Luo, A. Tsias, and T. Peter, Water activity as the determinant for homogeneous ice nucleation
in aqueous solutions, Nature, 406, 611-614, 2000.
Liu, L., and M. I. Mishchenko, Constraints on PSC particle microphysics derived from lidar observations, J.
Quant. Spectrosc. Radiat. Transfer, 70, 817-831, 2001.
Liu, Y., W.P. Arnott, and J. Hallett, Particle size distribution retrieval from multispectral optical depth:
Influences of particle nonsphericity and refractive index, J. Geophys. Res., 104, 31,753-31762, 1999.
Luo, B.P., T. Peter, S.A. Fueglistaler, H. Wernli, R.M. Hu, K.S. Carslaw, C.A. Hostetler, L. A. Poole, T. J.
McGee and J.F. Burris, Large stratospheric particles observed by lidar during the SOLVE/THESEO
2000 mission, J. Geophys. Res, submitted, 2001.
Martonichik, J. V., D.J. Diner, R. Kahn, T.P. Ackerman, M. M. Verstraete, B. Pinty, H. R. Gordon, Technique
for retrieval of aerosol properties over land and ocean using multi-angle Image. Submitted to IEEE
transaction on Geoscience and Remote sensing, 48pp, 1997.
McCormick, M.P., G.S. Kent, W.H. Hunt, M.T. Osborne, L.R. Poole, and M.C. Pitts, Arctic polar stratospheric
cloud observations by lidar. Geophys. Res. Lett., 17, 381, 1990.
McGee, T., M. Gross, U. Singh, J. Butler, and P. Kimvilankani, An improved stratospheric ozone lidar, Opt.

24

Eng., 34, 1421, 1995.
Mishchenko, M.I., Light scattering by size-shape distributions of randomly oriented axially symmetric particles
of a size comparable to wavelength, Appl. Opt., 32, 4652-4666, 1993.
Mishchenko, M.I., K. Sassen, Depolarization of lidar returns by small ice crystals: An application to contrails,
Geophys. Res. Lett., 25, 309-312, 1998.
Peter, T., Microphysics and heterogeneous chemistry of polar stratospheric clouds. Ann. Rev. Phys. Chem.,
48, 779-816, 1997.
Poole, L.R., M.P. McCormick, W.H. Hunt, M.T. Osborn, S. Schaffner, and M. C. Pitts, Dual-polarization lidar
observations of polar stratospheric cloud evolution. Geophys. Res. Lett., 17, 389, 1990.
Riviere, E.D., et al., Role of lee waves in the formation of solid polar stratospheric clouds: Case studies from
February 1997, J. Geophys. Res., 105, 6845-6853, 2000.
Rodgers, C.D., Retrieval of atmospheric temperature and composition from remote measurements of thermal
radiation, Rev. Geophys., 14, 609-624, 1976.
Tanre, D., D., M. Herman, Y.J. Kaufman, Information on aerosol size distribution contained in solar reflected
spectral radiances, J. Geophys. Res., 101, 19043-19060, 1996.
Toon, O. B., E.V. Browell, S. Kine, and J. Jordan, An analysis of lidar observations of polar stratospheric
clouds, Geophys, Res. Lett., 17, 393-396, 1990.
Toon, O.B., A. Tabazadeh, E.V. Browell, and J. Jordan, Analysis of lidar observations of Arctic polar
stratosphericclouds during January 1989, J. Geophys. Res., 105, 20589-20615, 2000.
Trautmann, T., B.P. Luo, T. Peter, Application of T-matrix light scattering theory to polar stratospheric
clouds: a database for interpreting lidar backscattering observations, Applied Optics, submitted, 2001.
Tsias, A., M. Wirth, K.S. Carslaw, J. Biele, H. Mehrtens, J. Reichardt, C. Wedekind, V. Weib, W.Renger, R.
Neuber, U. von Zahn, B. Stein, V. Santacesaria, L. Stefanutti, F. Fierli, J. Bacmeister, and T. Peter,
Aircraft lidar observation of an enhanced type Ia polar stratospheric clouds during APE-POLECAT. J.
Geophys. Res., 94, 23,961-23969, 1999.
Twomey, S.A., The influence of pollution on the short wave albedo of clouds, J. Atmos. Sci., 34,1149-1152,
1977a.
Twomey, S.A., Introduction to the mathematics of inversion in remote sensing and indirect measurements, in

25

Developments in Geomathematics, vol.3, 243pp., Elsevier Sci., New York, 1977b.
Wirth, M., A. Tsias, A. Dornbrack, V. Weiss, K.S. Carslaw, M.Leutbecher, W. Renger, H. Volkert, T. Peter,
Model-guided Lagrangian observation and simulation of mountain polar stratospheric clouds, J. Geophys.
Res, 104, 23971-23981, 1999.
Wu, D.L., and J.W. Waters, Gravity-wave-scale temperature fluctuations seen by the UARS Microwave Limb
Sounder, Geophys. Res. Lett., 23, 3289-3292, 1996a.

J. F. Burris, T. J. Mcgee, Laboratory for Atmospheres, Code 916, NASA Goddard Space Flight
Center, Greenbelt, MD20771, USA (e-mail: mcgee@aeolus.gsfc.nasa.gov)
K. S. Carslaw, R. M. Hu, School of the Environment, University of Leeds, Leeds LS2 9JT, U.K.
(e-mail: carslaw@env.leeds.ac.uk; hu@env.leeds.ac.uk).
S. Fuglistaler, B. P. Luo, T. Peter, Laboratory for Atmospheric Physics (LAPETH),
ETH, Zurich, Switzerland. (email: stefanf@atmos.umnw.ethz.ch; luo@atmos.umnw.ethz.ch;
tom@atmos.umnw.ethz.ch,)
Chris Hostetler, Lamont R. Poole, NASA Langley Research Center, Mail Stop 417, Hampton, VA
23681-2199, (e-mail: c.a.hostetler@larc.nasa.gov; l.r.poole@larc.nasa.gov)
Received

26

Table 1. Classification of PSC particle types, based on 1064 nm lidar backscatter ratio and 532
nm aerosol depolarization
Particle type

Backscatter ratio (S)

Aerosol depolarization (δ)

Refractive index

Background aerosol

S < 0.1

δ< 0.03

1.42

S > 0.1

δ< 0.03

1.42

0.1 < S < 50

δ> 0.03

1.48

S > 50

δ > 0.03

1.32

Liquid PSC
NAT
ice

27

Table 2. Summary of retrieved microphysical properties of ice and nitric acid trihydrate wave clouds
ice

NAT

N

m

rm

N

m

rm

(cm−3 )

(ppm)

(µm)

(cm−3 )

(ppb)

(µm)

0.03-0.4

0.4-1.0

0.4-1.5

December 10, 1999
January 14, 2000

4.0-6.0

1.2 − 1.4

1.5-2.0

0.4-5.0

2.5-4.0

0.2-0.7

January 23, 2000

5.0-10.0

1.0 − 2.0

1.7-2.0

0.5-5.0

1.0-5.0

0.5-1.0

January 25, 2000

5.0-10.0

2.0 − 5.0

1.6-2.0

0.3-2.0

1.0-3.5

0.4-1.0

Note that the stated ranges do not imply positive correlation between the different retrieved quantities.

dN(r)/d(logr) ( /cm3)

28

102
100

(a)

10-2
10-4
10-6
10-8
0.01

0.10

1.00

10.00

dN(r)/d(logr) ( /cm3)

dN(r)/d(logr) ( /cm3)

radius (micron)

100

(b)

10-2
10-4
10-6
10-8
0.1

1.0
radius (micron)

10.0

100

(c)

10-2
10-4
10-6
10-8
0.01

0.10

1.00

10.00

radius (micron)

Figure 1. Retrieval of particle size distributions for synthetic particle distributions. (a) Liquid aerosols
only, with mode radii of 0.05 and 0.15 µm; (b) mixed liquid/NAT particles, assuming a liquid mode
radius of 0.11 µm and NAT mode radii of 0.5, 1.5 and 3.0 µm, each with a total mass of 1 ppbv
condensed HNO3 ; (c) the effect of varying the liquid aerosol mass assumed to be present when retrieving
the properties of the solid particles. Solid line – the input synthetic size distribution; dashed line – the
retrieved size distribution.

29

-90
90
-80
80
-70
70
84
-60
60
-50
50
-40
40
-30
30
-20-10 76
0 10 20

68

(a)

60

latitude

80
70
60
50

(b)

40
0

5

10

15

longitude

time (UT)
50
0
-50

(c)

-100
0

5

10

15

time (UT)

Plate 1. Flight paths of the DC-8 aircraft analysed in this paper. January 25, 2000 (black line); January
23, 2000 (red line); January 14, 2000 (green line); December 10, 1999 (blue line).

30

25

5.00

25

10.00

2.69

5.01

22

0.77
0.42
0.22
0.12
0.06
0.03
0.02
0.01

21

7.10

7.20

7.30

7.40

7.50

24
Aerosol BSR 532

23

2.51

geometric altitude (km)

Aerosol BSR 355

geometric altitude (km)

1.44

24

23

22

7.20

7.30

7.40

7.50

7.60

25

1.00
0.63

22

6.31
2.51
1.00
0.40
0.16
0.06
0.03
0.01

21

24

Depolarization 532

23

0.40

geometric altitude (km)

Aerosol BSR 1064

geometric altitude (km)

15.85

24

7.50

0.08

time (UT hours)
39.81

7.40

0.16

0.01

7.10

100.00

7.30

0.32

0.02

21

7.60

25

7.20

0.63

0.04

time (UT hours)

7.10

1.26

23

22

0.25
0.16
0.10
0.06
0.04
0.03
0.02
0.01

21

7.60

7.10

time (UT hours)

7.20

7.30

7.40

7.50

7.60

time (UT hours)

geometric altitude (km)

25

24

23

22

21

7.10

7.20

7.30

7.40

7.50

7.60

time (UT hours)

Plate 2. Lidar observations of polar stratospheric clouds during a flight on January 14, 2000. (a)
backscatter ratio S at 354 nm laser wavelength; (b) backscatter ratio at 532 nm laser wavelength; (c)
backscatter ratio at 1064 nm laser wavelength; (d) aerosol depolarization δ at 532 nm; (e) particle type
assumed in the retrieval: background aerosol (blue), liquid PSC (green), NAT (red), ice (black).

31

23.9
23.4
23.0
22.5
22.0
21.5

56.23
21.54
8.25
3.16
1.21
0.46
0.18
0.07
0.03
0.01

21.0

24.9

23.9
23.4
23.0
22.5
22.0
21.5

7.1

7.2

7.3

7.4

7.5

20.1
7.0

7.6

2.99
1.59
0.84
0.45
0.24
0.13
0.07
0.04

7.1

7.2

7.3

7.4

7.5

7.6

time (UT hours)

1.33

23.9

1.25
Aspect Ratio

23.4
23.0
22.5
22.0

1.17
1.08
1.00
0.92
0.83
0.75
0.67

21.5

0.58
0.50

21.0
20.6

24.9

3.00
2.75

24.4

2.50

23.9

Mode Radius (um)

1.50
1.42

24.4

geometric altitude (km)

24.9

geometric altitude (km)

5.63

20.6

time (UT hours)

20.1
7.0

20.00
10.62

0.02
0.01

21.0

20.6
20.1
7.0

24.4

Aerosol Number Density (#/cm3)

146.78

geometric altitude (km)

1000.00
383.12

24.4

Condensed mass (ppb)

geometric altitude (km)

24.9

23.4
23.0
22.5
22.0

2.25
2.00
1.75
1.50
1.26
1.01
0.76
0.51

21.5

0.26
0.01

21.0
20.6

7.1

7.2

7.3

7.4

time (UT hours)

7.5

7.6

20.1
7.0

7.1

7.2

7.3

7.4

7.5

7.6

time (UT hours)

Plate 3. Retrieved aerosol properties from lidar measurements on January 14, 2000. (a) Condensed
mass; (b) number density; (c) spheroidal aspect ratio; (d) mode radius. The speckled regions in (b),
(c) and (d) indicate solution ambiguity. Non-speckled regions of the cloud indicate regions where the
retrieved properties are most reliable.

32

26

5.00

26

10.00

24

22

0.77
0.42
0.22
0.12
0.06
0.03
0.02

20

0.01

15.5

16.0

16.5

2.51
24

Aerosol BSR 532

geometric altitude (km)

5.01

1.44
Aerosol BSR 355

geometric altitude (km)

2.69

22

17.0

0.08

0.01

16.0

16.5

17.0

26

1.00
0.63

Aerosol BSR 1064

15.85
24

22

6.31
2.51
1.00
0.40
0.16
0.06
0.03

20

0.01

0.40
Depolarization 532

geometric altitude (km)

39.81

geometric altitude (km)

0.16

time (UT hours)
100.00

16.5

0.32

0.02

15.5

26

16.0

0.63

0.04
20

time (UT hours)

15.5

1.26

24

22

0.25
0.16
0.10
0.06
0.04
0.03
0.02

20

17.0

0.01

15.5

time (UT hours)

16.0

16.5

17.0

time (UT hours)

geometric altitude (km)

26

24

22

20

15.5

16.0

16.5

17.0

time (UT hours)

Plate 4. Lidar measurements of polar stratospheric clouds during a flight on January 25, 2000. (a)
backscatter ratio S at 355 nm laser wavelength; (b) backscatter ratio at 532 nm laser wavelength; (c)
backscatter ratio at 1064 nm laser wavelength; (d) aerosol depolarization δ at 532 nm; (e) particle type
assumed in the retrieval: background aerosol (blue), liquid PSC (green), NAT (red), ice (black).

33

24.2
23.4
22.7
21.9
21.1
20.4

56.23
21.54
8.25
3.16
1.21
0.46
0.18
0.07
0.03
0.01

19.6

25.7

24.2
23.4
22.7
21.9
21.1
20.4

15.34

15.66

15.99

16.31

16.64

18.1
15.01

16.97

1.84
0.96
0.50
0.26
0.14
0.07
0.04

15.34

15.66

15.99

16.31

16.64

16.97

time (UT hours)

1.33

24.2

1.25
Aspect Ratio

23.4
22.7
21.9
21.1

1.17
1.08
1.00
0.92
0.83
0.75
0.67

20.4

0.58
0.50

19.6
18.9

25.7

5.00
4.58

25.0

4.17

24.2

Mode Radius (um)

1.50
1.42

25.0

geometric altitude (km)

25.7

geometric altitude (km)

3.54

18.9

time (UT hours)

18.1
15.01

6.79

0.02
0.01

19.6

18.9
18.1
15.01

25.00
13.03

25.0

Aerosol Number Density (#/cm3)

146.78

geometric altitude (km)

1000.00
383.12

25.0

Condensed mass (ppb)

geometric altitude (km)

25.7

23.4
22.7
21.9
21.1

3.75
3.34
2.92
2.51
2.09
1.67
1.26
0.84

20.4

0.43
0.01

19.6
18.9

15.34

15.66

15.99

16.31

time (UT hours)

16.64

16.97

18.1
15.01

15.34

15.66

15.99

16.31

16.64

16.97

time (UT hours)

Plate 5. Retrieved aerosol properties from lidar measurements on January 25, 2000. (a) Condensed
mass; (b) number density; (c) aspect ratio; (d) mode radius. The speckled regions in (b), (c) and
(d) indicate solution ambiguity. Non-speckled regions of the cloud indicate regions where the retrieved
properties are most reliable.

34

26

5.00

26

10.00

20

0.42
0.22
0.12
0.06
0.03
0.02
0.01

18

16.2

16.4

16.6

16.8

2.51
24
Aerosol BSR 532

22

0.77

5.01

geometric altitude (km)

1.44
24
Aerosol BSR 355

geometric altitude (km)

2.69

22

20

16.4

16.6

16.8

17.0

26

1.00

20

2.51
1.00
0.40
0.16
0.06
0.03
0.01

18

0.40
24

Depolarization 532

22

6.31

0.63

geometric altitude (km)

15.85
24

Aerosol BSR 1064

geometric altitude (km)

0.08

time (UT hours)
39.81

16.8

0.16

0.01

16.2

100.00

16.6

0.32

0.02
18

17.0

26

16.4

0.63

0.04

time (UT hours)

16.2

1.26

22

20

0.25
0.16
0.10
0.06
0.04
0.03
0.02
0.01

18

17.0

16.2

time (UT hours)

16.4

16.6

16.8

17.0

time (UT hours)

geometric altitude (km)

26

24

22

20

18

16.2

16.4

16.6

16.8

17.0

time (UT hours)

Plate 6. Lidar measurements of polar stratospheric clouds during a flight on January 23, 2000. (a)
backscatter ratio S at 355 nm laser wavelength; (b) backscatter ratio at 532 nm laser wavelength; (c)
backscatter ratio at 1064 nm laser wavelength; (d) aerosol depolarization δ at 532 nm; (e) particle type
assumed in the retrieval: background aerosol (blue), liquid PSC (green), NAT (red), ice (black).

35

23.87
22.90
21.94
20.97
20.01
19.04

1.78
1.00
0.56
0.32
0.18
0.10
0.06
0.03
0.02
0.01

18.08

25.80

23.87
22.90
21.94
20.97
20.01
19.04

2.99
1.59
0.84
0.45
0.24
0.13
0.07
0.04

17.11
16.17

16.33

16.49

16.65

16.82

16.15
16.01

16.98

16.17

16.33

16.49

16.65

16.82

16.98

time (UT hours)

1.33

23.87

1.25
Aspect Ratio

22.90
21.94
20.97
20.01

1.17
1.08
1.00
0.92
0.83
0.75
0.67

19.04

0.58
0.50

18.08
17.11

25.80

3.00
2.75

24.83

2.50

23.87

Mode Radius (um)

1.50
1.42

24.83

geometric altitude (km)

25.80

geometric altitude (km)

5.63

0.02
0.01

time (UT hours)

16.15
16.01

20.00
10.62

18.08

17.11
16.15
16.01

24.83

Aerosol Number Density (#/cm3)

3.16

geometric altitude (km)

10.00
5.62

24.83
Condensed mass (ppb)

geometric altitude (km)

25.80

22.90
21.94
20.97
20.01

2.25
2.00
1.75
1.50
1.26
1.01
0.76
0.51

19.04

0.26
0.01

18.08
17.11

16.17

16.33

16.49

16.65

16.82

16.98

16.15
16.01

16.17

time (UT hours)

16.33

16.49

16.65

16.82

16.98

time (UT hours)

Plate 7. Retrieved aerosol properties from lidar measurements on January 23, 2000. (a) Condensed
mass; (b) number density; (c) aspect ratio; (d) mode radius.

36

Aerosol BSR 1064

geometric altitude (km)

24
22

20

18

13.5

14.0
14.5
time (UT hours)

15.0

Depolarization 532

geometric altitude (km)

24
22

20

18

13.5

14.0

10.00
5.01
2.51
1.26
0.63
0.32
0.16
0.08
0.04
0.02
0.01

14.5

15.0

14.5

15.0

0.50
0.34
0.23
0.15
0.10
0.07
0.05
0.03
0.02
0.01
0.01

time (UT hours)

geometric altitude (km)

24
22

20

18

13.5

14.0
time (UT hours)

Plate 8. Lidar measurements of polar stratospheric clouds during a flight on December 10, 1999. (a)
backscatter ratio S at 1064 nm laser wavelength; (b) aerosol depolarization δ at 532 nm; (c) particle
type assumed in the retrieval: background aerosol (blue), liquid PSC (green), NAT (red), ice (black).

37

22.2
21.5
20.7
20.0
19.2
18.4

1.78
1.00
0.56
0.32
0.18
0.10
0.06
0.03
0.02
0.01

17.7

23.8

22.2
21.5
20.7
20.0
19.2
18.4

13.65

14.00

14.36

14.72

16.1
12.93

15.08

1.59
0.84
0.45
0.24
0.13
0.07
0.04

13.29

13.65

14.00

14.36

14.72

15.08

time (UT hours)

1.33

22.2

1.25
Aspect Ratio

21.5
20.7
20.0
19.2

1.17
1.08
1.00
0.92
0.83
0.75
0.67

18.4

0.58
0.50

17.7
16.9

23.8

5.00
4.58

23.0

4.17

22.2

Mode Radius (um)

1.50
1.42

23.0

geometric altitude (km)

23.8

geometric altitude (km)

2.99

16.9
13.29

time (UT hours)

16.1
12.93

5.63

0.02
0.01

17.7

16.9
16.1
12.93

20.00
10.62

23.0

Aerosol Number Density (#/cm3)

3.16

geometric altitude (km)

10.00
5.62

23.0
Condensed mass (ppb)

geometric altitude (km)

23.8

21.5
20.7
20.0
19.2

3.75
3.34
2.92
2.51
2.09
1.67
1.26
0.84

18.4

0.43
0.01

17.7
16.9

13.29

13.65

14.00

14.36

14.72

15.08

16.1
12.93

13.29

time (UT hours)

13.65

14.00

14.36

14.72

15.08

time (UT hours)

Plate 9. Retrieved aerosol properties from lidar measurements on December 10, 1999. (a) Condensed
mass; (b) number density; (c) aspect ratio; (d) mode radius.