First Confirmation that Water Ice is the Primary Component of Polar
Mark E. Hervig
Department of Atmospheric Science, University of Wyoming, Laramie
Robert E. Thompson, Martin McHugh, and Larry L. Gordley
G&A Technical Software Inc., Newport News, VA
James M. Russell III
Center for Atmospheric Sciences, Hampton University, VA
Michael E. Summers
School of Computational Sciences and Department of Physics and Astronomy, George Mason
University, Fairfax, VA
Abstract. Polar mesospheric clouds (PMCs) have been measured in the infrared for the first
time by the Halogen Occultation Experiment (HALOE). PMC extinctions retrieved from
measurements at eight wavelengths show remarkable agreement with model spectra based on ice
particle extinction. The infrared spectrum of ice has a unique signature, and the
HALOE–model agreement thus provides the first physical confirmation that water ice is the
primary component of PMCs. PMC particle effective radii were estimated from the HALOE
extinctions based on a first order fit of model extinctions.
Polar mesospheric clouds occur in both hemispheres near the summer solstice at high
latitudes and altitudes near the mesopause (∼82 km) [e.g., Alpers et al., 2000]. From the ground
clouds appear in reflected sunlight against the twilight sky at solar
depression angles between
7° and 16°. These sightings can be brilliant, leading ground–based observers to call them
noctilucent or ”night–shining” clouds (NLCs). Wegener  was the first to suggest that these
clouds could be composed of water ice, and this theory has persisted because PMCs have been
with ice supersaturations and
because their growth appears to be an exponential function
Rusch et al.
Lübken et al
., 1996]. Until now
, this assertion has
by observation because PMCs had
only been measured at ultraviolet to near–infrared
wavelengths, where unique absorption features or “fingerprints” do not exist.
Since the existence of PMCs is apparently related to temperature and water vapor, these
clouds may be an excellent indicator of small changes in the global atmosphere. Thomas et al.
 suggested that the first documented appearance of PMCs in 1885 [Backhouse, 1885] may
have been a consequence of increasing global methane (CH
) following the start of the industrial
oxidizes to H
O in the stratosphere, increased CH
would eventually result in
O in the upper atmosphere, and thus provide enhanced condensible vapor for PMC
Indeed, a recent report by
 shows that the number of PMC sightings over
northern polar regions has increased by nearly 100% during the past 35 years, and satellite
observations [e.g., Evans et al., 1995; Debrestian et al., 1997] revealed an increase in PMC
occurrence over southern high latitudes since the 1980’s [Olivero et al., 1996]. Furthermore, a
example occurred on June 22–23, 1999 when PMCs were observed for the first time as far
south as Colorado and Utah. This unexpected low–latitude display prompted dozens of news
accounts in the media. The reported increases in PMC occurrence may suggest that the upper
is becoming cooler and/or more humid over time. Cooling of the mesopause region has
been suggested because of CO
build–up due to anthropogenic activities [Roble and Dickenson,
While increased CO
warms the lower atmosphere,
it cools the upper atmosphere by direct
space. As a result, mesospheric cloud formation has been proposed as an important
of global change [
, 1996]. It now appears that PMC occurrence
is a sensitive
climate indicator that has been conveying climate change information for over a century.
PMCs have been measured at eight infrared wavelengths (2.4 to 10
m) by the Halogen
during nine southern and nine northern polar summer seasons. A
recent analysis of these measurements has isolated the PMC signals and produced particulate
ficients from all eight
HALOE channels. This work reports the analysis of these data,
and provides comparisons with model spectra based on ice particle extinction.
Table 1. HALOE bandpass center wavelengths, bandwidth, designated absorbing species,
and absorbing species at mesospheric altitudes.
2.45 2.80 3.40 3.46 5.26 6.26 6.61 9.87
Bandwidth (nm) 36 53 76 50 89 43 67 773
Aerosol Aerosol NO,
2. HALOE PMC Observations
HALOE has been recording solar occultation measurements from the Upper Atmosphere
Satellite since its activation on October 1
1, 1991 [
Russell et al
., 1993]. HALOE contains
eight infrared channels and produces transmission profiles for 30 occultations per day. Table 1
summarizes the HALOE bandpass characteristics and lists the designated absorbing species in
to the important absorbers at mesospheric altitudes. HALOE provides four to six weeks of
coverage at PMC latitudes near each summer solstice as shown in Figure 1. The instrument field of
view (FOV) is 2 arcmin in elevation by 6 arcmin in azimuth, corresponding to about 1.5 km in
altitude by 4.5 km horizontally at typical PMC tangent point altitudes. The effective vertical
z) is increased to about 1.8 km due to optical ef
fects and low–pass filter smoothing, and
point path length (along–limb,
305 km) is a function of the vertical resolution.
Figure 1. An example of
HALOE latitude coverage for
– 1996. Plotted points are
daily mean latitudes for
typically 15 sunrise or sunset
occultations per day.
measured signal profiles are sampled at 0.3 km vertical spacing, which is determined by
the telescope scan rate and detector sampling frequency. normalized by the average observed
solar intensity to yield transmission,
), at each wavelength
. The profiles are
registered in altitude by comparing the CO
channel transmissions with simulations based on
temperature/pressure profiles from the National Center for Environmental Protection (NCEP) at
altitudes near 30 km [Hervig et al., 1996]. Solar ephemeris calculations and image pointing
references are used to accurately determine relative pointing and therefore tangent height for all
samples. During normal production data processing, the transmission profiles are smoothed in
altitude to increase the signal–to–noise ratio for the gas mixing ratio retrievals. However, this
degrades thin cloud signals by smearing them over altitude. For
the PMC analysis, the
transmissions were reprocessed without vertical smoothing to retain sensitivity and achieve the
highest possible vertical resolution (∼1.8 km). Transmission is related to optical depth by
( ) =
( )], where optical depth is ( ) = ( )∆x and ( ) is extinction coefficient. For the low
) associated with PMCs, optical depth can be approximated by:
), to much
less than one percent uncertainty.
HALOE profiles at PMC altitudes
indicate that the limb–path optical depth due
to molecular absorption changes gradually over altitude. In contrast, the change in optical depth
the top edge of a PMC is relatively sharp (Figure 2). These
characteristics were used to isolate
PMC signal in each
HALOE channel by removing the average optical depth measured above a
PMC from the average optical depth measured within a PMC. This approach assumes that the
absorption changes little over
, and attributes the optical depth change to PMC
particles. In this approach the average PMC optical depth, ( )
, was estimated by:
= [ ( )
+ ( )
+ ( )
] – [ ( )
+ ( )
+ ( )
where the subscript i refers to altitude (e.g., altitude i–1 is 0.3 km above altitude i). Three–point
smoothing is employed to reduce noise, and changes are then characterized over the vertical
resolution (1.8 km between centroids). Due to the optically thin nature of PMCs, any absorption
is assumed to be nearly equal to the tangent path optical depth. Assuming that PMCs occupy
entire tangent path, the average extinction is therefore
305 km for 1.8 km
thickness). Since PMC layers can be less than 1.8 km thick and possibly non–uniform across
the 305 km path length, this approach may overestimate the actual extinctions. However,
inaccuracies in path length will not affect the relative spectral signatures deduced from HALOE.
This approach provides a first order removal of contaminant gaseous absorption and unknown
zero–level offsets. Because the HALOE channels are temporally and spatially identical, the
volume of air is sampled simultaneously by each channel. Therefore, by finding cloud tops
using equation 1 to estimate
) for many profiles, an average
) can be obtained with
small uncertainty in the relative extinction between channels.
Figure 2. HALOE optical depth
from sunrise on July 30, 1997,
at 70°N, 327°E, without the
smoothing done in production data
A PMC signature at 81–83
is evident as a rather sharp increase
optical depth. CO
absorption in the
channel, and H
in the 6.61 m channel are evident as
increases in optical depth. The
5.27 and 9.87
m channels were
omitted for visual clarity.
3. Model PMC Extinction Spectra
current understanding of PMCs suggests that
these clouds are composed of ice particles.
test this assertion, PMC extinction spectra were modeled based on ice particle characteristics for
comparison with the HALOE observations. The extinction cross sections of individual PMC
particles were calculated using Mie theory with the complex refractive indices of ice. Extinction
were then calculated by integrating the single particle cross sections over an appropriate
types of ice known to exist at atmospheric pressures are ordinary hexagonal ice, cubic
which is formed by vapor condensation at low temperatures and remains stable below about
and amorphous ice which forms by condensation at even lower temperatures and remains
below roughly 110 to 140 K. While cubic and amorphous ice transform into hexagonal ice upon
hexagonal ice will maintain its form upon cooling to very low temperatures. Since PMCs
form at temperatures below about 150 K, they could be composed of hexagonal, cubic, or
ice. The optical properties of hexagonal and cubic ice can be considered identical [e.g.,
Warren, 1984]. The optical properties of amorphous ice are found for selected wavelengths,
the existing data were
ficient for use in this work. This work considered refractive
of hexagonal ice from
Bertie et al.
 (measured at 100 K temperature and
from 1.2 –
 (266 K; 0.05 – 2000
oon et al.
 (163 K; 1.4 – 20
Gosse et al.  (251 K; 1.4 – 7.8 m). Gosse et al. report only the imaginary index, and
using the real indices
arren. The various refractive indices are in generally good
agreement at the HALOE wavelengths, with some differences near 2.5 and 10 m (Figure 3).
The size distributions of PMC particles have been inferred from a variety of remote
measurements at ultraviolet to near infrared wavelengths. Rusch et al.  present PMC size
retrieved from reflectance measurements at three visible and ultraviolet wavelengths
from the Solar Mesosphere Explorer (SME) satellite. Their measurement inversions assumed ice
refractive indices and inferred the size distribution shape but not the total particle concentration.
results suggest that the ef
fective radius (R
) of PMC particles is typically less than about
with values occasionally near
von Cossart et al.
 present 1
1 lognormal PMC size
derived from three–color lidar observations over Norway
. Their inversions assumed
particles, and give R
between 38 and 86
nm. The average lognormal size distribution from their
sample set is described by a median radius (r) of 51 nm, distribution width (s) of 1.42, and total
(N) of 82 cm
= 70 nm).
Alpers et al.
 present unimodal lognormal size
Figure 3. The imaginary refractive indices of ice from Bertie et al. , Warren ,
Toon et al. , and Gosse et al. . The real refractive indices of ice from Bertie et
al., Warren, and Toon et al. The locations of HALOE wavelengths are indicated by vertical
lines on the abscissa.
distributions retrieved from five–color lidar observations of PMCs over Germany. Their size
were characterized by r
= 20 – 28 nm, s = 1.5 – 1.6, and N = 260 – 610
= 30 –
49 nm). In general, these clouds had higher concentrations of smaller particles compared to the
results of von Cossart et al. .
calculations considering the above size distributions and refractive indices were used
to construct PMC extinction spectra for comparison to the HALOE measurements. Similar
were then used to derive relationships that can be used to infer
PMC particle ef
radii from the HALOE measurements. These analyses are presented below.
preliminary treatment of PMC signals (equation
1) was applied to northern high latitude
measurements from July 25 to August 4, 1997. PMC layers were identified using the NO
able 1), because this bandpass is nearly free of gaseous absorption at mesospheric
and because it has one of the lar
particle extinction cross sections. A set of 16
distinct PMC measurements was assembled by identifying the largest optical depth change in a
at mesopause altitudes (between 78 and 90 km). These measurements were then used to
PMC extinctions (equation 1) and to construct an average extinction for each HALOE channel.
PMC extinction spectra were calculated using lognormal PMC size distributions
ice refractive indices (see section 3) for comparison to the HALOE data. In the first scenario,
HALOE was compared to model results based on the average PMC size distribution from von
 with ice refractive indices from
Bertie et al.
al.  (results using the Gosse et al.  indices were similar to those using the Warren
indices). To focus this comparison on spectral shape, each model curve was scaled to match the
m wavelength. The results show remarkable agreement between
the HALOE measurements and the normalized model PMC spectra (Figure 4a). The model
are similar for using the dif
ferent refractive index data, except near 10
m where the cold
temperature indices result in less extinction than the warm temperature indices, and thus more
closely match the HALOE measurement. Additional differences are found near 2.5
m, with no
clear separation between results based on the warm and cold temperature indices. For the second
scenario, spectra were calculated using the Bertie et al. indices with 11 PMC size distributions
von Cossart et al.
. This comparison evaluates the extinction magnitude as well as
shape by comparing unscaled model results to HALOE (Figure 4b). In this comparison, the
HALOE standard deviations are generally within the range of model extinctions based on the 1
distributions considered. The excellent agreement between HALOE and the modeled ice PMC
provides the first observational confirmation that the primary
component of PMCs is water
ice. While the current understanding of PMCs suggests that these clouds are composed of ice
this assertion has remained a subject of debate until now
. Uncertainties remained because
PMC measurements were only available at ultraviolet to near–infrared wavelengths, where
particulate spectra are void of unique absorption features or “fingerprints.”
fective radii were estimated from the measured extinctions using a first order
By fixing the width
(s) of a lognormal size distribution, a range of distributions can be
generated that depend only on the effective radius, R
= r exp[ 2.5 ln
(s)]. This range of size
Figure 4. HALOE PMC extinctions compared to various modeled extinction spectra. The
data are averages based on 16 PMC measurements from July 25 to August 4, 1997,
between 62°N and 72°N latitude. Vertical bars on each HALOE point represent the standard
of these measurements. a)
HALOE extinctions compared to model spectra calculated
using the average PMC size distribution from von Cossart et al.  with ice refractive
Bertie et al.
oon et al.
. The model spectra
scaled to match the HALOE extinction at 6.61
m wavelength. These scale factors were
0.76, 0.76, and 0.65 for results based on the Bertie et al., Warren, and Toon et al. indices.
HALOE extinctions compared to model PMC extinction spectra calculated using the
et al. refractive indices. Two model curves encompass the range of extinctions based on 11
PMC size distributions reported by von Cossart et al.
distributions was used to model relationships between R
and the ratio of extinction at two
wavelengths, independent of total particle concentration. These results indicate that HALOE
ratios based on the HF channel,
uniquely dependent on ef
> ∼20 nm (Figure 5). As a result, the measured ratios can be used to determine R
interpolating from modeled curves. The HALOE PMC extinctions from Figure 4 were used with
relationships based on the Bertie et al. refractive indices to estimate R
for the seven HALOE
ratios (Figure 5). These results give R
14 nm (average
standard deviation) for s =
Using s = 1.5 gives R
18 nm and using s = 2.1 gives R
1 nm. The standard
of the ef
fective radii estimates
from seven ratios is only 15%, indicating that the HALOE
PMC measurements are internally consistent. The effective radii determined from HALOE are
ger than indicated by previous studies. This may be a result of selecting only PMCs with
very strong signatures for this study, so chosen to more accurately characterize their spectrum.
clouds may have been more mature than average, and thus characterized by lar
according to simple condensational growth theory
. Another factor is that the typical response
a PMC by the 2.45
m channel is expected to be very low (Figure
4). As a result, the average 2.45
extinction from the 16 PMC sample could be biased toward high PMC extinctions, which would
the estimates of R
ger sizes. Nevertheless, previous results [e.g.,
, 1999] were based on visible measurements, and it is
possible that the infrared measurements
HALOE will shed new light on the properties of PMCs. Results using the other ice refractive
are similar to those above using the
Bertie et al.
 indices. While the initial estimates of
leave room for improvement, they suggest potential for the HALOE observations to reliably
PMC size distribution moments, and possibly size distributions. Reduced
future HALOE PMC extinction retrievals combined with more rigorous inversions will improve
and expand upon the analysis presented here.
Figure 5. Effective radius (R
) versus extinction
from model calculations (curves) based on ice
refractive indices from Bertie et al.  and
lognormal size distributions with fixed width (s =
Measured PMC extinction ratios based on the
HALOE data in Figure 4 were overplotted on the
model curves to give estimates of R
, and the
(94 nm) is indicated. Results for the
m channel were eliminated for visual clarity
fers the first measurements of the infrared signature of PMCs between 2.45
m wavelength. PMC extinctions derived from eight HALOE channels show remarkable
agreement with model PMC spectra based on ice particle extinction, and thus provide the first
confirmation that water ice is the
primary component of PMCs. A first order fit to the
PMC extinctions suggests the cloud particle ef
fective radius was between 69 and 128 nm.
The preliminary analyses offered here suggest the great potential for HALOE measurements to
the knowledge of PMCs, and future
forts will more rigorously assess these observations.
Acknowledgements. This paper combines efforts funded under under three NASA grants:
and NAG–7001 under
the NASA MTPE program and NAG5–9669 under the NASA
program. The authors thank Ellis Remsber
g, Lance Deaver
Anju Shah for many helpful
discussions, continued HALOE operation, and production of HALOE data.
M., M. Gerding, J.
, and U. von Zahn, NLC particle properties from a five color lidar
observation at 54°N, J. Geophys. Res., 105, 12,235–12,240, 2000.
Backhouse, T. W., The luminous cirrus cloud of June and July, Meteorol. Mag., 20, 33, 1885.
J. E., H. J. Labbe, and E. Whalley
, Absorptivity of ice I in the
Range 4000–30 cm
Phys., 50, 4501–4520, 1969.
Debrestian, D. J., J. D. Lumpe, E. P. Shettle, R. M. Bevilacqua, J. J. Olivero, J. S. Hornstein, W.
Glaccum, D. W. Rusch, C. E. Randall, and M. D. Fromm, An analysis of POAM II solar
observations of polar mesospheric clouds in the southern hemisphere,
Res., 102, 1971–1981, 1997.
Evans, W. F. J. , L. R. Laframboise, K. R. Sine, R. H. Wiens and G. G. Shepherd, Observation of
mesospheric clouds in summer
, 1993 by the WINDII instrument on UARS,
Lett., 22, 2793–2796, 1995.
Gadsden, M. The secular change in noctilucent cloud occurrence: Study of a 31–year sequence to
clarify the causes, Adv. Space Res., 20, 2097–2100, 1997.
S., D. Labrie, and P
, Refractive index of ice in the 1.4–7.8–
m spectral range,
Applied optics, v 34 n 28, p6582, 1995.
Hervig M. E., J. M. Russell III, L. L. Gordley, S. R. Drayson, K. Stone, R. E. Thompson, M.
Gelman, I. S. McDermid, A. Hauchecorne,
. Keckhut, T
. J. McGee, U. N. Singh, and M.
of temperature measurements
from the Halogen Occultation Experiment,
J. Geophys. Res., 101, 10,277–10,285, 1996.
Lübken, F–J., K. H. Fricke, and M. Langer, Noctilucent clouds and the thermal structure near the
Arctic mesopause in summer, J. Geophys. Res., 101, 9489–9508, 1996.
Olivero, J. J., and G. E. Thomas, Climatology of polar mesospheric clouds, J. Atmos. Sci., 43,
J.J., G. E. Thomas, W
. Evans, D. Debrestian and E.
Shettle, Three satellite comparison of
mesospheric cloud properties: A first look,
rans. American Geophys. Union, 77
Roble, R. G., and R. E. Dickenson, How will changes in carbon dioxide and methane modify the
mean structure of the mesosphere and thermosphere?, Geophys. Res. Lett., 16, 1441–1444,
., G. E. Thomas,
and E. J. Jensen, Particle size distributions in polar mesospheric clouds
from solar mesosphere explorer measurements,
J. Geophys. Res., 96,
J. M. III, L. L. Gordley
J. H. Park, S. R. Drayson, W
. D. Hesketh, R. J. Cicerone, A. F
J. E. Frederick, J. E. Harries, and P. J. Crutzen, The Halogen Occultation Experiment, J.
Geophys. Res., 98, 10,777–10,797, 1993.
G. E., Is the polar
mesosphere the miner
s canary of global change?.
. Space Res., 18
Thomas, G.E. and J.J. Olivero, Climatology of Polar Mesospheric Clouds. 2. Further analysis of
Solar Mesosphere Explorer Data, J. Geophys. Res., 94, 14,673–14,681, 1989.
Toon O. B., M. A. Tolbert, B. G. Koehler, A. M. Middlebrook, and J. Jordan, Infrared optical
O ice, amorphous nitric acid solutions, and nitric acid hydrates,
J. Geophys. Res
99, 25,631–25,645, 1994.
Warren, S. G., Optical constants of ice from the ultraviolet to the microwave, Appl. Optics, 23,
Cossart, G., J. Fiedler
, and U. von Zahn, Size distributions of NLC particles as determined from
observations of NLC by ground–based lidar
Geophys. Res. Lett., 26,
Wegener, A., Die Erforschung der obersten Atmospharenschichten, Gerlands Beitr. Geophys., 11,