Spatial Infrared Imaging Telescope (SPIRIT) III radiometer on the Midcourse Space Experiment (MSX) satellite measured highly structured infrared, IR, emission from polar mesospheric cloud (PMC) ice particles at northern latitudes above 51 degrees on 22 July 1996 in the 11.1 to 13.2 and 18.2 to 25.1 μm radiometer channels, bands C and E, respectively. Measurements of the PMC thermal emissions included the observation of an extended cloud at 84.8°N and 325.6°E at 0313:25 UT, a local solar time of approximately 0056. In this Earth limb observation, the radiance due to the PMC has been isolated from other sources-atmospheric emission, nonrejected off axis radiation from the terrestrial surface and zodiacal radiance-and inverted to determine the volume emission rates of the ice particles at a spatial resolution of 0.3 km in the altitude range from 83.4 to 86.4 km. The band C PMC volume emission rate profile has a maximum value at 84.0 ± 0.3 km and decreases to one half the peak value at 85.0 and 83.5 km. Temperatures in the range from 143 ± 7 to 130 ± 8 K and ice volume densities from 1.5 to 0.5 × 10−13 cm3 per cm3 were determined from the LWIR volume emission rates at altitudes from 83.4 to 86.4 km. The PMC ice densities are equivalent to an enriched gas phase water mixing ratio of 8 to 16 parts per million by volume, ppmv, and a vertical column mass density of 3.3 × 10−8 gms cm−2 in this observation.
 The microphysics, global distributions and frequency of occurrence of polar mesospheric clouds have been the focus of much interest in recent years stimulated in part by the suggestion that PMCs, a sensitive diagnostic of atmospheric temperature and composition, may serve as a monitor of anthropogenic effects on climate [Thomas, 1996, 2003]. The microphysics of PMC cloud ice particles have been described in a number of reports [Jensen and Thomas, 1988; Klostermeyer, 1998; von Zahn and Berger, 2003; Rapp and Thomas, 2006]. Stevens et al.  analyze the PMCs measured by the solar backscatter ultraviolet (SBUV) experiment, calculate the total PMC cloud mass, and conclude the space shuttle water exhaust from the main launch engines is the source of 22% of the PMC ice density at latitudes from 65° to 75°N during an eight day period in the Arctic summer of 1999. Grossman et al.  provide the first description of the infrared spectra of PMC ice particles as measured in early August 1997 with the shuttle based CRISTA-2 sensor at latitudes from 65 to 72°N. Their analysis concludes the ice cloud thermal emission is equivalent to a gas phase water concentration of 2.9 ppmv. The magnitude and significance of long term trends in the occurrence frequency of PMCs have been debated by von Zahn  and Thomas et al. . Space based observations of the latitude dependent occurrence frequency and brightness of PMCs and their secular trends are summarized by DeLand et al.  based on 27 years of measurements from the Solar Backscatter Ultraviolet (SBUV) satellite instruments. Their review concludes that after correcting for the anticorrelation with solar cycle the SBUV data indicate a long term increase in both PMC occurrence frequency and brightness in both hemispheres with an increase in albedo over the 27 years of 12% to 20% depending on hemisphere and latitude.
 The Midcourse Space Experiment (MSX), a Missile Defense Agency research program designed to support the development of advanced infrared space-based sensors, was launched on 24 April 1996 into a nearly sun-synchronous orbit at an altitude of 900 km. The MSX program objectives, sensors and experiments are described by Mill et al.  and O'Neil et al. . The characterization of high latitude infrared emissions from aurora [Sharma et al., 2001; O'Neil et al., 2007] and PMCs was part of the MSX Earth limb background measurements program. Polar mesospheric cloud observations by the MSX ultraviolet and visible (UVISI) sensors are described by Carbary et al. [1999, 2001, 2002, 2003]. Corbin et al.  and E. Bauer (unpublished report, 1980) estimated the infrared radiance of PMCs for representative particle sizes and densities and concluded the thermal emission of the cloud ice particles would exceed the emission from atmospheric molecular species in selected infrared spectral bands. Charles H. Humphrey (unpublished report, 1995) and other MSX team members designed and scheduled MSX experiments to measure the infrared radiance from PMCs. Preliminary results of these measurements were presented by Humphrey et al.  and O'Neil et al.  and are briefly described by Stair et al. . A more detailed description of the characteristics of PMC ice particles based on the observation of their LWIR thermal radiance at wavelengths in the range from 11.1 to 25.1 μm measured by two spectral bands of the SPIRIT III radiometer on 22 July 1996 is reported here. The following sections present: the MSX observation (section 2), the PMC limb radiance and volume emission profiles (3), the ice particle temperatures, densities and equivalent water vapor mixing ratios (4) and finally, the conclusions are given in section 5.
2. MSX-Infrared Measurements of Polar Mesospheric Clouds
 An experiment designed to observe PMCs and conducted on 22 July 1996 is illustrated in Figure 1. The line of sight (LOS) of the SPIRIT III optical axis was oriented with an azimuth angle 150° from the spacecraft heading and at a tangent point altitude of approximately 95 km. The experiment was initiated at 0257 UT with a tangent point located at 37°N and 209°E, continued as the tangent point approached the pole at 0312:30 UT (86°N, 290°E), and was completed at 0323:30 UT with a tangent point location of 52°N, 4°E. Figure 1 illustrates the location of the day night terminator at altitudes of 0, 100, and 900 km and indicates that the Earth limb region observed by the SPIRIT III sensor was sunlit during this measurement.
 The SPIRIT III cryogenic radiometer, described by Bartschi et al. , contained six spectral bands in the wavelength range from 4.2 to 25.1 μm as specified in Table 1. The B1 and B2 bands shared a single focal plane detector array. Each of the five focal planes contained from two to eight columns of 192 pixels with each column providing a one degree field of view (FOV), equivalent, as vertically oriented in this event, to a tangent height range from approximately 65 to 125 km and a spatial resolution of 0.3 km at the tangent point. In addition to bands C and E, measurements from band A (6.8–10.8 μm) were used in the current analysis. Infrared emission from PMCs were prominent features in bands C and E and band A was used to estimate the component of band C radiance that was due to atmospheric emission from the 9.6 μm ozone band. Earth limb radiance measured at apparent tangent heights from 65 to 100 km by band C of the SPIRIT III radiometer is shown in Figure 2. PMC emissions were observed as discrete features with highly structured forms at LOS tangent heights from approximately 68 to 87 km. PMCs are concentrated in thin layers, several kilometers in thickness, with maximum densities at altitudes from approximately 82 to 84 km [e.g., Carbary et al., 2001]. Hervig et al.  confirmed the cloud particles are composed of water ice. The PMC features in Figure 2 at apparent tangent heights below 80 km are due to discrete cloud forms located before or beyond the tangent point of the LOS and thus appear at lower tangent altitudes in the Earth limb view. Figure 3 presents a more detailed view of the band C measurements in the period from 0313:00 to 0316:00 UT with in-scan radiometric profiles at tangent heights of 112 and 84 km and a cross-scan profile at 0313:25 UT over the full range of tangent heights from 69 to 129 km. Figure 3 indicates a fairly constant radiance level of approximately 1.5 × 10−8 W cm−2 sr−1 at tangent heights above 97 km due primarily to nonrejected off-axis radiation from the Earth’s surface. The nonrejected Earth radiance (NRER) is produced by scatter from particulate contamination accumulated on the primary mirror of the SPIRIT III telescope during the period of prelaunch operations and during launch of the MSX satellite [O'Neil et al., 2006]. The arc-like features in Figures 2 and 3 are a consequence of maintaining the LOS at an oblique angle from the spacecraft heading and provide information on the location of the cloud along the LOS. Carbary et al.  describe similar arc-like images of PMCs measured with the MSX ultraviolet imager. In the measurement reported here, the SPIRIT III sensor, operating in a push-broom scan mode with an off-track heading of 150°, intercepted discrete cloud forms at ranges before or beyond the tangent point and, as the SPIRIT III sight lines traversed obliquely through the extended horizontal dimensions of the cloud layer, the cloud location along the sight lines approached, coincided with, and receded from the tangent point producing the arcs illustrated in Figures 2 and 3. The peak of a fully formed arc provides a measure of the cloud tangent height and, additionally, insight into the cloud size and this feature can be used to identify regions which support the assumption that the cloud layer characteristics within the path length intercepted along the LOS are uniform and vary only in the radial or vertical dimension, a necessary condition to apply the Abel inversion algorithm to the PMC radiance profile and determine the volume emission rates of the cloud ice particles. The fully formed symmetric arc observed at 0313:25 UT in Figures 2 and 3 fits this description and is the focus of the present analysis.
3. LWIR Limb Radiance and Volume Emission Profiles
 A series of band C and E limb radiance profiles at LOS tangent height altitudes from 70 to 90 km are shown in Figure 4 at selected times from 0257:00 to 0323:30 UT illustrating the variability in the atmospheric background and the enhanced radiance produced by the presence of PMCs. These and similar limb radiance profiles, based on data from two columns of 192 pixels offset by a half pixel, sampled at 72 Hz and averaged over one second, were used in this analysis. The limb radiance profiles at 0313:25 UT (labeled 73:25 in Figure 4) show enhancements of factors of approximately 30 and 6 above the non-PMC profiles in bands C and E, respectively, and a peak radiance of 2.5 × 10−7 W cm−2 sr−1 in band C. Consistent with preflight simulations (C. H. Humphrey, private communication, 1993), PMC radiance was masked by atmospheric emission in the SPIRIT III radiometer bands other than bands C and E.
 The PMC components were extracted from the Earth limb radiance profiles measured at 0313:25 UT in a sequence of steps. The radiance component due primarily to NRER together with a small contribution from the zodiacal background was determined by fitting the slowly varying high altitude profiles and extrapolating the profile to lower altitudes using the NRER radiance profiles described by O'Neil et al. . The NRER and zodiacal components of limb radiance were subtracted from a series of limb profiles including those illustrated in Figure 4 to yield limb measurements composed of atmospheric and, if present, polar mesospheric cloud radiance for bands C and E. The dominant source of atmospheric radiance in band C is the long wavelength tail of the 9.6 μm O3 vibrational band complex and in band E the atmospheric radiance is due to water vapor rotational state transitions. The 9.6 μm O3 vibrational band complex results from emission from the ν3 mode of O3 with the O3 (v3 = 1) vibrational state produced directly from the O3 ground state by collisional excitation and earthshine absorption and vibrational populations of O3 (v3 ≤ 7) resulting from the three body recombination of atomic and molecular oxygen. A comparison of bands A and C showed the atmospheric radiance levels and variability were highly correlated in the absence of PMCs at tangent heights from 65 to 90 km. Emission from O3 is the primary source of atmospheric emission in both bands and the ratio of the emission from the lower vibrational states (band A) to the higher vibrational states of O3 in the long wavelength tail (band C) was relatively constant in this sunlit measurement at high latitudes in regions with no evidence of PMCs. The band C to A ratio of atmospheric radiance was determined for a number of tangent height profiles in these regions and the average ratio applied to the band A measurement to estimate the atmospheric emission component in the corresponding band C profile at 0313:25 UT. This method to distinguish the band C components of limb radiance due to ozone and the PMC ice particles may be subject to some error if the water vapor density coincident with the PMC is substantially different from the levels in the regions used to determine the band C to A ratio of atmospheric radiance. Von Zahn and Berger  propose that water vapor is depleted in the region above the peak of the PMC layer due to freeze drying and enhanced below the layer due to sublimation of the ice particles. Siskind et al.  show mesospheric ozone and water vapor densities are inversely correlated by the effects of HOx chemistry. Changes in water vapor density induced by the formation or sublimation of the ice particles and the inverse effect on the density of ambient O3 impacts the spectra of the 9.6 μm O3 vibrational band complex by changing the excitation rate of the O3 (v3 = 1) state produced by the direct excitation of O3 (v3 = 0). The change in the water vapor density would have little effect on the production of vibrationally excited ozone from the chemiluminescent three body process including the emission from the lower vibrational states that radiate within band A populated directly or by radiative cascade from upper levels formed in this process. Thus a depleted H2O density in the upper region of the cloud layer would somewhat enhance the band A radiance and could result in an overestimate of the O3 atmospheric emission and an underestimate of the PMC cloud radiance components in the upper altitude region of the band C profile. Conversely, at lower altitudes an enhanced water vapor level below the peak of the cloud layer could result in an overestimate of the PMC radiance component of band C. However, the magnitude of the PMC radiance enhancement mitigates the potential impact of this uncertainty in the current analysis. The atmospheric ozone component of the band C radiance profile is estimated to be approximately three percent of the PMC component at a tangent height of 84 km and a slight error in the component of limb radiance attributed to ozone would be substantially less than the other sources of error in this analysis and would not significantly impact the results reported here.
 The atmospheric component in band E was based on the average of six high latitude band profiles recorded at times during the event with no evidence of enhanced radiance due to PMCs in band E or C. This approach assumes the enhanced band E radiance observed at 0313:25 UT is due to both PMC ice particles and water vapor emission at levels that are comparable to other high latitude regions with no detectable ice particles. Observations of water vapor and PMC ice particles are described by Summers et al.  who report enhanced levels of water vapor (10–15 ppmv) coincident with PMC ice particles at tangent heights from 82–84 km based on an analysis of measurements by the Middle Atmospheric High Resolution Spectrograph Investigation (MAHRSI) and the Halogen Occultation Experiment (HALOE) in mid August 1997. Additional analysis of the MAHRSI data from this time period by Stevens et al.  shows the water vapor mixing ratio is highly variable in the Arctic summer with values in the range from 3 to 23 ppmv at altitudes between 82 and 85 km and the PMCs are present in both saturated and unsaturated air. Although the band E PMC enhancement was a factor of approximately 6 greater than the other components of limb radiance (the atmosphere, NRER and zodiacal emission) at 84 km, the estimate of the magnitude of the atmospheric water vapor component of limb radiance in band E was the largest source of error in the ice particle temperatures and number densities determined in the present analysis. Removing the atmospheric, NRER and zodiacal components of limb radiance from bands C and E in this manner produced the PMC radiance profiles shown in Figure 5.
 An Abel inversion algorithm for optically thin limb radiance [e.g., Hays et al., 1973] was used to calculate the volume emission rates shown in Figure 5b from the PMC radiance profiles in Figure 5a. The PMC volume emission rate for band C has a maximum value of 8.6 × 10−15 W cm−3 sr−1 at 84.0 km and decreases to half this level at 83.5 and 85.0 km, forming a 1.5 km cloud layer by this definition. Use of the Abel inversion algorithm in this application requires that the PMC ice particle emission uniformly filled the LOS over the 280 km path length through the 1.5 km thick cloud layer at 84 km and the PMC volume emission rate varied only in the altitudinal dimension. These conditions appear likely for the SPIRIT III PMC measurement at 0313:25 UT assuming the cloud dimensions along the LOS are similar to or greater than the horizontal cloud dimensions illustrated in Figures 2 and 3 at this time. The uncertainties in the PMC radiance profiles illustrated in Figure 5a (1.7 and 4.9% at 84 km in bands C and E, respectively) are due to the combined effects of the photon noise in the measurements, the pixel to pixel responsivity (flat fielding) errors, and errors introduced in accounting for the atmospheric, NRER, and zodiacal components of limb radiance. The uncertainties in the volume emission rates were determined from 100 radiance profiles created by randomly adding the noise and estimated radiance uncertainties to a nominal smoothed PMC radiance profile for each band and inverting the profiles to determine the standard deviation in the volume emission rates for the 100 samples [see Ramsey and Diesso, 1999]. The standard deviation in the volume emission rates determined in this manner is 5 and 24% in bands C and E, respectively, at 84 km as illustrated in Figure 5b.
4. Ice Particle Temperatures and Densities
 The LWIR signatures of the ice particles comprising the PMC are due to both thermal emission and scatter of upwelling radiation from the lower atmosphere and terrestrial surface. The magnitude of these processes may be calculated based on the absorption (emission), σa, and scattering, σs, cross sections and assuming the ice particles may be represented as spherical particles. Using Mie theory for spherical particles with dimensions small compared to the observing wavelength, Van de Hulst  shows the cross sections may be expressed as:
where σa and σs are the cross sections in cm2,
 a is the particle radius in cm,
 x, dimensionless, is defined as x = 2 π a/λ, where λ is the wavelength of the radiation,
 Im is the imaginary component of this term, and
 m = n − ik, is the complex index of refraction of ice.
 The infrared complex index of refraction of water ice has been reported by: Bertie et al.  for wavelengths from 1.3 to 330 μm at a temperature of 100 K; Tsujimoto et al.  from 2.5 to 22 μm at 77, 150 and 180 K; Warren  from 0.045 to 167 μm at 266 K; Hudgins et al.  from 2.5 to 200 μm at 10 to 140 K; Toon et al.  from 1.4 to 20.7 μm at 163 K; and Clapp et al.  from 2.5 to 12.5 μm at temperatures from 130 to 210 K. Curtis et al.  present indices from high resolution measurements in the wavelength range from 15 to 200 μm at 10 K temperature increments from 106 to 176 K and provide a critical review of the large disparities in previously reported measurements in the wavelength region that includes the SPIRIT III E band (18.2–25.1 μm). In this analysis, the complex index of refraction for ice given by Toon et al.  at 163 K was used for band C with the assumption the index was constant over the temperature range of interest in this measurement and the results of Curtis et al.  at temperatures from 106 to 176 K were used for band E.
 The component of the PMC LWIR emission due to scatter of upwelling radiance from the lower atmosphere and terrestrial surface was calculated using expression (2) to determine the cross section for scattered radiation from a spherical ice particle with a 100 nm radius. The upwelling radiance was simulated using the SAMM2 (SHARC-4 and MODTRAN4 Merged) [Dothe et al., 2004] model for the time and location of this PMC observation. The calculated scattered earthshine radiance level was less than one percent of the measured volume emission rate for both bands C and E. This estimate is an upper limit since PMC particles sizes are nominally less than 100 nm [see review article by Deland et al., 2006, and references therein] and the scattering cross section is proportional to the 6th power of the radius, (2). Thus earthshine scatter is not a significant consideration for the present analysis.
Fiocco et al.  calculate equilibrium temperatures for small particles in the Earth's atmosphere at altitudes from 50 to 110 km and show the power emitted by small spherical particles may be represented as:
where BB (λ, T) represents the Plank blackbody function at particle temperature, T, and is given by:
where c1 and c2 are the first and second radiation constants, respectively.
where ∣m2+2∣ represents the absolute value of this term. Thus the power emitted by a spherical particle, (5), in the Rayleigh limit (dimensions less than the measurement wavelength) is proportional to its volume. The volume emission rates presented in Figure 5b are given by:
where Vd is the ice particle volume density in cm3 per cm3 and SR (λ) is the system spectral response function for band C or E.
 Using the indices of Toon et al.  and Curtis et al.  in (6), the ratios of the C to E band volume emission rates were calculated at 10K temperature increments in the range from 106 to 176 K with the assumption that the measurements of Toon et al.  at 163 K represent the complex index of refraction for band A over this temperature range. The temperatures of the PMC ice particles were determined by comparing the measured and modeled ratio of the C to E volume emission rates and using a linear interpolation between the model calculations at 10 K temperature increments. The ice particle temperatures determined in this manner at altitudes from 83.4 to 86.4 km are given in Table 2 and illustrated in Figure 6. The uncertainties in the absolute calibration of the SPRIT III radiometer, 3 and 6 percent in bands C and E, respectively, [T. Murdock, private communication, 1999; see also Price et al., 2004] contributed no significant error to the results in Table 2. The errors given in Table 2 are based on the propagation of the uncertainties in the ice particle volume emission rates illustrated in Figure 5b. Recent studies [Eremenko et al., 2005; Rapp et al., 2007] indicate PMC ice particles are nonspherical. It was assumed no significant additional errors were contributed to this analysis by uncertainties in the indices of refraction or the use of Mie theory based on spherical symmetry.
Table 2. Polar Mesospheric Cloud Ice Particle Temperatures, Volume Density, Equivalent Gas Phase Mixing Ratios, and Vertical Column Mass Density
 The equilibrium temperatures of mesospheric spherical ice particles have been calculated by Grams and Fiocco  and Espy and Jutt ; their results show the particle temperatures are elevated above the atmospheric kinetic temperature with increasing differentials at higher altitudes and with larger particles size. Rapp and Thomas  have applied the results of Espy and Jutt  to determine ice particle temperature differentials from the ambient atmosphere as a function of particle radius and altitude for an accommodation coefficient of one, that is complete thermal accommodation (particle cooling) in particle and air molecule collisions. Rapp and Thomas  indicate the temperature differential is inversely proportional to the accommodation coefficient and the temperature of a 50 nm size particle is elevated 1 and less than 2 K above the ambient atmospheric temperature at altitudes of 84.0 and 86.4 km, respectively, for an accommodation coefficient of one. Thus these model results indicate the ice particle temperatures determined in this analysis may be slightly warmer than the atmospheric temperature but within the uncertainties shown in Table 2. The ice particle temperatures presented in Table 2 are in excellent agreement (a few degrees) with the atmospheric temperatures in the MSIS 2000 model for this time and location and also with the high latitude (80-90° N) zonal mean temperature model result of von Zahn and Berger  for 21 June that is benchmarked with midsummer measurements at 28, 54, 69, and 78°N from a number of investigators.
 Given the ice particle temperature determined in this manner, the temperature was then used in (6) to solve for the volume of ice particles producing the band C emission. Ice volume densities in the range from 0.5 to 1.5 × 10−13 cm3 per cm3 were determined at altitudes from 83.4 to 86.4 km as indicated in Table 2 and Figure 7. The ice volume density is a direct result of the measurement of the SPIRIT III LWIR radiometer bands and represents the total volume of particles of all sizes. The ice volume density profile shows a broad peak value in the range from 1.4 to 1.5 × 10−13 cc per cc at altitudes from 83.6 to 84.8 km suggesting a region where the microphysical processes of ice crystal growth and decay approach a local steady state to conserve the total ice volume. Integrating the ice volume density over the altitude of this measurement and converting to mass density results in a vertical mass column density of 3.3 × 10−8 gms cm−2. Stevens et al.  compare the annual average vertical column mass of PMC ice particles measured by the Student Nitric Oxide Explorer (SNOE) and the Solar Backscattered Ultraviolet (SBUV) satellite experiments at 70 ± 2.5°N and 11:24 ± 1:00 LT from 1998 to 2002. The data were limited to periods of the brightest 10% of the PMCs observed by the SNOE sensor and the comparable SBUV measurements for this location and time. A vertical mass column of PMC ice particles of up to approximately 2.5 × 10−8 gms cm−2 was derived from an analysis of limb viewing measurements of the Student Nitric Oxide Explorer (SNOE) experiment that converted the data to a vertical view and assumed a Gaussian ice particle size distribution with a characteristic radius of 60 nm and distribution width of 15 nm. The SNOE results were 80 % greater than the average column ice mass measured by the SBUV instrument in comparable data samples analyzed with the same particle size assumptions. Stevens et al.  conclude the derived SNOE and SBUV vertical column densities agree within their uncertainties and indicate the disparity may be due to the assumptions made in the particle radius and size distribution. Rapp and Thomas  model the time dependent development of PMC ice particles and include a sensitivity study of the ice column mass density after 24 hours of ice cloud development. Their model indicates the ice particle volume is more sensitive to water vapor density than temperature and the ice column density reported here, 3.3 × 10−8 gms cm−2, is toward the upper end of the range of conditions considered in their sensitivity study for 24 hours of cloud development.
 The ice volume densities are also expressed in equivalent gas phase water vapor mixing ratios in Table 2 based on the total density given by the MSIS 2000 model atmosphere for this measurement. At 84.0 km the ice density is equivalent to a water vapor gas phase mixing ratio of 14 parts per million by volume, ppmv. As noted, Grossman et al.  describe early August 1997 limb measurements at latitudes from 65 to 72°N of the infrared spectra of a PMC at wavelengths near 12 μm from the shuttle based CRISTA-2 mission and analysis that concludes the ice cloud thermal emission is equivalent to a gas phase water concentration of 2.9 ppmv. A comparison of the radiant intensity of the CRISTA-2 spectra and the MSX measurement has been made by integrating the product of spectral radiance presented by Grossman et al.  and the MSX band C system spectral response function. The CRISTA-2 PMC radiance determined in this manner is 3.1 × 10−8 W per cm−2 sr−1, a factor of 7.6 less than the MSX band C value at 84 km. This compares with value of 5.5 for the ratio of the maximum ice particle equivalent water vapor levels reported here at 84.6 km (16 ppmv) to that given by Grossman et al.  at 82 km (2.9 ppmv) in the analysis of the MSX and CRISTA-2 experiments, respectively. The ice particle equivalent water vapor levels derived from the CRISTA-2 and MSX observations are proportional to the infrared radiance in the measurements within the uncertainties in the reported water vapor mixing ratios. The mixing ratios in Table 2 range from 9 to 16 to 8 ppmv at increasing altitudes from 83.4 to 86.4 km with a maximum value at altitudes near 84.6 km. These values are similar to the results of von Cossart et al.  who report a mean value of the PMC ice density equivalent to a gas phase water vapor mixing ratio of 12 ppmv based on ground based lidar measurements above the ALOMAR research site in Norway (69°N, 16°E). Stevens et al.  infer water vapor densities from analysis of hydroxyl (OH) ultraviolet emission measured by the MAHRSI (Middle Atmosphere High Resolution Spectrograph Investigation) sensor and report mid August 1997 enhanced water vapor levels at 82 km and latitudes above 65°N with mixing ratios in the range from 3 to 23 ppmv and an average value of 12 ppmv. As noted, Summers et al.  report water vapor mixing ratios of 10–15 ppmv at 82–84 km for July and August 1997 measurements from the MAHRSI and the Halogen Occultation Experiment (HALOE) that are coincident with PMC ice particles.
 Thus enriched water levels in both the gaseous and solid state have been observed in the summer polar mesosphere at attitudes in the range from 82 to 86 km that in some instances are comparable to the ice particle density reported here. Vertical transport of water vapor from the lower atmosphere and water vapor sequestering by nucleation and freeze drying at higher altitudes by small ice particles followed by sedimentation and sublimation have been proposed as mechanisms for water enrichment at these altitudes. Sublimation occurs at the higher ambient temperatures at altitudes below approximately 81 to 83 km and above these altitudes in temperature fluctuations driven by atmospheric tides and waves. Siskind et al.  present results for a two-dimensional coupled ice chemistry and atmospheric dynamics model to study the interaction of PMCs and atmospheric temperature as well as water vapor and ozone densities in the summer mesopause region. Their results indicate PMCs increase the kinetic temperature several degrees above the cloud layer by thermal radiation and impact the ozone density above and below the cloud layer by the PMC effects on the water vapor concentration and its inverse correlation with ozone density. The simulation of Siskind et al.  shows an ozone increase above the cloud layer due to dehydration associated with PMC formation and a decrease below due to water vapor enrichment from particle sublimation. Atmospheric limb radiance due to ozone and water vapor in the bands C and E, respectively, are variable in this measurement as illustrated in Figure 4. Change in ozone and water vapor density and radiance above and below a cloud layer correlated with the formation, presence and loss of the PMC ice particles as distinguished from atmospheric dynamics or other non-PMC related sources of ozone variability remains an intriguing subject of further study.
 The enhanced limb radiance profiles of a PMC were measured in two LWIR bands of the SPIRIT III radiometer on the MSX spacecraft at 84.8°N, 325.6°E, and 1313:25 UT on 22 July 1996 in a push-broom scan orientation that allowed the altitude and spatial extent of the PMC to be determined. The PMC limb radiance profiles, factors of 30 and 6 greater than the other components of limb radiance at a tangent height of 84 km in the 11.1–13.2 and 18.2–25.1μm bands, respectively, were isolated and inverted to determine the volume emission rates of the PMC in each band at a spatial resolution of 0.3 km in the altitude range from 83.4 to 86.4 km. The band C PMC volume emission rate profile has a maximum value at 84.0 ± 0.3 km, decreases more steeply with altitude below the peak, and reduces to one half the peak value at 85 and 83.5 km. The ratio of the volume emission rates provides a direct measure of the cloud ice particle temperature and the temperature, in turn, allows the ice volume density to be determined from the absolute radiance. The present analysis used the complex index of refraction for water ice measured by Toon et al.  at 163 K for the 11.1–13.2 μm band and by Curtis et al.  at 10 K increments from 116 to 176 K for the 18.2–25.1 μm band. The later results appear to resolve the large disparities reported in earlier measurement in this wavelength region and the consistency of the present results for ice temperature and density with other measurements based on different techniques and with models of atmospheric temperature validated with experimental results [i.e., von Zahn and Berger, 2003] lend credence to the methodology used in this analysis.
 PMC ice particles result from heterogeneous nucleation in the mesosphere as described in microphysical models that simulate the nucleation, growth, transport, and sublimation of the ice particles and describe their relationship to the smaller particles responsible for polar mesospheric echoes. The details of the nucleation process and the subsequent growth and decay of the ice particles whether driven by a combination of sedimentation and atmospheric dynamics [Jensen and Thomas, 1988, 1994; von Zahn and Berger, 2003; Rapp and Thomas, 2006] or primarily by atmospheric tides and gravity wave mechanisms [Klostermeyer, 1998] as well as the implications of PMCs on mesospheric HOx chemistry [Siskind et al., 2007] are topics of numerous recent studies and will be further advanced as new evidence is gathered. The high resolution altitude profiles of PMC ice temperatures and volume densities presented here result from the direct measurements of LWIR thermal emission, are independent of illumination conditions, particle size, and particle size distributions and contribute to a more complete characterization the ice particles forming PMCs.
 The support and encouragement of John Mill, the MSX Project Scientist, currently of SpaceX Consulting and Bruce Guilmain, the MSX Program Manager, currently with Utah State University, are respectfully and thankfully acknowledged.
 Zuyin Pu thanks Gary Thomas and another reviewer for their assistance in evaluating this paper.