Polar stratospheric clouds observed by the ILAS-II in the Antarctic region: Dual compositions and variation of compositions during June to August of 2003

Authors


Abstract

[1] Compositions and effective radii of polar stratospheric clouds (PSCs) in the Antarctic region from 5 June to 28 August of 2003 are determined using transmittance data from the Improved Limb Atmospheric Spectrometer-II (ILAS-II). Dual compositions are derived for 83 Antarctic cases. The primary components are β-NAT (β form of nitric acid trihydrate), NAW (nitric acid water, or SBS, supercooled binary solution), and ICE. Other minor composition components are LTA (liquid ternary aerosol, or STS, supercooled ternary solution), α-NAT (α form of NAT), NAD (nitric acid dihydrate), and SAW (sulfuric acid water). Three single compositions are found; that is, β-NAT particles were observed from 11 June to 12 August, NAW particles were observed from 28 June to 24 July, and ICE particles were observed from 28 July to 25 August. During this period, dual compositions are also found, i.e., NAW + β-NAT, β-NAT + ICE, NAW + ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW. For this observation period, temperatures varied from 195 K to 180 K while measurements were made progressively in time as latitude varied from 65°S to 80°S. This mixture of compositions is assumed to be either two separate patches of PSCs or a mixture of two different types along the line of sight. The feasibility of the coexistence of dual compositions in a PSC and importance of determination of the PSC particle type for the study of heterogeneous chemistry for ozone are briefly discussed.

1. Introduction

[2] The importance of PSC composition and phase in relation to heterogeneous chemical processes and ozone depletion has been reported in various studies [Solomon, 1990; Molina, 1991; Ravishankara and Hanson, 1996]. Measurements from balloon- and aircraft-based instruments and lidar instruments, as well as theoretical models, have identified various composition types of PSCs in the polar regions, i.e., Type Ia PSCs (nonspherical, crystalline particles), Type Ib PSCs (spherical, liquid particles), Type II PSCs (ICE particles), nitric acid trihydrate (NAT, HNO3(H2O)3), and supercooled ternary solution (STS, H2SO4/HNO3/H2O) [Crutzen and Arnold, 1986; Toon et al., 1986, 1990; Poole and McCormick, 1988; Kinne et al., 1989; Fahey et al., 1989; Browell et al., 1990; Dye et al., 1992, 1996; Carslaw et al., 1994; Drdla et al., 1994; Tabazadeh et al., 1994a, 1994b; Toon and Tolbert, 1995; Gobbi et al., 1998; Schreiner et al., 1999; Larsen et al., 2000; Biele et al., 2001]. Satellite-based observations of PSCs are reported by Fromm et al. [1999] using the 1060 nm extinction profiles from the Polar Ozone and Aerosol Measurement II (POAM II) experiment. Strawa et al. [2002] used measurements of aerosol extinction coefficients at 0.603 and 1.018 nm from POAM III to discriminate Type 1a and 1b PSCs, while Poole et al. [2003] used a ratio method, applied to 1022 and 449 nm extinction data, to discriminate Type 1a and 1b PSCs.

[3] Lee et al. [2003] reported satellite-based remote sensing of PSCs in the Arctic region in 1997 using ILAS-I occultation spectra which included 44 spectral bins from 6.21 to 11.76 μm. The compositions of PSCs were derived by a nonlinear least squares fit (NLSF) method. Observed normalized aerosol extinction spectra were compared to those derived from Mie theory simulations. As discussed by Lee et al. [2003], and in the text below, PSCs have distinctly different spectra in the infrared region. Lee et al. [2003] found three types of PSCs from the Arctic measurements: SBS (supercooled binary solution, HNO3/H2O, or Nitric Acid droplets), α-NAT, and β-NAT (α and β form of NAT) particles.

[4] We report the results of dual composition and size distributions of 83 Antarctic PSCs from ILAS-II transmittance data obtained in 2003. In addition to the ILAS-I spectral range from 6.21 to 11.76 μm, ILAS-II has an additional wavelength range (i.e., channel 2, 3.0–5.7 μm, 3,333–1,754 cm−1), so that additional element spectral bins are available to our analyses. As discussed below, various combinations of compositions are found for the 83 cases we analyze; i.e., nitric acid water, nitric acid trihydrates, supercooled ternary solution, nitric acid dihydrate, sulfuric acid water, and ICE.

[5] In the following sections of this paper we discuss the analysis technique, input data, and computational results. An error analysis and the significance of the current results for heterogeneous chemical reactions are discussed at the end of the paper.

2. Observed Data

[6] ILAS-II was launched on 14 December 2002 on the ADEOS-II platform, which has an orbital altitude of 802.9 km, an inclination of 98.62°, and latitudinal coverage of 54–71°N and 64–88°S [Nakajima et al., 2006]. The ILAS-II solar occultation grating spectrometer measured the upper atmosphere between 10 and 60 km with a vertical field of view of 1 km at nearly 14 longitude points a day (at constant latitude) in both polar regions. ILAS-II had 4 wavelength channel spectral regions: channel 1 (6.21–11.76 μm (1,610–850 cm−1), 44 elements), channel 2 (3.0–5.7 μm (3,333–1,754 cm−1), 22 elements), channel 3 (12.78–12.85 μm (782–778 cm−1), 22 elements), and channel 4 (753–784 nm (13,280–12,755 cm−1), 1024 elements). Profiles of O3, H2O,N2O, CH4, HNO3, NO2, ClONO2, aerosols, temperature, and pressure are archived by the ILAS team. The first two channels have 66 detector elements altogether with an equal spectral interval, 0.129 μm. By avoiding strong gaseous absorption bands, 22–26 spectral elements can be used to derive aerosol properties.

[7] As described by Lee et al. [2003], a transmittance-ratio technique is used to compute the relative extinction coefficients. The transmittance-ratio technique utilizes two adjacent transmittance profiles. The first transmittance profile contains a PSC signature while the second profile contains no PSC signature (or less than the first). The transmittance τpsc at an altitude z of an aerosol layer can be written as

equation image
equation image

where τpscg and τrefg are the transmittances by the gaseous absorptions along the limb path of PSCs and reference cases, respectively, βpsc is the aerosol extinction coefficient (measured for a limb path in the unit of km−1), Lpsc is the horizontal path length of the PSC (typically 200 km for a 1 km depth layer), and λ is the wavelength.

[8] Even though the reference profile can contain other types of aerosol, such as sulfate aerosol, PSC extinction values are usually much greater than that of the background sulfate extinction. When the aerosol extinction at the reference location βref is nearly zero (or much smaller than βpsc) and τpscg is similar to τrefg, then the relative optical depth of the PSC can be approximately determined by the ratio of τpscg and τrefg.

equation image

The ratio of the optical depths at two wavelengths λi and λj is defined as

equation image

αmea is hereinafter referred to as the relative extinction coefficient.

[9] Figures 1a–1c present the case of 1 July 2003; Figure 1a shows transmittance profiles of the reference (no. 401, dotted lines) and the PSCs (no. 391, solid lines) for 7 spectral elements (0, 6, 17, 29, 38, 56, each colored differently); Figure 1b shows the relative optical depths profiles (equation (3)); and Figure 1c shows the normalized average profile of the relative optical depths (after each is normalized by the peak value near 20 km). Profiles in Figure 1b have nearly the same peak layer and furthermore, the normalized average profile has a standard deviation of less than 0.015 relative to the peak value of unity. This feature of the profiles is due to the PSC absorption which influences all of the spectral channels.

Figure 1.

PSC case of 1 July 2003: (a) two vertical transmittance profiles of occultation numbers 391 (PSC profile) (solid lines) and 401 (reference profile) (dotted lines) for ILAS-II elements 0, 6, 17, 29, 38, 56, 65; (b) the optical depths for each element obtained from the transmittance-ratio of these two profiles for 7 elements; and (c) the optical depths normalized by its own peak value at 19 km of each element. The average profile of the normalized profiles of 24 elements and their standard deviations are shown as horizontal bars. The gray bars in the three frames indicate aerosol peak layer.

[10] The relative extinction coefficients (see equation (4)) at 20.5 km are shown in Figure 2 as a function of wavelength (and also element number on the top scale), using the element number 31 (7.113 μm) as the reference. The vertical bars on each of elements indicate estimated 1-sigma random errors primarily caused by tangent height uncertainties of ±340 m [Nakajima et al., 2002].

Figure 2.

For the case shown in Figure 1 the relative extinction coefficients (normalized by the value of the element number 31) shown as a function of wavelength (or element number) at 20.5 km.

3. Simulations of Relative Extinction Coefficients Using the Mie Theory

[11] The relative extinction coefficients are simulated using Mie theory and the properties of the aerosols (composition and effective radius) are derived by an NLSF technique. Theoretical extinction coefficients are computed for dual compositions for each occultation measurement.

[12] Theoretical extinction coefficients (km−1) for dual component PSC aerosols (with compositions comp1 and comp2) are represented by

equation image
equation image

where r is the radius of the PSC particles (μm), Qe(x, mλ) is the Mie efficiency factor, which depends upon the size parameter x = 2r/λ and the refractive index mλ = mre + mim, and dn(r)/dr is the size distribution (particles cm−3 μm−1). It is assumed that the PSC size distribution can be calculated by a lognormal distribution given by the expression

equation image

The k index is 1 or 2 for composition components 1 or 2. N is the total number density (particles cm−3), R is the median radius (μm), and σ is the geometric mean standard deviation (dimensionless) of r. Following Lee et al. [2003], the simulated relative extinction coefficient, αcal, is defined using the formula of equation (4),

equation image

Analysis results of R and σ are presented in terms of the conventional term, effective radius, Reff, defined by

equation image

4. NLSF Analysis for the 83 Cases of Antarctic PSCs

[13] By using the NLSF method, composition types and values of R, and σ are derived for components 1 and 2 by minimizing the difference function, f, between the measured and simulated values of the relative extinction coefficient,

equation image

4.1. Calculation of αcal Using Laboratory Optical Data

[14] Several refractive index data sets of different compositions and phases are used to compute αcal in our analyses and are summarized in Table 1. A discussion of the uncertainties in the optical data and their influences in the analyses of PSC composition is presented by Lee et al. [2003].

Table 1. Names of Various Aerosols Considered in the Analyses of ILAS-II PSCsa
Symbol (in Literature)Symbol (in This Text)PhaseDefinitionData Source
  • a

    The first column is the aerosol type commonly used in the literature; the second column cites the symbols defined in this text for purposes of discussion. The third column is the aerosol phase. The fourth and fifth columns are aerosol definitions (in terms of N% and S% composition weights) and the literature references.

α-NATNatrsolidα-type of NAT (nitric acid trihydrate) (HNO3/3H2O)Richwine et al. [1995]
β-NATNatbsolidβ-type of NAT (HNO3/3H2O)Toon et al. [1994]
NADNadnsolidnitric acid dihydrate (HNO3/2H2O)Niedziela et al. [1998a]
ICE Waterice8solidtype II (crystallized water vapor)Clapp et al. [1995]
LTA (STS)tnliquidliquid ternary aerosol (supercooled ternary solution) (H2SO4/HNO3/H2O)Norman et al. [2002]
LTA (STS)tn45liquidH2SO4: 10 wt%, HNO3: 45 wt%Norman et al. [2002]
LTA (STS)tn35liquidH2SO4: 22 wt%, HNO3: 35 wt%Norman et al. [2002]
LTA (STS)tn31liquidH2SO4: 28 wt%, HNO3: 31wt%Norman et al. [2002]
LTA (STS)tn26liquidH2SO4: 29 wt%, HNO3: 26 wt%Norman et al. [2002]
LTA (STS)tn24liquidH2SO4: 33 wt%, HNO3: 24 wt%Norman et al. [2002]
LTA (STS)tn15liquidH2SO4: 40 wt%, HNO3: 15 wt%Norman et al. [2002]
NAW (SBS)Nwliquidnitric acid water (supercooled binary solution) (HNO3/H2O)Norman et al. [1999]
NAW (SBS)nw35liquidHNO3: 35 wt%Norman et al. [1999]
NAW (SBS)nw40liquidHNO3: 40 wt%Norman et al. [1999]
NAW (SBS)nw45liquidHNO3: 45 wt%Norman et al. [1999]
NAW (SBS)nw50liquidHNO3: 50 wt%Norman et al. [1999]
NAW (SBS)nw54liquidHNO3: 54 wt%Norman et al. [1999]
NAW (SBS)nw63liquidHNO3: 63 wt%Norman et al. [1999]
SAWSwliquidsulfuric acid water (binary solution) (H2SO4/H2O)Niedziela et al. [1998b, 1999]
SAWsw32liquidH2SO4: 32 wt%Niedziela et al. [1998b, 1999]
SAWsw43liquidH2SO4: 43 wt%Niedziela et al. [1998b, 1999]
SAWsw55liquidH2SO4: 55 wt%Niedziela et al. [1998b, 1999]
SAWsw66liquidH2SO4: 66 wt%Niedziela et al. [1998b, 1999]

[15] Figure 3 displays calculated relative extinction coefficients for aerosols of 8 individual compositions (i.e., α-NAT, β-NAT, NAD at 190 K, ICE at 180 K, LTA N45% S10% (=tn45), LTA N15% S40% (=tn15), NAW (=nw54), and SAW (=sw55)), to demonstrate that their wavelength-dependent shapes are distinct as a function of wavelength. These are representative cases based upon available refractive index data. The open circles on the element axis indicate the elements where relative extinctions are determined from ILAS-II data. The extinction coefficients are calculated for a set of selected conditions (i.e., the geometric mean standard deviation σ = 1.6 and five different values of the mean radius Ri). All extinctions are normalized to that of element 31 (wavelength = 7.113 μm).

Figure 3.

Relative extinction coefficients of aerosol particles, computed using the Mie theory for compositions (i.e., (a) α-NAT, (b) β-NAT, (c) NAD at 190 K, (d) ICE at 180 K, (e) LTA N45 (=tn45), (f) LTA N15 (=tn15), (g) NAW (=nw54), and (h) SAW (=sw55)). The laboratory measurements of refractive indices are summarized in Table 1. For each case, relative extinction coefficients are computed for σ = 1.6 and 5 different values of the mean radius R.

[16] The wavelength-dependent characteristics of the extinction features (especially the positions of the maxima and minima) allow us to identify the composition type of the PSCs observed by ILAS-II. Generally, ILAS-II spectral bins are available for elements in the spectral regions that are important for defining the shapes of aerosol extinctions (especially the maxima and minima points). Solutions for PSC properties are sought using a limited number of refractive index data sets with an emphasis on the fitting of the general shapes of the relative extinctions, and for some cases, fitting of the extinction magnitudes. It is important to note a limitation of our simulations, in that the available refractive index data may not be correct for the observed conditions of temperature and wt% of H2SO4 or HNO3 (hereinafter referred to as N% or S%). Results from the current NLSF analyses, therefore, can contain errors that are caused by uncertainties in the refractive indices, as well as by lack of optical data for correct values of temperature and N% or S%.

4.2. Calculation of αmea From the ILAS-II Data

[17] The relative extinction coefficients at peak levels of total 83 cases are computed as described earlier. Spectral profiles are displayed in 8 groups (a–h) in Figure 4 for the purpose of later discussion, each having similar spectral shapes; however, it will be shown later that although profiles within a group have similar spectral shapes as well as locations of local maxima and minima, every profile within a group does not necessarily have the same dual compositions.

Figure 4.

Measured relative extinction coefficients for PSCs as a function of wavelength (i.e., element number) for 83 cases. Each profile is plotted relative to the value of the element 31, which corresponds to a wavelength of 7.113 μm. (a–h) Categorization of the extinction coefficients into eight groups according to wavelength-dependent shapes.

5. Computational Results

[18] Each profile of the total 83 cases of ILAS-II PSCs is simulated of the NLSF method with a dual composition formula, and the final selection is made after many trials of various refractive indices of Table 1, and also using various combinations of dual compositions. The NLSF technique enables us to identify the dual compositions from the observed data primarily based on the spectral shapes. The local maximum and minimum values, in particular, define the values of Reff and σ.

5.1. Sample Spectral Simulations

[19] The results for nine representative profiles are shown in Figure 5: i.e., three cases for which a single composition has the dominant spectral signal and 6 cases for which dual compositions have definite spectral signals as well as distinct spectral shapes. Residuals between the measured and simulated coefficients (yellow circles) are attributed to instrument noise. (1) Cases, a, b, and c are for single compositions, β-NAT, NAW, and ICE, respectively; spectral signal from the primary composition is dominant and signal from the secondary composition is marginally identifiable (defined hereinafter as the single composition case). (2) Cases, d, e, f, g, h, and i are for combinations of dual compositions, NAW + β-NAT, β-NAT + ICE, NAW+ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW, respectively; signals from the primary and the secondary compositions have distinct shapes and definite signal level, and therefore each composition type can be identified (defined hereinafter as dual composition case). Two cases of g and h have weak signals and spectral shapes for the secondary components, but each belongs definitely to laboratory data simulations of NAD and LTA that are shown in Figure 3. Residuals between the measured and simulated coefficients are shown as yellow circles and their values in the average are around less than 3%. We used a criterion that signal from a secondary composition should be more than 5 times larger than the residual for a dual composition case.

Figure 5.

Simulations of relative extinction coefficients for nine cases of single compositions and dual compositions. Note that shown in each frame are date, peak layer, the primary component composition (blue) and the secondary component composition (green) with their respective values of R, σ (=S), and Reff. The ILAS relative extinction coefficients are shown as green crosses, and the simulated values are shown as a solid red line. The residuals between the two are shown as yellow circles (at the level of 0. of the relative extinction coefficients). Shown also are contributions to the total extinctions from each of two compositions: blue solid line for the primary composition and green solid line for the second composition.

5.2. Identification of Composition Types

[20] Results of the NLSF analyses of 83 cases are summarized in Table 2 for peak layers. Composition types can be grouped (A to I) for β-NAT, NAW, ICE, NAW + β-NAT, β-NAT + ICE, NAW + ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW, respectively. Listed in the Table 2 are dates, profile numbers (PSC and reference), temperatures, heights, composition types and Reff for the 1st and 2nd compositions, and number density ratio.

Table 2. Results of Composition Types and Values of Effective Radii and Number Densities of Dual Components Derived From Relative Extinction Coefficients Computed From ILAS-II Dataa
GroupDate 2003PSC/REFT, Kz, kmComp1 (/Comp2)Reff1 (/Reff2)N1/N2
  • a

    The first column is PSC category. The second through fifth columns denote observation dates, event numbers for PSCs and reference profiles, temperature, and the peak layer. The sixth through eighth columns are computational results for components 1 and 2, the composition type, effective radius, and number density ratio.

A11 June371/341194.220.5β-NAT0.58
A12 June521/531188.221.5β-NAT0.52
A13 June081/051189.420.5β-NAT1.45
A14 June211/231189.321.0β-NAT1.21
A16 June511/531190.021.0β-NAT2.56
A16 June551/561194.020.5β-NAT1.41
A18 June251/241193.118.5β-NAT3.67
A19 June431/421193.122.5β-NAT2.13
A20 June541/011189.222.5β-NAT2.06
A21 June111/151189.621.5β-NAT2.48
A23 June301/351195.319.0β-NAT2.05
A23 June371/351192.019.5β-NAT8.82
A26 June231/241190.620.5β-NAT0.55
A6 July481/521186.121.0β-NAT1.25
A9 July361/311184.622.0β-NAT0.95
A10 July491/541188.818.0β-NAT1.25
A13 July311/301191.518.0β-NAT1.34
A13 July371/401187.721.0β-NAT0.52
A23 July031/521184.820.0β-NAT0.99
A25 July411/391187.024.0β-NAT0.52
A30 July551/011184.922.5β-NAT0.57
A30 July561/011187.721.0β-NAT1.02
A31 July041/021185.623.5β-NAT0.77
A31 July151/471190.422.0β-NAT2.32
A3 Aug.531/021184.822.0β-NAT1.20
A6 Aug.401/431185.422.0β-NAT1.18
A6 Aug.411/431189.822.0β-NAT1.79
A12 Aug.051/041180.422.5β-NAT0.62
B29 June121/131190.019.5NAW N45%0.42
B1 July391/401187.621.0NAW N50%0.39
B8 July201/241188.220.0NAW N40%0.31
B12 July211/221187.421.0NAW N45%0.37
B14 July481/441184.722.5NAW N45%0.36
B16 July281/171189.619.5NAW N45%0.43
C27 July151/141188.322.5ICE0.40
C28 July291/301189.221.5ICE1.04
C19 Aug.561/571183.419.5ICE1.54
C23 Aug.441/451185.218.5ICE0.94
C24 Aug.011/561192.425.0ICE0.98
C25 Aug.251/241182.917.5ICE1.42
D5 June031/021191.718.5β-NAT/NAW N50%4.47/0.452.41e+03
D25 June091/101187.521.5β-NAT/NAW N45%4.32/0.981.45e+01
D27 June351/371189.221.0β-NAT/NAW N50%3.74/0.832.64e+01
D28 June541/561187.820.0β-NAT/NAW N45%3.89/0.822.81e+01
D29 June091/071190.219.5β-NAT/NAW N45%3.89/1.087.39e+00
D10 July511/541185.422.5NAW N45%/β-NAT0.30/2.494.08e-04
D11 July081/111186.422.0NAW N40%/β-NAT0.13/1.517.87e-05
D19 July101/041188.819.5β-NAT/NAW N54%2.62/0.071.68e+04
D20 July231/261188.419.0β-NAT/NAW N45%3.35/0.465.51e+02
D23 July141/101189.918.5β-NAT/NAW N50%2.92/0.449.56e+00
E27 July121/141184.220.0ICE/β-NAT4.28/1.533.38e+01
E4 Aug.011/021187.523.5ICE/β-NAT6.45/1.571.20e+02
E8 Aug.051/531184.318.0ICE/β-NAT3.13/0.501.39e+01
E12 Aug.141/041183.319.0ICE/β-NAT2.08/0.256.25e+01
E13 Aug.191/041181.019.5ICE/β-NAT4.50/1.371.19e+01
E15 Aug.561/231182.122.5ICE/β-NAT1.95/0.572.46e+00
F7 July071/031188.720.5NAW N40%/ICE0.40/6.368.03e-05
F9 July371/411189.220.0ICE/NAW N45%2.31/0.484.83e+03
F10 July531/551187.321.5NAW N40%/ICE0.50/6.821.89e-04
F11 July101/121189.220.0NAW N40%/ICE0.39/7.385.96e-05
F14 July521/501185.421.5NAW N40%/ICE0.23/5.266.96e-06
F15 July041/051189.519.5NAW N40%/ICE0.25/7.468.16e-06
F19 July091/131184.121.5NAW N45%/ICE0.51/8.842.71e-04
F24 July161/521187.918.5ICE/NAW N50%9.42/0.534.67e+03
F7 Aug.511/531180.121.0ICE/NAW N35%5.06/1.329.88e+00
F17 Aug.271/291180.421.5ICE/NAW N35%3.00/0.878.14e+00
F28 Aug.101/151181.619.5ICE/NAW N45%8.07/0.424.19e+03
G19 June391/421191.719.5β-NAT/NAD2.08/2.901.60e-02
G19 July111/131187.222.0β-NAT/NAD2.30/3.144.61e-02
H22 June221/271190.520.0β-NAT/LTA N15% H40%1.16/1.489.58e-02
H22 June241/271190.420.5β-NAT/LTA N45% H10%3.30/2.624.45e-01
H22 June251/271192.819.5β-NAT/LTA N24% H33%4.17/4.043.59e+00
H23 June401/421187.922.5β-NAT/LTA N45% H10%2.47/2.531.40e-01
H24 June501/481196.718.0β-NAT/LTA N45% H10%4.86/3.853.71e+01
H26 June201/181191.920.5β-NAT/LTA N35% H22%3.74/2.721.95e+00
H29 June101/131186.220.0β-NAT/LTA N35% H22%1.71/2.471.01e-01
H30 June231/211185.920.0β-NAT/LTA N35% H22%1.25/2.069.30e-02
H1 July361/401188.319.5β-NAT/LTA N35% H22%1.31/1.633.03e-01
I10 July471/451185.922.0β-NAT/SAW H55%0.74/0.252.74e+01
I11 July041/011187.319.0β-NAT/SAW H55%0.93/0.621.20e+00
I27 July101/141184.522.5β-NAT/SAW H55%0.71/0.331.32e+01
I28 July231/301185.820.5β-NAT/SAW H55%1.02/0.898.09e-01
I29 July331/311186.223.0β-NAT/SAW H55%0.49/0.811.71e-02

5.3. Time Variation of Compositions

[21] Figure 6a shows the PSC occurrences as a function of time for compositions, β-NAT, NAW, ICE, NAW + β-NAT, β-NAT + ICE, NAW+ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW. It appears that the primary compositions (i.e., β-NAT, NAW, and ICE) existed separately (β-NAT for the most of June, July and August, NAW for the 1st part of June, followed by ICE for the month of August). Dual composition cases are found as a combination of either the two primary compositions or a primary and any of the 2nd compositions, NAD, LTA, and SAW. Figure 6b shows temperature variation (from UKMO) with time for each case of Figure 6a. Although there are a few extreme cases (e.g., β-NAT at T = 196.7 K on 6/24 (case H), β-NAT at T = 180.4 K on 8/12 (case A), and ICE at T = 192.4 K on 8/24 (case C)), PSCs are generally observed within the following temperature ranges that are indicated as vertical bars for each combination group type (with corresponding colors): (1) β-NAT, 195–185 K; (2) NAW, 190–185 K; (3) ICE, 189–183 K; (4) β-NAT + NAW, 192–186 K; (5) β-NAT + ICE, 187–181 K; (6) ICE+NAW, 189–180 K; (7) β-NAT + NAD, 192–187 K; (8) β-NAT + LTA, 196–186 K; and (9) β-NAT + SAW, 186–185 K.

Figure 6.

Summary of results as a function of observation date: (a) PSC compositions shown as a function of date from the top to the bottom rows for β-NAT, NAW, ICE, NAW + β-NAT, β-NAT + ICE, NAW+ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW along with temperature range; (b) local temperature variations shown as a function of time. The temperature range for each combination type (A–I) is indicated as a colored vertical bar. Points are designated as the letters A–I and by colors corresponding to PSC types shown in Figure 6a. Latitude progression is from 65 to 80°S with dates from 5 June to 28 August of observation points.

[22] For these cases in the figure, there are observational and theoretical evidence for the coexistence of dual composition types (see subsequent discussion). It is of interest to note that the ICE cases occur at temperatures generally colder than those of the NAT and NAW composition types. This time variation of the different compositions is further studied and discussed in the following section by using a method trajectory of air mass.

5.4. Other Note

[23] The NLSF signal-ratio technique is accurate for identifying the PSC layer altitude and the composition types, but it cannot be used to derive the individual N1 and N2 values nor the vertical extinction profile for each composition. Furthermore, for a given limb measurement, it is difficult to tell whether the derived compositions exist as separate patches or coexist in the same air mass. Finally, the 1 km vertical resolution and the effective horizontal path length of 300 km of an ILAS-II tangent ray needs to be kept in mind; in other words, the ILAS-II measurements cannot resolve small spatial scale variation of PSCs.

6. Discussion

6.1. Trajectory Study

[24] The variation of the different compositions found in our study may have occurred within the same air mass as a function of temperature and time as a consequence of the poleward drift in the observational latitude from 60°S to 80°S, causing local temperature variation. To investigate this possibility, air mass trajectories on isentropic surfaces are calculated to investigate whether air masses experienced unusual temperature histories. Steele et al. [2002] suggested from analyses of POAM aerosol data that factors other than temperature and temperature history are partially responsible for the formation of NAT PSCs although the occurrence probability at the NAT condensation temperature could be higher in air that has a previous cold temperature history.

[25] We examined the temperature histories of NAT and NAW PSCs at peak layers to see if they are exposed to temperatures below Tice. Although PSCs could exist anywhere along the limb path of transmittance data, we assume the PSCs were located at the tangent point. Daily UKMO analysis data were used to interpolate wind and temperature at PSCs locations and occultation measurement times. Randomly selected 13 cases of group A (NAT PSCs of Figure 6) are studied. Among these 13 cases, 8 cases have been exposed below Tice (189 K) within 3 days. Temperature histories for a few case of group D (NAT + NAW PSCs of Figure 6) are also studied. Temperatures 2 days previous to the observation date remain between Tsts (185 K) and Tnat (196 K). In summary, although there are uncertainties of the UKMO temperature data, cold temperature histories are associated with the PSCs that we report in Figure 6. This is consistent with the suggestion of Steele et al. [2002] based on their statistical analyses.

6.2. Importance of PSC Type for the Study of Heterogeneous Chemistry

[26] In the southern and northern polar stratosphere, three dimensional chemistry-transport calculations need to be accurately specified. It is important to be able to determine the PSC particle type because the rate of a heterogeneous chemical reaction is proportional to the product of the reaction probability and the surface area density (A, μm2 cm−3). As discussed elsewhere [e.g., Massie et al., 1998; Steele et al., 1999; Massie et al., 2000], surface area density can be estimated from PSC extinction (β, km−1) measurements. The reaction probability is dependent upon PSC composition type, and therefore studies such as ours, which develop and apply techniques by which the evolution of PSC composition type is derived from multiwavelength satellite observations, are developmentally important. In the future it is expected that satellite data will be used in model assimilations that specify the locations of PSCs, their surface area densities, composition types, and reaction probabilities.

7. Error Budget

[27] The wavelength-dependent characteristics of the extinction features allow us to identify the dual compositions of the PSCs along the line of sight. In the retrieval of PSC parameters (i.e., Reff and σ), there are a several error sources.

[28] 1. Some uncertainties are associated with the measured relative extinctions, the theoretical simulations, and uncertainties in the transmittance values. We assume that the gas profiles at the two adjacent occultation events are the same. If this assumption is violated, then random errors are introduced for aerosol extinction coefficients for some of the spectral elements which contain major gaseous absorption. Other potential problems include the possibility that the reference profile is contaminated by other types of aerosol particles, the presence of uncertainties in the altitude registration, and the assumption of a single mode particle size distribution. Error magnitudes arising from these sources are discussed by Lee et al. [2003] and are applicable to our analyses. We estimate that these uncertainties present a random error of ∼25%.

[29] 2. Another error source is the uncertainty in the values of the refractive indices. We use refractive index data that are reported for a certain range of temperature and for a limited range of wt% of nitric acid and/or sulfuric acid droplets and particles. These conditions may or may not be close to the actual measurement conditions of the ILAS-II observations. To estimate these uncertainties, we consider the case of 17 August (ICE/NAW N35%). In order to estimate the error caused in R and σ of the ICE primary component and the NAW secondary component by refractive indices of different temperature and N%, a Monte Carlo method was used to estimate errors by using the random uncertainty of 1.5% (1-sigma) in the real part of the refractive indices and 10% (1-sigma) uncertainty in the imaginary part of the refractive indices. From the NLSF retrievals of the 30 different cases of random errors, mean standard deviations are found ∼36% for R and ∼15% for σ for both components.

[30] 3. For the case of the derivation of dual compositions, the other significant error can be caused by the inability of distinguishing how two patches of PSCs are mixed along the limb path. The main effect will appear indirectly through an uncertainty of temperature where each component exists.

[31] It is estimated that the random error for the effective radius is in the range of ∼40% (RSS) in the average for the cases shown in Table 2. However, as indicated earlier, the composition types can be identified because of the distinct wavelength-dependent shapes of the relative extinction coefficients of the individual composition types.

8. Comparison With Other Observations

[32] There is observational and theoretical evidence for the coexistence of dual composition types in a PSC. Koop et al. [1997] discuss the thermodynamic stability and the particle formation pathways of PSCs. They state that liquid particles can coexist with solid particles as long as available H2SO4 remains in the aqueous phase. Coexistence of NAT and ternary droplets is one allowed possibility. Other possibilities [see Koop et al., 1997, Figure 1] include (with SAT denoting Sulfuric Acid Tetrahydrate, H2SO4/4H2O) (1) for T < Tice (189 K), ICE+STS, NAD+STS, NAT + STS, NAT + ICE+STS, NAT + SAT, NAT + ICE+STS, NAT + ICE+SAT; (2) for Tice < T < Tnat (196 K), SAT, SAT + NAT, NAT + STS, NAD+STS, NAT + SAT, NAT + STS, NAT + SAT, NAT + STS; and (3) for Tnat < T < Tsat (213 K), SAT, SAT + NAT. All possible combinations of liquid and solid particles, however, are not likely to exist, since kinetic considerations are also important (e.g., although a transformation pathway is possible, the timescale for the conversion is too long to be stratospherically important). Koop et al. [1997] indicate three conversion pathways (e.g., from aqueous H2SO4/H2O to solid H2SO4·H2O) that are kinetically unlikely.

[33] Biele et al. [2001] discuss Ny Alesund, Spitzbergen lidar observations of mixed phase liquid/solid Arctic PSCs. The mixed phase clouds are observed on the smallest spatial and temporal scales resolvable by the lidar, implying that their properties are not a result of spatial averaging of different cloud types. Biele et al. [2001] also summarize the observations of PSCs by Gobbi et al. [1998], Stein et al. [1999], Rosen et al. [1997], Dessler et al. [1999], Shibata et al. [2003], and Toon et al. [2000], that are supportive of the contention that solid and liquid PSC particles can coexist.

9. Summary and Discussion

[34] Using a transmittance-ratio technique and the NLSF method, dual compositions are derived for each of 83 Antarctic PSC cases observed by ILAS-II during the period of June to August, 2003. The wavelength-dependent characteristics of the extinction features allow us to identify the dual compositions of the PSCs. The primary single composition components observed during this time period are β-NAT, NAW, and ICE. Other secondary compositions are LTA, NAD, and SAW. Dual compositions found are NAW + β-NAT, β-NAT + ICE, NAW+ICE, β-NAT + NAD, β-NAT + LTA, and β-NAT + SAW for temperatures varying from 195 to 180 K during June to August. It is possible that the composition changes could be simply a latitudinal-dependent local phenomenon associated with the temperature changes occurring for the same air mass for the time period, since measurements for this period were made progressively with latitude from 65 to 80°S. Since heterogeneous chemistry reaction rates are dependent upon PSC composition type, our determinations of PSC type (see Figure 6) illustrates the potential of satellite remote sensing measurements to provide useful information for chemistry-transport model studies in which PSC composition type is assimilated into time-dependent calculations.

Acknowledgments

[35] The authors are grateful to ILAS-II Team who provided transmittance data for this study. The analysis work in Korea has been supported by KOSEF through a grant (R01-2003-000-10131-0) and Climate Environment System Research Center and by the BK21 program. Jae Park was supported by the Brainpool Program of the Korean Ministry of Science and Technology. NCAR is sponsored by the National Science Foundation. ILAS-II data were processed at the ILAS-II Data Handling Facility of the National Institute for Environmental Studies, Japan. The ILAS-II project is funded by the Ministry of the Environment, Japan.

Ancillary