The SALTRACE field campaigns performed in the summer of 2013 and in the winter and summer of 2014 belong to the SAMUM–SALTRACE field campaign series. As shown and illustrated in Fig. 7 of , six comprehensive dust field campaigns have been conducted since 2006: SAMUM-1, SAMUM-2a and 2b, and SALTRACE-1, 2, and 3. The first SAMUM project, SAMUM-1 , took place in southern Morocco (May–June 2006) to investigate the role of freshly emitted dust in the climate system. SAMUM-2 Cabo Verde; SAMUM-2a, January–February 2008; SAMUM-2b, May–June 2008; investigated the dust physicochemical, optical, and radiative properties of mixtures of biomass burning smoke and mineral dust (SAMUM-2a, winter transport regime) and of pure dust (SAMUM-2b, summer transport mode) after an atmospheric transport over 1000–3000 km (1–3 days after emission). During SALTRACE, we investigated the dust properties after atmospheric travel over 5–12 days and 5000–8000 km .

As the logistically favorable field site for lidar observations, the Caribbean Institute for Meteorology and Hydrology was selected (CIMH, 13.1∘ N, 59.6∘ W, 110 m above sea level, m a.s.l.), located in Husbands, in the northern part of the capital Bridgetown on the west coast of Barbados. The station is not influenced by any local (island) anthropogenic pollution because of the steady northeasterly airflow and the absence of strong pollution sources in the northern part of Barbados, upwind of the lidar station. In the summer months of June and July (SALTRACE-1 and 3), transported dust layers were observed. The SALTRACE lidar activities were complemented by shipborne observations along the main Saharan dust transport route over the tropical North Atlantic in April–May 2013 ; in situ observations of microphysical (size distribution, mass concentration, particle shape), chemical, and optical aerosol properties at Ragged Point (20 km east of CIMH) at the east coast of Barbados ; and airborne in situ aerosol observations and Doppler lidar measurements of aerosol layering and atmospheric wind fields . The SALTRACE in situ observations include studies of the efficacy of aged desert dust to serve as cloud condensation nuclei as well as modeling studies of dust transport across the Atlantic and the impact of the Caribbean island on the airflow and downward mixing of dust .

In , the dust transport from Africa towards the Caribbean is discussed. The main features of dust layering across the Atlantic described by the conceptual model are illuminated and compared with the shipborne SALTRACE lidar observations. According to the conceptual model hot, dry, dust-laden air masses emerge from the western coast of Africa as a series of large-scale pulses in the summer months. Associated with easterly wave activity, Saharan dust outbreaks occur as discrete episodic pulses, which generally last 3–5 days. These dust outbreaks are mostly confined to a well-mixed layer, the Saharan air layer (SAL), that often extends to 5–6 km in height over West Africa due to intense solar heating in summer months. The airborne dust is carried westward by the prevailing easterly flow in the latitude belt of 10–25∘ N. As the dust plumes are advected further west in the predominantly easterly flow, the base of the SAL rises rapidly as it is undercut by the relatively clean northeasterly trade winds. The well-mixed SAL resides above the trade wind inversion layer which is on top of the marine aerosol layer (MAL). The dust transport takes usually 5–7 days across the Atlantic. The strong temperature inversion at the base of the SAL limits convective activity and consequently precludes the possibility of strong wet deposition, except during periods with deep convection and precipitation.

We begin with a short historical overview of polarization lidar observations of tropospheric dust. The polarization lidar technique was applied to tropospheric aerosols for the first time in the early 1970s . Systematic measurements of the depolarization ratio in desert dust layers started in the 1980s in eastern Asia and demonstrated the importance of polarization lidar for dust monitoring. Consequently, the Asian Dust Lidar Network was established in the 1990s . In Europe, systematic Saharan dust studies with polarization lidars began in the 1990s . Dust was investigated later on by means of polarization lidar in the framework of the European Aerosol Research Lidar Network e.g.,. All of the ground-based observations, conducted before SALTRACE, were based on single-wavelength lidar observations, although several single-wavelength-lidar systems were combined during dedicated field campaigns such as SAMUM-1 and 2 . In the majority of applications the laser wavelength was 532 nm. presented the first ground-based dual-wavelength polarization lidar observations of dust performed at 532 and 1064 nm. Airborne dual-wavelength polarization lidar observations (at 532 and 1064 nm) were realized in Saharan dust aboard the Falcon of the Germany Aerospace Center during SAMUM-1 in 2006 . During SALTRACE, dual-wavelength polarization lidars were operated simultaneously at 355 and 532 nm . And recently the first triple-wavelength polarization lidars were developed and performed dust observations at 355, 532, and 1064 nm aboard an aircraft and on the ground .

The schematic structure of the lidar system BERTHA is shown in Fig. . Two Nd:YAG lasers transmit linearly polarized laser pulses at 355 and 1064 nm (first laser) and at 532 nm (second laser) with a repetition rate of 30 Hz. Two linear polarizers are installed behind the laser and before the beam expander to further clean the polarization of the outgoing light. The pulse energies can be as high as 1000 mJ (1064 nm), 800 mJ (532 nm), and 120 mJ (355 nm) in the ideal case of well-working optical elements in the transmission unit of the lidar. However, the pulse energies were only about 50 % of these maximum values during the SALTRACE campaigns. Two lasers are used for two reasons: firstly to have a frequency-stabilized 532 nm laser for the implementation of the HSRL-channel and secondly to have a backup laser in the field campaign. The laser beams are aligned on an optical axis and directed through a beam expander. The beams are expanded tenfold and afterwards pointed into the atmosphere at an off-zenith angle of 5∘. By using a tilt angle of > 3∘, specular reflection by falling and horizontally aligned ice crystals no longer influences the measurements, as our experience shows.

A Cassegrain telescope with 53 cm diameter collects the backscattered light. The receiver field of view is 0.8 mrad.

The receiver unit was completely re-designed to measure the linear depolarization ratio at all three wavelengths. In addition, a high spectral resolution (HSR) channel was added (see Fig. ). All detectors are photomultiplier tubes (PMTs) working in the photon counting mode (H10721P-110 from Hamamatsu). But for the 1064 nm channels the PMT R3236 from Hamamatsu is used at a temperature below -30 ∘C to reduce signal noise in the near infrared. The elastic backscatter signals (total signals) as well as the so-called cross-polarized signal components are detected at the three emitted wavelengths (355, 532, 1064 nm). Polarization filters, each adjusted orthogonal to the plane of linear polarization of the outgoing laser pulses, are placed in front of the detectors which enable the detection of the cross-polarized laser radiation at the three wavelengths. For these six elastic channels, interference filters with 1 nm FWHM (full width at half maximum) are placed in front of each PMT. The vibrational–rotational Raman signals at 387 and 607 nm (nitrogen) and 407 nm (water vapor, night time only) are detected as well. Interference filters with 3 nm FWHM are used at the Raman wavelengths. A double-grating monochromator enables the detection of pure rotational Raman signals from nitrogen and oxygen (J0, J6, and J12) from the 532 nm emission wavelength. Neutral density filters are placed in front of each detector to adjust the signal to the linear range of the detector to avoid dead time effects of the photomultipliers. Nevertheless the PMTs are tested in the lab to correct high counting rates for dead time effects if necessary .

The signals are detected with a range resolution of 7.5 m and a time resolution of 5 to 30 s. A camera is used to visualize the overlap between the 532 nm laser beam and the receiver field of view (RFOV). The camera is permanently placed in the position of a receiving channel see. Complete overlap is reached at 800–1000 m for the 532 nm related channels and approximately at 1500 m for the 355 nm related channels. The 355 and 532 nm backscatter coefficient is derived from the ratio of the elastic backscatter signal to the respective nitrogen Raman signal (387 or 607 nm) and is therefore not much affected by the overlap profile. Even at 1064 nm we used the nitrogen Raman signal at 387 nm as the reference signal in the 1064 nm backscatter retrieval. Both signals are caused by backscattered laser-1 photons (see Fig. ) so that almost the same overlap characteristics hold for both signals, and the overlap effects widely cancel out when forming the signal ratio. We used the sun photometer observations (Sect. ) of the spectral slope of aerosol extinction for the correction of minor differential (387 nm vs. 1064 nm) particle extinction effects in the backscatter retrieval. On clear (dust-free) days we checked the overlap function according to . Trustworthy results can be obtained at altitudes above about 400 m for the backscatter coefficient and above about 1000 m for the extinction coefficient, derived from the 387 and 607 nm nitrogen Raman signal profiles. To control the correct alignment, a telecover test with eight segments (four inner and four outer) has been performed during SALTRACE-2 and in Leipzig in the autumn of 2014.

The focus of this article is on depolarization-ratio observations. In order to ensure a high quality of the data, the polarization sensitivity of the lidar system was characterized carefully. The polarization-sensitive transmission of the optical elements in the emitter and the receiver has been characterized. A detailed description of the polarization characterization can be found in Appendix .

The basic lidar-derived quantity is the volume linear depolarization ratio defined as the ratio of cross- to co-polarized signal components . The prefixes “co” and “cross” denote the planes of polarization (for which the receiver channels are sensitive) parallel and orthogonal to the plane of linear polarization of the transmitted laser pulses, respectively. In the case of BERTHA we measure the cross-polarized and total (cross + co-polarized) signal components and thus determine the co-polarized signal component from the cross-polarized and total signal components (more details are given in the Appendix).

The volume depolarization ratio at 355 and 532 nm is influenced by light depolarization by air molecules, aerosol, and cloud particles. To obtain the particle depolarization ratio a correction for molecular depolarization effects has to be applied . To account for molecular backscatter, extinction and depolarization contributions to the measured lidar signals we used the SALTRACE radiosonde observations at CIMH, which were performed twice a day (Sect. ). The radiosonde height profiles of air temperature and pressure profiles permit the computation of the actual height profile of air molecule number concentration over Barbados.

The presence of cirrus is routinely used to check the consistence of the depolarization ratios at different wavelengths among each other. Because the size of the ice crystals is usually much larger than the laser wavelength, the measured optical properties are close together for the 355 to 1064 nm wavelength range. This can be used to evaluate the quality of the depolarization-ratio observations for each of the three wavelengths. Case study II in the next section will be an example for these routine checks.

An extended analysis of systematic uncertainties in the retrieved optical properties can be found in , , , and . The error bars of the retrieval products given in the next sections show standard deviations considering the overall uncertainty.

During the SALTRACE-1 campaign in the summer of 2013, the dual-wavelength polarization lidar POLIS (Portable Lidar System) of the Munich University was operated at CIMH, about 50 m north of the BERTHA lidar. POLIS is a well-designed and characterized six-channel polarization/Raman lidar and provides volume linear depolarization-ratio profiles at 355 and 532 nm with high accuracy . POLIS is used as the reference polarization lidar system in EARLINET calibration and quality-assurance activities. The deployment of both lidars at the same site was motivated by the fact that the 13-channel BERTHA lidar, which integrates the HSR lidar technique, Raman aerosol, water vapor, temperature profiling methods, and now in addition the multiwavelength depolarization-ratio profiling option in one system, is a complex lidar system with a large number of potential sources for uncertainties (see Appendix A). Therefore to avoid any risk and to guarantee a high-quality SALTRACE data set of multiwavelength depolarization-ratio profiles, we decided to run POLIS and BERTHA side by side during the entire SALTRACE-1 campaign at CIMH.

The full overlap of the laser beams of POLIS with the RFOV is at about 200 to 250 m above ground , so well within the marine boundary layer (MBL) and below the lofted Saharan air layer. The range resolution of the raw data is 3.75 m, the temporal resolution 5–10 s depending on atmospheric conditions. The repetition rate of the frequency doubled and tripled Nd:YAG laser is 10 Hz with a pulse energy of 50 mJ at 355 nm and 27 mJ at 532 nm (see Table  for more details).

Spaceborne lidar observations of the 532 nm particle linear depolarization ratio are used for comparison with the 532 nm particle depolarization ratios from the ground-based lidar observations during SAMUM-1, SAMUM-2, and SALTRACE. We analyzed all CALIOP observations for well-defined areas over Morocco (26–31∘ N, 3–8∘ W), in the Cabo Verde region (13–18∘ N, 21–26∘ W), and around Barbados (10–15∘ N, 55–60∘ W), performed in June 2013 (11–13 overpasses), July 2013 (15–17 overpasses), June 2014 (13–16 overpasses), and July 2014 (11–14 overpasses). We checked all day and night overpasses (scenes) for the presence of dust and averaged all dusty signal profiles within the defined areas. In order to retrieve the dust depolarization-ratio profiles for each overpass, only the observations characterized as dust from the CALIPSO subtype algorithm are used. In these profiles, the particle linear depolarization ratio is recalculated from L2 perpendicular and total backscatter profiles, to improve the accuracy compared to the original CALIPSO L2-Version 3 product, which has a known error . Furthermore, several quality control procedures and filtering criteria are applied in the data set as described in .

In the next step, we selected those height ranges (below 6 km height) of the monthly mean profiles in which the depolarization ratios were almost height-independent and computed the column-averaged 532 nm particle depolarization ratio for these specific height ranges. These column values are used for comparison in Sects. 4 and 5. The selected height ranges with almost height-independent depolarization ratios extended from 800–1000 to 3700–5600 m a.s.l. (Morocco), from 1500–2300 to 4800–5400 m a.s.l. (Cabo Verde), and from 2500 to 3200–4200 m a.s.l. (Barbados). At lower heights, contamination with aerosol pollution and/or marine particles caused a significantly lower depolarization ratio. Therefore these lower heights were not considered in the dust-related depolarization data analysis.

Three sun photometers were run during the SALTRACE campaign at CIMH lidar station in 2013. Besides two CIMEL sun–sky photometers of AERONET Aerosol Robotic Network; from TROPOS and the University of Valladolid (see Barbados_SALTRACE, ), an automatic sun–sky radiometer of the Meteorological Institute of the University of Munich measured the spectral aerosol optical thickness (AOT) and sky radiances . The photometers covered a wavelength range from 340 to 1640 nm. The TROPOS photometer was operated from June 2013 to July 2014 (with an interruption from October 2013 to February 2014 caused by a damage of the sun photometer). Another photometer of AERONET is installed at Ragged Point (east coast of Barbados) in the vicinity of the Barbados Cloud Observatory . The Ragged Point photometer has performed measurements since 2007.

As during the SAMUM-1 and 2 campaigns, we regularly performed radiosonde observations. The Vaisala RS92 radiosondes measuring height profiles of temperature, air pressure, relative humidity, wind speed, and direction up to heights above 20 km were launched around local noon (15:00–16:00 UTC, 11:00–12:00 local time, LT) and after sunset (23:00–24:00 UTC, 19:00–20:00 LT). In total 133 radiosonde ascends were conducted at CIMH, 56, 35, and 42 during the SALTRACE-1, 2, and 3 campaigns, respectively.

A strong and long-lasting dust outbreak occurred from 9 to 13 July 2013. Figure  shows the BERTHA observations of the lofted SAL in the evening of 10 July 2013 (19:15–20:45 LT). The African air mass crossed Barbados with 15–20 m s-1 wind speed from east to west according to the radiosonde profiles (Fig. a). The relative humidity ranged from 30 to 50 % in the dust layer between 1.75 and 4.6 km height (Fig. c). The moist marine aerosol layer (MAL), indicated by high relative humidity around 80–90 %, reached to 1.75 km height on this evening. The MBL is the convective part of the MAL and is often topped with trade wind cumuli. The MAL extends up to the base of the SAL, which coincides with the trade wind inversion zone. Downward mixing of dusty air into the upper part of the marine aerosol layer is visible in Fig. d (green colors between 0.5 and 1.5 km a.s.l.). This layer is also called the intermediate layer , due to its location between the convective boundary layer and the SAL. The heterogeneous structures in the 532 nm volume depolarization height–time display in Fig. d below 1.5 km height are caused by island effects. Differences in orography and heat release over land and ocean surfaces disturb the air mass flow in the lowest part of the atmosphere . Such vertical mixing features were not observed during the SALTRACE shipborne lidar observations over the open Atlantic in May 2013 . The backward trajectories in Fig.  at 3000 m height indicate dust uptake over desert areas of northwestern Africa so that contamination with anthropogenic pollution was probably low. The dusty air masses traveled 5–7 days across the Atlantic to Barbados.

Figure  presents the particle optical properties obtained with the conventional Raman lidar technique . Typical features of Saharan dust were observed . The backscatter and extinction coefficients at 355 and 532 nm are similar and the dust 1064 nm backscatter coefficient is significantly lower than the respective 532 nm backscatter coefficient in the SAL. The observed Saharan dust lidar ratios accumulate in the 50–60 sr range at 355 and 532 nm. Below 1.75 km height a mixture of marine particles and dust particles prevailed so that the lidar ratio decreased. For pure marine conditions, the lidar ratio would be close to 15–25 sr . The SAL AOT was about 0.3 at 355 and 532 nm on this day. Figure  shows the particle linear depolarization-ratio profiles obtained with POLIS and BERTHA on 11 July 2013, 00:00–00:45 UTC. The POLIS backscatter coefficients in Fig. a are computed by applying the Klett method with a dust lidar ratio of 55 sr within the SAL and 30–40 sr below the SAL . The Raman lidar method is used in the computation of the backscatter profiles from the BERTHA observations.

As can be seen, very good agreement is obtained regarding the volume linear depolarization ratio at 355 and 532 nm (Fig. b). In the computation of the particle depolarization ratio, the particle backscatter coefficients are required and cause further uncertainty. This impact is most sensitive at 355 nm. The apparent noise in the 355 nm particle depolarization-ratio profiles is caused by the backscatter coefficients used. At 355 and 532 nm the SAL mean particle depolarization ratio can be compared between BERTHA (0.25 and 0.28 at 355 and 532 nm, respectively) and POLIS (0.26 and 0.27). Both systems agree very well. Only BERTHA measured at 1064 nm, and derived a SAL mean particle depolarization ratio of 0.22. Below the SAL, all particle depolarization ratios decrease. The down-mixed dust prevents a decrease towards pure marine values of 0.02–0.03 .

In Fig. c, profiles of the ratio of particle depolarization ratios (355 nm over 532 nm in blue, 1064 nm over 532 nm in red) are shown. The mean values are 0.81 (1064 / 532 nm) and 0.88 (355 / 532 nm). The observed height independence of the particle depolarization ratios at all three wavelengths and of the less noisy ratio of the 1064-to-532 nm depolarization ratios implies vertically homogeneous dust size–shape characteristics. An impact of gravitational settling leading to a decrease of coarse-mode dust concentration in the SAL top region after 5–10 days of travel, which would show up in a significant change in the spectral slope of the depolarization ratio (especially at 1064 nm), is not visible. This corroborates the hypothesis proposed by that heating of the dust particles and turbulent mixing of the SAL air masses during daytime hours may widely reduce coarse-mode dust removal by gravitational settling of particles.

During SALTRACE-3 in 2014, POLIS was not available. Cirrus depolarization measurements were used to check the quality of triple-wavelength depolarization observations over time. Ice crystals are very large compared to the laser wavelengths so that the spectral dependence of backscattering, extinction, and depolarization properties is rather weak.

Figure  presents a cirrus measurement performed on 20–21 June 2014. The cirrus layer between 12 and 14 km height was optically thin with an AOT of 0.1. The wavelength-independent backscatter coefficients of up to 3.5 Mm-1 sr-1 at cirrus center indicate peak particle extinction values of 100–120 Mm-1. The extinction values are obtained by applying a multiple-scattering-corrected cirrus lidar ratio of 30–35 sr to the cirrus backscatter coefficients . The extinction measured with the Raman channels would need a too-large smoothing length in the thin cirrus.

As shown in Fig. b, depolarization ratios within the cirrus (above 12.2 km height) for 532 and 1064 nm are almost equal up to cloud top. The 1064 nm particle linear depolarization ratio is close to 0.5 and height-independent from 12.2 to 13.3 km height. The noisy 355 nm particle depolarization ratio is less trustworthy, but close to the 532 and 1064 nm depolarization ratios at least in the cirrus backscatter center from 12.4 to 12.8 km height. This consistent cirrus measurement of wavelength-independent cirrus backscatter and depolarization corroborate that BERTHA was performing well and that our dust observations below 5 km height are trustworthy.

Figure  shows the aerosol layers in the lower troposphere on this cirrus day. A 3 km thick SAL was present above the marine aerosol layer. Relative humidities of 40–50 % in the SAL were comparably high and suggest some mixing with moist marine air.

The HYSPLIT backward trajectories in Fig.  corroborate this hypothesis. A total of 4 days before arriving at Barbados, the air masses at 2.5 km height had the chance of vertical mixing with marine particles or African pollution. Only the uppermost dust layer (3–4 km height over Barbados, trajectories not shown) seems to contain pure dust. The air masses arriving at 3.5 km height were above 6 km over desert areas in western Africa.

Figure  presents the particle optical properties derived from the BERTHA observations. A steady and almost monotonic decrease of the backscatter coefficients with height was found. The extinction coefficients at 355 and 532 nm are again very close and the SAL AOT was about 0.25. The lidar ratios range from 40 to 50 sr in the layer from 1 to 3 km height, which probably contained some marine and anthropogenic haze particles, and were higher with values of 50–60 sr in the uppermost pure dust layer (3–4 km height).

The height profiles of the 532 and 1064 nm particle depolarization ratio show slightly different profile shapes (Fig. c). The 1064 nm depolarization ratio decreases slightly with height whereas the 532 nm increases with height. Since the 1064 nm depolarization ratio is very sensitive to coarse-mode particles, the decrease of the 1064 nm depolarization ratio (and of the ratio of 1064-to-532 nm depolarization ratios, Fig. d) may be related to a decreasing coarse-mode mean radius of the particles with height. The 532 nm depolarization ratio seems to be relatively insensitive against small changes in the coarse-mode size distribution . The 532 nm depolarization ratio for coarse mode particles is always in the range of 0.35–0.4. The slightly lower 532 nm depolarization values around 2 km height may again indicate a different history of the dust transport. The more fine-mode sensitive depolarization ratios (355 and 532 nm) and the respective ratio of 355-to-532 nm depolarization ratios show an almost height-invariant behavior when ignoring the noise in the 355 nm depolarization-ratio profile (Fig. d).

A unique case was observed during SALTRACE-3 on 6 July 2014. A 3 km deep dust layer crossed Barbados and traveled westward towards the United States (see backward trajectories in Fig.  and time-height display of the BERTHA observations in Fig. ). Coincidentally, this aged dust layer was observed with an airborne triple-wavelength polarization lidar (high-spectral-resolution lidar HSRL-2) 1 week later . We use this unexpected opportunity to compare the triple-wavelength depolarization observations of aged dust after long-range transport over 6000 km (Barbados) and 12 000 km (Missouri, midwestern USA). Rather low relative humidities around 20 % were measured with radiosonde in the lofted SAL (see Fig. c), suggesting almost no interference with cloud formation and associated upward mixing of marine air into the lower part of the SAL. High wind speeds around 18 m s-1 prevailed above 2.5 km height over Barbados. The 12.5-day backward trajectories indicate that the dust observed over Barbados at 2.5 and 3.5 km height descended towards 1.6 and 2.4 km height over Missouri, after crossing Yucatán (Mexico), Texas, and Oklahoma. After leaving the African continent, the dust layers arrived after 5 and 12 days over Barbados and Missouri, respectively.

Figure  shows the aerosol optical properties derived from the BERTHA measurements. The triple-wavelength particle linear depolarization observations performed over Missouri are added. The backscatter, extinction, and lidar ratio profiles again show typical dust optical properties. The SAL AOT was about 0.3 over Barbados on this day. The backscatter and extinction coefficients were roughly 25–30 % lower in the lofted dust layer over Missouri compared to the values measured over Barbados 1 week before.

An excellent agreement between the two lidar data sets of depolarization-ratio profiles was found (Fig. c). When comparing the values in the layers from 2.5 to 4.0 km height over Barbados with the values in the layer from 1.5 to 2.0 km height over Missouri, both lidars found similar depolarization ratios. At 532 nm (0.28, BERTHA; 0.30, HSRL-2) and 1064 nm (0.26, BERTHA; 0.27–0.28, HSRL-2), the depolarization ratio over Barbados is slightly lower. The 355 nm depolarization ratio measured with BERTHA with a larger systematic uncertainty is around 0.24. It is higher than over Missouri (0.21, HSRL-2). But the spectral behavior (Fig. d) of the 1064 nm to the 532 nm depolarization ratio was the same over Barbados and Missouri (0.91). The 355 / 532 ratio of depolarization ratios decreased between Barbados and the US.

Note also the higher depolarization ratios in the planetary boundary layer over Missouri (Fig. c). Probably, new soil dust particles were injected into the continental planetary boundary layer and then mixed upward into the lofted aged SAL over the United States. This entrainment of fresh coarse dust especially influences the 1064 nm depolarization ratio. Fine-mode pollution is released as well over the United States, and upward mixing and entrainment of pollution aerosol into the SAL affects the 355 nm depolarization ratio most sensitively. Nevertheless, the agreement is surprisingly good. Even after 7 days of travel from Barbados to the midwestern United States, the Saharan aerosol widely preserved its dust characteristics. No significant height dependence of the optical properties (lidar ratio, depolarization ratio, ratio of depolarization ratios) is visible in the main parts of the layer over both sites.