Injection of mineral dust into the free troposphere during fire events observed with polarization lidar at Limassol, Cyprus

Four-year observations (2010–2014) with EARLINET polarization lidar and AERONET sun/sky photometer at Limassol (34.7◦ N, 33◦ E), Cyprus, were used to study the soil dust content in lofted fire smoke plumes advected from Turkey. This first systematic attempt to characterize 5 less than 3 days old smoke plumes in terms of particle depolarization ratio (PDR), measured with lidar, contributes to the more general effort to properly describe the life cycle of free-tropospheric smoke–dust mixtures from the emission event to phases of long-range transport (> 4 days after 10 emission). We found significant PDR differences with values from 9–18 % in lofted aerosol layers when Turkish fires contributed to the aerosol burden and of 3–13 % when Turkish fires were absent. High Ångström exponents of 1.4–2.2 during all these events with lofted smoke layers, occuring 15 between 1 and 3 km height, suggest the absence of a pronounced particle coarse mode. When plotted vs. travel time (spatial distance between Limassol and last fire area), PDR decreased strongly from initial values around 16–18 % (one day travel) to 4–8 % after 4 days of travel caused by depo20 sition processes. This behavior was found to be in close agreement with findings described in the literature. Computation of particle extinction coefficient and mass concentrations, derived from the lidar observations, separately for finemode dust, coarse-mode dust, and non-dust aerosol compo25 nents show extinction-related dust fractions of the order of 10 % (for PDR =4%, travel times> 4 days) and 50 % (PDR = 15%, one day travel time) and respective mass-related dust fractions of 25 % (PDR =4%) to 80 % (PDR =15%). Biomass burning should therfore be considered as another 30 source of free tropospheric soil dust. Correspondence to: A. Nisantzi (argyro.nisantzi@cut.ac.cy)


Introduction
Biomass burning smoke affects air quality, visibility, and climate directly and indirectly.Hot fire smoke plumes eas-35 ily reach the free troposphere (FT) (Amiridis et al., 2010) and can be transported from continent to continent within one week (Fiebig et al., 2002;Mattis et al., 2003;Murayama et al., 2004;Müller et al., 2005;Petzold et al., 2007;Ansmann et al., 2009;Baars et al., 2011) and around 40 the globe within 2-3 weeks (Damoah et al., 2004).These plumes partly reach the upper troposphere (Mattis et al., 2008;Dahlkötter et al., 2014).Fire smoke must be regarded as an important source of the free-tropospheric background aerosol.It is well known that strong winds inside the com-45 bustion zones can raise considerable amounts of soil particles into the atmosphere (Palmer, 1981;Gaudichet et al., 1995;Maenhaut et al., 1996).Fires around the world may thus also be significant sources of FT dust particles which influence climate directly and indirectly as favorable cloud condensa-diameters of 200-400 nm compared to smoke-free background aerosol observations (Fiebig et al., 2002;Petzold et al., 2007;Dahlkötter et al., 2014).Particle depolarization ratios of 6-8 % and sometimes up to 15 % in North American fire smoke after long range transport observed with polarization lidars (Fiebig et al., 2002;Murayama et al., 2004), point to fine-mode dust particles contributions of about 5-40 % to the observed particle optical properties of the aged smoke layers.
Polarization lidar observations of the particle linear depolarization ratio allow us to separate fine-mode and coarsemode dust and to distinguish dust from non-dust aerosol contributions to the measured particle mass concentration, optical depth, and extinction coefficients (Mamouri and Ansmann, 2014).Only irregularly shaped partices such as mineral dust particles (soil, road, and desert dust) cause strong depolarization of emitted linearly polarized laser light, whereas spherical or almost spherical particles such as anthropogenic haze, biomass-burning smoke, and marine particles do not produce strong depolarization of backscattered light (Groß et al., 2011(Groß et al., , 2013;;Ansmann et al., 2011Ansmann et al., , 2012)).Our dust detection method may be biased by lightdepolarizing ash particles.However, these large particles are assumed to fall out quickly, within one day after injection.The observations, discussed in Sect.3, at least do not indicate the presence of large particles during fire smoke events.
To better characterize free-tropospheric fire smoke plumes and mixtures of dust and smoke as well as changes in the microphysical and optical properties of these mixed aerosol layers with travel time, lidar measurements of the particle depolarization ratio in fires smoke plumes especially during the first 3 days after emisson are required.A first systematic attempt to fill this gap is undertaken in this paper.We analyzed polarization lidar observations performed at the EAR-LINET site of Limassol, Cyprus, from April 2010 to February 2014.These results are presented in Sect.3. In our study, we concentrated on air masses advected from Turkey and regions further north of the Black Sea area during the main burning season (summer half year).We separated cases with strong impact of smoke events (occuring over Turkey during 1-3 days before arrival at Limassol) from observation with more background-like aerosol signatures (not influenced by Turkish fire smoke).We found significant differences in the depolarization characteristics of the backscattered layer light.
The particle depolarization ratio (the defintion is given in the Sect.2) was typically 10-15 % when Turkish fires contributed to the aerosol burden in the free troposphere and considerably lower with values of 3-8 % when fires over Turkey were absent while the air masses crossed this country.AERONET photometer observations showed high Ångström exponents (mainly from 1.4-2.0)during all these events with lofted smoke layers suggesting the absence of a pronounced particle coarse mode.Depolarization ratios of 10-15 % then point to fine-mode dust contributions of the order of 50 % to the observed optical particle properties as will be discussed  (Pappalardo et al., 2014).The site is located about 150 km south of Turkey and 400 km west of Syria.The lidar transmits linearly polarized laser pulses at 532 nm and detects the parallel-and cross-polarized 140 signal components at this wavelength.From the calibrated ratio of the cross-to-parallel-polarized lidar signals the volume linear depolarization ratio can directly be determined (Freudenthaler et al., 2009).Calibration of the polarization channels is performed by rotating the box with the po-145 larization sensitive channels following the methodology of Freudenthaler et al. (2009).The transmission properties of the receiver (for parallel and perpendicularly polarized light) required for an accurate determination of the particle linear depolarization ratio are known from measurements.The un-150 certainty in the volume depolarization ratio is ≤ 5 % The full overlap of the laser beam with the receiver field of view of the 20 cm Cassegrain telescope is obtained at heights around 300 ma.s.l.(Mamouri et al., 2013).The measured volume depolarization ratio is reliable to about 50 m above 155 ground.Overlap effects widely cancel out here because the depolarization ratios are calculated from signal ratios.However, for this study we only analyzed data for heights above 300 m and set the depolarization ratio below 300 m to a height-independent value (see discussion in Sect.3).

160
Our study presented in Sect. 3 is based on height profiles of the particle backscatter coefficient and the particle depolarization ratio at 532 nm.The determination of the particle backscatter coefficient is described in detail by Mamouri et al. (2013).The particle depolarization ratio is 165 computed from the volume depolarization ratio by means of the determined particle backscatter coefficient (Freudenthaler et al., 2009).Uncertainties in the retrieval products (particle backscatter coefficient, particle depolarization ratio) are discussed by Mamouri et al. (2013) and are typically of the order of ≤ 10 %.
In the retrieval of the particle depolarization ratio, the Rayleigh depolarization ratio, i.e., the depolarization ratio for particle-free air must be known.The Rayleigh depolariza-tion ratio is estimated from the volume depolarization ratio during clear days in aerosol-free air. Figure 1 presents two examples.Although the shown volume linear depolarization ratio is noisy at heights above the main aerosol layers reaching up to about 2-3 km height one can see that the volume depolarization ratio assumes values around 1 % at heights above 4 km.The volume depolarization is equal to the depolarization ratio for pure Rayleigh backscattering (in cases with negligible particle backscattering).In the data analysis, discussed in Sect.3, we used a fixed Rayleigh depolarization ratio of 1.3 % for all measurement cases.The uncertainty introduced by a wrong Rayleigh value of ±1 % is of the order of 1-2 % in the particle depolarization ratio.
Figure 1 also shows that the volume depolarization ratio in pronounced aerosol layers can be as low as about 2 % in cases with westerly winds (right panel) when maritime particles and aged anthropogenic haze from western, southern, and central Europe dominate the aerosol conditions over the eastern Mediterranean.Values around 2-3 % are typical for maritime particles (Groß et al., 2011(Groß et al., , 2013)).As mentioned above, road and soil dust emitted around Limassol and even dried marine particles (dry sea salt) may contribute to the measured light depolarization, but these effects are assumed to be small and mainly confined to the boundary layer (lowest 400 m).
Recently, a new polarization-lidar-based method was introduced by Mamouri and Ansmann (2014) that allows us to separate spherical (marine and continental fine-mode and marine coarse-mode particles), fine-mode dust particles causing particle depolarization ratios around 16 % (Sakai et al., 2010), and coarse-mode dust particles causing particle depolarization ratios around 39 % (Sakai et al., 2010).The fine mode includes all particles with radii < 500 nm.We used the new technique to estimate the contribution of finemode and coarse-mode dust to the mass concentrations and particle extinction coefficients in the detected smoke-dust plumes in Sect.3.
As described by Mamouri and Ansmann (2014), a twostep approach is applied to the measured 532 nm backscattercoefficient profiles.In our study, we explicitely assume that free-tropospheric spherical particles cause depolarization ratios of 1 %, that the fine mode aerosol mixture of smoke and dust causes a particle depolarization ratio of 8 % (assuming a mixture of roughly 50 % smoke and 50 % dust), and that, as mentioned, the coarse-mode dust particles cause a depolarization ratio of 39 %.According to the error discussions by Mamouri et al. (2013) and Mamouri and Ansmann (2014) the overall uncertainty in the separation of the backscatter and extinction coefficients for the different aerosol types is of the order of 20-40 %, and the uncertainty in the retrieved mass concentration profiles is about 50 %.

AERONET sun/sky photometer
The lidar is collocated with a sun/sky photometer of the Aerosol Robotic Network (AERONET, CUT-TEPAK site, Limassol, Cyprus, http://aeronet.gsfc.nasa.gov)(Holben et al., 1998).The CUT AERONET photometer measures 230 the aerosol optical thickness (AOT) at eight wavelengths from 339 to 1638 nm.From the spectral AOT distribution, the Ångström exponent AE ( Ångström, 1964), and the fine mode fraction FMF (fraction of fine-mode AOT to total AOT) (O'Neill et al., 2003) are retrieved.AOT errors are of the or-235 der of 0.01-0.02 in the absence of unfiltered cloud contamination (Chew et al., 2012).

Combined lidar/photometer data analysis
Mamouri et al. ( 2013) provides a detailed description of the lidar data analysis to obtain height profiles of particle 240 backscatter and extinction coefficients at 532 nm which are in good agreement with the column-integrated photometer observations of AOT.In this approach, the particle depolarization ratio is used to distinguish dust and non-dust contributions to the particle optical properties.Besides height pro-245 files of the particle optical properties, essential products are the derived extinction-to-backscatter ratios (lidar ratios) for the entire tropospheric column S col , for the free troposphere S FT , and lidar ratios for dust S FT,D in the free troposphere.Climatological values (obtained from the long-term observa-250 tions at Limassol) for the lidar ratio S PBL of 20-35 sr in the planetary boundary layer (PBL), i.e., for the lowermost 300-400 m of the tropospheric column, where the overlap of the laser beam with the receiver field-of-view is incomplete, and for the non-dust free-tropospheric lidar ratio S FT,S of 40-255 60 sr are required in this retrieval as assumed input.Index S denotes fine-mode spherical particles in the free troposphere such as fire smoke and anthropogenic haze particles.

HYSPLIT backward trajectories
In order to investigate the influence of fire activities on 260 the observed optical properties of the lofted aerosol plumes crossing Cyprus we studied the air mass transport by means of backward trajectory analysis.The calculations were performed with the HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model.Access is provided 265 via the NOAA ARL READY Website (http://www.arl.noaa.gov/HYSPLIT.php).HYSPLIT is described in detail in Draxler and Hess (1997, 1998) and Draxler (1999).

MODIS fire products
In order to identify the areas where biomass burning aerosols 270 were generated, MODIS (Moderate Resolution Imaging Spectroradiometer) active fire products were used (Flynn et al., 2002;Giglio et al., 2003).MODIS comprises a multispectral sensor with 36 spectral bands and covers the wave-with 1 km spatial resolution.MODIS is flown aboard the two NASA Earth Observing System (EOS) satellites Terra and Aqua.Both satellites are polar orbiting.Fires at the 1 km scale can be measured up to four times per day.The MODIS algorithms (including the fire algorithm) are updated periodically, leading to different versions, which are used to generate a series of Collections of the data products.The latest (Collection 5) fire data were used in this study.
We show seven-day fire spots, i.e., all spots detected within a week before the lidar observation.Data from University of Hawaii (http://modis.higp.hawaii.edu/)are taken.MODIS hot spots are provided with a level of confidence.For the retrievals here we used the confidence level of 80 %, so we excluded all values with lower confidence values.The fire events were then combined with six-day HYSPLIT backward trajectories in order to estimate the presence and age of smoke plumes detected with the EARLINET lidar.

Results
The focus of our study is the investigation of the potential of open fires (wild fires and controlled biomass burning) to trig-295 ger injection of soil dust into the free troposphere.Another goal was to concentrate on young smoke plumes (not older than 1-3 days).Therefore, we only consider measurement cases of air mass transport from northerly directions crossing Turkey.We observed 45 respective cases from April 2010 to February 2014.In 24 out of the 45 cases presented here, the backward trajectories together with the MODIS fire maps indicated that the measured lidar profiles were influenced by Turkish fire smoke.
We begin with two case studies.The measurement examples in Figs. 2 and 4 together with Figs. 3 and 5 (HYSPLIT backward trajectories and MODIS fire information) provide a contrasting impression of the differences in the observed particle depolarization ratios when no fires were observed by MODIS over Turkey (Fig. 2) and when several fires over central Turkey influenced the lidar measurements over Limassol significantly (Fig. 4).In the smoke-free case, the particle depolarization ratio was less than 10 % throughout the entire free troposphere from 500 m to 4 km height.The aerosol plume was well mixed.The moderately enhanced particle depolarization ratios (compared to low values of 1-3 % for haze pollution and fine-mode smoke) may indicate the influence of soil dust from arid regions, from remote deserts, and dust injected during fires at times 3-7 days before the arrival of the air masses at Limassol from areas north of the Black 320 Sea.
The Ångström exponent of AE A = 1.67 and fine mode fraction FMF A = 0.76 from the AERONET observations indicate the dominance of fine mode particles throughout the entire troposphere.The particle optical depth of AOT A = 0.31 and extinction-to-backscatter ratios S FT around 35 sr in the free troposphere indicate a high aerosol load of a mixture of aged, less absorbing anthropogenic haze, fire smoke, and background aerosols from rural areas.
Figure 4 represents a measurement case with strong influ-330 ence of Turkish fire smoke (see also Fig. 5).Again, a well mixed lofted layer of smoke and dust was found above 1 km height.The layer-mean particle depolarization ratio of almost 18 % is the highest depolarization value we observed during the 4 years of lidar measurements.However most Figure 6 provides an overview of the geometrical properties of all aerosol layers in the lower free troposphere observed during the four-year period from February 2010 to April 2014.Most layers were found between 1 and 3 km 345 height with layer centers around 1.9-2 km height.On average, the smoke layers showed a vertical extent of 1.4-1.5 km.
Based on spaceborne CALIPSO lidar data, Amiridis et al. (2010) found most top heights (injection heights) of smoke layers between 2-4 km for the summer months of 2006-2008. 350 The average top height was 3.1 km.This is in good agreement with our observations (mean top height at about 2.9 km height).
Figure 7 presents the time series of derived layer-mean particle depolarization ratios.Most values were > 10 % 355 when the air masses crossed fire areas over Turkey.For cases free of Turkish smoke we observed values from 3-13 %.Depolarization ratios below < 4 % may indicate dust-free conditions in the free troposphere.The variability in the depolarization ratio may reflect the influence of numerous dust 360 sources around the eastern Mediterranean.A seasonal cycle in the depolarization ratio time series is not visible because most northerly flows occur during the summer season.In winter the release of soil dust may be generally reduced by enhanced precipitation (increased wash out), wet soils (pro-365 hibit dust emission), and the presence of snow covers (in the Turkish mountains and further north and east).
An overview of the optical properties (range of values) of all layers for northerly air mass advection is presented in Fig. 8. Layers with comparably fresh fire smoke from Turkey 370 (red symbols, depolarization values from 9-18 %) showed low backscatter coefficients around 1 ± 0.5 Mm −1 sr −1 .Multiplication with a tyical lidar ratio of 40-50 sr yields smoke-dust layer particle extinction coefficients of the order of 25-75 Mm −1 .For the smoke-free cases, several 375 large backscatter coefficients > 3 Mm −1 sr −1 were observed which may indicate the influence of marine particles.Even during northerly airflows, sea breeze effects at the south coast of Turkey and the north coast of Cyprus (including mountaininduced circulation effects over Cyprus) and corresponding the atmosphere can never be completely excluded when interpreting free-tropospheric lidar observations at Limassol.But as shown, these events are rare.
The 532 nm AOT for smoke-dust plumes was found to be in the range from 0.05-0.25.According to Mattis et al. (2003), Murayama et al. (2004) 2007) also reported Ångström exponents > 1.4 during fire smoke events (within the first 2-3 days after emission).

405
Figure 9 shows the dependence of the layer-mean particle depolarization ratio on travel time (i.e., the spatial distance between Limassol and the last fire area upwind of the lidar site).The travel time is calculated from the HYSPLIT backward trajetories for the smoke layer centers.As can be seen, the depolarization ratio drops from 14-18 % (day 1) to about 5 % (day 4).Several literature values are shown in addition and corroborate the observed trend.Our fire smoke data (< 4 days old smoke) together with literature values provide, for the first time, an impression how fast the depolarization ratio and thus the dust fraction decreases with time.The exponential fit curve in Figure 9 indicates an 1/e decay time of 4 days.Our data clearly fill an observational gap and contribute to the effort to properly describe the life cycle of free-tropospheric smoke-dust mixtures from the emission event to phases of long-range transport.
Assuming mean wind speeds of 5-10 ms −1 in the lower free troposphere over Turkey and areas north of Turkey, the air masses travelled 450-900 km per day.At greater heights wind speeds of > 15 ms −1 are typical so that smoke plumes may travel > 1350 km per day (as is the case for the shown literature cases in Fig. 9).
The observed scatter in the data in Fig. 9 can be related to many reasons.First of all, there are uncertainties in the assignment of the fire events (after MODIS) to our lidar observations by using uncertain backward trajectories and fire information integrated over five days.Second, the soil characteristics and fire-induced dust injection efficiency may differ significantly from site to site, even for favorable arid regions such as Turkey.The literature values in Fig. 9 include Siberian as well North American biomass-burning smoke plumes.Third, free-tropospheric meteorological conditions can vary strongly along the transport path ways and thus also the aerosol mixing, diffusion, deposition, and particle growth processes.It was observed that (spherical) smoke particles 440 increase their sizes during long-range transport because of gas-to-particle conversion and water uptake (Müller et al., 2007).The increasing backscatter coefficient for spherical particles then leads to a decrease of the observed particle depolarization ratio.Fourth, flaming fires as well as smoldering 445 fires can occur.Flaming fires may be more efficient regarding the mobilization of surface soil dust.Fifth, some plumes observed with lidar may have been influenced by many (flaming and smoldering) fires, others by just one (flaming or smoldering) burning event.Sixth, it can also not be excluded 450 that large, irregularly-shaped fire particles affected the lidar depolarization measurement, especially in very fresh smoke plumes.Finally, the nearby deserts in the Middle East and North Africa may have also contributed to the free tropospheric dust load.

455
In Figs. 10 and 11 two cases with high depolarization ratio in the smoke plume (PDR = 14 %, travel time < 2 days) and low depolarization ratio (PDR = 4.6 %, travel time of 4 days) are shown to quantify the range of dust contributions to the observed optical properties in the free tropospheric layer.

460
The profiles of particle extinction coefficient were computed from the retrieved backscatter coefficients by using a lidar ratio S FT,S of 45 sr for fine-mode spherical particles and the retrieved dust lidar ratios S FT,D were close to 30 sr in these specific two cases.As mentioned in Sect.2, the retrieval is 465 explained in detail by Mamouri et al. (2013) and Mamouri and Ansmann (2014).
On 25 August 2011 (Fig. 10), smoke layer mean extinction coefficients were 18 Mm −1 (fine-mode smoke), 10 Mm −1 (fine-mode dust), and 8 Mm −1 (coarse-mode dust).The dust 470 fraction (DF) of particle extinction and optical depth was thus 50 %.The fine-mode fraction, i.e., the ratio of the sum of the AOTs of fine-mode dust and fine-mode spherical particles to the total AOT (at 500 nm) was 0.7 for the total tropospheric column (AERONET) and 0.77 in the FT on the basis of the 475 lidar-derived profiles shown in Fig. 10.
On 25 June 2012 (Fig. 11) the FT aerosol layer was almost smoke free.The mean extinction coefficients in the lofted layer from 400-3000 m were 86.5 Mm −1 for spherical particles (probably mainly fine-mode urban haze) and 480 12.5 Mm −1 for fine-mode dust.The dust fraction was in this case DF L = 12 %.The fine-mode fraction after AERONET was 0.79, and 1 in the free troposphere according to the lidar extinction profiles in Fig. 11.Ångström exponents of 1.41 and 1.56 indicate the dominance of fine-mode particles 485 in both cases.
In terms of particle mass concentration, the dust fraction in the mixed aerosol plumes is even higher.This is shown in Fig. 12. Mass concentrations were determined from the extinction coefficient profiles after Ansmann et al. (2012) and particle densities of 2.6 and 1.55 gcm −3 for mineral dust particles and non-dust fine-mode particles, respectively.For the required particle volume-to-extinction conversion factors we applied values of 0.2×10 −6 m, 0.3×10 −6 m, and 0.8×10 −6 m for fine-mode spherical particles, fine dust, and coarse dust, respectively (Mamouri and Ansmann, 2014).
As can be seen in Fig. 12, the layer mean particle mass concentrations were about 5.5 µgm −3 (fine-mode spherical smoke particles), 8 µgm −3 (fine-mode dust), and 500 16.5 µgm −3 (coarse-mode dust) for the smoke case observed on 25 August 2011.The dust fraction was about 80 % on this day.Even on 25 June 2012 (case with background-level smoke and dust) the dust mass fraction was 25 % for the observed case with the low PDR = 4.6 %.We can conclude that dust mass fractions were of the order of 30-70 % in aged smoke-dust plumes observed over Limassol after travel times of 2-4 days.Fire smoke aerosol plumes in the free troposphere must be regarded as a non-negligible reservoir for dust and thus for cloud condensation and ice nuclei in the 510 free troposphere.

Conclusions
Four-year observations (2010-2014) with EARLINET polarization lidar and AERONET sun/sky photometer at Limassol (34.7 • N, 33 • E), Cyprus, were used to study the soil dust 515 content in lofted fire smoke plumes advected from Turkey.This first systematic attempt to characterize less than 3 days old smoke plumes in terms of particle depolarization filled an observational gap and contributes to the more general effort to properly describe the life cycle of free-tropospheric 520 smoke-dust mixtures from the emission event to phases of long-range transport.
We found significant differences in PDR.PDR was 9-18 % in lofted Turkish smoke-dust-haze aerosol layers and considerably lower with values of 3-13 % when Turkish fires 525 were not detected by MODIS.AERONET photometer observations showed high Ångström exponents (1.4-2.2) during all these events with lofted smoke layers suggesting the absence of a pronounced particle coarse mode.The lofted aerosol layers typically occurred between 1 and 3 km height.
When plotted vs. the travel time (defined as the spatial distance between Limassol and last fires area), PDR decreased strongly from initial values around 16-18 % (1 day travel) to 4-8 % after 4 days of travel, in agreement with the literature.Computation of particle extinction coefficient and mass concentrations, separately for fine-mode dust, coarse-mode dust, and non-dust aerosol components shows extinction-related dust fractions of the order of 10 % (PDR = 4 %, after times of > 4 days) to 50 % (PDR = 15 %, after travel times of 1 day) and mass-related dust fractions of 25 % (PDR = 4 %) to 540 80 % (PDR = 15 %).Biomass burning in arid regions must obviously be regarded as a significant source of soil dust that contributes to the background aerosol burden of the free troposphere.
Our study must however be regarded as a first step.More 545 efforts of polarization lidar monitoring of fresh and aged smoke-dust plumes are required at very different places around the world to support our findings.The arid regions in the southeastern part of Europe and in the western part of Asia together with the deserts in the Middle East and North-550 ern Africa may provide rather favorable conditions for dust injection into the atmosphere (free troposphere).Such almost optimum conditions may not be given at higher latitudes in the Northern Hemisphere as well as in large parts of the Southern Hemisphere.Freudenthaler, V., Esselborn, M., Wiegner, M., Heese, B., Tesche, M., Ansmann, A., Müller, D., Althausen, D., Wirth, M., Fix, A., Ehret, G., Knippertz, P., Toledano, C., Gasteiger, J., Garhammer, M., and Seefeldner, M.: Depolarization ratio profiling at several wavelengths in pure Saharan dust during SAMUM  Atmos. Meas. Tech. Discuss., 7, 2929-2980, 10.5194/amtd-7-2929-2014, 2014. Petzold, A., Weinzierl, B., Huntrieser, H., Stohl, A., Real, E., Cozic, J., Fiebig, M., Hendricks, J., Lauer, A., Law, K., Roiger, A., Schlager, H., and   Fig. 2. 532 nm particle backscatter coefficient (left, green) and particle linear depolarization ratio (right, black) in the free troposphere during northerly airflow.No fires occurred over Turkey (smokefree case).The grey-shaded area indicates the identified lofted aerosol layer.AEA, AOTA, and FMFA in the left panel denote Ångström exponent, particle optical depth, and fine-mode fraction derived from AERONET photometer observations (for the total tropospheric column).Retrieved lidar ratios SFT for the free troposphere and SCOL for the total column obtained after Mamouri et al. (2013) are given in the right panel.The boundary layer lidar ratio SPBL (for the lowest 400 m) is required as input in this retrieval.In addition, the mean particle depolarization ratio δFT for the greyshaded lofted aerosol layer is given.In the case of the blue vertical lines (21 smoke-free cases) no fire events over Turkey occured and influenced the lidar measurements.In the case of the red layers (24 fire smoke cases), the air masses crossed fire areas over Turkey before arriving at the lidar site at Limassol.The average bottom and top heights (plus one standard deviation) of all detected layers, separately for smoke and smoke-free cases, are given in addition.

Depolarization ratio [%] S(FT) [sr]
Angstrom exponent Fine mode fraction Fig. 8. Range of observed values of the particle backscatter coefficient (mean value of the lofted aerosol layers), free-tropospheric aerosol particle optical thickness AOT(FT), lidar ratio S(FT) in the lofted aerosol layers, Ångström exponent for the total vertical column, and fine mode fraction (regarding the total tropospheric 500 nm AOT) for the Turkish fire-smoke cases (red circles) and the cases without smoke from Turkey (blue circles).All values (circles) are given as a function of layer-mean particle depolarization ratio to better distinguish between fire-smoke and smoke-free cases.smoke (north of Turkey) smoke (Turkey) Mattis et al., 2003-Mueller et al., 2005Murayama et al., 2004Fiebig et al., 2002Dahlkoetter et al., 2014  .532 nm particle backscatter coefficient (green profile, left panel), particle linear depolarization ratio (black profile, left panel), and lidar-derived particle extinction coefficients (right panel) for spherical fine-mode particles (blue profile, index SF), fine-mode dust particles (orange profile, index DF), and coarse-mode dust particles (dark red profile, index DC).The extinction coefficient profiles are obtained by using the method of Mamouri and Ansmann (2014).The respective free tropospheric AOTs (above 300-400 m height), the AERONET-derived Ångström exponent AEA and finemode fraction FMFA (for total tropospheric column), and the lidarderived dust fraction DFL (with respect to the free tropospheric 532 nm AOT) are given as numbers together with the layer-mean depolarization ratio PDRL.The measurement was taken on 25 August 2011.Ansmann et al. (2012).Fine-mode and coarse-mode dust mass retrieval is described by Mamouri and Ansmann (2014).

335
of the dust seems to be again fine-mode dust because the Ångström exponent of AE A = 1.45 and the fine mode fraction FMF A = 0.79 are again comparably high.Lidar ratios S FT in the free troposphere of around 50 sr point to a mixture of moderately absorbing smoke particles (60 sr) and less 340 absorbing, non-spherical dust particles (35-45 sr).

Fig. 1 .
Fig. 1. 532 nm particle backscatter coefficient (BC, green) and volume linear depolarization ratio (DV, black) measured on two days with almost particle-free conditions above 2-3 km height.The vertical lines indicate DV = 1.3 %.This value was used in the data analysis as Rayleigh depolarization ratio (clear air DV).figure

Fig. 9 .
Fig. 9. Decrease of the layer-mean particle depolarization ratio of smoke-influenced aerosol layers with increasing travel time, defined as the temporal distance between Limassol and the last fire area which the observed air masses crossed before arriving at Limassol.Ten of the 21 blue circles (cases without Turkish smoke) were influenced by smoke generated in areas north of the Black Sea.The exponetial fit indicates an 1/e decay time (dust decrease) of 4 days.Grey symbols show observations as published in the given literature.Vertical bars show the range of observed depolarization ratios (literature values) or show the standard deviation of the all values determined within the individual layers (red and blue values).