Wintertime Saharan dust transport towards the Caribbean: airborne lidar observations during EURECA

Wintertime Saharan dust plumes in the vicinity of Barbados are investigated by means of airborne lidar measurements. The measurements were conducted in the framework of the EURECA field experiment (Elucidating the Role of Cloud-Circulation Coupling in Climate) upstream the Caribbean island in January/February 2020. The combination of the water vapor differential absorption and high spectral resolution lidar techniques together with dropsonde measurements aboard the German HALO (High Altitude and Long-Range) research aircraft enable a detailed vertical and horizontal characterization of 5 the measured dust plumes. In contrast to summertime dust transport, mineral dust aerosols were transported at lower altitudes and were always located below 3.5 km. Calculated backward trajectories affirm that the dust-laden layers have been transported in nearly constant low-level altitude over the North Atlantic Ocean. Only mixtures of dust-particles with other aerosol species, i.e. biomass burning aerosol from fires in West Africa and marine aerosol, were detected by the lidar. No pure mineral dust regimes were observed. Additionally, all the dust-laden airmasses that were observed during EURECA came along with 10 enhanced water vapor concentrations compared to the free atmosphere above. Such enhancements have already been observed during summertime and were found to have a great impact on radiative transfer and atmospheric stability.

this season, dust particles are then frequently transported towards the South American continent and the Amazon region (e.g. Yu et al., 2015). During boreal summer, however, strong solar insolation over the Sahara causes a deep boundary layer (e.g. Messager et al., 2009) which can reach up to 6 km altitude (≈500 hPa). This allows dust particles to be mixed upward to higher altitudes where they get caught by the trade winds. Due to the northward shift of the ITC during boreal summer, they then get carried westwards as far as the Caribbean as well as Central and North America along a more northerly transportation route (e.g. Prospero and Carlson, 1972;Prospero et al., 2010). 30 Summertime dust advection can frequently be observed in the Caribbean. During that time of the year the dust particles are usually advected in elevated and decoupled layers -so called Saharan air layers (SALs; Prospero and Carlson, 1972;Carlson and Prospero, 1972;Prospero et al., 2021). Wintertime dust plumes, however, reach the Caribbean less frequently and only in connection with favorable synoptic situations. They are mostly transported at lower levels and are sometimes mixed with other aerosol types like biomass burning aerosols or marine aerosols in the marine boundary layer (MBL; Chiapello et al., 1995;35 Ben- Ami et al., 2009;Groß et al., 2011).
Recent studies have shown, that summertime SALs may come along with enhanced concentrations of water vapor compared to the surrounding dry free troposphere (Gutleben et al., 2019b(Gutleben et al., , 2020Ryder, 2021). It was shown that not the aerosol but the water vapor inside the SALs is the dominating driver for net radiative heating during transatlantic dust transport. In this way, water vapor associated with dust layers has the potential to modify the atmospheric stability and consequently to impact the 40 evolution as well as macrophysical properties of shallow marine trade wind clouds.
In addition to that, the transportation of dust particles at low atmospheric levels within/atop of the MBL during wintertime implies aerosol radiative effects. The particles may not only have a direct effect on the evolution of shallow marine clouds via absorption and scattering, but also have an indirect effect as dust particles can act as cloud condensation and ice nucleating particles (Karydis et al., 2011;Bègue et al., 2015;DeMott et al., 2015;Boose et al., 2016). However, before an investigation of 45 these effects can be performed, wintertime transatlantic dust transport towards the Caribbean has to be investigated in detail, which is the focus of this study.
Up to now, the concurrence of transatlantic Saharan dust and water vapor transport during boreal winter has never been studied. In early 2020, extensive wintertime mineral dust plumes could be observed upstream the Caribbean island of Barbados during the EUREC 4 A field experiment (Elucidating the Role of Cloud-Circulation Coupling in Climate; Stevens et al., 2021).

50
Several research flights with the High Altitude and Long Range Research Aircraft (HALO; Krautstrunk and Giez, 2012) were carried out over these regions and an extensive lidar and dropsonde data set was collected. This unique data set now enables a detailed investigation of the plumes. This paper is structured as follows. In section 2 an overview of the EUREC 4 A project as well as of the performed research flights is given. Moreover, the capabilities and features of the airborne lidar system WALES (Water Vapor Lidar Experiment 55 in Space; Wirth et al., 2009)  The EUREC 4 A field campaign aimed at investigating the driving factors for the evolution of trade-wind cumulus clouds in the winter trades (Stevens et al., 2021). As one of several employed research platforms during EUREC 4 A, the German highflying research aircraft HALO conducted remote sensing measurements east of the Caribbean island of Barbados (Konow et al., 2021). In the period from 18 January to 18 February 2020, the modified Gulfstream G550 research aircraft performed a total of 15 scientific flights (13 local flights from and to Barbados and two transfer flights from and to Germany). An overview of the HALO flight tracks is shown in Figure 1. The circular flight patterns of the research flights were flown for dropsonde-based analyses of the respective prevailing large-scale divergences (Bony and Stevens, 2019).
On three measurement days, Saharan mineral dust plumes were advected to the research area, i.e. on 30 Jan 2020 during HALO-0130, on 31 Jan 2020 during HALO-0131 and on 2 Feb 2020 during HALO-0202. Collected airborne lidar data sets during measurement flights on these days enable a characterization of winter-time dust transport, although the flights tracks 70 have not been specifically designed for dust observations. The enhanced total column aerosol optical depth (AOD) over the dust-affected research areas was also captured by MODIS (Moderate-resolution Imaging Spectroradiometer) and is shown in  In total, approximately 18.5 hours of lidar data could be collected during the three flights. Additionally, a total of 164 dropsondes were launched (Vaisala RD-41; George et al., 2021;Vaisala, 2020). 157 of them worked properly and collected thermodynamic data from aircraft to ground level.

80
During EUREC 4 A the WALES lidar instrument ) was operated aboard HALO. It is an airborne water vapor differential absorption lidar (DIAL) system with depolarization and high spectral resolution lidar (HSRL; Esselborn et al., 2008) capabilities.
The DIAL module operates at four wavelengths (three online and one offline wavelength) around the H 2 Oabsorption band at 935 nm. This setup allows for measurements of water vapor mass mixing ratios (r m ) that cover the whole extent from aircraft 85 to ground level. Due to its high pulse-repetition rate of 0.01 s between online and offline pulses, it enables horizontally and vertically highly resolved measurements of r m with relative uncertainties of less than 5 % (Kiemle et al., 2008). To obtain the high pulse repetition rate, two Q-switched Nd:YAG (Neodymium-doped Yttrium Aluminum Garnet; Nd : Y 3 Al 5 O 12 ) ring lasers, that generate pulses at 1064 nm wavelength, are operated temporally interleaved. The required frequencies for DIAL measurements are generated via frequency-doubling of parts of the emitted pulses and subsequent pumping of injection-seeded 90 optical parametric oscillators (Mahnke et al., 2007). In addition to the DIAL capability, the integrated HSRL-module and depolarization sensitive channels allow for highly resolved measurements of particle linear depolarization ratios (δ p(532) ), backscatter ratios (R 532 = 1 + β p(532) /β m(532) ; with β p(532) and β m(532) being the particle backscatter and molecular backscatter coefficients), particle extinction coefficients (α p(532) ) as well as the lidar ratios (S) at 532 nm wavelength. Relative uncertainties in measurements of δ p(532) , R 532 and 95 α p(532) are 10 % to 16 %, 5 % and 10 % to 20 %, respectively (Esselborn et al., 2008).
WALES data is spatially and temporally averaged to improve the signal to noise ratio. At typical aircraft speed of 200 m s −1 , the horizontal resolutions amount to 3000 m (DIAL) and 200 m (HSRL). The vertical resolution is 15 m.

Saharan dust research flights during EUREC 4 A 100
Transported Saharan mineral dust was present in the research area on three measurement days. Times of take-off and landing of the flights conducted on these days, as well as the number of launched dropsondes are summarized in Table 1. In the following, a detailed overview of the measurements during these three flights is given.   Groß et al., 2013). This value does not change with transatlantic transportation (Groß et al., 2015). However, δ p(532) between 0.7 km to 3.5 km altitude took slightly smaller values that ranged from 15 % to 30 %.
Depolarization ratios in this range are typical for aerosol regimes where mineral dust particles are mixed with less depolarizing particles, i.e. marine aerosol or biomass burning aerosol. Mixing of different aerosol types is also evident HALO's second dust-flight aimed at measurements of large-scale divergence from dropsondes. This is why a total of seven circles were flown upstream of Barbados and dropsondes were launched at a very high rate. After 3.5 circles a short excursion towards the Northwest Tropical Atlantic Station -a meteorological buoy -was performed. The sampled atmosphere below the circles was characterized by an almost dust-free region in the Northwest and a dusty regime in 140 the Southeast (see Figure 2 (b)). As a result the lidar depolarization data shows recurring features (Figure 3 (b)). Highly depolarizing and dust-laden regions were observed between 0.7 km and 3 km altitude (15 %<δ p(532) <30 %), with greatest depolarization ratios in 2 km altitude. In these altitudes the lidar ratio S took values around 60 sr, indicating the presence of both mineral dust and biomass burning aerosol. At lower levels (0.0 km to 0.7 km) S was smaller and around 20 sr.
This points towards an additional contribution of marine aerosol in these altitudes.

145
Similar to HALO-0130, backward trajectories with starting points at low altitudes originate from more northern latitudes than the trajectories starting from altitudes affected by long-range-transported Saharan mineral dust, i.e. at 1.2 km and 2.0 km altitude. The low latitudes together with low transportation altitudes can again explain that not only mineral dust,  The MBL was again characterized by a well-mixed convective layer from surface to approximately 0.7 km altitude (Θ = const.). However, water vapor mixing ratios were slightly smaller compared to the other two flights (∼ 16 g kg −1 ) .

160
The MBL was capped at approximately 2.5 km altitude by a pronounced TWI. Depolarizing aerosol was found along the whole vertical extent of the MBL. However, δ p(532) -values were highest at the very top of the MBL (∼30 % at 2.3 km altitude). Lidar ratios again indicate that the aerosol layer was characterized by a particle mixture of mineral dust aerosol, biomass-burning aerosol and marine aerosol. S took highest values inside the aerosol layer atop of the MBL (∼60 sr) and was lowest near ground level (∼30 sr). This indicates that the contribution of biomass burning aerosol to the aerosol 165 mixture was greatest at the top levels and marine aerosol mostly contributed to the aerosol mixture at lower levels.
Regions inside the MBL that were impacted by this aerosol mixture came along with reduced atmospheric moisture content and were characterized by small water vapor mixing ratios that ranged from 5 g kg −1 to 10 g kg −1 .
7-d backward trajectories for this research flight again indicate that the dust-laden air masses in 1.2 km and 2.0 km altitude traveled at nearly constant altitude and that they took a southerly route that favored mixing processes with 170 biomass burning aerosols and marine aerosols.

Aerosol and water vapor composition of the observed dust layers
Measurements of lidar ratios and particle linear depolarization ratios during the three research flights over transported Saharan mineral dust indicated that dust particles have been mixed together with biomass burning and marine aerosols. An aerosol regime containing only mineral dust particles was never observed during EUREC 4 A. To investigate these mixed regimes in 175 more detail, the contribution of Saharan mineral dust aerosol to the aerosol mixture is calculated. The well-established aerosolseparation technique for two-component aerosol mixtures shown by Tesche et al. (2009) and Groß et al. (2011) allows for the calculation of the contribution (in percent) of mineral dust to both the measured particle linear depolarization ratio δ p(532) and the measured lidar ratio S. The percentage that Saharan mineral dust is contributing to a measured particle linear depolarization ratio of a two-component 180 aerosol mixture δ p(532) with biomass burning aerosol or marine aerosol is calculated as follows, with the coefficients D A and D B , (2) 185 D B := δ p(532) − δ p(532),BB|MA S BB|MA (1 + δ p(532),BB|MA ) .
Here, S DU and S BB|M A are the known lidar ratios of Saharan mineral dust (50±4) as well as of biomass burning aerosols (63±7) and marine aerosol (18±5) (mean values of observations by Groß et al., 2013). δ p(532),DU and δ p(532),BB|M A are the corresponding known particle linear depolarization ratios (Saharan mineral dust: 27±2 %; biomass burning aerosol: 14±2 %; marine aerosol: 3±1 %). For S the percentage is calculated using, Using these equations, one can derive mixing lines between the pure Saharan mineral dust aerosol regime and the marine and biomass burning aerosol regime in a two-dimensional space of S and δ p(532) (Figure 5). This can also be seen in the joint distributions of r m with δ p(532) and S (see Figure 6). It indicates that regions affected by mineral dust are not completely dry, but always come along with enhanced water vapor concentrations. Only in mixed regions,

210
where biomass-burning aerosol dominates the dust aerosol (δ p(532) ≈ 20 %; S ≈ 60 sr), water vapor mixing ratios drop to values smaller 4 g kg −1 . In the lower MBL, where a mixture of mineral dust with marine aerosol predominates, water vapor concentrations are high in general and take values between 8 g kg −1 and 18 g kg −1 .

Discussion and Conclusions
While during the summer months Saharan dust particles are predominantly transported westwards in SALs at altitudes as great 215 as 6 km (e.g. Prospero and Carlson, 1972;Prospero et al., 2010), Saharan dust transport in the winter months happens at lower atmospheric levels (Chiapello et al., 1995). This agrees well with the findings in this study, as mineral dust particles were never observed in altitudes higher than ∼ 3.5 km. Low level transport also favors mixing processes of mineral dust particles with other aerosol species like biomass burning aerosol or marine aerosol. As a consequence, pure dust aerosol regimes were never observed during EUREC 4 A and mineral dust particles could only be observed in mixed aerosol regimes. At lowermost 220 altitudes inside the MBL the dust particles were predominantly mixed with marine aerosol. Above the MBL in altitudes from 2 km to 3.5 km the dust particles were mixed with biomass burning aerosol from fires in West Africa.
During their travel over the Atlantic Ocean these dusty aerosol regimes can impact the Earth's radiation budget by directly interacting with radiation via absorption and scattering, by changing cloud microphysical properties or by modifying the atmospheric stratification. However, up to now the radiative impact of long-range Saharan mineral dust transport over the 225 Atlantic Ocean during boreal winter has never been investigated in detail. Needed highly resolved observations of the vertical and horizontal aerosol distribution were missing. For summertime transport Gutleben et al. (2019bGutleben et al. ( , 2020 highlighted that not the aerosol, but the water vapor embedded in long-range-transported SALs is the dominant driver for radiative heating and the subsequent modification of the atmospheric stability. They used a radiative transfer model together with airborne lidar data collected during NARVAL-II (Next-generation aircraft remote sensing for validation studies II) for their calculation. During 230 summertime the dust transport usually occurs in elevated layers. Due to this spatial separation of the different aerosol layers, the impact of enhanced concentrations of water vapor in SALs on atmospheric heating could easily be quantified. However, wintertime low-level transport -as observed during EUREC 4 A -hampers such a separation as the water vapor concentration inside the MBL is high on principle. In addition, the indirect radiative effect of Saharan mineral dust and biomass burning aerosols (i.e. the modification of marine cloud microphysics by aerosol particles) has to be considered in future simulations 235 of radiative transfer in the winter season. Haarig et al. (2019) demonstrated that smoke particles play a crucial role for cloud condensation especially during boreal winter as they dominate the concentration of available cloud condensation nuclei in the mixture. Moreover, the potential of mixed dust aerosol regimes during winter-time transport for ice nucleation should be investigated in furture studies.
This study highlighted the characteristics of the observed dusty aerosol layers during EUREC 4 A. In near future the compo-240 sition of the dust layers and their radiative impact on the subtropical environment should be investigated in more detail. Finally, it is highly recommended that future and ongoing analyses with focus on radiative transfer, cloud physics or cloud occurrence based on observations during EUREC 4 A consider the impacts of the pronounced aerosol layers described in this paper.
Data availability. The data used in this publication were collected in the framework of the field study EUREC 4 A and are publicly available online in the AERIS database (https://observations.ipsl.fr/aeris/eurec4a/)