Measurements of polar mesospheric clouds in infrared emission by MIPAS/ENVISAT

Authors


Abstract

[1] We report on the infrared emission (10–13 μm) of polar mesospheric clouds (PMCs) as measured by Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on board the Environmental Satellite (MIPAS/ENVISAT) during 19–21 July 2005 in the summer Northern Hemisphere. The spectral distribution of MIPAS radiances is well described by that simulated for ice particle emission at low temperatures and hence provides further evidence of the water ice nature of the PMC particles. The vertical profiles of integrated radiance clearly show an emission peak at tangent heights of 81–83 km, as expected from PMC emission. Ice particles' volume density retrieved from MIPAS spectra shows a layer of ice particles extending from 80.5 to 87 km and from about 60°N to the North Pole, well confined within the region where temperature is below approximately 150 K. The upper part of the ice particle layer then extends to altitudes higher than that observed by UV/visible measurements, which confirms the current understanding that PMCs cover the entire polar mesopause region. The altitude of the layer peak changes from 82.5–83 km at 70°N–90°N to slightly higher altitudes (83–85 km) at lower latitudes (60°N–70°N). MIPAS ice particle altitude/latitude distribution suggests that they are formed mainly close to the pole and then are transported upward and to lower latitudes by the ascending branch of the meridional circulation, confirming previous modeling.

1. Introduction

[2] Polar mesospheric clouds (PMCs), usually called noctilucent clouds (NLCs) when observed from the ground, occur at the coldest regions of the atmosphere near the summer high-latitude mesopause. PMCs normally form a layer, a few kilometers in vertical extent, peaking near 83 km, located at latitudes poleward of 50°, where temperature frequently drops below frost point which, for mesospheric pressures and humidities, is as low as about 150 K. They mainly consist of water ice particles with radii that usually range from about 25 nm to about 100 nm [Hervig et al., 2001; Rusch et al., 1991; Gumbel and Witt, 1998; von Savigny et al., 2005].

[3] PMCs are being discussed as potential early indicators of global change [Thomas et al., 1989; von Zahn, 2003] since they are very sensitive to temperature and water vapor concentration. Since enhanced CO2 amounts would lead to an eventual cooler upper mesosphere/lower thermosphere, and higher CH4 amounts may lead to enhanced H2O near the mesopause [Roble and Dickinson, 1989], they could both lead to an increase of PMC occurrence, which might then be seen as an effect of climate change in the upper atmosphere. There is not, however, a unanimous opinion in the scientific community on this aspect [von Zahn, 2003; Thomas et al., 2003].

[4] PMCs were intensively studied by observations from ground, rockets, and space (SNOE, SBUV, ODIN, SCIAMACHY, AIM) [Baumgarten and Fiedler, 2008; Fiedler et al., 2009; Gumbel and Witt, 1998; Bailey et al., 2005; DeLand et al., 2003; Petelina et al., 2006; von Savigny et al., 2005; von Savigny and Burrows, 2007; von Savigny et al., 2007a; Russell et al., 2009], and by sophisticated models [Berger and von Zahn, 2002, 2007; Rapp and Thomas, 2006]. Evidence for a quasi 5-day planetary wave has been inferred from SCIAMACHY NLC occurrence data by von Savigny et al. [2007b], and their apparent disappearance during the solar protons events of January 2005 is discussed by von Savigny et al. [2007c]. Baumgarten and Fiedler [2008] have shown a study of 8 years of Rayleigh/Mie/Raman (RMR) lidar measurements of NLCs above ALOMAR (69°N, 16°E), providing a very solid data set about the vertical structure of the ice particle properties including their backscatter coefficient at 532 nm, volume density, mean radius and number density.

[5] An assessment of the altitude where ice particles start nucleating and how they grow and evolve in the polar mesosphere region has recently been performed on the basis of a detailed 3-D model study by Berger and von Zahn [2007]. Our current knowledge of PMCs has been presented very recently in a review paper by Rapp and Thomas [2006].

[6] While PMCs emit thermal radiation, their observation by infrared emission techniques is very difficult because of the low icy particle volume density and the very cold mesopause temperatures, thus requiring very sensitive instruments for their detection. Actually, only two observations in the infrared, in emission, have been so far reported: that taken by CRISTA [Grossmann et al., 2006] and by the SPIRIT [O'Neil et al., 2008] instruments. Infrared emission spectroscopy, however, has the advantage that it can also measure PMCs under dark conditions and hence might provide a better knowledge of their spatial distribution.

[7] NLCs are only the optically visible (and lower) part of the layer of icy particles covering the entire polar mesopause region [Berger and von Zahn, 2002; Rapp and Thomas, 2006]; while the whole layer modifies the ambient plasma of the D region and gives rise to intense radar echoes called PMSE [Rapp and Lübken, 2004]. As mentioned above, several studies have been carried out about the vertical distribution of particle sizes [von Savigny et al., 2005; Baumgarten and Fiedler, 2008], and it has been found that the larger particles are located near the bottom of the layer, while smaller ones are more likely to be near the top [Berger and von Zahn, 2002; von Savigny et al., 2005; Baumgarten and Fiedler, 2008]. All of these observations are performed by measuring the scattered light, in the visible or UV, of the solar radiation in the case of instruments from space or of the lidar light in case of ground instruments. These techniques usually observe the ice particles with radii larger than about 20 nm but are very little sensitive to particles with radii below this value [see, e.g., Rapp and Thomas, 2006]. An exceptional case is the recent measurements by the SOFIE/AIM instrument [Hervig et al., 2009a].

[8] The infrared radiance emitted by the ice particles whose diameter is much smaller than the wavelength is proportional to their volume density (under the assumption of constant temperature). Thus, although infrared observations are more sensitive to larger particles, implying larger volume densities, they can also provide information about the presence of small ice particles, if in sufficient concentration.

[9] The aim of this paper is twofold. First, we report on the detection of PMCs from their emission in the infrared taken by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) instrument on board ENVISAT (Environmental Satellite), and hence provide further evidence of the water ice nature of the PMC particles. Second, by using the simultaneously measured temperatures from MIPAS, we present 3D distributions of the ice particle volume density retrieved from MIPAS measurements for 19–21 July 2005. These distributions are then discussed on the light of our current understanding of PMCs as obtained from both modeling and UV/visible observations.

2. MIPAS Measurements

[10] MIPAS is a high-resolution limb sounder on board the ENVISAT satellite, launched on 1 March 2002. It has a wide spectral coverage, high spectral resolution (0.025 cm−1 during 2002–2004 and 0.0625 cm−1 from 2005 to present), and high sensitivity, which allows measurement of most of the atmospheric emissions in the midinfrared in a wide altitude range [Fischer et al., 2008]. MIPAS also operates with a global latitude coverage (pole to pole) and performs measurements indistinctly at daytime or nighttime. The instrument spends most of the time observing in the 6–68 km altitude range (nominal mode) but occasionally also looks at higher altitudes in its middle atmosphere, NLC, and upper atmosphere modes [De Laurentis, 2005].

[11] MIPAS was measuring during 19–21 July 2005 in its NLC mode, a variant of the middle atmosphere mode adopted for the summer in the Northern Hemisphere [De Laurentis, 2005]. In this mode, the spectra are taken at tangent heights ranging from 39 km up to 78 km at 3 km steps; then from 78 km up to 87 km at 1.5 km steps, and finally from 87 km up to 102 km again in 3 km steps. These spectra were taken with the optimized spectral resolution of MIPAS, i.e., at 0.0625 cm−1. During this period, a total of 40 orbits were taken in the NLC mode, with nearly 100 scans (profiles) per orbit, hence accounting for a total of nearly 4000 scans. Here we have analyzed the L1b (geolocated, and spectrally and radiometrically calibrated) spectra in the region from 730 to 950 cm−1, for this period. The noise equivalent spectral radiance in this band at the optimized spectral resolution is about 20 nW/(cm2 sr cm−1).

3. PMC Infrared Spectra

[12] Figure 1 shows MIPAS coadded spectra at a tangent height of ∼81 km (1) for a selected set of scans where signatures of PMCs are more prominent (red) and (2) for the rest of the scans taken in this period (blue). In the first case, a total of 357 scans with significant presence of PMCs were coadded, dropping the noise from ∼20 to ∼1 nW/(cm2 sr cm−1), while 3421 spectra free of PMCs were coadded in the second case, lowering the noise down to ∼0.35 nW/(cm2 sr cm−1). Due to fluctuations of the actual tangent altitudes, coadding was performed for tangent altitude bins of 78–79.5 km, 79.5–81 km, and 81–82.5 km. Note, however, that this coadding does not affect the results since we are interested here only in the PMC spectral signature, which is independent of the averaging.

Figure 1.

MIPAS coadded spectra at a tangent height of ∼81 km for conditions when PMCs are present (red) and absent (blue). The measurements cover all latitudes from pole to pole. The geolocations of the spectra with PMCs signatures are shown in Figure 3a. Note that the fine-scale fluctuations are noise.

[13] Limb scans with PMC signal were distinguished from PMC-free measurements on the basis of the altitude profile of radiances integrated from 770 to 920 cm−1, particularly the contrast of radiances at tangent altitudes where PMCs are expected to occur and tangent altitudes supposed to be PMC-free. Specifically, those limb scans where selected as PMC measurements where the integrated radiance at tangent altitudes of 78–82.5 km exceeded that at tangent altitudes of 88.5–96 km by at least 20%. Although the spectral noise is still large, and there seems to be a problem with zero-level offset calibration, the enhanced signal in the PMC spectrum is clearly noticeable. The coadded spectra at tangent heights of ∼79.5 km and ∼82.5 km show very similar spectral dependencies.

[14] To show more clearly the PMC signal and remove the residual offset calibration, we subtracted the coadded non-PMC spectrum (blue) from the coadded PMC spectrum (red) in each tangent altitude bin. Further, we degraded the difference spectrum to a spectral resolution of 25 cm−1 resulting in a noise of ∼0.035 nW/(cm2 sr cm−1). Figure 2 shows the coadded spectrally degraded net-PMC spectrum of the 81–82.5 km tangent altitude bin. The spectral shape of the PMC emission is more clearly seen now which resembles the weak spectral dependence of the ice particle absorption at these wave numbers [see, e.g., Hervig et al., 2001; Eremenko et al., 2005; Grossmann et al., 2006].

Figure 2.

MIPAS PMC spectrum (red), obtained by the difference of spectra with presence of PMCs and without PMCs (see Figure 1), degraded to 25 cm−1 and averaged for the tangent heights of 81 and 82.5 km. The related simulated spectrum, calculated for the same tangent heights for the mean of the kinetic temperature retrieved from MIPAS for each of the scans coadded in the PMC spectrum and an ice particle volume density layer similar to those in Figure 7 with a peak value of 1.7 × 10−14 cm3/cm3 peaking at 83 km (see section 5.2), is also shown (black line).

[15] For the tangent altitude bins defined above, limb emission spectra have been calculated with the KOPRA (Karlsruhe Optimized and Precise Radiative Transfer Algorithm) radiative transfer algorithm [Stiller et al., 2002] using the ice refractive indices reported by Toon et al. [1994]. The mean temperature profile measured by MIPAS for this period at latitudes from 65°N to 90°N was used, as well as the best fitting ice particle volume density profile (see section 5.2). The spectral shape of the simulated signal closely resembles the MIPAS coadded spectrally degraded PMC spectra (Figure 2). This provides further evidence (after that of Hervig et al. [2001], Eremenko et al. [2005], Grossmann et al. [2006] and O'Neil et al. [2008]) of the water ice nature of the PMC particles.

[16] As a further check for the evidence of PMC signatures in MIPAS spectra we have also compared with the observations of PMCs taken by SCIAMACHY (also on the ENVISAT satellite) during the same days. Figure 3a shows the geolocations of the 357 selected scans with the larger PMC signal in the 40 orbits of measurements taken by MIPAS during 19–21 July 2005. Figure 3a suggests that PMCs during these days are very frequent at all latitudes poleward of 60°N, although with a tendency to be more frequent at latitudes north of 80°N. A large fraction is also observed between 70°N and 80°N, and fewer at lower latitudes. In comparison with all measured scans, MIPAS observed very clear signal of PMCs in about 55% of the scans measured north of 65°N, and in about 60% of the scans measured northernmost of 70°N. The MIPAS observations of PMCs are consistent with the low temperatures regnant in those regions as retrieved from MIPAS spectra in the CO2 15 μm lines in band A (see solid black line in Figure 3a). This clearly shows the good spatial coincidence of PMCs occurrence and temperatures below approximately 150 K, for the whole polar region at latitudes north of around 60°N. The frost temperature, i.e., the temperature at which PMCs start forming, depends on the water vapor saturation pressure, that is, on atmospheric pressure and water vapor concentration. We do not yet have water vapor concentrations retrieved for this period and hence we took an approximated frost temperature of 150 K. Note that this temperature slightly decreases with altitude in the 80–90 km altitude region [Rapp and Thomas, 2006].

Figure 3.

Geolocations of the PMCs detected by (a) MIPAS and (b) SCIAMACHY on 19–21 July 2005 in the Northern Hemisphere. Red (MIPAS) and blue (SCIAMACHY) symbols denote detections of PMCs by the corresponding instruments. Green symbols denote measurements with absence of PMCs. The solid black line is the 150 K temperature contour at 84 km as measured by MIPAS (see Figure 6b).

[17] The geolocations for the NLCs detected by SCIAMACHY for the same period are shown in Figure 3b. The derivation of NLCs from SCIAMACHY limb-scatter measurements is reported in detail by von Savigny et al. [2007c]. As in the case of MIPAS, SCIAMACHY NLC locations are also very well spatially correlated with the temperatures below about 150 K. SCIAMACHY observations also show that NLCs are spread over all longitudes at latitudes northernmost of about 60°N. We should note, however, that the percentage of scans with observed NLCs versus those with absence of NLCs is larger in the case of SCIAMACHY than in MIPAS. While in MIPAS the percentages are 55% and 60% for latitudes north of 65°N and 70°N, respectively, in SCIAMACHY the percentages are 86% and 94%. At least two reasons could be responsible for this. First, we used a rather strict PMC detection limit for the MIPAS data shown in Figure 3a, while SCIAMACHY can discriminate thinner NLCs from noise. Second, while MIPAS and SCIAMACHY have similar along-track spatial resolutions, about 400 km, SCIAMACHY also scans horizontally, leading to an across-track spatial resolution of about 1000 km, much larger than that of 30 km for MIPAS, and hence sampling a larger air volume.

4. Radiance Profiles

[18] After having confirmed the water-ice nature of PMCs on the basis of coadded spectra, there is no particular advantage in further using spectrally resolved information for analysis of single measurements whose signal to noise ratio is so close to unity. Instead we use PMC radiances integrated over the 770–920 cm−1 wave number range. The radiance profiles thus obtained are plotted in Figures 4a and 4b binned in latitude bands 10° wide from 40°N to the North Pole. The engineering tangent heights recorded in the L1b were corrected with the altitudes retrieved in the temperature/instrumental line of sight elevation pointing retrieval discussed in section 5.1.

Figure 4.

MIPAS radiance profiles integrated over the 770–920 cm−1 interval averaged over several latitude bands (±5° of the shown latitudes). (a) Raw radiances without offset or instrumental effects corrections. (b) After correction for the mean slope for all altitudes and the offset at the uppermost altitudes corresponding to each latitude box. The estimated noise error in these profiles is about 4 nW/(cm2 sr).

[19] The raw profiles, prior to offset correction, are shown in Figure 4a. It is evident that there is an offset (also revealed in the spectra; see Figure 1) which seems to have two components: (1) a constant value, which depends very much on latitude, and (2) an altitude-dependent contribution with a slope which seems to be latitude-invariant. The constant component is attributed to the instrument calibration, which, although it seems large, is small in terms of spectral radiance (well below the specified noise error). Regarding the altitude-dependent component, its origin is not clear and it is still under investigation by ESA. At latitudes with no-PMC signal, for example, 40°N–50°N, we should expect, according to gas-phase radiative transfer calculations, a negligible cloud-free atmospheric radiance contribution. One possible reason for the altitude dependence of this offset might be stray light in the instrument which would also be expected to increase at lower tangent heights.

[20] For our purposes here, we corrected the integrated radiances by a linear altitude-dependent offset. Its slope was calculated from the average of all PMC-free radiance profiles shown in Figure 4a at altitudes above 75 km, and was assumed the same for all scans. The altitude-constant component was calculated from the mean of the four uppermost altitudes for each individual scan, once the slope component was removed. The offset-corrected profiles for the five latitudes bands are shown in Figure 4b, which can be considered as the PMC net radiance emission. These radiance also contain the atmospheric gas phase (CO2) contribution but this component is very small, with values of about 10 nW/(cm2 sr) at 70 km, 1 nW/(cm2 sr) at 75 km, and negligible at higher altitudes.

[21] The noise equivalent spectral radiance of 20 nW/(cm2 sr cm−1) translates into an error of ∼60 nW/(cm2 sr) for a single integrated radiance value. Since about 240 radiances are averaged together in each of the profiles in Figure 4, the noise error in those profiles is about 4 nW/(cm2 sr).

[22] The evidence of the PMC emission is again apparent at the altitude region where they are expected to occur, between approximately 78 and 86 km. The emission at the peak altitude is about 100–120 nW/(cm2 sr), which, after taking into account the integration interval of 770–920 cm−1, results in a mean spectral radiance of about 0.67–0.8 nW/(cm2 sr cm−1). Grossmann et al. [2006], who showed the only PMC infrared emission measurement so far reported, obtained values ranging from 0.6 to 1 nW/(cm2 sr cm−1), with a mean value close to 0.8 nW/(cm2 sr cm−1), over the 777–861 cm−1 spectral range (see their Figure 1 and note that radiance units given there should be W/(cm2 sr cm−1) × 10−9 and not W/(cm2 sr) × 10−9 as stated). Hence, the MIPAS measurements are in very good agreement with previous measurements by CRISTA.

5. Retrieval Method

[23] The emission of PMCs in the infrared depends on the ice particle volume density and, via the Planck function, on temperature. Since MIPAS spectra in band A covers, in addition to the PMC signal at 10–13 μm, the CO2 15 μm region, we can also derive the temperature. Thus we first retrieve the temperature from MIPAS spectra in the CO2 15 μm region, and use this in a second step for the retrieval of ice particle volume densities from the radiance integrated in the 770–920 cm−1 (∼1–13 μm) region.

5.1. Temperature Retrieval

[24] The inference of the kinetic temperature and/or pressure from spaceborne limb emission measurements in the infrared, and particularly the CO2 15 μm emission, is a very well known and widely used technique [see, e.g., von Clarmann et al., 2003, and references therein]. It is assumed that the abundance of CO2 is known, and from the measurement of its 15 μm emission at the local kinetic temperature, the latter is inferred. In our particular application, we have used the IMK/IAA (Institut für Meteorologie und Klimaforschung/Instituto de Astrofísica de Andalucía) scientific data processor for retrieving the instrumental line of sight elevation pointing, pressure, and temperature from the MIPAS spectra [von Clarmann et al., 2003] with the following modifications. First, the retrieval scheme was adapted for nonlocal thermodynamic equilibrium (non-LTE) emissions, and was coupled with the Generic RAdiative traNsfer AnD non-LTE population Algorithm (GRANADA) model [Funke et al., 2002] to compute the non-LTE populations of the CO2 15 μm emitting levels. This was necessary since we need to retrieve temperatures up to about 100 km, where the LTE conditions are not further fulfilled. Another update is the extension of the spectral microwindows to the Q branches of the strongest CO2 hot bands and a few lines of the strongest CO2 15 μm fundamental band. These stronger lines are needed for obtaining temperature information at the highest altitudes.

[25] The random error for a single scan, arising mainly from the measurement noise, is smaller than ±1.5 K below 70 km, ±5 K at 80 km, ±8 K at 95 km, and larger than ±10 K above 105 km.

[26] The most important systematic errors in the retrieved kinetic temperature arise from the errors in the calculated CO2 vibrational populations which, in turn, originate from uncertainties in the collisional processes considered in the non-LTE model. The largest errors come from uncertainties in the rates of CO2v2-quanta exchange, kvv, and quenching of the CO2v2 levels in collisions with N2 and O2, kair, and with atomic oxygen, kO. The error in the MSIS atomic oxygen abundance, used for the latter process, is an additional source of error. Current uncertainties in the rest of the collisional processes considered in the model do not produce significant errors in the v2 levels vibrational temperatures and, thus, not in the retrieved kinetic temperature.

[27] Since the nominal collisional rates of all processes considered in the Tk retrieval are the same as those indicated by García-Comas et al. [2008, their Table 1], we have assumed the uncertainties they report based on the values of the rates available in the literature, i.e., ±30% in the rate of collisions with N2 and O2, and ±50% in collisions with O. The only exception is the rate of v2-quanta exchange, which, based on comparisons of SABER and falling spheres kinetic temperatures [Kutepov et al., 2006], has been set to the laboratory value of Dang et al. [1983], assuming their ±20% measurement error. Further, we have also considered a ±50% error in the MSIS atomic oxygen number density, according to considerations from García-Comas et al. [2008].

[28] Another systematic source of Tk error is the uncertainty in the CO2 abundance, taken from the WACCM model and assumed to be 15% [see Remsberg et al., 2008].

[29] A summary of all estimated errors for polar summer conditions is given in Table 1. The estimated overall systematic error in the retrieved kinetic temperature is smaller (in absolute values) than ±1 K below 75 km, ±5 K at 85 km, ±4 K at 90 km, and larger than ±15 K above 95 km. Even if the uncertainties in the collisional rates are symmetric, the induced errors in Tk are not. This is due to nonlinearity. At 90 km, a 50% increase in kO results in a 2.3 K warmer Tk, whereas a 50% decrease cools Tk in 0.4 K. These errors are similar to those reported for SABER kinetic temperature for similar atmospheric conditions [García-Comas et al., 2008]. It is worth noting that the errors reported here are not globally representative, but are somewhat larger than for midlatitude or winter conditions due to the unfavorable cold polar summer mesopause.

Table 1. Estimated Errors in Retrieved Kinetic Temperaturea
z (km)kair (30%)kO (50%)[O] (50%)kvv (20%)[CO2] (15%)Tot SysNoise
  • a

    Temperatures given in kelvin; z denotes altitude; Tot Sys denotes total systematic errors.

70.00.10.00.00.10.10.21.5
75.00.40.10.10.60.71.03.4
80.01.60.30.31.61.32.64.9
85.02.02.52.52.42.35.25.1
90.00.71.51.50.73.24.07.3
95.02.013132.42.8198.3
100.02.621212.91.6306.6

[30] The MIPAS line of sight elevation pointing was retrieved simultaneously with temperature. The absolute error in the tangent altitudes is about 200 m [von Clarmann et al., 2003].

5.2. Ice Particle Volume Density Retrieval

[31] The thermal emission from a cloud of ice particles as measured by an instrument in the limb at tangent altitude zt, assuming optically thin conditions and that particles are small compared to the emitting wavelength (and then scattering effects can be neglected), is given by

equation image

where ν is wave number, B(T, ν) is the Planck function, the integrals are carried out along the instantaneous field of view (FOV) of MIPAS, ω, which is 3 km at the tangent point, and along the line of sight (LOS) of MIPAS, x, respectively; and T and β(x, ν) are the temperature and the absorption coefficient, of the ice particles. Assuming spherical particles, their absorption coefficient is given by [e.g., Harwit, 1998]

equation image

where N and V are the number density and volume of the ice particles, respectively, mv is the complex refractive index of ice, and equation image represents the imaginary part of quantity in parentheses. Since we do not have information from the MIPAS spectra on the particle size, we retrieve the volume density, v = N V. Inserting equation (2) into equation (1) and integrating over the wave number range Δν, we get

equation image

where A(x) is given by

equation image

[32] From the integrated net PMC radiances L(zt), water ice volume density profiles v(x) were retrieved by pseudoinversion of the discretized (Δz = 0.5 km) and linearized equation (3). Similar as for the temperature retrieval described in section 5.1, this was done by linearly constrained least squares fitting, where the Jacobians were calculated using the KOPRA radiative transfer algorithm [Stiller et al., 2002] using ice refractive indices reported by Toon et al. [1994], and assuming a monomodal lognormal size distribution of particles with mean radius of 30 nm and a distribution width of σ = 1.5. For simplicity, these parameters were assumed to be homogeneously distributed along the MIPAS line of sight for each altitude/layer of the atmosphere. While these assumptions certainly are not realistic in all aspects, they do not affect the results, since, as discussed above, the particles are much smaller than the wavelength and the signal thus depends only on temperature and volume density and not on the assumed particle size distribution. The volume density and the kinetic temperature were also assumed to be homogeneously distributed along the line of sight. The inversion was constrained by a Tikhonov-type scheme [Tikhonov, 1963] using a squared first-order differences matrix to obtain a reasonably smoothed vertical profile of volume densities. Otherwise, since the retrieval grid was chosen finer than the measurement grid, the inverse problem otherwise would have been underdetermined. In addition, for further stabilization of the retrieval, both the uppermost volume densities (100 and 120 km) were, in agreement with all we know about polar mesospheric clouds, constrained toward zero by a diagonal regularization matrix. The columns of the averaging kernel for a typical PMC radiance profile for several altitudes are shown in Figure 5. From Figure 5 we deduce that the altitude resolution in terms of the half-width of the columns of the averaging kernel matrix [Rodgers, 2000] varies from ∼2.5 km at 81–82 km to ∼3 km at 86 km altitude and to 3.5–4 km at and above 90 km.

Figure 5.

Columns of the averaging kernel corresponding to the retrieval of ice particle volume density profile from a typical radiance profile for altitudes of 80, 82, 85, 88, and 91 km.

[33] Models have shown [Rapp and Thomas, 2006] that the ice particles are not completely in thermal equilibrium with the air atmosphere, having a slightly larger temperature. Hence we applied a temperature correction of the emitting particles that varies linearly from 1 K at 80 km to 2 K at 90 km. We assumed that this correction is representative for a particle distribution with a mean radii between 30 and 50 nm and an accommodation coefficient of 0.5 [Rapp and Thomas, 2006].

[34] The retrieval scheme above was applied to all the individual radiance profiles available for 19–21 July 2005, after the offset correction described in section 4 had been applied. An atmospheric gas phase (CO2) contribution was also subtracted from the radiance profiles prior to the retrieval. This component is however very small, with values of about 10 nW/(cm2 sr) at 70 km, 1 nW/(cm2 sr) at 75 km, and negligible at higher altitudes.

[35] The precision of a single volume density profile at the altitudes of the PMC layer is estimated at about 60% at the PMC peak height, being the major error contributions the mapping of instrumental noise and propagation of random temperature retrieval errors (∼5 K; see Table 1). The systematic error is estimated at about 27%, being dominated by the kinetic temperature systematic error of about 4 K in this region (see Table 1).

6. Results and Discussion

[36] We have performed the retrieval of the volume density from all the offset-corrected radiance profiles for 19–21 July 2005 at latitudes northernmost of 40°N.

[37] The zonal mean distribution of the retrieved volume density, along with similar plots for the measured PMC integrated radiance (Figure 6a) (as a function of limb tangent height in this case) and retrieved kinetic temperatures (Figure 6b), is shown in Figure 7. We observe in Figure 7 that the ice particles are present in a layer extending from 80.5 to 87 km in altitude and from about 60°N to the North Pole. While they abruptly disappear below 80.5 km, their upper limit altitude seems to increase slightly with decreasing latitude. Berger and Lübken [2006] have shown that the natural variability of ice particles is large and they can appear much equatorward of 60°N. MIPAS measurements also suggest that weak PMCs appear at latitudes equatorward of 60°N at around 85 km for the observed period (see Figure 7).

Figure 6.

Zonal mean (a) PMC radiances and (b) temperatures for 19–21 July 2005 for the Northern Hemisphere as measured by MIPAS.

Figure 7.

Zonal mean volume density of ice particles for 19–21 July 2005 for the Northern Hemisphere as measured by MIPAS. The solid red line indicates the 150 K temperature contour, and the red dashed line is the mesopause as measured by MIPAS (from Figure 6b). The estimated noise error of the volume density is about 0.16 × 10−14 cm3/cm3.

[38] The altitude of the layer peak does not change significantly at latitudes 70°N–90°N, remaining at 82.5–83 km, which is in agreement with earlier findings from lidar observation and modeling [Lübken and Berger, 2007]. It is, however, located at slightly higher altitudes (83–85 km) at lower latitudes (60°N–70°N). The ice particle volume density at the layer peak increases monotonically from around 60°N until about 80°N and then remains nearly constant until the North Pole. Thus they show the maximum abundance around 82.5–83 km at latitudes north of 80°N.

[39] The location of the layer peak agrees very well with previous measurements. Baumgarten and Fiedler [2008] have found in their study of 8 years (1998–2005) of Rayleigh/Mie/Raman (RMR) lidar measurements of NLCs above ALOMAR (69°N, 16°E), a mean altitude of the peak of 82.6 km. This altitude is slightly higher, 83.3 km when considering all, weak and strong, PMC events (J. Fiedler, personal communication, 2009). This altitude interval agrees very well with the altitude of the peak volume density of our measurements. However, the agreement on its extension is not so good, since we observe a significant ice particle volume density up to about 87 km at 70°N, which is about 2 km higher than the uppermost altitude of the NLC layer measured by lidars [see, e.g., Baumgarten and Fiedler, 2008]. The reason for the discrepancy could be the use of different techniques. Baumgarten and Fiedler [2008] used the lidar technique, which is more sensitive to the larger particles, located at the lower part of the layer [Berger and von Zahn, 2002; von Savigny et al., 2005; Baumgarten and Fiedler, 2008]. Actually, NLCs are now understood as only being the optically visible part of a layer of icy particles covering the entire polar mesopause region and extending in altitude well above the visible NLC layer [Berger and von Zahn, 2002; Rapp and Thomas, 2006; Berger and von Zahn, 2007], while the whole layer modifies the ambient plasma of the D region and gives rise to intense radar echoes called PMSE [Rapp and Lübken, 2004]. Our MIPAS infrared emission measurements support that idea. The lidar measurements are less sensitive to the small particles (backscatter coefficient, β, is proportional to rα with α = 5–6), located at higher altitudes, than infrared measurements, whose emission is independent on particle size. Also, very recently, PMC measurements taken with the SOFIE instrument have been reported, which detects cloud signatures between about 78 km and all the way up to 89 or 90 km [Russell et al., 2009; Hervig et al., 2009a]. MIPAS observations at high altitude are also in agreement with those measurements and provides an independent confirmation of the extended ice particle layer.

[40] Regarding the absolute values retrieved from MIPAS, the maximum zonal mean volume density at 70°N is 1.3 × 10−14 cm3/cm3, which is significantly smaller than the value measured by Baumgarten and Fiedler [2008] of about 6 × 10−14 cm3/cm3 above ALOMAR (69°N). The PMC volume densities integrated over altitude agree better: from Figure 3 (left) of Baumgarten and Fiedler [2008] we estimated a column volume density of about 18 × 10−14 cm3/cm3 km, while from MIPAS data (Figure 7) we obtain ∼5.2 × 10−14 cm3/cm3 km, about a factor 3.5 smaller. This smaller concentration derived from MIPAS might be due to the fact that we assume a homogeneous volume density distribution in the tangent layer. Also sampling issues might explain the differences since we are comparing a zonal mean quantity with values measured above ALOMAR only during NLC events. In addition, natural variability might cause those differences. The model calculation of Berger and von Zahn [2007], however, predicts a peak value for the β backscatter coefficient of 6 × 10−10 m−1 sr−1 at 70°N (their Figure 12), while Baumgarten and Fiedler [2008] measured 15.6 × 10−10 m−1 sr−1 (Table 1), that is, nearly 3 times less in the model. This indicates that MIPAS measurements are in rather good agreement with the model estimations of Berger and von Zahn [2007] and Lübken and Berger [2007].

[41] Our results agree much better with those made from the space with the SOFIE instrument. Multiplying the mean volume density derived with MIPAS by the bulk density of ice at mesopause temperatures (0.93 g/cm3), we obtain a maximum zonal mean ice mass density of 12.1 ng/m3, similar to the mean value derived from SOFIE measurements (14.2 ng/m3) during the northern summer of 2007 at an average latitude of 68.9°N [Hervig et al., 2009b]. However, these mean values are somehow smaller, although within the reported range, than the mean mass density measured from ALOMAR during 1998, 36–102 ng/m3 [von Cossart et al., 1999], and with HALOE from 1992 to 2005, 24–89 ng/m3 [Hervig et al., 2003].

[42] Our estimation of column volume density can also easily be translated into column mass density (or ice water content, IWC) by multiplying by the ice density. This gives a mean IWC of 48 μg/m2, which is also somehow smaller than the value modeled by Rapp and Thomas [2006] (135 μg/m2) and the estimations of Collins et al. [2009] in Alaska (65°N) during one night in 2005 (190–240 μg/m2) (although the latter already report the presence of an unusually dense NLC during their measurements).

[43] As stated above, the small mean ice densities we obtain as compared to the ground-based measurements are most likely due to MIPAS averaging over events with no PMCs and the consideration of a homogeneous cloud at the tangent altitude. Also, it is important to note that the HALOE mass density threshold of PMC detection was set to 23 ng/m3, neglecting thus fainter clouds that are averaged in MIPAS zonal means.

[44] The small increase of the upper altitude of the layer with decreasing latitude at 80°N–90°N, suggesting an upward and equatorward motion of the ice particles, seems to confirm the model predictions of Berger and von Zahn [2002, 2007]. As these authors discuss, the sedimentation speeds of small (a few nanometers) particles near the North Pole at ∼85 km is less than 20 mm/s, while the upward component of the background meridional circulation winds is 70 mm/s and hence all small particles are expected to be transported upward and equatorward after nucleation. This is in agreement with MIPAS observations of the slight increase of the upper limit of the layer equatorward since this upper layer, as discussed above, is thought to be mainly composed of small particles.

[45] The sharp lower limit at 80.5–81 km seems to be well determined by the ∼150 K temperature limit for all latitudes (see Figure 6b). Figure 6b shows, however, significant ice particles below the 150 K contour at 81.5 km. This is very plausible caused by the vertical resolution of the measurements, about 2.5 km at this altitude, and hence smearing the lower limit of the layer in about 1–1.5 km.

[46] At about 60°N–70°N, the lower limit of the layer closely follows the 150 K temperature contour and the temperature field might be the responsible for the shift to higher altitudes of the layer peak. That is, the 150 K temperature contour rises with altitude equatorward at these latitudes, preventing the occurrence of PMCs at the lower altitudes (82–83 km), but still allowing their presence at 84 km and above. This will then cause a shift in the layer peak.

[47] For the whole region studied, the mesopause is located around 86 km. MIPAS measurements then suggest that significant amounts of icy particles are present at the mesopause altitude, although the particles at this altitude are most likely the smallest ones.

[48] Latitude/longitude maps for the retrieved volume density of ice particles at altitudes of 83 and 85 km are shown in Figures 8a and 8b. It is also clear, as in Figure 7, that the PMC's occurrence is confined to the regions where temperature is below about 150 K.

Figure 8.

Distribution of ice particle volume density at altitudes of (a) 83 km and (b) 85 km, retrieved from MIPAS radiance in the 770–920 cm−1 interval. The solid black line is the 150 K temperature contour. The fields are constructed from the observation by distance-weighted averaging of all measurements within ±10° latitude/±25° longitude around each grid point. The estimated noise error of the volume density is about 0.6 × 10−14 cm3/cm3.

[49] At about 83 km (Figure 8a) the larger volume densities are confined to latitudes northward of 70°N. It is also observed that the longitudinal distribution of volume density is not homogeneous, with smaller values located at longitudes 30°E–150°E. This situation changes with altitude and at about 85 km (Figure 8b), the PMCs are more tenuous near the North Pole and extend to lower latitudes. This is also in agreement with the idea mentioned: that PMC mainly nucleates close to the North Pole and then is transported upward and equatorward by the ascending branch of the meridional circulation.

7. Summary and Conclusions

[50] We analyze the spectra taken by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) instrument in its NLC mode (39–102 km) during July 2005. The spectral distribution of radiances measured by MIPAS in the 770–930 cm−1 region is very similar to that simulated for ice particle emission at low temperatures (below 150 K). This provides further evidence (after that of Hervig et al. [2001], Eremenko et al. [2005], Grossmann et al. [2006] and O'Neil et al. [2008]) of the water ice nature of the PMC particles. As a further check, the regions where MIPAS detected PMCs were compared to those derived from SCIAMACHY data and it was found that they coincide very well, although the percentage of scans with PMCs in the 60°N–90°N region is larger in SCIAMACHY than in MIPAS, probably because of the better sensitivity of SCIAMACHY (measuring in the UV) compared to MIPAS (observing in the infrared).

[51] The spectra were integrated over the 770–920 cm−1 wave number range, a spectral region where the PMCs have the largest emissivity and which is free of atmospheric gases components. The radiance profiles show also very clearly the PMC signal, with an increase at 78–85 km tangent heights, being larger for latitudes closer to the North Pole.

[52] By using the kinetic temperatures derived from MIPAS spectra in the 15 μm spectral region, the volume densities of the PMCs were retrieved for all measurements taken by MIPAS during 19–21 July 2005 at latitudes north of 40°N. The results show a layer of ice particles extending from 80.5 to 87 km in altitude and from about 60°N to the North Pole in latitude. It abruptly disappears below about 80 km, but it goes up to about 88 km at latitudes close to 70°N, although with very low abundances. The altitude of the layer peak remains at 82.5–83 km at 70°N–90°N, but it moves to slightly higher altitudes (83–85 km) at lower latitudes (60°N–70°N).

[53] Regarding its latitudinal distribution, the ice particle volume density increases monotonically from around 60°N until about 80°N and then remains nearly constant up to the North Pole. That is, they show the maximum abundance around 82.5–83 km at latitudes north of 80°N. The ice particles are well confined within the region where temperature is below approximately 150 K.

[54] The altitude of the peak of the layer agrees very well with previous measurements and model simulations [Baumgarten and Fiedler, 2008; Rapp and Thomas, 2006]. However, it extends to altitudes higher than the uppermost altitude of UV/visible measurements. This supports the current understanding of PMCs, whereby, they are thought to be the optically visible part of a layer of icy particles covering the entire polar mesopause region and extending in altitude well above the visible NLC layer [Berger and von Zahn, 2002; Rapp and Thomas, 2006; Berger and von Zahn, 2007]. MIPAS measurements, measuring in the infrared and therefore sensitive to smaller particles located at higher altitudes, confirm that understanding.

[55] The zonal mean volume densities are about a factor of 6 smaller when compared to single ground-based measurements (ALOMAR). This factor decreases to 3.5 when comparing column volume densities. This discrepancy might be due to the horizontal resolution involved in MIPAS limb geometry or the fact that we compare zonal means with observations taken during NLC events. However, our results are in reasonable agreement with model predictions [Berger and von Zahn, 2007]. Also, MIPAS mean ice mass densities are in excellent agreement with results from SOFIE [Hervig et al., 2009b], although both underestimate measurements at ALOMAR and from HALOE. The latter can be explained by the larger lower limit of mass density used to identify HALOE PMC events.

[56] MIPAS ice particle altitude/latitude distribution suggests that they are formed mainly at latitudes close to the pole, where they are more abundant, and then they are transported upward and to lower latitudes by the ascending branch of the meridional circulation, a picture which is consistent with the model predictions of Berger and von Zahn [2007].

[57] MIPAS has already measured during other PMC periods in the past years and is still operating. The extension of this study to other years will be done in the future to find out if the PMC features shown here are systematic or not.

Acknowledgments

[58] The IAA team was supported by the Spanish MICINN, under project AYA2008-03498/ESP, and EC FEDER funds. The authors acknowledge ESA for providing MIPAS L1b spectra. SCIAMACHY is jointly funded by Germany, Netherlands, and Belgium.

Advertisement