Quasi-coincident Observations of Polar Stratospheric Clouds by Ground-based Lidar and CALIOP at Concordia (Dome C, Antarctica) from 2014 to 2018

. Polar stratospheric clouds (PSCs) have been observed from 2014 to 2018 from the lidar observatory at the Antarctic Concordia station (Dome C), included as a primary station in the NDACC (Network for Detection of Atmospheric Climate Change). Many of these measurements have been performed in coincidence with overpasses of the satellite-borne CALIOP (Cloud Aerosol Lidar with Orthogonal Polarization) lidar, in order to perform a comparison in terms of PSC detection and composition classiﬁcation. Good agreement has been obtained, despite of intrinsic differences in observation geometry and 5 data sampling. This study reports, up to our knowledge, the most extensive comparison of PSC observations by ground-based and satellite-borne lidars. The PSCs observed by the ground-based lidar and CALIOP form a complementary and congruent dataset, and allow to study the seasonal and interannual variations of PSC occurrences at Dome C. Moreover a strong correlation with the formation temperature of NAT (Nitricacidtrihydrate), T NAT , calculated from local temperature, pressure and H 2 O and HNO 3 concentrations 10 is shown. PSCs appear at Dome C at the beginning of June up to 26 km, and start to disappear at the second half of August, when the local temperatures start to rise above T NAT . Rare PSC observations in September coincide with colder air masses


Introduction
Long term ground-based and satellite-borne lidar observations provide valuable climatological data, and allow monitoring of 15 the state of the polar stratosphere and comparison with Chemistry Climate Models (CCMs). Several Antarctic stations have been equipped with lidar for PSC observations since the 1980s. The longest time records are from Dumont D'Urville (1989-1998, 2006-present) (Santacesaria et al., 2001;David et al., 1998David et al., , 2010 and McMurdo (1990McMurdo ( -2010 (Adriani et al., 1992(Adriani et al., , 1995(Adriani et al., , 2004Di Liberto et al., 2014;Snels et al., 2019). The McMurdo stratospheric lidar transferred to Concordia station at Dome C and has been operational since 2014. Concordia station has the advantage of being on the Antarctic plateau, far from the coast and well within the polar vortex during most of the winter (see Figure 1). Meteorological conditions are in general more stable than those at the lidar stations located on the coast (McMurdo, Dumont D'Urville, Davis, Syowa, Rothera, Belgrano II) and tropospheric clouds rarely obstruct the PSC observations. South Pole station is also located far from the coast and as such shares the advantages of Dome C, but unfortunately all three lidar systems that have been operated there 5 Fua et al., 1992;Simpson et al., 2005;Huang et al., 2007;Campbell and Sassen, 2008) were without a depolarization channel, and thus are limited in classifying the composition of the observed PSCs. The CALIPSO (Cloud-Aerosol Lidar and Infrared  (Stephens et al., 2002(Stephens et al., , 2017. With an orbit inclination of 98.2 o , it provides extensive daily measurement coverage over the polar regions of both hemispheres, up to 82 o in latitude. The primary instrument of CALIPSO is the Cloud Aerosol Lidar with 10 Orthogonal Polarization (CALIOP ). CALIOP has extensively been used for observing PSCs (Pitts et al., 2009(Pitts et al., , 2011(Pitts et al., , 2013(Pitts et al., , 2018.
The comparison of ground-based and satellite-borne lidars is useful for calibration purposes and intercomparison, but presents many difficulties, which are mainly due to the fact that perfect coincident measurements are not feasible. The reason is very simple: the footprint of the satellite overpass is never exactly coincident with the position of the ground-based lidar. 15 Also the different observation geometries pose serious difficulties. For instance the air masses examined have different shapes; the ground-based lidar is observing an air mass determined by the divergence of the laser and the field of view of the telescope, resulting in round pixels with increasing diameter with altitude, while CALIOP produces a rectangular cross section moving along the flight track, by averaging all the laser shots along a 5 km orbit segment. Moreover, the ground-based lidar signal suffers 2-way attenuation between the ground and detected PSCs, especially the attenuation due to tropospheric clouds, while 5 the satellite-borne lidar is hardly attenuated down from the orbit up to the first PSCs encountered between 25 and 30 km. Then of course the operating conditions are very different; while the satellite-borne lidar is autonomously operating and acquiring data most of the time, the ground-based lidar can not be operated under severe weather conditions, and needs an operator for cleaning the view port from the snow and for trouble shooting when faults occur.
In a previous paper  the full dataset of PSC obervations by ground-based and satellite-borne lidar above 10 McMurdo has been statistically compared in terms of detection and composition classification of PSCs. In the McMurdo study all ground-based lidar and CALIOP data within a longitude-latitude box (7 o × 1 o ) centered on McMurdo, without further constraints, were taken into account. This implies that e.g. ground-based data recorded on days without CALIOP overpasses in the longitude-latitude box were included in the statistical analysis, as well as all CALIOP profiles within the box. Here we consider only quasi-coincident observations, which implies a much smaller set of ground-based and satellite-15 borne PSC observations above Concordia station, by comparing the ground-based observations with single CALIOP profiles acquired at the closest possible distance from the ground station and within 30 minutes of the ground-based observation. For the CALIPSO PSC product, a single vertical profile consists of an average over a 5 km long segment of the flight track. The quasi-coincident approach has not been applied to the McMurdo data since too few quasi-concidences were available. At Dome C we consistently pursued to record all possible coincidences with CALIOP leading to a relatively large number of 20 quasi-coincidences.
If we want to use ground-based and CALIOP data together, in a complementary and congruent way, we need to verify if both lidars observe essentially the same scene, in terms of PSC detection and composition. Much depends on the spatial extent of the PSC fields and their homogeneity in terms of particle composition. If the PSC's spatial extent is generally small with respect to the average closest distances of the footprint of the satellite-borne lidar with respect to the ground station, our approach is 25 condemned to fail. If, however, PSC fields extend over many tens or hundreds of km, our approach may yield valuable results.
The position of Dome C on the central plateau, with no important orographic features present, implies that the temperature fields should be mainly of a synoptic nature, with few local perturbations. This would cause the PSCs to have a large extension and a predominantly homogeneous composition. We will show by studying PSCs observed during CALIOP overpasses close to Dome C, that in most cases the PSC fields have sufficiently large extension to allow for a comparison of almost coincident 30 observations. Of course this is valid for the detection, while a comparison of the composition would require that the PSCs be also of an homogeneous composition over these scales. We will also address this problem and show that in many cases the PSC composition is substantially homogenous, i.e. more than two thirds of the CALIOP overpasses in the box around Dome C at a specific vertical level have at least 75 % of the PSCs of the same composition. It should be stressed that the results presented here are restricted to the area around Dome C, and might be very different for other locations in Antarctica.

35
The formation of PSCs depends on the the availability of condensation nuclei, the temperature and number densities of water and nitric acid. Nitric acid trihydrate (NAT) particles are is in thermodynamic equilibrium with the gas phase at about 6 K above the ice point (Hanson and Mauersberger, 1988), while liquid PSCs in the form of supercooled ternary solutions (STS) may form below temperatures of about 3 K above the frost point. Finally ice PSCs form below the frost point. While ice evaporates once the temperature exceeds the frost point, NAT and STS may survive for some hours/days above their equilibrium temperature.

5
Here we compare the occurrence of PSCs, as detected by the ground-based lidar and CALIOP, with the formation temperature of NAT. All species discussed here, NAT, STS, ice and their mixtures, form below T NAT , and one would expect to observe PSCs when the local temperature is below T NAT , although NAT and STS might also survive some time above this temperature.
We observe how PSC occurrences drop rapidly in the third and fourth week of August, when the local temperatures exceed T NAT . Seasonal and interannual variations of PSC occurrences are seen to be dependent on the position of the cold polar 10 vortex. Here we use T NAT as a delimiter of the area where PSCs may be formed.

PSC Observations at Dome C by Ground-based and Satellite-borne Lidar
A lidar has operated in the Antarctic station of Dome C since 2014, as a continuation of the long time series observations at McMurdo (1991McMurdo ( -2010, with the goal to measure the atmospheric backscatter with parallel and perpendicular polarization with respect to the emitted laser radiation. The lidar observatory uses most of the hardware previously used at McMurdo, with 15 some small improvements. In addition, the instrument has been adapted to the harsher environment at Dome C. In particular a triple glass view port has been mounted above the telescope in order to have a better insulation from the outside temperature. Recently, the observatory has been equipped with a remote control of the laser and the data acquisition system, allowing for a complete control of the measurements from the main buildings at Dome C, at a distance of about 400 meters from the observatory, or from our home Institute in Italy. 20 Dome C is well within the stratospheric polar vortex from mid-June to the end of September, although climatologically the coldest part of the vortex is migrating towards the Antarctic Peninsula starting from the second half of August. In general the weather conditions are rather stable with respect to McMurdo and other coastal lidar stations and the lidar has been operated satisfactorily from 2014 on, during the Antarctic winter, with the exception of 2019, when severe instrumental problems occurred. The lidar is operated by science winter-overs of the PNRA (Piano Nazionale della Ricerca in Antartide) during the 25 Antarctic winter, typically from the end of May until the end of September to cover the whole period of PSC occurrence.
The lidar is operated once or twice per day, when meteorological conditions are favorable. If possible, the observations are synchronized with overpasses of the CALIPSO satellite, when its footprint is within 100 km distance from Dome C. Single vertical profiles with a vertical resolution of 60 m have been recorded by averaging 30 minutes of acquisition. All lidar data have been deposited with the NDACC database (Snels, 2019) , where they are available to the scientific community. 30 CALIOP is a two wavelength lidar, measuring backscatter at wavelengths of 1064 nm and 532 nm, the latter signal separated into parallel and cross polarization, with respect to the polarization of the outgoing laser beam. Details on CALIOP can be found in Winker et al., 2009). The orbital period of CALIPSO is about 98 min, which results in about 14-15 orbits per day. This results in two overpasses per day at distances ranging from 0 to 400 km from the ground-based lidar at Dome C, one on an ascending and the other on a descending orbit. Here we consider all overpasses within a longitude-latitude box The different sampling times and observation geometries of the ground-based lidar and CALIOP imply that it is extremely difficult to obtain "real" coincidences. An interesting approach has been suggested by David et al. (2012), who used trajectories calculated from wind velocities and directions to connect the air masses observed by ground-based lidar and CALIOP. Of course this method is limited to few intersections, and depends strongly on the altitude, since the wind velocity increases with altitude.

10
The average windspeed ranges from about 50 to 110 km/h between 10 and 20 km of altitude, but maximum wind speeds exceed 200 km/h. The wind direction at Dome C is mostly between NE and SE. We've explored the possibility to apply the trajectory approach to our data, but the number of coincidences is very low, and not homogeneously distributed for all altitudes and this method has thus been discarded. Instead we compared ground-based data with the closest profile on each CALIPSO flight track (within 100 km) and with an overpass time within 30 minutes of the ground-based observation. This is a good compromise for 15 obtaining a significant number of comparisons and having a reasonable probability that both lidars observe similar air masses, although not perfectly coincident. The comparison is made considering the detection and composition classification of PSC clouds.

PSC Detection and Classification Criteria for the CALIPSO v2 Data
The CALIOP v2 PSC detection and composition classification algorithm has been used to create the recently released CALIOP 20 v2 PSC mask database covering the period from June 2006 to October 2019. Here we compare these v2 data with groundbased observations at Dome C from 2014 to 2018. Major enhancements in the v2 algorithm over earlier versions include daily adjustment of composition boundaries to account for effects of denitrification and dehydration, and estimates of the random uncertainties u(β ⊥ ) and u(R) due to shot noise in each data sample, which are used to establish dynamic detection thresholds and composition boundaries. The CALIOP v2 algorithm is represented pictorially in Figure 2 and is described in more detail 25 in (Pitts et al., 2018;Snels et al., 2019).

PSC Detection and Composition Classification Criteria for the Ground-based Lidar Data
In order to compare the ground-based lidar data to the CALIOP data we have adopted a similar algorithm which follows the same approach and uses the same optical parameters as the v2 CALIOP algorithm (see Figure 2). The v2 background aerosol thresholds β ⊥,thresh and R thresh have been calculated in a different way, to take into account a series of errors due the small  troposphere, have been taken into account to calculate u(β ⊥ ) and u(R) and create the dynamic thresholds for detection and classification.

Data processing
Raw data consist of photon counts recorded in 400 ns bins, corresponding with a vertical resolution of 60 m and are accumulated in records with a duration of two minutes. First the data are averaged and the background count is determined from the first 40 5 bins, before the laser fires. The background is then subtracted from the signals. The lidar records two channels, one collecting the signal with the polarization parallel to the laser emission and the other with perpendicular polarization. The two components are separated by using two polarizing beamsplitter cubes. The perpendicular polarization is in theory due to the depolarization by molecules and aerosols. In practice some instrumental factors may contribute to the perpendicular polarization, such as a small perpendicular component of the laser emission, the residual transmission/reflection of the unwanted component of 10 6 https://doi.org/10.5194/acp-2020-972 Preprint. Discussion started: 12 October 2020 c Author(s) 2020. CC BY 4.0 License. the polarizing beamsplitter cubes, and other effects. In our case there is a substantial contribution due to the triple viewport.
Starting from the lidar equation we can express the signals on the two detectors, if we neglect the extinction, while only considering the crosstalk as originating from the optical elements (polarizer, and viewport), as where g 1 and g 2 are the gain factors of the two detectors and CT is the crosstalk from the parallel channel to the perpendicular channel. We neglect the crosstalk from perpendicular to parallel channel.
In order to facilitate the interpretation of the signals and the detection of clouds, we divide both signals by the molecular backscatter coefficient and multiply by z 2 and we get 10 The molecular backscatter coefficient has been calculated by using local temperature and pressure provided by radiosoundings and, where these were not available, from NCEP.
We can normalize these expressions to 1, where no aerosols are present (typically between 26 and 30 km) The normalized With some algebra we can now obtain expressions for R and β ⊥ 20 and where δ mol = β mol⊥ /β mol . In our case, using an optical bandpass filter centered at the laser wavelength (532 nm) with a FWHM of 2 nm, δ mol is 0.007 (Behrendt and Nakamura (2002)).
Now we can see that the crosstalk CT can be written as The two parameters rn and rn ⊥ can be determined from the calibration process for aerosol free regions, so we only need 5 the ratio of the two gain constants of the two detection channels, which can be determined e.g. by switching the detectors, or by more sophisticated methods (Snels et al. (2009)).
Now we perform the correction for extinction, using Klett and the proportionality between β and the extinction coefficient as reported by Gobbi (1995) and use the ratios obtained after the correction to calculate R(z) and β ⊥ (z).

Error processing 10
The statistical errors deriving from photon counting process, u(β ⊥ ) and u(R), have been determined from the raw signals, and are thus dependent on z. The background aerosol thresholds β ⊥,thresh and R thresh have been determined mainly by comparing with clear sky profiles and have been expressed in terms of the ratios r (z) and r ⊥ (z). They are estimated to be 1.1 for r (z) and for r ⊥ (z). For z < 16 km, these values increase gradually to 1.2 in order to take into account an insufficient correction of saturation effects.

PSC Detection and Composition Classification
PSC detection and classification from lidar measurements with orthogonal polarization is based on two optical parameters derived from the optical signals with parallel and perpendicular polarization with respect to the laser. Here we use a method that approximately follows the v2 classification and detection scheme (see also ), proposed by Pitts et al. (2018) for the classication of the CALIOP PSC data, and using the backscatter ratio R and the perpendicular backscatter The backscatter ratio R and the perpendicular backscatter coefficient β ⊥ have been determined from the raw data as described above. The optical parameters obtained in this way, as well as their errors were smoothed to the vertical scale of the CALIOP profiles, with a vertical resolution of 180 m per layer or pixels as we will call the single bins on a profile from now on. The detection thresholds for backscatter ratio R was thus determined to be R thresh + u(R) and the threshold for β ⊥ as 25 β ⊥,thresh +u(β ⊥ ) . This results in dynamic thresholds that vary from profile to profile, for instance due to attenuation of the lidar signal by cirrus clouds, and vary with altitude, mostly because of the statistical errors in the photon counting process.
In order to detect a PSC, it is sufficient that either the backscatter ratio R or the perpendicular backscatter coefficient β ⊥ , exceeds the respective threshold. A final step of the processing requires that at least 5 consecutive points on a vertical profile are identified as PSCs, in order to avoid the appearance of "spikes" in the profiles. Sequences of less than five PSC points are 30 thus considered to be non PSCs. This procedure is similar to the coherence criterium used for the CALIOP data.

PSC Composition
Composition classification for ground-based PSCs is nearly identical to the CALIOP v2 procedure, the exception being that we use values of R NAT|ice reported for the closest profiles in the v2 CALIOP data files. The borderline value to discriminate between STS and NAT is equal to the detection threshold for β ⊥ , exactly the same as for the CALIOP data.  Table 1 shows some statistics illustrating the number of data acquired and actually used for comparison. Note that Table 1 includes all CALIOP tracks passing within a 200×200 km square around Dome C, including also some tracks with a closest distance of more than 100 km. This explains why the numbers of CALIOP overpasses might be slightly larger than the 80 overpasses mentioned before. Throughout this paper, the comparison for detection and composition classification will be performed by comparing vertical bins, with the same height and a thicknes of 180 m, one of the ground-based lidar profile and 15 the other of the nearest CALIOP profile. From now on we will refer to these bins as pixels, since each bin will be color coded in the figures in correspondence with the composition of the PSC. In our analysis we will consider only pixels between 12 and 26 km, since very few PSCs have been observed above 26 km and inclusion of the pixels above 26 km would not be a good measure for comparison.

20
A second test has also been performed to quantify how much the PSC composition varies along the track. For this test we consider 4 PSC classes; STS, NAT mixtures, enhanced NAT mixtures and ice. We count the number of PSC segments along each orbit track at each altitude that belong to one of these four classes. The results are displayed in Table 2. The numbers in this    on one-dimensional tracks. However, the overpasses have different directions, due to the ascending and descending orbits of CALIPSO and thus many overpasses will ultimately fill a two-dimensional field.

Comparison of ground-based and CALIOP data for 2015-2018.
Now that we have some confidence that a comparison between ground-based and CALIOP data at an average distance of 50 km and within 30 minutes of the observation is a reasonable way of proceeding, we will illustrate the full procedure of detection The detection and composition classification of the PSCs in CALIOP data is based directly on the CALIOP v2 PSC Mask database, while for the ground-based data we have applied the detection and classification scheme as has been discussed above. were detected in the coincident profiles. Although these "clear sky" profiles are a minority of the measurements, the two data sets agree rather well also for these clear sky profiles and thus they have been included in the detection statistics (see Table 1).  and PSC detected by the ground-based lidar only. The first two categories are listed in the column "detected" (see Table 1), because there is agreement between the (non) detection of a PSC. There is a small fraction (≈ 5%) of coincidences where only CALIOP detects a PSC, but a larger fraction (11-25 %) of coincidences where only the ground-based lidar detects a PSC.
The larger fraction of coincidences with ground-based lidar only detection of PSCs may be partly due to the larger sensitivity of the Dome C lidar below 15-16 km. This can be also seen by comparing figures of all coincident measurements and those 15 observations where both lidars detected a PSC (left and right columns in Figures 4,5,6 and 7).
For comparison of PSC composition, we restrict the analysis to only those measurement pixels where both lidars detected a PSC (see the right columns of Figures 4, 5, 6 and 7). The results of this comparison are reported in Table 3. The first four columns report the percentage of pixels with a certain PSC composition as determined by CALIOP. The columns under the STS, red= enhanced NAT mixtures, blue =ice. The circles indicate measured profiles without PSCs.  However, we want to explore how critically the PSC composition of the ground-based data depends on the choice of the thresholds, also considering that the calculated threshold for β ⊥ in the classification of the ground-based data may be affected by instrumental errors, and also R NAT|ice may be slightly different at Dome C with respect to the value for the closest CALIOP profile. To take into account these uncertainties in both β ⊥threshold + u(β ⊥ ) and R NAT|ice we allow a 10% tolerance on both.
This implies that all PSCs with a value of β ⊥ between β ⊥threshold + u(β ⊥ ) ± 10 % are possibly STS or NAT and all PSCs 5 with a value of the backscatter ratio R between R NAT|ice ± 10 % are possibly enhanced NAT mixtures or ice. This additional tolerance gives slightly better scores for the comparison (see Table 3, column gb method 2), but doesn't change the results in a significant way.
While performing ground-based measurements with a duration of several hours, we have observed that PSC layers may move up and down on the time scale of 30 minutes. So we allowed also a small vertical displacement of the air mass observed 10 by the ground-based lidar with respect to the CALIOP observation, which is in average at a distance of 50 km from Dome C.
This implies that each CALIOP pixel is now compared with three ground-based pixels, one at the same height and the other two at the next upper or lower layer (± 180 m) (method 3). This last method leads to an improvement of about 10 % of the overall agreement and is shown in the column gb method 3 of Table 3. The largest effect can be observed for STS and NAT mixtures. The last three columns in Table 3 report the sum of the pixels (for each of the four composition classes), where the 15 ground-based lidar identifies the same composition as CALIOP, for each for the three methods applied.
The results of the comparison show that in average for 58 % of all observations both lidars observe PSCs of the same composition, which becomes 71 % when tolerances on the thresholds are applied as well as on the altitude (± 1 layer  producing the more extreme low or high values of the optical parameters. For instance if the ground-based lidar observes STS for 10 minutes and NAT mixtures for 20 minutes, the average value of β ⊥ will be probably higher then the detection threshold and thus the observation will be classified as NAT mixtures, while CALIOP on the corresponding overpass track might identify both STS and NAT mixtures, and the closest profile represents the statistical distribution of STS and NAT mixtures. In any case, considering the expected composition homogeneity derived from the analysis made for all CALIOP overpasses in the 10 box around Dome C (see Table 2), the overall result is satisfactory.

PSC occurrence as Observed and Classified by Ground-based and Satellite-borne Lidar
PSC occurrence and composition classification has been performed for all ground-based data and all satellite-borne lidar profiles at the shortest distance from Dome C (thus not limited to the quasi-coincident measurements), using detection and composition classification criteria as mentioned before. The ice frost temperature T ice and the formation temperature for NAT    The result of the detection comparison is that about 75 % of the (non) PSCs were detected by both lidars, while about 5 % was detected only by CALIOP and 20 % only by the ground-based lidar. Probably the better detection efficiency of the groundbased lidar at lower altitudes might explain the latter. If we consider only 2016, 2017 and 2018 these values are even better and reach 76 to 84 % of agreement. The composition of the detected PSCs has been compared in a strict, pixel-to-pixel way, and also by introducing some more permissive criteria, such as a small variation of the classification thresholds and a comparison with the next higher or lower layer (± 180 m). It can be concluded that the observation of NAT mixtures by CALIOP is confirmed in most cases (83 %) by the ground-based lidar, while the identification of the minor species by CALIOP, was confirmed in average for 59 , 32 and 67 % of the cases, for STS, enhanced NAT mixtures and ice, respectively, by the ground-based lidar.

5
Our explanation is that the ground-based data acquisition produces averaged values of the optical constants, by integrating over 30 minutes, corresponding with a spatial integration of 15-30 km. This integration process favours the classification as NAT mixtures, at the cost of reducing classification of the other species. On the other hand CALIOP takes a "snapshot" during its overpass of about 30 seconds, and is more sensitive to the other species.
The results presented here are providing a solid basis for the comparison of ground-based and space-borne lidar observations 10 of polar stratospheric clouds (PSC). This is the most extensive comparison of such data, to our knowledge, and may provide a means to produce a standard PSC product for ground-based lidars, with a good compatibility with CALIOP and other space-borne instruments, although with some caveats. It opens new possibilities of including ground-based validated PSC data in CCM models and microphysical studies. The method proposed here is shown to be valid for polar regions with rather uniform temperature fields, and absence of important orographic structure, but might be used with some constraints to different 15 situations.
It has also been shown that observations obtained by the ground-based lidar and CALIOP are complementary and congruent and can be used to study seasonal and inerannual variations of the presence of PSC clouds at Dome C. The PSCs observed by both systems are generally observed for local temperatures below T NAT , although some observations at higher temperature are reported. These are mostly NAT mixtures, that are known to persist some days even above T NAT . During the winter season, 20 PSCs slowly descend and are rarely observed from the second half of August, in agreement with the warming of the vortex at Dome C.
For all five years concerned here, few PSCs have been observed during the second half of August and during September, most probably because a displacement of the cold pool versus the Antarctic pensinsula. Presently a climatological study for the Dome C area is underway by combining data of both lidars, which, based on this study, are compatible up to a large degree.

Data availability
The raw data of the ground-based lidar at McMurdo are publicly available on the NDACC data base (ftp://ftp.cpc.ncep.noaa.gov/ndacc/station/mcmurdo/ames/lidar/). The CALIOP v2 data are available on request from Michael Pitts and Lamont Poole.