Groundbased validation of CALIPSO ... - Wiley Online Library

JOURNAL OF GEOPHYSICAL RESEARCH: ATMOSPHERES, VOL. 118, 2889–2902, doi:10.1002/jgrd.50248, 2013

Ground-based validation of CALIPSO observations of dust and smoke in the Cape Verde region M. Tesche,1 U. Wandinger,2 A. Ansmann,2 D. Althausen,2 D. Müller,2,3,4 and A. H. Omar5 Received 19 September 2012; revised 28 January 2013; accepted 4 February 2013; published 11 April 2013.

Ground-based Raman lidar measurements during the second Saharan Mineral Dust Experiment (SAMUM-2) in 2008 were used for validation of measurements of the lidar aboard the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite within the dusty environment of the Cape Verde region. SAMUM-2 featured two one-month campaigns in January/February and May/June 2008 to cover different modes of aerosol transport to the tropical Atlantic: dust from northern Africa and biomass-burning smoke from western Africa during winter, and pure Saharan dust during summer. During the investigated time period, 33 CALIPSO overflights occurred at a distance of less than 500 km from the location of the ground-based lidar. Fifteen out of these 33 cases were found suitable for comparing the findings of the two instruments. The parameters for this comparison are the particle backscatter coefficient at 532 and 1064 nm, the extinction coefficient, the lidar ratio (aerosol type), and the particle depolarization ratio at 532 nm, as well as the backscatter-related Ångström exponent for the wavelength pair 532/1064 nm. Best agreement was found for the 532 nm backscatter coefficient, while the 532 nm extinction coefficient is underestimated by up to 30%. The latter is due to the use of an effective dust lidar ratio that gives reliable backscatter coefficients but is not suitable to transform these to extinction coefficients. CALIPSO particle depolarization ratios provided in the current (version 3.01) aerosol profile product were found to be affected by a computing error and should be calculated from the perpendicular and total particle backscatter coefficients provided in the same data file. CALIPSO aerosol classification was found to be mostly correct but a demand for homogeneous aerosol layers could improve the retrieval. Suggestions for the improvement of the CALIPSO retrieval by introducing iterative procedures are provided. [1]

Citation: Tesche, M., U. Wandinger, A. Ansmann, D. Althausen, D. Müller, and A. H. Omar (2013), Ground-based validation of CALIPSO observations of dust and smoke in the Cape Verde region, J. Geophys. Res. Atmos., 118, 2889–2902, doi:10.1002/jgrd.50248.

1. Introduction [2] Since June 2006, the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) aboard the CloudAerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) [Winker et al., 2009] satellite provides the scientific community—for the first time—with continuous and vertically resolved observations of the global atmospheric 1 Department of Applied Environmental Science (ITM), Stockholm University, Stockholm, Sweden. 2 Leibniz Institute for Tropospheric Research (TROPOS), Leipzig, Germany. 3 Science Systems and Applications Inc., NASA Langley Research Center, Hampton, Virginia, USA. 4 Department of Physics, Astronomy, and Mathematics, University of Hertfordshire, Hatfield, UK. 5 NASA Langley Research Center, Hampton, Virginia, USA.

Corresponding author: M. Tesche, Department of Applied Environmental Science (ITM), Stockholm University, Svante Arrhenius väg 8, SE-11418 Stockholm, Sweden. ([email protected]) ©2013. American Geophysical Union. All Rights Reserved. 2169-897X/13/10.1002/jgrd.50248

aerosol distribution. CALIOP is an elastic-backscatter lidar and does not allow for a direct measurement of the particle extinction coefficient, which is the parameter of relevance for an assessment of aerosol radiative effects. A priori knowledge of the aerosol type present is essential in the data analysis. This is a challenging task of any automated retrieval algorithm and a significant error source. [3] Previous studies of the unique CALIPSO data set dealt with the investigation of the vertical extent of aerosol layers [Liu et al., 2008a; Devasthale et al., 2011] or certain transport regimes and events [Liu et al., 2008b; Ben-Ami et al., 2009, 2010]. The CALIPSO data retrieval was also evaluated without using external data [Liu et al., 2011]. Constant efforts were taken to validate CALIPSO measurements from the first moment of operation. Studies available in the literature assess the identification of vertical features [McGill et al., 2007] and the accuracy of level 1 products [Mamouri et al., 2009; Mona et al., 2009; Pappalardo et al., 2010; Rogers et al., 2011]. Kittaka et al. [2011] and Redemann et al. [2012] use observations of the MODerate resolution Imaging Spectroradiometer (MODIS) aboard the Aqua satellite for a global assessment of CALIPSO-derived

2889

TESCHE ET AL.: SAMUM-2 CALIPSO COMPARISONS

aerosol optical depth. Note that case studies that examine the level 2 products of the CALIPSO retrieval [e.g., Burton et al. 2010; Chazette et al. 2010; Kacenelenbogen et al. 2011; Mielonen et al. 2009] were usually performed during dedicated campaigns or at established lidar sites at midlatitudes, for example, in the framework of the European Aerosol Research Lidar Network (EARLINET; Pappalardo et al. 2010). These sites are mainly located in urban areas over which complex aerosol mixtures prevail during most of the time. Little is known about the performance of the CALIPSO data retrieval in environments that are dominated by mineral dust or mixtures of dust with other aerosol types [Campbell et al., 2012]. However, it has already been discussed that problems can be encountered in the CALIPSO retrievals of mineral dust properties [Wandinger et al., 2010]. [4] Validation studies during field campaigns are of importance in regions where ground-based measurements are sparse and which are key areas for climate studies. One of these areas is the Cape Verde region that undergoes strong seasonal changes in aerosol transport patterns with pure dust and mixed dust/smoke layers during winter and deep pure dust layers during summer [Ben-Ami et al., 2009; Schepanski et al., 2009; Tesche et al., 2009b, 2011a, 2011b]. Measurements with the Backscatter Extinction lidar-Ratio Temperature Humidity Apparatus (BERTHA) [Althausen et al. 2000; Tesche et al. 2009a] multiwavelength lidar of the Leibniz Institute for Tropospheric Research, Leipzig, Germany, during the second Saharan Mineral Dust Experiment (SAMUM-2) [Ansmann et al., 2011; Tesche et al., 2011a] at Cape Verde provide the unique opportunity to validate CALIPSO level 2 products with ground-based lidar measurements at this location. This paper presents the findings of the comparison of BERTHA measurements during SAMUM-2 at Cape Verde with CALIPSO overpasses at distances of less than 500 km from the ground site.

2. Instruments 2.1. BERTHA [5] A detailed description of the six-wavelength aerosol lidar BERTHA can be found in Althausen et al. [2000] and Tesche et al. [2009a]. Measurements during SAMUM-2 are described in Tesche et al. [2011a]. The instrument measures profiles of the particle backscatter coefficient at 355, 532, and 1064 nm. Extinction coefficients are measured at 355 and 532 nm and enable a direct calculation of the extinctionto-backscatter (lidar) ratio at these two wavelengths. In addition, Ångström exponents [Ångström, 1964] related to particle backscatter and extinction can be calculated as

åX1 /2 =

ln (X(1 )/X(2 )) ln (2 /1 )

(1)

exponent for backscatter and extinction coefficients together with air-mass transport calculations [Tesche et al., 2009b, 2011a]. 2.2. CALIOP 2.2.1. Instrument [6] CALIOP is an elastic-backscatter lidar that emits linearly polarized light at 532 and 1064 nm. The system features three measurement channels. To allow for depolarization-ratio profiling, backscattered light at 532 nm is split into a parallel and a cross-polarized signal with respect to the plane of polarization of the emitted laser light. The total backscattering signal is detected at 1064 nm. Hunt et al. [2009] give an overview of the instrument architecture and performance, while a summary of the data retrieval algorithms is provided by Winker et al. [2009]. 2.2.2. Aerosol Model and Data Retrieval [7] When lacking an independent optical depth measurement to constrain the retrieval [Young, 1995], the analysis of measurements with elastic-backscatter lidar requires a priori knowledge of the aerosol-type-specific lidar ratio [Klett, 1981; Fernald, 1984; Ansmann and Müller, 2005; Müller et al., 2007]. The CALIPSO aerosol retrieval provides this information by identifying one out of six different aerosol types depending on the location of a measurement (on the globe and in the atmospheric column), on the magnitude of the integrated attenuated backscatter coefficient, and on an approximate particle depolarization ratio [Omar et al., 2009]. The aerosol types used in the CALIPSO aerosol model were identified from cluster analysis of AERONET measurements and include clean marine, smoke (biomass burning), dust, polluted dust (mixtures of dust and smoke), polluted continental, and clean continental [Omar et al., 2005, 2009]. Each aerosol type is characterized by a set of lidar ratios at 532 and 1064 nm (see Table 1) that were calculated from typical particle size distributions and complex refractive indices taken from AERONET observations (polluted continental, smoke, and polluted dust) or experimental data (clean continental and clean marine) and are in good agreement with measurements [Omar et al., 2009]. The CALIPSO dust model is based on theoretical scattering calculations using discrete dipole approximation with realistic particle composition and irregular shapes [Omar et al., 2009]. The CALIPSO dust lidar ratio of 40 sr at 532 nm is much lower than the value of 55 sr found for pure dust layers over Morocco and Cape Verde during SAMUM [Tesche et al., 2009a, 2011a]. For mineral dust from the western Sahara, previous research suggests that a dust lidar ratio of 40 sr can be regarded as an effective lidar ratio that includes Table 1. Aerosol Types and Associated Lidar Ratios at 532 and 1064 nm as Used in the CALIPSO Data Retrieval [Omar et al., 2005, 2009] Aerosol Type

with X referring to the backscatter or extinction coefficients. Furthermore, particle depolarization ratios are measured at 710 nm and can be transformed to 532 nm [Tesche et al., 2011a]. Aerosol-type classification from BERTHA measurements is performed with the help of the intensive parameters particle depolarization ratio, lidar ratio, and Ångström 2890

Clean marine Dust Polluted continental Clean continental Polluted dust Smoke

Lidar Ratio (sr) 532 nm

1064 nm

20 40 70 35 65 70

45 55 30 30 30 40

TESCHE ET AL.: SAMUM-2 CALIPSO COMPARISONS Table 2. Aerosol Parameters Used for the Comparison of Measurements with BERTHA and CALIPSOa Parameter

Name in CALIPSO Files

ˇ532 ? ˇ 532 ˇ1064 ˛532 S532 file ı532

Particle backscatter coefficient Perpendicular particle backscatter coefficient Particle backscatter coefficient Particle extinction coefficient Particle lidar ratio (aerosol feature type) Particle depolarization ratio

Total_Backscatter_Coefficient_532 Perpendicular_Backscatter_Coefficient_532 Backscatter_Coefficient_1064 Extinction_Coefficient_532 Atmospheric_Volume_Description, bits 10-12 (see Table 1) Particulate_Depolarization_Ratio_Profile_532

S532 åbsc 532/1064

Particle lidar ratio, calculated as ˛532 /ˇ532 Backscatter-related Ångström exponent, calculated according to equation (1)

?/k ı532

Particle depolarization ratio, calculated according to equation (2)

a

The wavelength of observation is given as index.

the effect of multiple scattering by large dust particles [Wandinger et al., 2010]. Note that the complex mineralogy of mineral dust varies for different source regions and might cause a regional variation of the dust lidar ratio with values down to 40 sr at 532 nm [Schuster et al., 2012]. However, actual measurements of the dust lidar ratio close to a source region are only available for Saharan dust during SAMUM-1 in Morocco [Tesche et al., 2009a]. [8] The pre-defined lidar ratios are used in the CALIPSO level 2 retrieval algorithm to obtain backscatter-coefficient profiles of aerosol and cloud layers at 532 and 1064 nm from the respective level 1 attenuated backscatter profiles (i.e., calibrated range-corrected elastic backscatter signals) [Omar et al., 2009; Vaughan et al., 2009]. It is assumed that the retrieval algorithm selects the lidar ratio with an uncertainty of no more than 30%. Note that this uncertainty is aerosol-type dependent and is lower than 30% for sea salt and as high as 50% for mineral dust. Extinctioncoefficient profiles are obtained by multiplying backscatter coefficients with lidar ratios used in the retrieval. Note that the CALIPSO data retrieval algorithm works gradually, i.e., individual results are not used in iterative loops to improve outcomes of earlier stages of the retrieval. Note also that the particle depolarization ratio as an intensive property would be a more suitable choice for aerosol classification rather than the volume depolarization ratio or an approximate particle depolarization ratio. However, the particle depolarization ratio is an end product of the CALIPSO data retrieval and could only be applied in a later iteration (starting at the selection of aerosol type), if such were included in the retrieval algorithm. [9] For the comparison presented here, we use level 2 version 3.01 products with a vertical resolution of 60 m (below 20.2 km height) and a horizontal resolution of 5 km [Vaughan et al., 2009]. Version 3.01 of the CALIPSO data retrieval provides better screening of small-scale clouds in aerosols within the planetary boundary layer, an improved cloud-aerosol discrimination (CAD) algorithm [Liu et al., 2010], and a new procedure for layer-base determination for boundary layer aerosols [Vaughan et al., 2010]. Furthermore, the level 2 version 3.01 aerosol profile product now provides the user with profiles of the perpendicular backscatter coefficient and the particle depolarization ratio at 532 nm as well as with uncertainty profiles for all parameters. Improvements in the CAD algorithm led to a better discrimination between clouds and dense aerosol layers [Liu et al., 2010]. This is of particular importance for dense dust layers observed during SAMUM-2 [Tesche et al., 2011a].

2.2.3. Compared Parameters [10] The parameters considered in this comparison study are summarized in Table 2 together with their respective name in the CALIPSO level 2 version 3.01 files or the equation used for calculation. The particle depolarization ratio was taken from the CALIPSO level 2 version 3.01 files file (referred to as ı532 ); a second version is calculated from the individual profiles of the cross-polarized (? ˇ 532 ) and total backscatter coefficients as ?/k ı532 =

?

ˇ 532 . ˇ532 – ? ˇ 532

(2)

[11] For the comparison, we only considered highquality CALIPSO profiles with Atmospheric_Volume_ Description bits 1–3 equal to 3 (feature type = aerosol), a CAD_Score below –20 (screen artifacts from data [Liu et al., 2010]), and an Extinction_QC_Flag_532 [Young and Vaughan, 2009] of either 0 (unconstrained retrieval; initial lidar ratio unchanged during solution process) or 1 (constrained retrieval). A description of the CALIPSO lidar level 2 5 km cloud and aerosol profile products can be found in the CALIPSO Users Guide [2011].

3. Comparison Methodology [12] The BERTHA measurements during SAMUM-2 were usually performed at noon and after sunset [Tesche et al., 2011a]. Additional measurements for CALIPSO validation were conducted when the satellite’s ground track passed the measurement site at Cape Verde at a distance of less than 500 km. For ocean sites like Cape Verde, high correlation between satellite and ground-based measurements can still be expected for such distances [Kovacs, 2006; Liu et al., 2008b]. Even for a comparison over smaller distances, it is essential that ground-based measurements are properly connected to a given CALIPSO overpass. In this study, HYSPLIT trajectories [Draxler and Rolph, 2010] were used to determine the correct temporal delay between the two measurements and to find the ground-track segment that is most suitable for comparison. [13] The methodology for relating BERTHA measurements to the appropriate along-track latitudinal range of a CALIPSO overpass is illustrated in Figure 1 using the examples of 11 and 13 February 2008, 31 May 2008, and 3 and 15 June 2008. CALIPSO overpasses to the west and east of Cape Verde were connected to the ground station through

2891


Figure 1. (a) Examples of CALIPSO ground tracks (bold lines) for overflights in the Cape Verde region on 11 February 2008 (red, 46 km west of Praia), 13 February 2008 (magenta, 280 km east of Praia), 31 May 2008 (orange, 362 km west of Praia), 3 June 2008 (blue, 171 km east of Praia), and 15 June 2008 (green, 478 km west of Praia). Backward trajectories starting at Praia (red star) at 2 km (thin solid lines and small circles), 3 km (thin dashed lines and small squares), and 4 km (thin dotted lines and small triangles) were used to connect CALIPSO measurements with the ground-based observations, if the overflight occurred east of the measurement site (e.g., 3 June 2008). For overflights west of Praia (e.g., 15 June 2008) forward trajectories were used to investigate the comparability and to determine the latitudinal interval for averaging of individual CALIPSO profiles. The numbers on the trajectories denote the start/arrival heights and the time interval between Praia and the intersection with the CALIPSO ground track. (b) The vertical displacement of air parcels traveling along the individual trajectories. The symbols mark the time/height after/at which the trajectories cross the CALIPSO track.

averaging CALIPSO profiles that are most likely measured in the same air parcels as was done with BERTHA at Cape Verde. It enabled matching of CALIPSO overpasses to BERTHA measurements over distances of several hundred kilometers and time periods of up to 18 h. During SAMUM2 (winter and summer), 33 CALIPSO overpasses occurred at a distance of less than 500 km from the BERTHA site. Fifteen out of these 33 overpasses were found suitable for the comparison of the aerosol profiles of BERTHA and CALIPSO. Conditions that generally inhibited a comparison during the remaining 18 CALIPSO overpasses were the occurrence of low and high clouds which attenuate the signal of the ground-based and spaceborne system, respectively, as well as low aerosol load which results in a low signal-to-noise ratio. [15] A detailed overview of the suitable cases is given in Table 3. Besides the measurement times of BERTHA and CALIPSO, the table also gives the distance and temporal delay between the respective measurements and the height range and along-track latitudinal interval selected for individual comparison cases. Within the respective latitudinal range, CALIPSO profiles were chosen with respect to the quality-assurance criteria given in section 2.2.3, averaged, and smoothed with a window length of 660 m (11 height bins). The same smoothing length is applied to the BERTHA data presented in Tesche et al. [2011a]. The availability of aerosol parameters as measured with BERTHA during the comparison days is given in Table 4. Because the overflight on 3 February 2008 is compared to a coinciding BERTHA daytime measurement for which Raman signals cannot be detected, extinction coefficients and lidar ratios are not available for comparison. No observations of the 1064 nm backscatter coefficient could be performed with BERTHA in the beginning of the summer campaign due to a broken photomultiplier tube. Laser problems prevented measurements of the depolarization ratio toward the end of the summer campaign [Tesche et al., 2011a].

4. Case Studies forward and backward trajectories starting and arriving over BERTHA, respectively. The along-track latitudinal interval of the CALIPSO measurement that seemed most appropriate for comparison is given by the intersections between CALIPSO ground track and trajectories at different height levels (thus accounting for along-track shear). Figure 1b indicates the vertical displacement of air parcels traveling along the trajectories shown in Figure 1a. In most cases, small vertical displacement was found along the trajectories. Hence, aerosol profiles of BERTHA and CALIPSO were not expected to show a large height shift of features, and vertical displacement was not accounted for in the comparison. Figure 1b suggests that larger disagreements in the comparison might be introduced when aerosol at different height levels is transported with different velocity (as can be seen for the case of 3 June 2008). This across-track shear was accounted for by averaging several hours of groundbased measurements for the comparison to an individual CALIPSO overpass. [14] The approach described above is necessary to find the respective along-track latitudinal intervals for

[16] Very different aerosol conditions prevail over the eastern Atlantic during winter and summer [Ben-Ami et al., 2009; Schepanski et al., 2009; Tesche et al., 2011a]. Hence, we will present a detailed CALIPSO-BERTHA comparison case study for each of the two SAMUM-2 campaigns before the general findings are presented. Aerosol conditions are better defined during summer which is why we start with a SAMUM-2b case [Tesche et al., 2011a]. 4.1. 15 June 2008 [17] Figure 2 presents the observations during a daytime CALIPSO overpass around 1528 UTC on 15 June 2008 (green line in Figure 1) at a distance of 478 km to the west (see Table 3). The level 1 attenuated backscatter coefficient along the track between 0ı and 30ı N versus height is shown in Figure 2a in the original CALIPSO resolution of 333 m. Level 2 version 3.01 products for the range marked by the black box in 2a are shown in Figures 2b–2e. Note that level 2 products are reported at 5 km resolution but might be retrieved at a resolution of up to 80 km

2892

TESCHE ET AL.: SAMUM-2 CALIPSO COMPARISONS Table 3. CALIPSO Overpasses (Within a Distance of 500 km from Cape Verde) That Could Be Compared to Ground-Based Measurements with BERTHA During SAMUM-2a CALIPSO Overflights Date Time

BERTHA Measurements Date Time

Distance to Praia

Delay

Comparison for Heights Latitudes

25 Jan 3 Feb 5 Feb 6 Feb 8 Feb 11 Feb 12 Feb 13 Feb

1515 1509 1422 0245 1457 0328 1503 0253

161 km to the west 4 km to the east 325 km to the east 441 km to the east 487 km to the west 46 km to the west 164 km to the east 280 km to the east


1215–1658 1120–1610 2000–2236 2146–2252 2127–2256 2332–0442 2010–2252 2010–2252

0h 0h 7h 5h 16 h 0h 6h 5h

1.3–3.0 km 0.5–4.8 km 3.0–4.5 km 3.3–5.0 km 2.4–4.5 km 2.6–4.6 km 1.2–3.7 km 1.4–3.8 km

12.0-15.0ı N 10.0-16.0ı N 14.0-16.0ı N 14.0-16.0ı N 12.0-15.0ı N 13.0-15.0ı N 14.0-17.0ı N 13.0-17.0ı N

28 May 31 May 3 June 11 June 15 June 16 June 17 June

0258 0327 1458 0309 1528 0341 1516

441 km to the east 362 km to the west 171 km to the east 127 km to the east 478 km to the west 351 km to the west 153 km to the west


1014–1121 2011–2305 2330–0220 2134–0001 2133–2317 2042–2332 0215–0505

8h 5h 12 h 4h 18 h 5h 12 h

2.0–4.0 km 1.5–4.5 km 1.1–5.3 km 0.8–5.0 km 1.5–4.5 km 1.8–5.9 km 1.1–5.8 km

14.0-16.0ı N 12.0-15.5ı N 13.5-16.0ı N 14.0-16.0ı N 16.0-18.0ı N 12.0-15.0ı N 14.0-16.0ı N

a Times and dates refer to UTC (local time –1) and 2008, respectively. The last two columns give the height ranges and the along-track latitudinal intervals identified to be most suitable for comparison.

[Vaughan et al., 2009], which results in the lines of coarse resolution visible in Figures 2d and 2e. [18] Figure 2b shows that the version 3.01 CAD retrieval properly separates aerosols in a lofted layer and close to the surface from clouds at the top of both the lofted aerosol layer and the marine boundary layer. The discrimination is similar to what can be obtained by eye from Figure 2a but is much more sensitive to the actual extent of the lofted aerosol layer. Only small and very dense parts of the aerosol plume seem to be misclassified as cloud instead of aerosol (blue lines at 4 km height and 14ı –16ı N). In these regions, backscatter is so intense that features can be detected at single-shot resolution. Note that by convention, anything that can be detected at single-shot resolution is immediately classified as a cloud in the CALIPSO retrieval. The aerosol subtype mask in Figure 2c shows that the major part of the lofted aerosol plume is classified as dust (yellow) with small fractions of polluted dust (brown). It seems like the latter is only identified in regions of low signal-to-noise ratio (SNR). The particles in the marine boundary layer are classified as clean marine (blue), dust, and polluted dust. Table 4. Availability of Parameters Measured with BERTHA During the Comparison Days Given in Table 3a ˇ532

ˇ1064

˛532

S532

åbsc 532/1064

ı532



date

a

All dates refer to 2008.

[19] The display of the particle backscatter coefficient shows two regions of dust concentration: a 2 km deep dense part to the south of 16ı N and a 5 km deep homogeneous layer of lower dust concentration to the north of 16ı N. The first can also easily be identified in the plot of the rangecorrected signal while the low SNR in the latter inhibits a quick-look assessment of the extent of the dust plume from level 1 data. The 532 nm depolarization ratio plot in Figure 2e shows the different aerosol layers: the marine aerosol in the lowermost kilometer shows low values while the nonspherical dust particles in the elevated layer show increased values. [20] Figure 3 gives a better view of the quantitative results of the CALIPSO observations between the white lines in Figure 2 and shows the comparison to aerosol profiles from a BERTHA measurement performed between 2133 and 2317 UTC on 14 June 2008. Even though profiles are shown from the surface to 6 km height, a quantitative comparison was only performed between 1.5 and 4.5 km (see section 3). The aerosol profiles measured with BERTHA and CALIPSO show very similar features that might be shifted a few hundred meters due to vertical motion along the trajectory of transport (see Figure 1). Very good agreement in terms of profile shape and magnitude is found for the 532 nm backscatter coefficient even though the CALIPSO lidar ratio of 40 sr is considerable lower than values of 50–60 sr measured with BERTHA. A linear fit through zero for a correlation plot of the measurements from BERTHA and CALIPSO for this parameter (50 points representing data for 60 m height intervals) gives a slope of 1.073 (with R2 = 0.99). In contrast to that, CALIPSO overestimates the 1064 nm backscatter coefficient by 35% and underestimates the 532 nm extinction coefficient by 20%. The latter finding is most likely due to multiple-scattering effects by large mineral dust particles [Wandinger et al., 2010] and the use of an effective lidar ratio that accounts for this effect in the backscatter retrieval. A detailed discussion is presented in section 5. Due to the increased 1064 nm backscatter coefficient, 532/1064 nm Ångström exponents vary between 0 and 0.3 for the CALIPSO measurement while BERTHA reveals values around 0.5. Special emphasis

2893


Figure 2. The CALIPSO measurement during an overpass 478 km to the west of Praia at around 1528 UTC on 15 June 2008. Besides the (a) latitude-height plots of the attenuated backscatter coefficient between 0ı and 30ı N, the figure shows the (b) vertical feature mask (bits 1–3 of the Atmospheric_Volume_Description), (c) the aerosol subtype mask (bits 10–12 of the Atmospheric_Volume_Description for feature type equals aerosol), (d) the 532 nm backscatter coefficient, and (e) the 532 nm particle depolarization ratio between 10ı and 23.2ı N (marked by a black frame in 2a). The white lines in the pictures mark the interval chosen for the comparison to BERTHA measurements (see Table 3). was put to the comparison of the 532 nm particle depolarization ratio. BERTHA showed typical dust values of 0.30–0.33 while CALIPSO values are slightly lower—for both cases of values from the CALIPSO file and calculated according to equation (2). The comparison of the other days given in Table 3 were performed accordingly. One more example is discussed in the following section while a comprehensive overview of the general findings is given in section 5. 4.2. 11 February 2008 [21] Figure 4 presents the observations during a CALIPSO overpass around 0328 UTC on 11 February 2008. Clouds above 5 km height occasionally attenuated the laser beam and caused the vertical black lines in the plots. Note that these cloudy profiles were not considered in this study. Areas of aerosol, clouds, and clear air are clearly identified in the vertical feature mask. The elevated layer was identified to consist mainly of mineral dust with minor contributions of polluted dust and traces of smoke. The lower layer is dominated by clean marine aerosol with small contributions of polluted dust and polluted continental.

The SAMUM observations showed that the boundary layer over Cape Verde was indeed dominated by marine aerosol [Tesche et al., 2011a; Groß et al., 2011]. However, it was also observed that this type of aerosol was mixed with mineral dust. While the classification seems appropriate in this layer, it is worth mentioning that a mixture of marine aerosol and mineral dust is not explicitly represented in the CALIPSO typing scheme [Omar et al., 2009; Winker et al., 2009]. Note that a mixture of marine aerosol and mineral dust has considerably different optical properties, in particular a smaller lidar ratio, than what is labeled as polluted dust in the CALIPSO retrieval [Müller et al., 2007; Groß et al., 2011]. Furthermore, investigations of the CALIPSO Science Team revealed that the classification of dust and polluted dust in marine boundary layers that lie beneath other cloud or aerosol layers is due to an error in the CALIPSO retrieval software that will be fixed in future data releases (Mark Vaughan, personal communication). The elevated aerosol plume over Cape Verde on the other hand usually consisted of a mixture of mineral dust and biomass-burning smoke and should be classified completely as polluted dust [Tesche et al., 2011a]. The CALIPSO retrieval fails to

2894


Figure 3. Height profiles of the backscatter coefficients at (a) 532 and (b) 1064 nm, the (c) extinction coefficient and (e) lidar ratio at 532 nm, the (e) backscatter-related Ångström exponent for the wavelength pair 532/1064 nm, and the (f) particle depolarization ratio at 532 nm as measured with BERTHA (red) between 2133 and 2317 UTC on 14 June 2008 and CALIPSO (blue) during an overpass at 1528 UTC on 15 June 2008 about 478 km to the west of Praia, Cape Verde (see Table 3). Thin and thick lines denote unsmoothed and smoothed (660 m) profiles, respectively. Particle depolarization ratio profiles measured with BERTHA are compared to the ones given in the CALIPSO level 2 files (light blue in f) and calculated according to equation (2) (dark blue in Figure 3f). The dotted lines mark the vertical range used for a comparison of the measurements of the two instruments (see column 7 in Table 3). do so. However, this is due to an error in the CALIPSO retrieval software that produces erroneous particle depolarization ratio profiles (Mark Vaughan, personal com-

munication). This point will be discussed in more detail later in the text. Note that problems in separating dust from polluted dust could also originate from an improper choice

Figure 4. Same as Figure 2 but for CALIPSO measurements between (b–e) 8ı and 18ı N during an overpass at 0328 UTC on 11 February 2008 about 46 km west of Praia. 2895


Figure 5. Same as Figure 3 but for the comparison of the BERTHA measurement averaged between 2332 and 0442 UTC on 10/11 February 2008 and a CALIPSO overpass at 0328 UTC on 11 February 2008 about 46 km to the west of Praia (see Table 3). of the threshold of the particle depolarization ratio used for this purpose. [22] The particle backscatter coefficients in Figure 4d show increased values in the center of the elevated aerosol plume and at the top of the boundary layer where lots of clouds are identified (light blue in Figure 4b). The particle depolarization ratio plot reveals that the boundary layer aerosol was weakly depolarizing and therefore classified as marine. The elevated plume shows much higher values, even though these seem to be dominated by noise. Again, the contrast between the two layers is best visible where they are in direct contact. [23] Figure 5 gives a better view in the quantitative findings of the CALIPSO measurements between the white lines in Figure 4. BERTHA measurements between 2332 and 0442 UTC in the night from 10 to 11 February 2008 are compared to CALIPSO observations during the overpass at a distance of about 46 km from the ground site around 0328 UTC on 11 February 2008 (see Table 3 and Figure 1). Note that very homogeneous aerosol conditions existed during this time period. Within the height range of comparison, slight overestimation is found for the backscatter coefficients at both 532 and 1064 nm but with good agreement of the wavelength dependence of the two parameters. The lidar ratio shown in Figure 5d is a mean value calculated from profiles that were classified as mineral dust and polluted dust (see Fig. 4c) and is 25–30% smaller than the BERTHA measurements. Consequently, the 532 nm extinction coefficient from CALIPSO is slightly lower than the BERTHA observations. The underestimation of the lidar ratio together with a reasonable agreement of the backscatter coefficient again points to a significant impact of multiple scattering by large coarse-mode particles in the case of the spaceborne lidar measurements [Wandinger et al., 2010]. [24] Special emphasis should be drawn to the particle depolarization ratios in Figure 5f. BERTHA measurements give low values around 0.05 for the top of the marine boundary layer. The elevated dust/smoke layer shows two regions that are dominated by dust (2.6–3.4 km, ı = 0.2) and smoke (3.4–4.6 km, ı = 0.06–0.10). The CALIPSO values calculated according to equation (2) are in good agreement

with BERTHA observations in the lower part of the elevated layer. Even though they do not show the decrease in the particle depolarization ratio with height that was observed in the ground-based measurements, the calculated values cover a realistic range. Realistic values are also found for the lowermost layer for which accurate CALIPSO depolarization ratio measurements are usually only available during night. The particle depolarization ratios provided in the CALIPSO files are similar to the calculated one within the lowermost layer. In the elevated aerosol plume, however, they tend to be significantly larger. The discrepancy between the two particle depolarization ratios will be discussed in more detail later in the text.

5. General Findings and Discussion [25] An overview of the comparison between measurements of BERTHA and CALIPSO for the 15 cases considered in this study can be found in Table 5. The table presents the slopes and correlation coefficients (linear fits through zero) from a comparison of results of BERTHA and CALIPSO within the height sections for comparison given in Table 3. Here, we give a discussion of the findings for the individual parameters. 5.1. Backscatter and Extinction Coefficients, Ångström Exponents [26] We generally find good agreement between 532 nm backscatter coefficient measurements with BERTHA and CALIPSO with R2 of 0.88–0.99 and an average slope of 1.000 and 1.146 for the considered winter and summer cases, respectively. Similarly good agreement was found for the 1064 nm backscatter coefficient for the winter cases while the three comparisons in summer showed that CALIPSO overestimates 1064 nm backscatter of pure dust by on average 50%. Note that due to the different aerosol conditions during winter (dust/smoke mixture) and summer (pure dust) over Cape Verde, such a difference in the performance of the CALIPSO 1064 nm backscatter coefficient is likely due to the changing choice of the lidar ratio in the retrieval (see Table 1). However, calibrating the

2896

TESCHE ET AL.: SAMUM-2 CALIPSO COMPARISONS Table 5. Parameters Derived from Linear Fits Through Zero (CALIPSO = a BERTHA) for Correlations of Measurements of BERTHA and CALIPSO for the Times, Along-Track Latitude Ranges, and Height Intervals Given in Table 3a

1064 nm signal still poses a big challenge to the CALIPSO Science Team. The disagreement in the backscatter coefficients amplifies in the calculation of the Ångström exponent. Consequently, even though reasonable values are obtained, this parameter shows a varying behavior with both underand overestimation of the CALIPSO measurements during winter. In case of the three summer comparisons, the overestimation in 1064 nm backscatter coefficients causes an underestimation in the 532/1064 nm backscatter-related Ångström exponent. [27] As discussed by Wandinger et al. [2010], the 532 nm dust lidar ratio of 40 sr is too low compared to measurements during SAMUM [Tesche et al., 2009a, 2011a] and points to multiple-scattering effects. Note that multiple scattering is mainly caused by the size of the mineral dust particles (i.e., the width of the forward scattering peak in the phase function) and does not primarily depend on AOD.

Using the value of 40 sr in the calculation of the extinction coefficient of pure dust leads to a significant underestimation of this parameter by up to 30%. For the extinction coefficient, slopes close to unity are only found in cases of an overestimation of the 532 nm backscatter coefficient. However, the CALIPSO 532 nm particle backscatter coefficients mostly agree with our ground-based measurements and observed differences are caused by varying dust conditions during the observations of BERTHA and CALIPSO (i.e., temporal and/or spatial inhomogeneities). Wandinger et al. [2010] proposed to calculate the 532 nm extinction coefficient with the mean dust lidar ratio of 55 sr measured during the SAMUM campaigns [Tesche et al., 2009a, 2011a]. [28] A view of the integrated backscatter and extinction coefficients in Figure 6 gives a similar impression as the findings in Table 5. The integrated 532 nm backscatter coef-

Figure 6. Comparison of integrated backscatter ((a) 532 nm and (c) 1064 nm) and extinction coefficients ((b) 532 nm) for the days and height intervals given in Table 3. Winter and summer cases are shown as white and black dots, respectively. The dashed black lines gives the 1 : 1 line. Black lines and numbers represent the linear correlation (forced through zero) of the shown data points. The gray line and numbers in Figure 6b show the linear correlation between measurements of BERTHA and CALIPSO, if summertime CALIPSO-derived extinction values are multiplied by a factor of 55/40 to account for the mean dust lidar ratio measured during SAMUM. 2897


Figure 7. Particle depolarization ratios at 532 nm as measured with BERTHA (red), taken from CALIPSO version 3.01 profile data (light blue), and calculated from CALIPSO level 2 backscatter profiles according to equation (2) (dark blue) for overpasses on (a) 12 February 2008 and (b) 3 June 2008. Thin lines give unsmoothed data while bold lines refer to a smoothing length of 660 m.

uct in the same CALIPSO file (version 3.01). They are far less noisy than the CALIPSO depolarization-ratio product and furthermore in better agreement with the BERTHA measurements. This can be seen clearly in Figures 5f and 7. The last columns in Table 5 also show that almost perfect correlation was found, if BERTHA measurements were compared to depolarization ratios calculated after equation (2) instead of the ones provided in the CALIPSO profile file. [30] The same is also evident in the comparison of the two CALIPSO-derived depolarization ratios in Figure 8. The calculated values were obtained by applying equation (2) to mean values within the selected latitudinal interval. The largest changes occur for particle depolarization ratios above 0.20 in winter and 0.25 in summer. Depolarization ratios provided in the CALIPSO files often show unrealistically high values. Calculating the particle depolarization ratio first and averaging second do not lead to significantly different results. Hence, the disagreement in CALIPSO particle depolarization ratios cannot be attributed to the treatment of single-shot profiles in the CALIPSO retrieval algorithm. The reason for the bias reported here is currently investigated by the CALIPSO Science Team

ficients are highly correlated and show a slope close to unity. The weak variability is attributed to actual changes in aerosol conditions. Combined winter and summer measurements show that the CALIPSO retrieval seems to overestimate the 1064 nm backscatter coefficient by 30–40%. However, this tendency is dominated by the summer measurements (high values, a = 1.433, R2 = 0.985) while winter measurements (low values, a = 1.179, R2 = 0.986) alone show reasonable agreement. CALIPSO 532 nm extinction coefficients calculated with the dust lidar ratio of 40 sr are lower than the ones from the BERTHA measurements (dots and black line in Figure 6b). If the values are corrected for the use of a too low dust lidar ratio (i.e., multiplied by 55/40, grey line in Figure 6b), the agreement of BERTHA and CALIPSO findings is improved to a slope similar to the one for the 532 nm backscatter coefficient, and the remaining variability (difference) is likely due to varying aerosol conditions. 5.2. Depolarization Ratios [29] The linear particle depolarization ratio is of crucial importance for the identification of mineral dust in lidar observations and reliable measurements of this parameter are needed to facilitate proper aerosol-type classification or improve present classification schemes [Burton et al., 2012]. Measurements during SAMUM showed values of 0.31˙0.03 for layers of pure dust over Morocco [Freudenthaler et al., 2009] and Cape Verde [Tesche et al., 2011a]. Similar values were also found in laboratory studies [Sakai et al., 2010]. From the comparison of CALIPSO observations to ground-based measurements, we found that particle depolarization ratios calculated from the total and perpendicular backscatter coefficients by means of equation (2) are more reliable than the ones given as a final prod-

Figure 8. Comparison of the 532 nm particle depolarization ratios taken from the CALIPSO level 2 version 3.01 files and calculated from CALIPSO level 2 version 3.01 backscatter coefficients according to equation (2) for the days given in Table 3.

2898


Figure 9. Relative contributions of dust (red) and smoke (green) to the 532 nm backscatter coefficient as determined using depolarization-ratio-based aerosol-type separation for data from BERTHA (thin) and CALIPSO (bold) for the comparison days during winter 2008. and is due to a software error in the CALIPSO retrieval algorithm (Mark Vaughan, personal communication). The particle depolarization ratio is obtained in the retrieval after calculating the total and perpendicular backscatter coefficients. Until the bias is resolved in the next data release, the particle depolarization ratio given in the CALIPSO files should not be used but calculated according to equation (2) instead. 5.3. Aerosol-Type Identification and Separation [31] As a last point, we want to take a look at the detection of aerosol types in the CALIPSO retrieval since this is crucial for the selection of a proper lidar ratio [Omar et al., 2009]. Groß et al. [2011] and Tesche et al. [2011a] show that the boundary layer over Cape Verde consisted of marine aerosol with traces of dust during winter and pure marine aerosol during summer. The CALIPSO aerosol subtype mask (see, e.g., Figures 2c and 4b) also identifies clean marine aerosol as being dominant in the boundary layer. However, there are also profiles for which the aerosol in the boundary layer was identified as polluted dust, dust, or even polluted continental during both winter and summer. [32] The elevated layer above the marine boundary layer consisted of pure dust during summer and a mixture of dust and smoke during winter [Tesche et al., 2011a]. The summer comparison cases showed that the CALIPSO retrieval selects dust to be the dominant aerosol type. However, there

are also regions for which the aerosol was classified as polluted dust, smoke, clean continental, and even clean marine. For the more complicated winter cases, dust was dominating as well followed by polluted dust, smoke, and polluted continental. [33] Using a method based on accurate measurements of the particle depolarization ratio, Tesche et al. [2009b, 2011b] derived the contribution of dust and smoke aerosol to the 532 nm particle backscatter coefficient measured during the SAMUM winter campaign. Encouraged by the investigation of the CALIPSO particle depolarization ratio presented in section 5.2, we tested whether the same approach can also be successfully applied to CALIPSO measurements. We used the particle depolarization ratio calculated according to equation (2) for the aerosol-type separation and the assumptions by Tesche et al. [2009b]. Figure 9 shows the results of applying the aerosol-type separation to measurements of BERTHA and CALIPSO for the eight comparison cases in winter 2008. Since the particle depolarization ratio is the most crucial input parameter, the outcome of the procedure is strongly dependent on the accuracy of this parameter. Noisy profiles like the ones for 25 January and 5 February 2008 lead to unreliable findings of the contributions of dust and smoke. For the other days, however, the agreement is good and suggests that CALIPSO observations can be used to derive more detailed information on dust/smoke contributions than the one currently provided by

2899


the label polluted dust (as a mixture of dust and smoke). A straightforward application would be to assess the role of the dust in the mixture, e.g., as dominating (dust contribution >60%), balanced (dust contribution of 40–60%), or secondary (dust contribution 0.20) from less depolarizing aerosol types. The classification of pure dust layers as polluted dust suggests that the approximated particle depolarization ratio still contains sig-

nificant molecular contributions and, thus, is lower than the actual particle depolarization ratio. This holds especially in case of low backscatter ratios (ratio of total, i.e., particles + molecules, to molecular backscatter). Reliable aerosol classification needs the utilization of intensive properties like the particle depolarization ratio for a discrimination between pure mineral dust and mixtures of dust and other aerosol types (referred to as polluted dust in the CALIPSO data retrieval). Aerosol-type separation for two-component mixtures as suggested by Tesche et al. [2009b] can be performed for CALIPSO measurements if quality-assured particle depolarization ratio profiles (i.e., values calculated according to equation (2) rather than the ones given in the CALIPSO level 2 version 3.01/3.02 files) are used and if the kind of contributing aerosols is known. For regions of increased particle depolarization ratios below the threshold value for pure dust, the information on geographical location, height (boundary layer or free troposphere), and time (which transport season) could be used to estimate the other aerosol types that most likely contribute to the dust mixture. [37] In the following, we would like to propose some improvements to the currently non-recurring CALIPSO algorithm by introducing a second retrieval stage. Note that we do not account for the feasibility of these suggestions or the resulting increase in computational cost. Reliable 532 nm backscatter coefficients are obtained with the current retrieval even for problematic aerosol conditions with dust and/or polluted dust. These could be used (1) to obtain the actual particle depolarization ratio (according to equation (2) or by the CALIPSO retrieval after resolving the bias described in section 5.2). Comparisons to ground-based measurements presented in this paper showed that such CALIPSO particle-depolarization ratios are reliable. Hence, they could be used (2) for a refined aerosol classification (i.e., a quality-assured aerosol-subtype mask) that should be able to properly distinguish desert dust from polluted dust. This second classification could furthermore (3) include a demand for homogeneous layers (in terms of aerosol type) rather than performing a profile-by-profile analysis. Afterward, the refined classification could (4) be used together with the mean dust lidar ratio of 55 sr measured during SAMUM for the calculation of the dust extinction coefficients. Note that the latter step holds for dust from source regions in western Africa but would not be applicable on a global scale, if the dust lidar ratio was showing a regional variability [Schuster et al., 2012]. A discussion of the variability of dust lidar ratios from lidar measurements is provided by, e.g., Müller et al. [2007], Omar et al. [2009], and Tesche et al. [2009a]. Finally, (5) the particledepolarization-ratio-based aerosol-type separation could be used to gain further insight into the contribution of desert dust to layers that were characterized as polluted dust.

6. Summary [38] We presented a validation of CALIPSO level 2 version 3.01 aerosol products for dust and smoke measured in the vicinity of Cape Verde with ground-based lidar measurements during SAMUM-2. For the 15 considered cases, it was found that the CALIPSO retrieval works best for the 532 nm backscatter coefficient. Differences to groundbased observations of this parameter are most likely due to

2900


aerosol inhomogeneities rather than retrieval errors. It was found that using the effective dust lidar ratio of 40 sr for the backscatter retrieval rather than the observed mean value of 55 sr in the calculation of the 532 nm extinction coefficient leads to an underestimation of this parameter by up to 30%. Introducing a second lookup table for the extinction retrieval can most likely resolve this discrepancy. 1064 nm backscatter coefficients from CALIPSO were in good agreement to ground-based values in case of the mixed dust/smoke layers during winter. CALIPSO overestimates the 1064 nm backscatter coefficient by up to 50% within the pure dust layers observed during summer. However, only three cases were available for this comparison and the parameter is rarely used in CALIPSO-based investigations available in the literature. [39] The CALIPSO depolarization ratio profiling was found to be of high accuracy as long as particle depolarization ratios are calculated from the perpendicular and total particle backscatter coefficients in the level 2 version 3.01 aerosol profile files. Values provided in the same file were found to be larger than both ground-based and calculated values and furthermore much noisier than the latter. The reason for this effect is not yet resolved but under investigation. The good agreement of calculated CALIPSO particle-depolarization ratios to ground-based measurements suggests that this parameter should be used for an improved aerosol-type classification. In case of aerosol layers labeled as polluted dust, the geographical information could be used together with the particle-depolarization ratio to gain further insight into the dust contribution within the respective layers by means of a depolarization-ratio-based aerosol-type separation. [40] Measures for an improvement of the CALIPSO retrieval by introducing a second loop to the currently nonrecurring algorithm were suggested. This could improve results of the 532 nm extinction coefficients for mineral dust, the 532 nm particle depolarization ratio, and the aerosol classification. [41] Acknowledgments. We like to thank Mark Vaughan for his help and criticism regarding the present paper. The SAMUM research group was funded by the Deutsche Forschungsgemeinschaft (DFG) under grant FOR 539. CALIPSO data used in this study were obtained from the NASA Langley Research Center Atmospheric Science Data Center (http://eosweb.larc. nasa.gov).

References Althausen, D., D. Müller, A. Ansmann, U. Wandinger, H. Hube, E. Clauder, and S. Zörner (2000), Scanning six-wavelength eleven-channel aerosol lidar, J. Atmos. Oceanic Techn., 17, 1496–1482. Ångström, A. (1964), The parameters of atmospheric turbidity, Tellus, 16, 64–75. Ansmann, A., and D. Müller (2005), Lidar and atmospheric aerosol particles, in LIDAR—Range-Resolved Optical Remote Sensing of the Atmosphere, edited by Weitkamp, C., 105–141, Springer, New York, NY, USA. Ansmann, A., A. Petzold, K. Kandler, I. Tegen, M. Wendisch, D. Müller, B. Weinzierl, and T. Müller (2011), Saharan Mineral Dust Experiments SAMUM-1 and SAMUM-2: What have we learned? Tellus, 63B, 403–429, doi:10.1111/j.1600-08892011.00555.X. Ben-Ami, Y., I. Koren, and O. Altaratz (2009), Patterns of North African dust transport over the Atlantic: Winter vs. summer, based on CALIPSO first year data, Atmos. Chem. Phys., 9, 7867–7875, doi: 10.5194/acp-97867-2009. Ben-Ami, Y., I. Koren, Y. Rudich, P. Artaxo, S. T. Martin, and M. O. Andreae (2010), Transport of North African dust from the Bodélé

depression to the Amazon Basin: A case study, Atmos. Chem. Phys., 10, 7533–7544, doi:10.5194/acp-10-7533-2010. Burton, S. P., et al. (2010), Using airborne high spectral resolution lidar data to evaluate combined active plus passive retrievals of aerosol extinction profiles, J. Geophys. Res., 115, D00H15, doi:10.1029/2009JD012130. Burton, S. P., R. A. Ferrare, C. A. Hostetler, J. W. Hair, R. R. Rogers, M. D. Obland, C. F. Butler, A. L. Cook, D. B. Harper, and K. D. Froyd (2012), Aerosol classification using airborne high spectral resolution lidar measurements—Methodology and examples, Atmos. Meas. Tech., 5, 73–98, doi:10.5194/amt-5-73-2012. CALIPSO Users Guide (2011), Lidar Level 2 5 km Cloud and Aerosol Profile Products. Available at: http://www-calipso.larc.nasa.gov/resources/ calipso_users_guide/data_summaries/profile_data.php. Campbell, J. R., et al. (2012), Evaluating nighttime CALIOP 0.532 m aerosol optical depth and extinction coefficient retrievals, Atmos. Meas. Tech., 5, 2143–2160, doi:10.5194/amt-5-2143-2012. Chazette, P., J.-C. Raut, F. Dulac, S. Berthier, S.-W. Kim, P. Royer, J. Sanak, S. Loaëc, and H. Grigaut-Desbrosses (2010), Simultaneous observations of lower tropospheric continental aerosols with a ground-based, an airborne, and the spaceborne CALIOP lidar system, J. Geophys. Res., 115, D00H31, doi:10.1029/2009JD012341. Draxler, R. R., and G. D. Rolph (2010), HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://ready.arl.noaa.gov/HYSPLIT.php). NOAA Air Resources Laboratory, Silver Spring, MD. Devasthale, A., M. Tjernström, and A. H. Omar (2011), The vertical distribution of thin features over the Arctic analysed from CALIPSO observations part II: Aerosols, Tellus, 63B, 86–95, doi: 10.1111/j.16000889.2010.00517.X. Fernald, F. G. (1984), Analysis of atmospheric lidar observations: Some comments, Appl. Opt., 23, 652–653, doi:10.1364/AO.23.000652. Freudenthaler, V., et al. (2009), Depolarization-ratio profiling at several wavelengths in pure Saharan dust during SAMUM 2006, Tellus, 61B, 165–179, doi:10.1111/j.1600-0889.2008.00396.X. Groß, S., et al. (2011), Characterization of the planetary boundary layer during SAMUM-2 by means of lidar measurements, Tellus, 63B, 695– 705, doi:10.1111/j.1600-0889.2011.00557.X. Hunt, W. H., D. M. Winker, M. A. Vaughan, K. A. Powell, P. L. Lucker, and C. Weimer (2009), CALIPSO lidar description and performance assessment, J. Atmos. Oceanic Technol., 26, 1214–1228, doi: 10.1175/2009JTECHA1223.1. Kacenelenbogen, M., M. A. Vaughan, J. Redemann, R. M. Hoff, R. R. Rogers, R. A. Ferrare, P. B. Russell, C. A. Hostetler, J. W. Hair, and B. N. Holben (2011), An accuracy assessment of the CALIOP/ CALIPSO version 2/version 3 daytime aerosol extinction product based on a detailed multi-sensor, multi-platform case study, Atmos. Chem. Phys., 11, 3981–4000, doi:10.5194/acp-11-3981-2011. Kittaka, C., D. M. Winker, M. A. Vaughan, A. Omar, and L. A. Remer (2011), Intercomparison of column aerosol optical depths from CALIPSO and MODIS-Aqua, Atmos. Meas. Tech., 4, 131–141, doi: 10.5194/amt-4-131-2011. Klett, J. D. (1981), Stable analytical inversion solution for processing lidar returns, Appl. Opt., 20, 211–220, doi:10.1364/AO.20.000211. Kovacs, T. (2006), Comparing MODIS and AERONET aerosol optical depth at varying separation distances to assess ground-based validation strategies for spaceborne lidar, J. Geophys. Res., 111, D24203, doi: 10.1029/2006JD007349. Liu, D., Z. Wang, Z. Liu, D. Winker, and C. Trepte (2008a), A height resolved global view of dust aerosols from the first year CALIPSO lidar measurements, J. Geophys. Res., 113, D16214, doi: 10.1029/2007JD009776. Liu, Z., et al. (2008b), CALIPSO lidar observations of the optical properties of Saharan dust: A case study of long-range transport, J. Geophys. Res., 113, D07207, doi:10.1029/2007JD008878. Liu, Z., R. Kuehn, M. Vaughan, D. Winker, A. Omar, K. Powell, C. Trepte, Y. Hu, and C. Hostetler (2010), The CALIPSO cloud and aerosol discrimination: Version 3 algorithm and test results, Proceedings of the 25th International Laser Radar Conference, 5–9 July 2010, St. Petersburg, Russia, 155–158. Liu, Z., D. Winker, A. Omar, M. Vaughan, C. Trepte, Y. Hu, K. Powell, W. Sun, and B. Lin (2011), Effective lidar ratios of dense dust layers over North Africa derived from the CALIOP measurements, J. Quant. Spectr. Rad. Transfer, 112, 204–213, doi:10.1016/j.jqsrt.2010. 05.006. Mamouri, R. E., V. Amiridis, A. Papayannis, E. Giannakaki, G. Tsaknakis, and D. S. Balis (2009), Validation of CALIPSO space-borne-derived attenuated backscatter coefficient profiles using a ground-based lidar in Athens, Greece, Atmos. Meas. Tech., 2, 513–522, doi: 10.5194/amt-2513-2009.

2901

TESCHE ET AL.: SAMUM-2 CALIPSO COMPARISONS McGill, M. J., M. A. Vaughan, C. R. Trepte, W. D. Hart, D. L. Hlavka, D. M. Winker, and R. Kuehn (2007), Airborne validation of spatial properties measured by the CALIPSO lidar, J. Geophys. Res., 111, D20201, doi: 10.1029/2007JD008768. Mielonen, T., A. Arola, M. Komppula, J. Kukkonen, J. Koskinen, G. de Leeuw, and K. E. J. Lehtinen (2009), Comparison of CALIOP level 2 aerosol subtypes to aerosol types derived from AERONET inversion data, Geophys. Res. Lett., 36, L18804, doi:10.1029/2009GL039609. Mona, L., G. Pappalardo, A. Amodeo, G. D’Amico, F. Madonna, A. Boselli, A. Giunta, F. Russo, and V. Cuomo (2009), One year of CNRIMAA multi-wavelength Raman lidar measurements in coincidence with CALIPSO overpasses: Level 1 products comparison, Atmos. Chem. Phys., 9, 7213–7228, doi:10.5194/acp-9-7213-2009. Müller, D., A. Ansmann, I. Mattis, M. Tesche, U. Wandinger, D. Althausen, and G. Pisani (2007), Aerosol-type-dependent lidar ratios observed with Raman lidar, J. Geophys. Res., 112, D16202, doi: 10.1029/2006JD008292. Omar, A. H., J.-G. Won, D. M. Winker, S.-C. Yoon, O. Dubovik, and M. P. McCormick (2005), Development of global aerosol models using cluster analysis of Aerosol Robotic Network (AERONET) measurements, J. Geophys. Res., 110, D10S14, doi:10.1029/2004JD004874. Omar, A. H., et al. (2009), The CALIPSO automated aerosol classification and lidar ratio selection algorithm, J. Atmos. Oceanic Technol., 26, 1994–2014, doi:10.1175/2009JTECHA1231.1. Pappalardo, G., et al. (2010), EARLINET correlative measurements for CALIPSO: First intercomparison results, J. Geophys. Res., 115, D00H19, doi:10.1029/2009JD012147. Redemann, J., M. A. Vaughan, Q. Zhang, Y. Shinozuka, P. B. Russell, J. M. Livingston, M. Kacenelenbogen, and L. A. Remer (2012), The comparison of MODIS-Aqua (C5) and CALIOP (V2 & V3) aerosol optical depth, Atmos. Chem. Phys., 12, 3025–3043, doi: 10.5194/acp-12-30252012. Rogers, R. R., et al. (2011), Assessment of the CALIPSO-lidar 532nm attenuated backscatter calibration using the NASA LaRC airborne high spectral resolution lidar, Atmos. Chem. Phys., 11, 1295–1311, doi: 10.5194/acp-11-1295-2011. Sakai, T., T. Nagai, Y. Zaizen, and Y. Mano (2010), Backscattering linear depolarization ratio measurements of mineral, sea-salt, and ammonium sulfate particles simulated in a laboratory chamber, Appl. Opt., 49, 4441–4449, doi:10.1364/AO.49.004441. Schepanski, K., I. Tegen, and A. Macke (2009), Saharan dust transport and deposition towards the tropical northern Atlantic, Atmos. Chem. Phys., 9, 1173–1189, doi:10.5194/acp-9-1173-2009. Schuster, G. L., M. Vaughan, D. MacDonnell, W. Su, D. Winker, O. Dubovik, T. Lapyonok, and C. Trepte (2012), Comparison of

CALIPSO aerosol optical depth retrievals to AERONET measurements, and a climatology for the lidar ratio of dust, Atmos. Chem. Phys., 12, 7431–7452, doi:10.5194/acp-12-7431-2012. Tesche, M., et al. (2009a), Vertical profiling of Saharan dust with Raman lidars and airborne HSRL in southern Morocco during SAMUM, Tellus, 61B, 144–164, doi:10.1111/j.1600-0889.2008.00390.X. Tesche, M., A. Ansmann, D. Müller, D. Althausen, R. Engelmann, V. Freudenthaler, and S. Groß (2009b), Vertically resolved separation of dust and smoke over Cape Verde by using multiwavelength Raman and polarization lidars during Saharan Mineral Dust Experiment 2008, J. Geophys. Res., 114, D13202, doi:10.1029/2009JD011862. Tesche, M., A. Ansmann, D. Müller, D. Althausen, R. Engelmann, V. Freudenthaler, S. Groß, and M. Esselborn (2011a), Profiling of Saharan dust and biomass burning smoke with multiwavelength polarization Raman lidar at Cape Verde, Tellus, 63B, 649–676, doi: 10.1111/j.16000889.2011.00548.X. Tesche, M., A. Ansmann, D. Müller, D. Althausen, V. Freudenthaler, S. Groß, M. Esselborn, B. Weinzierl, and A. Petzold (2011b), Microphysical properties of biomass-burning smoke over Cape Verde as inferred from multiwavelength polarization Raman lidar measurements, Tellus, 63B, 677–694, doi: 10.1111/j.1600-0889.2011. 00549.X. Vaughan, M. A., K. A. Powell, R. E. Kuehn, S. A. Young, D. M. Winker, C. A. Hostetler, W. H. Hunt, Z. Liu, M. J. McGill, and B. J. Getzewich (2009), Fully automated detection of cloud and aerosol layers in the CALIPSO lidar measurements, J. Atmos. Oceanic Technol., 26, 2034–2050, doi:10.1175/2009JTECHA1228.1. Vaughan, M., et al. (2010), Strategies for improved CALIPSO aerosol optical depth estimates, Proceedings of the 25th International Laser Radar Conference, 5–9 July 2010, St. Petersburg, Russia, 1340–1343. Wandinger, U., M. Tesche, P. Seifert, A. Ansmann, D. Müller, and D. Althausen (2010), Size matters: Influence of multiple scattering on CALIPSO light-extinction profiling in desert dust, Geophys. Res. Lett., 37, L10801, doi:10.1029/2010GL042815. Winker, D. M., M. A. Vaughan, A. Omar, Y. Hu, K. A. Powell, Z. Liu, W. H. Hunt, and S. A. Young (2009), Overview of the CALIPSO mission and CALIOP data processing algorithms, J. Atmos. Oceanic Technol., 26, 2310–2323, doi:10.1175/2009JTECHA1281.1. Young, S. A. (1995), Analysis of lidar backscatter profiles in optically thin clouds, Appl. Opt., 34, 7019-7031, doi:10.1364/AO.34. 007019. Young, S. A., and M. A. Vaughan (2009), The retrieval of profiles of particulate extinction from Cloud-Aerosol Lidar Infrared Pathfinder Satellite Observations (CALIPSO) data: Algorithm description, J. Atmos. Oceanic Technol., 26, 1105–1119, doi:10.1175/ 2008JTECHA1221.1.

2902