An eye-safe micropulse lidar (MPL; 0.523 μm) has operated at the Scott-Amundsen South Pole Station, Antarctica, since December 1999 to collect continuous long-term measurements of polar clouds. A 5-year data subset is presented here to describe macrophysical, optical, and thermodynamic properties of polar stratospheric clouds (PSC) in austral winters 2000 and 2003–2006. PSC cloud occurrence is examined relative to seasonal temperature and theoretical chemical structure. A linear relationship is established with high correlation between total integrated PSC scattering and ozone loss for 2000 and 2003–2005 when springtime overturning of the air mass occurred nominally. In 2006, ozone-depleted air persisted over the South Pole through the end of December. In this case, overturning of the air mass was limited temporally by vortex-related mechanisms, and any correlation with PSC occurrence was eliminated. PSC formed near and above 18.0 km above mean sea level (MSL) in late May and early June likely influence clouds formed at lower heights later in the season from sedimentation, evaporation/sublimation, and repartitioning of nitrogen and water vapor in the air mass bounded by the dynamic polar vortex. Conceptual profiles for seasonal PSC occurrence and thermal structure are described. PSC are common to near and above 20.0 km MSL through June. After this, they are most frequent near 15.0 km MSL through August.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Catalytic cycles for ozone destruction in the lower stratosphere involving anomalous reactive chlorine and bromine molecules were proposed in the early 1970s in response to both natural and increasing anthropogenic sources [e.g., Crutzen, 1974; Cicerone et al., 1974; Stolarski and Cicerone, 1974; Wofsy et al., 1975]. Not long afterward, the discovery of large ozone losses in spring over Antarctic [Farman et al., 1985] inspired over a decade of intensive field, laboratory and theoretical research into the physical mechanisms responsible. These efforts were summarized in the seminal review by Solomon . It was concluded that a primary mechanism driving the formation of the annual austral ozone “hole” involved the role of polar stratospheric clouds (PSC) in partitioning stable halogenated molecules relative to photochemically active ones.
 The second role played by PSC is in the causal repartitioning of water vapor and reactive nitrogen (NOy) concentrations through the sedimentation of cloud particles [e.g., Toon et al., 1986; Crutzen and Arnold, 1986; McElroy et al., 1986; Fahey et al., 1989; Kelly et al., 1989]. Denitrification promotes reactive halogen-bearing chemical mechanisms since nitrogen-based reactions are most efficient at neutralizing them [e.g., Solomon, 1990, 1999, and references therein]. Satellite measurements, combined with those done in situ, show that lower stratospheric concentrations of N2O and HNO3, for example, are either reduced substantially or removed altogether near the southern pole by as early as mid-June [e.g., Fahey et al., 1990; Santee et al., 2005]. Conversely, reactive species, such as ClO, increase during this time. Such conditions persist through the lifespan of the Ozone Hole, and only reverse with the breakup of the polar vortex in spring and the overturning of the polar air mass.
 Lidar instruments are linked to important historical data sets that have improved the understanding of PSC physical processes. The presence of small solution droplets at temperatures above the ice frost point, and eventual differentiation between frozen particles (type Ia) and liquid droplets (type Ib), was confirmed from differences in the polarization state of elastic lidar backscatter measurements [Browell et al., 1990; Toon et al., 1990]. These findings followed ground-based measurements of nonzero depolarization in weakly scattering Arctic PSC and high depolarization in relatively bright layers [Kent et al., 1986]. Iwasaka [1986a, 1986b] and Iwasaka et al.  reported similar findings along the northeast Antarctic coast. Poole and McCormick  correlated airborne-based backscatter measurements with temperature to discriminate between stages for type I and II growth.
 The seasonal PSC distribution at the South Pole was first described by Collins et al.  from episodic lidar measurements made at the Amundsen-Scott Station during 1990. Hayashi et al.  and Shiraishi et al.  described similar observations recorded at Ny Ålesund, Norway (78°55′N, 11°56′E) for the 1994/95 season. A longer PSC climatology for this site was reported by Massoli et al. . Lidar-based climatologies have been reported from two coastal Antarctic sites; McMurdo Station (77°48′S, 166°42′E [Adriani et al., 2004]) and Dumont d'Urville (66°24′S, 140°0′E [Santacesaria et al., 2001]). Gobbi et al.  grouped backscatter ratios (defined as the ratio between backscatter and that attributable to the molecular atmosphere), linear depolarization ratios (defined as the ratio of returned signals in the orthogonal versus parallel polarization states relative to a linearly polarized source [e.g., Sassen, 1991]) and volume concentrations collected with a 0.532 μm lidar in 1995 at McMurdo Station into seven categories for phase and composition. Isentropic ascent and gravity waves have also been studied as a means for the lift and cooling necessary to nucleate and sustain PSC [Shibata et al., 2003; Palm et al., 2005]. Satellite lidar data sets are now being investigated for regional PSC frequencies and distributions [Palm et al., 2005; Pitts et al., 2007], though their orbital tracks inhibit measurements nearest the poles.
 This paper describes 5 years of PSC measurements using data collected at the South Pole from May to September 2000 and May to October 2003–2006. This data set was compiled as part of the National Aeronautics and Space Administration Micropulse Lidar Network (MPLNET [Welton et al., 2001]) project using an autonomous and eye-safe micropulse lidar (0.523 μm; MPL [Spinhirne, 1993]). MPL instruments are capable of continuous full-time ground-based measurements and are compact and suitably rugged to withstand long-term remote deployment. The instrument has been maintained on site since 1999 at the National Oceanic and Atmospheric Administration Earth Systems Research Laboratory (NOAA ESRL) Atmospheric Research Observatory (ARO; 89.98°S, 24.80°W, 2.835 km above mean sea level; MSL).
 Situated atop the Antarctic plateau, and therefore near the climatological core position of the wintertime polar vortex [Simmonds and Law, 1995], the South Pole represents a fundamental point from which to observe and study PSC. Cloud properties there represent one end of a longitudinal cross section of the vortex where PSC dynamic and microphysical processes vary as a function of thermal and dynamic processes. Sufficiently cold temperatures are persistent from late May through September [Renwick, 2004]. Furthermore, the meridional mixing of air across the vortex edges, from baroclinic disturbances and planetary-scale wave breaking, which replenishes necessary cloud components (i.e., water vapor and nitric acid) lost through particle growth and fallout, is mostly limited.
 This paper is the first in a two-part series. Here, we describe the macrophysical, optical and thermodynamic properties of PSC observed at the South Pole with the MPL. The instrument and data sets are introduced in the following section. PSC heights and optical properties are then described in relation to ambient thermodynamic parameters, including theoretical concentrations for water vapor and nitric acid based on local radiosonde profiles. The result is a series of composite depictions of seasonal PSC presence and chemical partitioning. Total cloud presence and scattering is then examined relative to ozone loss at the South Pole to determine if any relationship exists. In a future paper, we will reinvestigate these data with respect to synoptic transport and the physical and dynamic character of austral polar vortex to determine how each influences cloud occurrence.
2. South Pole MPL Data Sets and Elevated Cloud Retrievals
 An MPL was first deployed to the NOAA ARO facility in December 1999. Three instruments have been used through December 2006. The MPL instrument is described by Campbell et al. . Eye safety is achieved by the expansion of a rapidly pulsed outgoing laser beam along a shared path with the receive optics coinciding through a 0.2 m diameter telescope. The pulse repetition rate is 2500 Hz. The combined package volume is less than 1.0 m3 and rests in an upper corner of the ARO lidar laboratory. The instrument points horizontally toward an integrated window mount holding a scanning 90° turning mirror. The transceiver is aligned in a plane 45° relative to the mirror. A Windows-based personal computer maintains the MPL, the scanner and all Internet communications (i.e., data transfer).
 MPL data collected at the South Pole are recorded at 60 s temporal and 0.030 km range resolution. The angle of the turning mirror is nominally set such that the propagating plane is within ±3° of zenith, though never directly at 0° to limit the effects of horizontally aligned ice crystals on the backscatter profile [e.g., Sassen, 1991]. However, this angle is varied at regularly timed intervals over the course of any given day. Nadir “tips” are done at multiple times to relieve possible snow accumulation. Low-angle scans are done along the horizon near 2000 UTC daily to analyze boundary layer conditions, such as the depth of the blowing snow layer [Mahesh et al., 2003], and accommodate slant path retrievals for cloud extinction cross-section profiles [Spinhirne et al., 1980]. This angle is recorded and later integrated into processed network Level 1.0 normalized relative backscatter (NRB) data files available for download from the MPLNET project (http://mplnet.gsfc.nasa.gov). The NRB data product and processing algorithm are described by Campbell et al. . Welton and Campbell  describe Level 1.0 signal uncertainties.
 We focus here on data collected during the 2000 and 2003–2006 winter seasons. Measurements during 2001 and 2002 were interrupted by instrument failures. Although the 2000 data set is similarly limited in length, data is available through early September and, therefore, a reasonable representation for that year. Data from 2003 to 2006 are mostly complete. Cloud heights were retrieved at 0.02 fractional day resolution (approximately 30 min averaging) using an algorithm described by Campbell et al. . This routine discriminates elevated particulate scattering structure on the basis of the signal uncertainties of the NRB product. To eliminate ambiguities arising from movement of the scanner, only those data recorded while the scanner was within ±3° of zenith were used. Campbell et al.  discuss the advantages of multitemporal resolution processing, and the discretion of the user in determining which layer height retrievals best fit the needs of the study. It was determined independently that 0.02 day (∼28 min) resolution produced the most stable retrievals for PSC identification.
 In Figure 1, we examine system performance across the data set using a proxy parameter derived during postprocessing. Aside from cloud heights, relevant retrievals from the Campbell et al.  algorithm include the instrument calibration factor and the attenuated lidar scattering ratio (ASR), each with uncertainties. ASR is defined as
Here, β(r) is the total backscatter cross section from both particulate and pure molecular scattering, T2(r) is the transmission term, βm(r) is the molecular scattering cross section, r is range and rb is the range where normalization of the lidar profile relative to a theoretical molecular scattering profile was calculated. When calibrating, a normalization step creates the threshold for cloud layer discrimination by superimposing the signal uncertainties at each range bin onto the molecular scattering profile. Therefore, a minimum detectable scattering ratio can be derived by doubling the error bar, since signal at each bin must exceed the threshold outside of its uncertainty to be considered cloud.
 For all clear-sky profiles derived in July of each season studied, the average minimum detectable scattering ratio is shown in Figure 1, with a corresponding averaged calibration value noted in the inset. The latter value is proportional to the instrument gain function, and, therefore, a proxy for optical efficiency. Campbell et al.  describe the calibration process. At the South Pole, the value is solved at an appropriate height above 3.0 km AGL and is accurate to within 1%. Performance in 2000 was highest, followed by that in 2003. Performance during 2004–2006 varied little. The differences can be attributed primarily to laser stability. Winter seasons in 2000 and 2003 followed instrument reset visits and the installation of new laser diodes. From Figure 1, detection efficiencies differ significantly beginning above roughly 12.0 km AGL, and are greatest above 20.0 km AGL.
 Previous measurements establish that, for all types of PSC and their embryonic subspecies, the lower range of lidar scattering ratio measurements at green wavelengths approach unity [Collins et al., 1993; Gobbi et al., 1998]. Therefore, it is implicit from this analysis that the MPL does not detect all PSC. For example, diffuse PSC layers above 20.0 km MSL have been reported from lidar measurements at the South Pole from as late as August [Collins et al., 1993], and below this height through October [Fiocco et al., 1992]. Instead, these data represent a reasonable and qualified sample of the clouds present each season.
 Shown in Table 1 are statistics relating to total cloudiness at the South Pole in each season studied with the MPL. The data are divided into two sections. The first describes the total number of 0.02 day signal averages from May to October analyzed (a total of 9200 possible for 184 days), and how many were missing, either from actual lack of data or scanner-induced ambiguity. The second section is divided into three parts. The first is the number of shots where low cloud inhibited normalization of the averaged lidar profile (i.e., attenuation-limited). Since the algorithm is only applied to the signal profile above 3.0 km AGL, the presence of low clouds, blowing snow, or, though less frequently, snow on the scanner window can limit the sampling range of the lidar. The second column is the number of profiles analyzed where calibration was successful and cloud detected. We do not discriminate between PSC and upper tropospheric cloud in this statistic at this time. Since it is most likely that attenuation-limited conditions are due to cloud, the combination of the two columns may be considered a general proxy for total cloudiness, including any stratospheric influence, at the site. The final column is the number of profiles where no cloud was detected.
Table 1. Sky Conditions for May–October 2000 and 2003–2006a
Processed algorithm statistics for cloudiness in May–October 2000 and 2003–2006. Percentage values for total and missing data points are relative to a possible total of 9200 data points for 0.02 fractional day resolution retrievals. Values for blocked, cloudy, and clear profiles are relative to the total available in the first column.
 Our analysis yields a relationship of 67% to 33% cloudiness versus clear skies at the South Pole from May to October. This compares with a previous report of 40% cloudiness due to tropospheric cloud alone during winter on site [Mahesh et al., 2003]. 2003, 2004 and 2005 were similar in number of clear profiles and close to the overall average. 2003 was a relatively high year for elevated cloudiness, however, compared to the sample mean, while the other 2 years exhibited the highest incidence of attenuation-limited conditions. 2006 was an unusually clear season. As discussed above, the fewest number of profiles were available from the 2000 data set. The season was also the cloudiest of the five studied. From our analysis of Figure 1, however, this number is likely biased high relative to instrument performance in other years. Since PSC frequencies decrease rapidly into September and October, overall cloudiness follows proportionally.
3. Seasonal PSC Results
 Shown in Figure 2 are raw algorithm retrievals of ASR at 0.02 day resolution for May–October 2000 and 2003–2006 from 3.0 to 28.0 km MSL. Gaps are either the result of missing data or profiles that could not be normalized. The PSC appear as deep and persistent signals, nominally from June through August and above 10.0 km MSL. Upper tropospheric clouds can be seen intermittently in the images, mostly in 2.0–3.0 km deep layers with high ASR near and below 10.0 km MSL. From these images, and Table 1, the combination of missing data and low-level blocked profiles were greatest in 2004 and 2005. This is especially clear when comparing those data to those from 2003, which was relatively free of attenuation effects.
 It is necessary to establish a delineating height to distinguish PSC from upper tropospheric clouds, which in many cases, as can be interpreted in Figure 2, may be related to PSC events. A traditional tropopause height, denoted by rapidly increasing static stability above it, can be approximated from December through March in South Pole sounding data near 8.0 km MSL. During other months, however, temperatures decrease with height, save for shallow folds/inversions, to nearly 25.0 km MSL. Dynamic coupling between the troposphere and stratosphere is more likely in such conditions. On the basis of visual inspection, we conservatively estimate 12.0 km MSL to be the lower limit for unambiguous PSC presence. The region from 8.0 to 12.0 km MSL is considered a transitional zone, where cloud type cannot be established with any certainty.
 The depth and persistence of the PSC layers in these data are characteristics that substantiate the use of data-smoothing filters to approximate values within the gaps in the raw data and limit noise. When applying these techniques, our goal is to create seasonal composite PSC profiles that retain as much macrophysical structure necessary to yield a meaningful climatology. The raw retrievals (Figure 2) have been reprocessed using a two-dimensional Hanning function (commonly referred to as the “Von Hann” window [Blackman and Tukey, 1959]). This formula is easy to implement and efficiently minimizes energy leakage from high-frequency spectra [Pfenninger et al., 1999].
 We have smoothed the raw data to a 0.5 day/0.250 km resolution grid using 5.0 day and 0.750 km temporal and spatial half widths. These settings were independently chosen after testing. Shown in Figure 3 are the results for the 2003 season. In this case, PSC are first observed in late May, and are persistent between 15.0 and 25.0 km MSL through early July. After which, aside from a brief episode in the second half of that month, clouds were observed episodically below 17.0 km MSL. Smoothing eliminates nearly all of the small-scale PSC structure, and these results should be regarded as generalized for qualitative discussion. For completeness, we qualify all of our results below using both the raw and smoothed data sets.
 Superimposed on these ASR data, with start points set to the seasonal maxima in early June near 25.0 km MSL, are theoretical 100-day fall trajectories for spherical particles of mass density 1.0 g/cm3, based on Kasten , assuming no evaporation. Trajectories are shown for particles of 10.0, 3.0, 1.0 and 0.1 μm radii. The larger subset corresponds to a fall rate of approximately 1.0 km d−1. Near 1.0 μm diameter, fall rates vary through the lower stratosphere from between a kilometer every 25 to 60 days. Submicron sized particles do not fall appreciably. These values were chosen on the basis of the measurement of multiple number concentration modes for similar particle size distributions in PSC [e.g., Dye et al., 1992; Toon et al., 2000].
 The bulk of the macrophysical cloud structure observed in 2003 (and many of the other seasons, as will be shown) slopes downward beginning from early June through the end July. This follows a progression of coldest temperatures that are observed at progressively lower heights through August (discussed below). By superimposing the fall velocities, however, we conceptualize the influence of large particles on denitrification and dehumidification [e.g., Salawitch et al., 1989; Toon et al., 1990; Tabazadeh et al., 2001]. To produce any meaningful effect throughout the vortex, particles with diameters on the order of 10.0 μm are necessary. Furthermore, the negative slope of cloud occurrence indicates the likelihood that PSC occurring early in the season redistribute nitrogen and water vapor through the lower stratosphere in advance of the occurrence of temperatures cold enough to support clouds. We return to this point below.
 ASR relates total backscatter to that of the molecular atmosphere. Therefore, it is ambiguous, since the molecular scattering cross-section changes by an order of magnitude from 10.0 km to 25.0 km MSL as a function of air density. Equation (1) can be simplified as
where βp(r) is the particulate backscatter cross section and Tp2(r) is the particulate transmission term. After subtracting for unity and multiplying by the molecular backscatter cross section, equation (2) becomes
which is the unattenuated particulate backscatter cross section, an absolute value relative to ASR.
 The assumption of negligible particulate transmission is not always appropriate, however. From testing the raw data in Figures 2a–2e using equation (3), and approximating a ratio of extinction-to-backscatter of 25.0, or that representing ice crystals [e.g., Ansmann et al., 1992], we find that optical depth of PSC regularly exceeds 0.03 from June to August, or the approximate threshold for cloud to a ground observer [Sassen and Cho, 1992] (not shown). The modeling study of Gobbi  shows that this is likely a low-end estimate for extinction-to-backscatter in the case for PSC because of the complexity of particle composition. In any event, it is safe to estimate that two-way transmission regularly approaches 0.90 at the green MPL wavelength. Therefore, we use the result of equation (3) and define it as an approximate attenuated backscatter cross section (AAB), where such nomenclature acknowledges that the value is not a traditionally attenuated one in the lidar sense (i.e., of the form XT2).
Figure 4 depicts a series of composite images for May to December 2000 and 2003–2006 from 5.0 to 28.0 km MSL relating smoothed MPL backscatter to ambient thermodynamic properties. First, smoothed ASR, having been converted to AAB, are shown from May to October for cases where signal exceeds 1.0 × 10−5 km−1 sr−1. This is an approximate lower limit for PSC based on Gobbi  and Gobbi et al. . Again, from Figure 1, the MPL is not sensitive to low backscatter targets above 15.0 km AGL. However, smoothing yields data points with values lower than algorithm thresholds used to produce the raw data sets. Therefore, we implement this new lower threshold as a practical means for limiting the influence of this noise on the data due to smoothing.
 Proxies for temperature are represented in Figure 4 by saturation isopleths derived from local rawinsonde data for NAT at 10.0 ppbv HNO3/4.0 ppmv water vapor concentrations [Hanson and Mauersberger, 1988] and ice at 6.0, 4.0 and 2.0 ppmv water vapor. NAT concentrations are approximated on the basis of historical measurements for lower stratospheric background values in an unperturbed (i.e., not denitrified) environment [e.g., Gille et al., 1993]. Similarly, water vapor values are approximated using previous in situ experimental data [Rosen et al., 1993]. Finally, ozone partial pressures derived from ozonesonde launches conducted at the South Pole are shown for 4.0, 2.0 and 1.0 mPa isobars from May through December. A Dobson unit (DU) timeline is shown below each composite image from May through January. A dashed line is included to represent 220 DU, which is the colloquial threshold representing Ozone Hole conditions [Stolarski et al., 1986].
 Despite daily launches, rawinsonde measurements at the South Pole during winter can be irregular in the lower stratosphere from early balloon bursts [Pfenninger et al., 1999]. Therefore, available data have been gridded and smoothed using the Hanning function to 1.0 day/0.050 km resolution using 14 day/2.0 km temporal and spatial half widths. Some missing data exist, and this can be seen as minor distortion in the images. Ozonesonde launches are typically more irregular, save for spring when the Ozone Hole is overhead and daily measurements are possible. Again, this data set was gridded and smoothed to the 1.0 day/0.050 km resolution using the same Hanning function settings applied as with the rawinsonde data. Surface DU measurements, therefore, represent the lowest grid point in this process, or 2.850 km MSL.
 NAT and ice frost point saturation isopleths were derived with respect to the saturation vapor pressure over both pure NAT and ice, respectively. Therefore, they are only rough approximations to reality, where ternary or hydrate-based solution droplets act as cloud particle embryos. Discussing the homogeneous nucleation of ice, Koop et al.  postulated that nucleation rates in aqueous solutions are independent of the nature of the solute and instead a function solely upon water activity. A recent review by Cantrell and Heymsfield  examines the practicality of the Koop et al.  model. Actual saturation temperatures for type I and II PSC will occur at lower values, likely on the order of 1 to 2 K. Therefore, these data are only approximate proxies for actual saturation scenarios, and are shown solely for qualitative analysis.
 For each year in Figure 4, PSC are first apparent above 10.0 km MSL after the onset of 10.0 ppbv HNO3/4.0 ppmv water vapor NAT-saturated air, but before saturation with respect to ice at 6.0 ppmv water vapor. An initial period of persistent cloud is present from 15.0 to 25.0 km MSL through June each year. After this, PSC occurrence is episodic, and varies greatly between seasons in spite of the existence of the coldest seasonal air from 15.0 to 20.0 km MSL, usually in early to mid-August. Total backscatter is greatest in 2000. Clouds were consistent at most heights through the end of the data set in early September. In 2003, clouds were not seen above roughly 15.0 km MSL beginning from mid-July suggesting that denitrified/dehumidified air persisted over the site. PSC detection was lowest in the 2004 data set. The 2005 and 2006 seasons exhibited relatively similar distributions of cloud and thermal properties. There are a few instances where cloud was detected above the upper bounds of the NAT saturation isopleth. We speculate that this is likely due, aside from the possible underestimation of NAT or water vapor concentrations present, to the enhancement of cloud nucleation and growth by inertial gravity waves [e.g., Shibata et al., 2003]. Potential energy gradients become stronger in the presence of increasingly higher static stability above 25.0 km MSL during the polar night [Pfenninger et al., 1999; Yoshiki and Sato, 2000].
 Satellite measurements of HNO3 made with the NASA Microwave Limb Sounder (MLS [e.g., Waters et al., 2006]) in mid-June 2005 (not shown) show total depletion of the former near 18.0 km MSL south of 80.0°S. The growth of liquid-phase PSC (type Ia and b) involves low supersaturation rates and the colloidal stability of small particles in relatively high concentrations. Solid phase particles, once nucleated, grow to larger sizes capable of yielding sufficient fall velocities necessary to redistribute nitrogen, and water vapor [e.g., Jensen et al., 2002]. Sublimation and evaporation of falling particles repartition these components. Combining these observations with the cloud and thermal structure of each season in Figure 4, the redistribution of nitrogen and water vapor in the lower stratosphere is likely occurring below 18.0 km MSL and influencing cloud nucleation processes and, therefore, heterogeneous chemistry and denitrification at these lower heights when temperatures gradually become favorable from late June through August. This may be notable, since ozone losses are greatest between 15.0 and 18.0 km MSL in September and October.
 DU values drop below 220 typically near 1 September each year and are restored typically by December. Notable exceptions to this are in 2004, when ozone replenishment had occurred by mid-November, and 2006, when persistently low values were measured through the end of the year. The height of near-zero ozone concentrations occurs, on average, near 16.5 km MSL. This level is slightly lower than that for seasonal thermal minimums, which are observed closer to 20.0 km MSL.
 AAB data, and therefore ASR, shown here are consistent, though in greater temporal detail, with previous lidar measurements reported from the South Pole. Fiocco et al. [1991, 1992] and Fuà et al.  report 0.532 μm measurements of PSC layers made episodically during winter 1988. Fiocco et al.  describe continuous layers frequently greater than 5.0 km in depth up to 21.0 km MSL, with ASR regularly approaching 4.0 and backscattering cross sections frequently on the order of 1.0 × 10−4 km−1·sr−1. Cacciani et al. [1997a] (0.532 μm) and Collins et al.  (0.589 μm) report similar characteristics in episodic measurements made during winter 1990, just before the Mt. Pinatubo eruption. Collins et al. , in particular, report diffuse PSC, with ASR just above 1.0, near and above 15.0 km MSL as late as October. Rosen et al.  report measurements from a series of balloon-borne dual-wavelength backscatter-sondes (akin to the lidar technique, though in situ) that, though limited, are also consistent in the development of cloud relative to temperature during winter 1991. Cacciani et al. [1997b] report measurements made post-Pinatubo during the winters of 1992 and 1993. The MPL data are also consistent with macrophysical structure reported from coastal Antarctic sites, including Syowa Station [Iwasaka et al., 1985; Iwasaka, 1986a, 1986b], Dumont d'Urville [Stefanutti et al., 1991; Santacesaria et al., 2001] and McMurdo Station [Gobbi et al., 1998].
 Shown in Figure 5 are rates of occurrence each season, including the subset mean, for smoothed and raw AAB from May to October, and ozone losses at the South Pole from August to December. Because of the 2000 data set being limited, those results are not included in the AAB displays. As a proxy for ozone loss, we calculate negative daily DU departures from 220 DU and integrate over time to produce a value representative of both temporal extent and depth of low ozone column concentrations directly at the South Pole (i.e., analogous to a “degree day,” and, henceforth, referred to as a “DU day” [Geer, 1996]). Corresponding totals for smoothed and raw AAB and DU days from 12.0–24.0 km MSL are shown in Table 2, as well as values for three specific depths: 12.0–16.0, 16.0–20.0 and 20.0–24.0 km MSL. The 2005 season stands as most typical in each set of results. Occurrence rates for PSC there mirror the 5-year average. Ozone losses also agree well with the mean. AAB totals were third highest of the subset and near average at each of the three height ranges.
Table 2. Seasonal Ozone Loss and PSC Backscatter Statisticsa
For 2000 and 2003–2006, integrated Dobson unit days from August to December, total integrated smoothed approximate attenuated backscatter (AAB) from 12.0 to 24.0 km MSL at 0.250 km/0.5 day resolution, total integrated raw AAB from 12.0 to 24.0 km MSL at 0.030 km/0.02 day resolution, total integrated smoothed AAB for three height ranges within the 12.0–24.0 km depth at 0.250 km/0.5 day resolution, and total integrated raw AAB for three height ranges within the 12.0–24.0 km depth at 0.030 km/0.02 day resolution.
 During 2000 and 2003, then-record Ozone Hole seasonal maximum areas were observed. These values stood until a new record was established during the 2006 (http://www.theozonehole.com/; attributed to R. McPeters, NASA/Goddard Space Flight Center) season. The highest totals for both raw and smoothed AAB were measured in 2000. Totals observed in 2003 were second-highest. In both years, abnormally large AAB totals were observed above 20.0 km MSL (Table 2). It must again be stressed here that instrument performance was higher during these years than subsequent ones. However, these findings are substantiated by relatively high corresponding DU day totals. AAB values from 12.0 to 16.0 km MSL were also higher these years than others, and it was determined (Figure 1) that the effect of instrument performance on cloud retrievals at these heights would only be slight. Therefore, we are confident than any instrument bias on these findings is minimal. The 2003 season was unique. The bulk of the cloud occurring that season was observed by late July. In a future paper, we will describe how the polar vortex was unusually deep in August of this season and speculate on how this influenced the development and transport of PSC within the polar air mass. Ironically, despite this latter observation, breakup of the vortex occurred quickly (Figure 5c), and DU days (Table 2) are relatively low despite the coldest winter temperatures of the subset (Figures 4c and 4d).
 AAB totals from 2006 were near the group mean, but DU day totals were very high. Figures 4i, 4j, and 5c show that depleted ozone concentrations were persistent through the end of the year despite a relatively average season for PSC occurrence. This is due to a later-than-normal breakup of the polar vortex and slow recycling of the air mass, which enhanced the temporal extent of ozone-depleted air over the South Pole. Therefore, the 2006 season is unique to this data set. Data from the remaining years show that replenishment of ozone, to levels above 220 DU, nominally occurs by the first week of November. The 2004 season, another very low PSC year, is only slightly anomalous in this regard. The lowest AAB and DU day totals in this subset were observed in 2004. In 2002 and 2004, the seasonal maximum area of the Ozone Hole was measured at 15-year lows. The 2002 season is well documented for the Ozone Hole having split into dual lobes because of unusual Rossby wave forcing [e.g., Hoppel et al., 2003]. Scarcely any PSC were observed in 2004 above 20.0 km MSL.
Figure 6 is a series of plots comparing both smoothed and raw AAB versus DU days for each of the five seasons. The plots are broken down into total AAB derived between 12.0 and 24.0 km MSL and then for each of the three discrete height ranges outlined in Table 2 (see caption of Figure 6 for order). In these plots, data from the 2006 season stands most pronounced. If the 2006 data points are removed, linear regression fits to the 2000 and 2003–2005 data yield correlation coefficients of 0.89 and 0.78 for the total smoothed and raw data, respectively. For the 12.0–16.0 km MSL region, values of 0.62 and 0.46 are found. For 16.0–20.0 km MSL, they are 0.87 and 0.88. Finally, for 20.0–24.0 km MSL, values of 0.75 and 0.62 are found. Interestingly, despite the caveats introduced when smoothing the raw data, both data sets yield similar results both here and in the comparisons from Figure 5.
 From these data, PSC presence and ozone depletion measured at the South Pole are highly correlated near the height of maximum ozone loss and for the scenario of typical seasonal decay of the polar vortex. Correlation values at upper levels are also strong. Since PSC at these heights are most commonly seen early in the season and believed to be influencing clouds formed later at lower levels, and that ozone losses are also tangible at up near 20 km MSL, this is not surprising. Correlation values at the lowest heights are lower, but the potential influence of upper tropospheric disturbances on these data, and a mostly static ozone partial pressure near 12.0 km MSL, substantiate this result.
 The correlated findings for the bulk sample, 12.0–24.0 km MSL, are consistent with model calculations made by Rex et al.  and Douglass et al. , albeit for the Arctic, and using PSC volume rather than AAB, and modeled ozone losses rather than DU days. Those studies estimate cloud volume using model reanalysis temperatures and NAT condensation and ice frost points for unperturbed background concentrations of HNO3 and water vapor. They do not consider the effects of denitrification or dehumidification, though their effects in the Arctic are typically localized and not as widespread as in the Southern Hemisphere [e.g., Fahey et al., 1990].
 In Figure 7, averaged AAB data from the 5-year smoothed data sets are shown, with averaged 4.0/2.0/1.0 mPa ozone partial pressures and averaged 185/190/195/200 K isotherms shown in the latter. Averaged AAB values are only calculated for points where cloud was observed above the 1.0 × 10−5 km−1 sr−1 detection threshold each season. PSC can be expected between 15.0 and 25.0 km MSL beginning in late May and lasting through June. Clouds with the highest optical significance occurring above 15.0 km MSL occur during this period. After this, PSC are more common, though episodic, below 20.0 km MSL, and especially from 12.0 to 15.0 km MSL, through August.
 Clouds are not infrequently observed at ozone-depleted heights beginning from 1 September, though the sensitivity of the instrument may bias these results for the case of optically thin clouds [Collins et al., 1993]. Heterogeneous chlorine activation may occur on PSC surfaces in the presence of sunlight irregardless of denitrification levels [Portmann et al., 1996; Chipperfield and Pyle, 1998]. This process is seemingly a minor one at the South Pole, which reinforces the role of preconditioning the air mass through denitrification and dehumidification by sedimenting cloud particles during polar night.
4. Summary and Conclusions
 The Ozone Hole remains a contemporary social issue [Newman et al., 2006; Weatherhead and Andersen, 2006], though most of the science on the topic is settled, on the basis of the bulk of research in the late 1980s through mid-1990s. The role of polar stratospheric clouds (PSC) in promoting catalytic ozone loss cycles is well understood [e.g., Solomon, 1999]. Cloud particles provide sites for heterogeneous reactions that free reactive chlorine and bromine molecules from inert forms. The uptake of nitric acid in intermediate stages of PSC growth, and subsequent sedimentation arising from the increase in particle fall velocity, remove nitrogen from nucleation heights [e.g., Toon et al., 1990]. Nitrogen otherwise reacts with and locks chlorine and bromine into relatively stable reservoir species that suppress catalytic ozone depletion [e.g., Turco et al., 1989; Solomon, 1999]. Still, models for ozone loss are not easily validated by current observations [e.g., Douglass et al., 2006]. Denitrification and dehumidification are not fully understood [e.g., Tabazadeh et al., 2001]. A primary reason these processes remain uncertain comes from a lack of continuous observations of PSC structure, type and phase over the austral pole, though chlorine and bromine loading are also poorly resolved [e.g., Douglass et al., 2006]. While satellite techniques and coverage have improved since the first observations of ozone loss in the early 1980s, limitations still exist, particularly with respect to their vertical resolution and coverage directly over the South Pole.
 This paper is an investigation of PSC macrophysical, optical and thermodynamic properties using data collected from 2000 and 2003–2006 using a micropulse lidar (MPL); a period important for it lacks any lingering stratospheric influence of the Mt. Pinatubo eruption of 1992 [e.g., Cacciani et al., 1997b]. Composite profiles for seasonal PSC occurrence, thermal evolution and ozone loss are described. These are based on smoothed representations of raw cloud parameter retrievals and yield generalized profiles of cloud scattering properties. Climatological characteristics are described. This paper is the first of a two-part series. In a future paper we will reexamine the MPL data within the context of synoptic-scale transport and polar vortex dynamic properties to more completely investigate their influence on clouds observed at the South Pole.
 Our primary finding is a high correlation between total integrated MPL backscatter and Dobson unit days (negative daily integrated values relative to the 220 DU Ozone Hole threshold [Stolarski et al., 1986; Geer, 1996]) for 2000 and 2003–2005. In these years, overturning of the polar winter air mass was observed in occur, on average, in November. In 2006, an average season for PSC occurrence, ozone-depleted air persisted over the South Pole through December from lack of significant overturning and chemical replenishment from lower latitudes. Although PSC play a first-order role in ozone loss, by passively conditioning the polar air mass to support catalytic chemistry, they still form as a function of the depth and size of the polar vortex since sufficiently cold temperatures would likely not persist otherwise (i.e., the boreal scenario). The dynamic and physical character of the polar vortex may then ultimately be the most important factor in reconciling the magnitude and temporal extent of seasonal ozone losses.
 Whereas traditional models for denitrification and dehumidification describe the irreversible removal of nitrates and water vapor from PSC cloud-nucleating heights, they do not consider the effects of molecular redistribution within the stratosphere [e.g., Jensen et al., 2002]. In late May and early June, when clouds are typically first observed, base heights are detected near and above 18.0 km MSL. Below these levels, settling and evaporation of cloud particles from above is likely occurring, and corresponds with reports of low nitrate concentrations by passive satellite measurements around the austral pole in June [e.g., Waters et al., 2006]. By midseason, PSC are increasingly common at lower heights as temperatures there become favorable with time. We believe that, in tandem with satellite measurements, the MPL data indicate the likelihood that nitrogen and water vapor are being redistributed downward through the lower stratospheric air mass in late May and June. Early season PSC at upper levels (near and above 20.0 km MSL) are likely enhancing the formation of PSC at lower levels in later months, assuming the air mass within the vortex remains unperturbed. Coincidentally, or not, these lower heights, 15.0–18.0 km MSL, are where maximum ozone losses are measured in spring.
 Given the diversity and advances in most lidar technologies, MPL instruments would be an unlikely choice for monitoring PSC. Over a decade earlier, much higher powered instruments were used to collect more robust observations of stratospheric clouds over Antarctica [e.g., Fiocco et al., 1992; Collins et al., 1993; Gobbi et al., 1998]. The advantages to the MPL technique, though, are eye safety and the ability to make autonomous full-time measurements without otherwise major safety concerns [Campbell et al., 2002]. Additionally, the MPL is easily deployed and maintained. Shipping to even the most remote field sites is possible with only minor modifications to basic climate-controlled working spaces [Campbell et al., 2003]. Full-time monitoring produces data sets necessary for long-term climate studies at high spatial and temporal resolutions, whereas sporadic or episodic observations may lead to an incomplete perspective. Despite its low-powered laser source, we have explicitly demonstrated the sensitivity of the instrument to PSC and qualified our results on the basis of the probability for missing optically thin clouds in the processed data.
 MPLNET operations at the South Pole continue, and we anticipate extending our investigation into future years. In tandem with passive radiometric measurements of chemical concentrations, particularly those perturbed by heterogeneous activity (e.g., ClO, HCl, etc…), as well as chemical modeling, we believe accurate forecasts for seasonal ozone losses and future trends are possible and will work toward integrating these data sets into such models. Furthermore, augmentation of the instrument to include polarization-sensitive channels would allow for discrimination of cloud phase to better determine the distribution and frequency of type I versus type II cloud particles and lead a more complete depiction of denitrification cycles.
 The MPLNET project is funded through the NASA Earth Observing System and the NASA Atmospheric Radiation Sciences program. This research was supported, in part, by a grant from the National Science Foundation (NSF ATM-0630506). The comments and encouragement of R. L. Collins are gratefully acknowledged. The authors thank E. J. Welton, J. D. Spinhirne, T. A. Berkoff, S. Valencia, S. Stewart, and L. Belcher for their support of the South Pole MPL at the NASA Goddard Space Flight Center. Author J.C. also thanks E. G. Dutton and B. Vasel at the NOAA Earth Systems Research Laboratory for their continuing support of the experiment, as well as the many NOAA technicians who have and continue to maintain the instrument on site at the South Pole.