Accepted Manuscript

Long-term Measurements of Aerosol Optical and Physical Properties over the Eastern
Mediterranean: Hygroscopic Nature and Source Regions

Ersin Tutsak, Mustafa Koçak

PII: S1352-2310(19)30169-4
DOI: https://doi.org/10.1016/j.atmosenv.2019.03.007
Reference: AEA 16599

To appear in: Atmospheric Environment

Received Date: 28 September 2018

Revised Date: 8 March 2019
Accepted Date: 11 March 2019

Please cite this article as: Tutsak, E., Koçak, M., Long-term Measurements of Aerosol Optical and
Physical Properties over the Eastern Mediterranean: Hygroscopic Nature and Source Regions,
Atmospheric Environment, https://doi.org/10.1016/j.atmosenv.2019.03.007.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
1 Long-term Measurements of Aerosol Optical and Physical Properties over the Eastern
2 Mediterranean: Hygroscopic Nature and Source Regions
3
4 Ersin Tutsak1 and Mustafa Koçak1*
5
1
6 Institute of Marine Sciences, Middle East Technical University, P.O. Box 28, 33731,

7 Erdemli-Mersin, Turkey

PT
*
8 Corresponding author (Tel: +90324-5213434; Fax: +90324-5212327;

9 mkocak@ims.metu.edu.tr

RI
10
11 Abstract

SC
12 Long-term aerosol optical and physical properties such as aerosol optical thickness and
13 angstrom exponent were obtained from AERONET ground based sun sky radiometer located
14 at the rural coastal Erdemli (36.6°N, 34.3°E) site in the Eastern Mediterranean from January

U
15 2000 to December 2014. Aerosol Optical Thickness (AOT) and Angstrom Exponent (α)
16 exhibited a great short-term variability, varying up to couple of fold from day to day. The
AN
17 lowest AOT values was found in winter due to the removal of particles by wet deposition.
18 Elevated AOT values with low α440-870 was observed in spring owing to sporadic mineral dust
19 transport from North Africa and the Middle East. High AOT at 440 nm and α440-870 were
M

20 characterized in summer because of high gas-to-particle conversion, sluggish air masses and
21 absence of rain. Enhanced AOT and low α (0.7) and fine mode fraction (47 %) were
22 associated with air flow from North Africa and the Middle East with a slight decrease from
440 nm to 1020 nm, showing importance of coarse particles. Air flow from Mediterranean
D

23
24 Sea, Eastern Europe, Western Europe and Turkey, on the other hand, showed comparatively
25 higher α (> 1.1) and fine mode fraction (> 62 %). Three classes of aerosol were identified
TE

26 namely non-hygroscopic dominated, moderately hygroscopic and hygroscopic dominated.

27 Non- hygroscopic particles were dust dominated aerosol population originated from North
28 Africa (Southern Tunisia, Western Libya, Libya/Chad border, southern Egypt northeastern
EP

29 Mauritania, north Mali, southern Algeria, Northwestern Niger) and the Middle East
30 (extending from Iraq to eastern Saudi Arabia). Moderately hygroscopic particles were a
31 mixture of mineral dust and anthropic aerosols when the dust reached receptor site after
32 passing through industrialized and populated sites (such as Spain, France and Italy).
C

33 Hygroscopic particles, however, were ascribed to nearly mesoscale formation of secondary

AC

34 aerosols under the prevailing summer conditions.

35

36
37 Keywords: Aerosol Physical Properties, hygroscopic nature, source regions, Eastern
38 Mediterranean
39
40
41

42

1
 ACCEPTED MANUSCRIPT
43

44 1. Introduction

45 Aerosols play a crucial role in various geophysical and geochemical processes.

46 Atmospheric particles can alter the earth’s radiation and hence climate directly by scattering

47 and absorbing both incoming solar radiation and re-emitted radiation from the surface of the

PT
48 Earth to the atmosphere (Charlson et al., 1992; Haywood and Shine 1995; Haywood and

RI
49 Shine, 1997). Aerosols can also play an essential role in hydrological cycle (Rosenfeld et al.,

50 2008). Atmospheric particles serve as ice nuclei (IN) or cloud condensation nuclei (CNN) and

SC
51 they can influence the concentration and size distribution of cloud droplets; alter the

52 distribution of rainfall and their radiative properties (Levin et al., 2005; Huang et al., 2006;

53 Chen et al., 2007).

U
AN
54 Direct and indirect aerosol forcing tightly depend on geographical distribution, optical
M

55 properties, hygroscopic nature and types of the aerosols in relation to history of air-masses,

56 aging process, external/internal mixing, interaction with gaseous species (heterogeneous),

D

57 gas-to-particle conversion and removal either by wet or dry deposition (Patterson et al., 1977
TE

58 and reference therein; Dubovik et al., 2002; Haywood et al., 2003; Eck et al., 2010; IPCC,

59 2013). High Aerosol Optical Thickness (AOT) values over the globe are also related to
EP

60 distinct processes in different regions and seasons of the year. There is a difference between

61 Northern and Southern Hemispheres, former having higher values compared to latter mainly
C
AC

62 due to deserts (extending from the west coast of North Africa over the Middle East, central

63 and south Asia to China, Prospero et al., 2002; Washington et al., 2003) and industrialized

64 and populated countries (Husar et al., 1997; Kaufman et al., 2002). Relatively higher aerosol

65 burden are found over Mediterranean particularly in MAM and JJA (up to 0.4) period because

66 of dust transport from Sahara and Middle East desert as well as increase in secondary aerosol

67 and biomass burning (Kubilay and Saydam, 1995; Mihalopoulos et al., 1997; Moulin et al.,

2
 ACCEPTED MANUSCRIPT
68 1998; Kubilay et al., 2000; Barnaba and Gobbi, 2004; Sciare et al., 2008; Koçak et al., 2005,

69 2012).

70 A number of studies have dedicated on aerosol optical properties in the atmosphere

71 over the Mediterranean region by applying ground-based real-time sun-photometric

72 measurements (Kubilay et al., 2003; Meloni et al., 2006; Fotiadi et al., 2006; Pace et al., 2006;

PT
73 Gerasopoulos et al., 2007; Santes et al., 2008; Toledano et al., 2009). The short and long term

RI
74 variability in the aerosol optical thickness as well as angstrom exponent and the aerosol type

75 have been the main foci of these articles. These studies have revealed that the change in the

SC
76 aerosol optical, physical properties and the aerosol type might be resulted from a combination

77 of (a) the history of air masses back trajectories, (b) source emission strength, (c) the presence

78
U
or absence of rain as well as (d) gas to particle conversion. However, to our knowledge there
AN
79 is a lack of scientific information about the degree of relationship between water vapor and
M

80 aerosol optical thickness as a function of angstrom exponent to assess the hygroscopic state of

81 aerosols and their possible source regions.

D

82 The current study aims to explore the optical properties of the aerosol burden in the
TE

83 atmospheric column over the Eastern Mediterranean. This will be achieved by using 15-year

84 long data from the AERONET sun/sky radiometer. The unique contribution of the present
EP

85 study will be to (i) explore the climatology of the aerosol optical properties regarding their

86 spectral behavior, (ii) assess the influence of meteorological parameters on the seasonal
C
AC

87 variability in aerosol optical properties, (iii) determine the effect of air flow on the aerosol

88 optical properties and (iv) clarify the hygroscopic state of aerosols and their source regions.

89
90 2. Material and method

91 2.1. Study site and climatology of the region

92
93 The rural IMS-METU-Erdemli (here after Erdemli) atmospheric sampling site is

94 located on the coastline of Eastern Mediterranean (EM), Erdemli, Turkey (36.57°N and

3
 ACCEPTED MANUSCRIPT
95 34.26°E, Figure S1). AErosol RObit NETwork (AERONET) sun photometer was installed at

96 the Institute of Marine Science, Middle Technical University Campus in December 1999 and

97 the instrument has been operated since then. Depending on regional atmospheric dynamics,

98 Erdemli aerosol population is mainly influenced by three aerosol sources namely; (i)

99 anthropogenic particles from industrialized and semi-industrialized regions situated at North,

PT
100 (ii) mineral dust from the Sahara and Middle East Deserts located to the South and (iii) sea

RI
101 salt from the Mediterranean Sea (Kubilay et al., 2000, Koçak et al., 2004; Koçak et al., 2007).

102 Climatically, the weather pattern of the Mediterranean region is characterized by mild,

SC
103 relatively wet winters and hot, dry summers (Lionello et al., 2006). Monthly average

104 atmospheric temperature and rainfall for a 15 year period (2000-2014) at the Erdemli site are

105
U
illustrated in Figure 5h,i. The temperature presents a strong seasonal cycle with a winter
AN
106 minimum and a summer maximum. The average monthly temperature commences to increase
M

107 from April and reaches its maximum in August (28.5 °C) and then decreases to its minimum

108 in January (9.8 °C). On annual average, the site receives 549 mm of rain with a strong
D

109 seasonal cycle. The majority of rainfall (304 mm) is observed in winter whereas very low
TE

110 amount of rain (11 mm) is found in summer. It is clear that the character of the sampling site

111 is representative of the Mediterranean climate.

EP

112

113 2.2. AERONET Measurements

C
AC

114 AERONET is a federated network of ground based sun sky radiometers operating

115 worldwide since the middle of the 1990’s. AERONET provides columnar aerosol properties

116 such as Aerosol Optical Thickness (AOT), Water Vapor (WV) and inversion aerosol

117 products. More detailed descriptions are given in Holben et al. (1998). To summarize, the sun

118 sky radiometer makes direct Sun measurements every 15 minutes at 340, 380, 440, 500, 675,

119 870, 940 and 1020 nm. Except for 940 nm (used for retrieving WV), these measurements are

4
 ACCEPTED MANUSCRIPT
120 then applied to calculate AOT, with an accuracy of ~ 0.01 for the wavelengths longer than

121 440 nm and ~ 0.02 for shorter wavelengths (Eck et al., 1999). The instrument deployed at the

122 Erdemli site has only 440, 500 (after 2004), 675, 870 and 1020 channels for AOT

123 measurements. AOT at 500 nm is decomposed into fine and coarse mode AOT according to

124 the Spectral De-convolution Algorithm by O’Neill et al. (2001, 2003). The Angstrom

PT
125 Exponent (α440-870) is defined by the logarithms of AOT and wavelength (see Equation 1) and

RI
126 it is calculated for the all wavelengths varying from 440 to 870 nm employing a lean fit

127 between AOT and wavelength (λ) (Holben et al., 1991; 2001). The value approaching zero

SC
128 denotes dominance of coarse particles whereas; the value larger than 1 implies the

129 preeminence of fine aerosol (Kaufman et al., 1994, Eck et al., 1999).

U

= −  
 ( )/ ( ) . 1
AN
130 The instrument also performs hourly sky radiance measurements in almucantar
M

131 geometry at 440, 675, 870 and 1020 nm. These sky radiance measurements in conjunction

132 with corresponding direct sun measurements are then used in inversion algorithms to derive
D

133 particle size distribution (PSD) (Dubovik and King 2000; Dubovik et al., 2006). The retrieval
TE

134 error in PSD is estimated to vary from 15 % to 35 % for the intermediate particle size range

135 (0.1 ≤ r ≤ 0.7 µm). However, the error in PSD for small (0.05 ≤ r ≤0.1 µm) and large (7 ≤ r ≤
EP

136 15 µm) particle size interval may reach up to 80 %.

Daily averaged aerosol optical and microphysical properties (Level 2.0, Version 2
C

137
AC

138 AERONET) have been obtained for Erdemli from January 2000 to December 2014. For 15

139 year period, the mean observational coverage of AOT, α and WV was around 50 % (n =

140 2712). The missing values was not only resulted from quality control but also emerged from

141 technical malfunction of the sun photometer instrument. However, the particle size

142 distribution was attained for 1414 days (observational coverage = 26 %).

143

5
 ACCEPTED MANUSCRIPT
144

145

146 2.3. Calculation of the 5 day Air Mass Back Trajectories

147 Five-day back-trajectories were calculated running the HYSPLIT Dispersion Model,

148 employing the meteorological data of the NCEP/NCAR Reanalysis Project (HybridSingle

PT
149 Particle Langrangian Integrated Trajectory; Draxler, 1999; Stein et al., 2015). The air masses

RI
150 arriving at Erdemli at 1, 2, 3 and 4 km above sea level for each day (06:00 UTC) was

151 computed during the period between January 2000 and December 2014 (99.9 % coverage).

SC
152 Five-day backward trajectories were selected since it extends far enough to represent the

153 synoptic scale flow. The vertical motion of air masses was calculated using three-dimensional

154
U
vertical velocity field. Based on the approach by Koçak et al. (2005, 2012), five day back
AN
155 trajectories arriving to Erdemli at the height of 1 km and 4 km were classified into six sectors:
M

156 (i) Western Europe (WE), (ii) Eastern Europe (EE), (iii) Turkey (TUR), (iv) Middle East

157 (MID), (v) Saharan (SAH) and (vi) Mediterranean Sea (MED) in order to evaluate the impact
D

158 of airflow on aerosol optical properties (see Figure S1). The air masses reaching at height of 2
TE

159 and 3 km were omitted since the former and later were quite similar to those arriving at 1 and

160 4 km, respectively.

EP

161

162 2.4. Potential Source Contribution Function

C
AC

163 Possible source regions of atmospheric aerosol or gases which have high concentration

164 at a receptor site could be determined using the spatial probability distribution of pathways of

165 air parcels that arrived at the receptor site over a time period. When the material is emitted

166 into the air parcel from a source, it can be transported to the receptor site throughout the

167 trajectory of the air parcel. If the endpoint of a trajectory stays at a cell of address (i,j), it is

168 assumed that the trajectory collects the emissions from the cell. The total number of

6
 ACCEPTED MANUSCRIPT
169 endpoints that are located in the cell is nij and there exists m ij points for which the measured

170 parameter rises above a criterion value defined for this parameter, then, potential source

171 contribution function (PSCF) is calculated using the following formula (Polissar et al., 2001).

172  =  ⁄ . 2

173 In order to identify the possible source regions of non-hygroscopic, moderately

PT
174 hygroscopic and hygroscopic aerosols (for more details see section 3.5), PSCF was applied.

RI
175 The residence time probability maps (P[Bij]) were constructed when AOT values at 440 nm

176 were larger than 0.4. Air masses back trajectories arriving at 1 km and 4 km altitudes were

SC
177 utilized to define source areas, by dividing the region into 2°X2° grids (910 cells) between

178 20°W-50°E and 20°N-70°N. As stated above, the air masses reaching at height of 2 and 3 km

179
U
were not considered during PSCF analysis (for more information see 2.3). The small number
AN
180 of endpoints in a grid may generate high PSCF values with high uncertainties. After the
M

181 application of binomial distribution at 95 % confidence level (Vasconcelos et al., 1996),

182 statistically significant PSCF values were retained. The PSCF values were subdivided into
D

183 three classes namely: (a) weak (0.0–0.4), (b) intermediate (0.4–0.6) and (c) strong (0.6–1.0).
TE

184

185 3. Results and Discussion

EP

186 3.1. General Characteristics of AOT, Angstrom Exponent, WV and Size Distribution at
C

187 Erdemli
AC

188 Before discussing aerosol optical properties and size distribution, it is worth

189 mentioning the probability distribution of aerosol optical thickness (AOT) measured at

190 Erdemli from January 2000 to December 2014. Kolmogorov-Smirnov test (at 95 %

191 confidence interval) demonstrated that AOT at 440 nm distribution was lognormal (see Figure

192 S2). Therefore, it should be highlighted that the AOT is best represented by geometric mean

193 value. O’Neill et al. (2000) have also shown similar distribution for AOT. Since there is a lack

7
 ACCEPTED MANUSCRIPT
194 of such values in the literature therefore arithmetic mean values will be applied when

195 comparing the current values obtained and those from previous studies.

196 Table 1 illustrates statistical summary for AOT and Angstrom Exponent (α) obtained

197 between January 2000 and December 2014 at Erdemli site. Figure 1 also shows corresponding

198 mean aerosol volume size concentration throughout the study period. As can be seen from the

PT
199 table, arithmetic mean of AOT exhibited a remarkable decrease from 440 nm (0.30) to 1020

RI
200 nm (0.12) and this reduction might be attributed to the dominance of fine particles. The

201 arithmetic mean of AOT (440 nm) was 0.30±0.18 with values ranging between 0.03 and 1.96.

SC
202 Geometric mean and median were identical with a value of 0.26, being 1.15 times less than

203 that of arithmetic mean. Angstrom exponent was found to be varied between 0.03 and 2.23

204
U
with an arithmetic mean value of 1.22±0.37. WV ranged from 0.19 cm to 4.54 cm with an
AN
205 arithmetic mean of 2.04 cm. Mean aerosol volume size concentration demonstrated bimodal
M

206 distribution. The first peak was ascribed to fine particles (radius < 1.25 µm), majority being

207 between 0.1 µm and 0.3 µm whilst the second peak was attributed to coarse particles, chiefly
D

208 existing between 1.5 µm and 6.0 µm. Fotiadi et al. (2006) has reported two pseudo-coarse
TE

209 modes centered around 1.7 and 3.9 µm at Finokalai station, representing sea salt and dust

210 particles, respectively. Unlike Finokalia, there were no pseudo-coarse modes at Erdemli rural
EP

211 site.
C

212
AC

213

214 Table 1. Statistical summary for aerosol optical thickness (AOT: 440, 500, 675, 870, 1020
215 nm), angstrom exponent (α440-870) and WV (cm) at Erdemli between January 2000 and
216 December 2014
Parameter Arithmetic Mean (±δ) Geometric Mean Median Minimum Maximum
AOT1020 0.12±0.11 0.10 0.10 0.01 1.85
AOT870 0.14±0.11 0.11 0.12 0.01 1.87
AOT675 0.18±0.13 0.15 0.15 0.01 1.90
AOT500 0.26±0.16 0.22 0.22 0.01 1.94
AOT440 0.30±0.18 0.26 0.26 0.03 1.96
α440-870 1.22±0.37 1.14 1.30 0.03 2.23
WV 2.04±0.91 1.82 1.90 0.19 4.54
217 δ refers to standard deviation

8
 ACCEPTED MANUSCRIPT
218

PT
RI
SC
219
220 Figure 1. Average volume size distributions from Erdemli station throughout the study period

U
221 (2000-2014).
222
AN
223 The AOT and α values from different geographical regions of the world including free
M

224 troposphere, natural, rural and urban sites were used in order to assess spatial variability (see

225 Figure S3). Figure 2 presents mean AOT440 and α440-870 for each site together with their
D

226 regions and character. Mauna-Lao (Yoon et al., 2012) had the lowest AOT (0.02) since the
TE

227 site is above the planetary boundary layer (altitude of 3397 m). Corresponding α was 1.2,

228 indicating contribution of stratospheric aerosols to the columnar AOT over the site. Despite
EP

229 Lampedusa and Forth CRETE (Mallet et al., 2013) are characterized as natural backgrounds,
C

230 AOT (~ 0.2) and α (~ 1.1) values at these sites were comparable to those of rural sites such as
AC

231 Erdemli/Turkey (this study), Blida/Algeria (Mallet et al., 2013) and Kaashidoo/Maldives

232 (Holben et al., 2001). This peculiarity might chiefly be attributed to the influence of anthropic

233 aerosols on AOT and α when airflow originated from industrialized and semi-industrialized

234 countries located at the northern border of the Mediterranean Sea. On the other hand, AOT

235 and α at San Nicolas (Holben et al., 2001) were respectively 0.05 and 1.2, exhibiting the

236 natural background character of the site. Being a natural site, AOT (0.4) and α (0.3) values at

9
 ACCEPTED MANUSCRIPT
237 Capoverde/Sal Island were remarkably similar to those observed for Dakar (rural coastal),

238 Banizoumbou (rural/semi-arid) and Solar Village (rural). It might be argued that AOT and α

239 values at Sal Island, Dakar and Banizoumbou were heavily modified by mineral dust transport

240 from Sahara desert (Holben et al., 2001, Yoon et al., 2012, Gama et al., 2015). Considering

241 rural sites, AOT (~ 0.5) and α (~ 0.3) at Dakar, Banizoumbou and Solar Village were

PT
242 distinctly different than those of the remaining sites, demonstrating the significant effect of

RI
243 mineral dust particles on columnar AOT and α. The highest AOT values were observed over

244 the urban sites in Asia, namely, Taihu (~ 0.9, Xia et al., 2007) and Beijing (~ 0.7, Yoon et al.,

SC
245 2012). It has been reported that these two Asian sites were under the heavy pressure of

246 regional pollution (Xin et al., 2007). Except for Ilorin (urban semi-arid), the remaining urban

247
U
stations had relatively high α values ranging from 1.12 to 1.68, denoting predominance of fine
AN
248 particles.
M
D
TE
C EP
AC

249

250 Figure 2. Average values of AOT (440 nm) and α (440/870) at the AERONET stations (TS
251 This Study; 1 Yoon et al., 2012, 2 Holben et al., 2001; 3 Mallet et al., 2013, 4 Xia et al., 2007).
252

253

10
 ACCEPTED MANUSCRIPT
254

255

256 3.2. Daily variability in Aerosol Optical Thickness and Angstrom Exponent

257 Figure 3 exhibits daily variability in aerosol optical thickness and angstrom exponent

258 together with corresponding rainfall for Erdemli site between January 2000 and December

PT
259 2014. As can be deduced from the figure, AOT and α denoted a large variability from one day

RI
260 to another with an order of magnitude change in their values (Holben et al., 2001; Smirnov et

261 al., 2002; Eck et al., 2005; Giles et al., 2011). The lowest AOT values was generally

SC
262 associated with rain or the day after these events since wet deposition removes aerosol

263 particles from the atmospheric compartment efficiently (Cadle and Dasch, 1988). For

264
U
instance, one of the lowest AOT was observed on 19th of January 2000 with a value 0.03.
AN
265 Corresponding rainfall for the same day was 15.1 mm. On 19th of January 2000 Angstrom
M

266 exponent was 1.79, indicating the dominance of submicron particles.

267
D
TE
C EP
AC

268
11
 ACCEPTED MANUSCRIPT
269 Figure 3. Time series of daily Aerosol Optical Thickness (a), Angstrom Exponent (b) and
270 amount of rain (c) from January 2000 to December 2014.
271

272 In winter, AOT440 values seldom reached up to 0.6, the highest being observed on 25th

273 of February 2007 (AOT440 = 0.20 ± 0.12; α440-870 = 1.29 ± 0.40). Elevated AOT440 values were

274 identified in spring (AOT440 = 0.31 ± 0.19; α440-870 = 1.06 ± 0.46) and summer (AOT440 = 0.40

PT
275 ± 0.17; α440-870 = 1.32 ± 0.24) however, their corresponding α440-870 values were distinctly

RI
276 different. Higher AOT440 and lower α440-870 were particularly detected during spring months

277 (March, April and May) when the air masses back trajectories originated from deserts situated

SC
278 at North Africa and the Middle East. Studies have clearly shown that aerosol population in the

279 Eastern Mediterranean is considerably affected by sporadic mineral dust pulses mainly

280
U
occurring in transitional seasons (Kubilay and Saydam, 1995; Kubilay et al., 2000, 2003).
AN
281 Enhanced AOT440 and α440-870 were identified throughout summer owing to prevailing
M

282 weather conditions. Higher gas to particle conversion, sluggish airflow and the lack of rain

283 allow accumulation of aerosol particles in the atmosphere over the Eastern Mediterranean
D

284 (Koçak et al., 2007, Dayan et al., 2017 and references therein). Consequently, higher values of
TE

285 AOT might be categorized into three main classes: (a) dust dominated, (b) mixed

286 (dust/pollution) and (c) pollution dominated. During the study period, 46 cases were identified
EP

287 as dust dominated episodes (AOT440 > 0.5; α440-870 <0.5) with mean AOT440 and α440-870 values

288 of 0.74 ± 0.30 and 0.25 ± 0.12, respectively. From 2000 to 2014, 65 and 253 cases were
C
AC

289 respectively characterized as mixed (AOT440 = 0.64 ± 0.12, α440-870 = 0.74 ± 0.15; AOT440 >

290 0.5; α440-870 = 0.5-1.0) and pollution dominated events (AOT440 = 0.62 ± 0.11, α440-870 = 1.40 ±

291 0.17; AOT440 > 0.5; α440-870 > 1.0). These events exhibited that such episodes were mainly

292 associated with pollution dominated aerosol type (~ 70 %). Corresponding examples for each

293 event are presented in Figures 4a, b and c (see Figure S4).

12
 ACCEPTED MANUSCRIPT

PT
RI
SC
294
295 Figure 4. Variability in Aerosol Optical Thickness and Angstrom Exponent (α) during dust
296 (a), mixed (b) and pollution events (c) at Erdemli.

U
297 AN
298 (a) Dust dominated: One of the highest AOT was observed on 2nd of April 2013 with

299 a value of 1.67 (Figure 4a). On March 30, AOT (440 nm) and α were respectively found to be
M

300 0.22 and 1.08. From March 30 to April 1, AOT (440 nm) tripled over the Erdemli in contrast,

301 α value decreased five times. The initial signal of this dust event was clearly detected by the
D

302 sun photometer on April 1. During the next day, AOT (α) incremented (reduced) 2.6 (1.5)
TE

303 times compared to former day, demonstrating intensification of the dust episode over the site.

Ozone Mapping Instrument Aerosol Index (OMI-AI) satellite image for April 1 (Figure S5a)
EP

304

305 demonstrated dense dust cloud over the Levantine basin however; the thickness of dust was
C

306 less over the Northeastern Levantine, OMI-AI being around 1. During this day, OMI-AI
AC

307 values reached up to 4 between coordinates 30˚N−37˚N and 27˚E−30˚E. Within a day (Figure

308 S5b), the dust cloud became less apparent over the South Levantine Basin whilst the intensity

309 of the dust cloud enhanced in the atmosphere over the Northeastern Levantine, especially at

310 Turkish coastline. For instance, AI on April 2 over Erdemli site doubled compared to that of

311 observed on April 1. Five-day back trajectories also supported mineral dust transport from

312 deserts located at the Middle East and North Africa. On April 1 (Figure S5a), trajectories at 1,

13
 ACCEPTED MANUSCRIPT
313 2 and 3 km denoted that Erdemli site was under the influence of airflow from the Middle East

314 whereas; trajectory at 4 km showed air mass transport from North Africa. Next day, lower

315 layer trajectories (1 and 2 km, Figure S5b) demonstrated that site was still affected by airflow

316 from south east. Therefore, prevailing air mass transport mainly from the Middle East and

317 lesser extent from North Africa as well as the absence of rain resulted in enhanced AOT over

PT
318 the Erdemli site. Another example for dust dominated event was observed on 7th of March

RI
319 2009 when AOT440 and α were respectively 1.1 and 0.19 (see Figure S6). Air masses back

320 trajectories exhibited air flow from the Middle East at altitudes of 1 km whilst the air masses

SC
321 reaching at 2, 3 and 4 km heights originated from the North Africa, suggesting the site was

322 under the influence of deserts located in the Middle East and the North Africa at the same

323
U
time. OMI-AI satellite image for March 7 also supported the presence of the dust plume over
AN
324 the Eastern Mediterranean, with AI values exceeding 2.5 particularly over the Erdemli.
M

325 (b) Mixed Event: Mixed dust event was characterized from February 25 to 27, 2007

326 (Figure 4b). On 25 February, the influence of the mineral dust first detected by sun
D

327 photometer with an AOT440 (α) value of 1.06 (0.58). The following day AOT440 was found to
TE

328 decrease to 0.81, together with a slight reduction in α value. On 27 February, AOT440

329 continued to decreased, attaining value of 0.52 with a twofold decrement compared to the
EP

330 initial AOT440 value of the case nonetheless, α value increased 1.9 times from previous two

331 days, reaching value of 0.97. OMI-AI satellite image for 25 February indicated a large dust
C
AC

332 plume over Eastern Mediterranean extending from 32˚N−38˚N to 25˚E−35˚E, which AI

333 values ranging from 2.0 to 3.0 were observed over the region (Figure S7). During the next

334 day, dust plume became less intense and on 27 February dust cloud disappeared over the

335 region. Considering the back trajectories (Figure S7), air masses originated from North Africa

336 (passing over Egypt) arrived to the site at 2, 3 and 4 km altitudes on 25th February, whilst the

337 back trajectory for 1 km altitude indicated airflow over Turkey. During the next day, air flow

14
 ACCEPTED MANUSCRIPT
338 for 1 km altitude showed a transport from region located over Ukraine, middle Black Sea and

339 Turkey whereas air masses at upper levels (3 and 4 km) advected to the site from North

340 Africa. Back trajectories for 27 February indicated air flow at upper levels from North Africa

341 terminated and started to come from northwest. The second example were identified on 14th

342 of May 2010 with AOT440 and α values of 0.54 and 0.71, respectively (see Figure S8). Air

PT
343 masses arriving at 1 and 2 km resulted from Western Sahara and reached the site after

RI
344 sweeping Italy, the Balkans and highly populated as well as industrialized the region of

345 Marmara, including the Mega City Istanbul. On the other hand, the air masses at 3 and 4 km

SC
346 altitude denoted air flow from Western Sahara without interacting urban and/or industrialized

347 areas. OMI-AI diagram also showed dust activity in the North Eastern Mediterranean Basin,

348 ranging from 1.5 to 2.5.

U
AN
349 (c) Pollution Dominated Event: The third event was described from August 25 to 27,
M

350 2014 (Figure 4c). On 25 August, AOT440 and α were respectively found to be 0.24 and 1.21

351 by sun photometer. Next day, AOT440 almost doubled over the Erdemli whereas α was found
D

352 to decrease by 10 %, reaching the value of 1.07. On 27 August, AOT reached the highest
TE

353 value of the event, attaining value of 1.15 with a fivefold increment from 25 to 27 August.

354 Nonetheless, α was found to be 1.08 and the change in Angstrom Exponent from 25 to 27
EP

355 August was only about 10 %, all values being higher than 1. MODIS-Terra (Moderate

356 Resolution Imaging Spectrometer) AOD satellite image for 25 August indicated low thickness
C
AC

357 over site (see Figure S9). Then, next day it started to increase and on 27 August, attained the

358 value of 0.8 with a four-fold increase from 25 to 27 August. Corresponding five-day back

359 trajectories exhibited that air masses arrived to the site at all altitudes from north and

360 northwest on 25 August (see Figure S9). Next day, lower layers (1 and 2 km) back trajectories

361 was related to air flows coming from northern sector whereas, upper layers (3 and 4 km) back

362 trajectories were associated with air mass transport from south. On 27 August, trajectory at 1

15
 ACCEPTED MANUSCRIPT
363 km indicated the site still remained under influence of air mass transport from the region

364 located over eastern Black Sea and Turkey. On the other hand, trajectory at 2, 3 and 4 km

365 pointed the air mass transport from southwest. It should be noted that on 26 and 27 August,

366 air masses reached the site at 3 and 4 km, descending vertically from further height levels

367 nonetheless, the trajectory at 1 km on 26 August and trajectories at 1 and 2 km on 27 August

PT
368 indicated that air masses arrived to the site at corresponding heights, ascending vertically

from ground levels (see Figure S10). The case observed on 14th of July 2013 can be given as

RI
369

370 another example for pollution dominated episodes with corresponding AOT440 and α values of

SC
371 0.60 and 1.43 (see Figure S11). Air masses back trajectories exhibited that all air flows were

372 arising from industrialized, semi- industrialized and populated areas located on the North of

373
U
the Mediterranean Sea. OMI-AI satellite image illustrated that the Eastern Mediterranean was
AN
374 not under the influence of dust particle, being less than 0.5.
M

375 3.3. Seasonal variability in Aerosol Optical Thickness and Angstrom Exponent.

376 Monthly arithmetic mean and standard deviations for aerosol optical thickness (440
D

377 and 870nm, coarse and fine at 500 nm), angstrom exponent, water vapor together with local
TE

378 rain amount and temperature are presented in Figure 5. The lowest values of AOT (440, 870

379 nm, fine and coarse) were found in winter. The minimum in winter might be ascribed to four
EP

380 main processes namely (a) efficient removal of aerosol burden from the atmospheric

381 compartment by rain, (b) reduced gas to particle conversion due to cloud cover and incoming
C
AC

382 solar flux, and (c) less frequent dust transport from the Middle East and North Africa (d)

383 emission strength of sources (Kubilay et al., 2003; Fotiadi et al., 2006; Gerosopoulos et al.,

384 2007). Notwithstanding, there was a distinct difference in the seasonal cycle of AOT440,

385 AOT870, AOTcoarse and AOTfine. To begin with AOT870 (AOTcoarse), monthly arithmetic mean

386 gradually increased from January to March and reached its maximum in April with a value of

387 0.22 (0.17). Arithmetic mean of AOT870 (AOTcoarse) then reduced 40 % (60 %) from April to

16
 ACCEPTED MANUSCRIPT
388 June. AOT870 (AOTcoarse) demonstrated 20 % (only 5 %) enhancement from June to August.

389 Mean AOT870 (AOTcoarse) abruptly declined from 0.18 (0.09) to 0.08 (0.05) between August

390 and December. On the whole, arithmetic mean of AOT440 (AOTfine) illustrated piecemeal

391 augmentation from January to August, attaining its maximum with a value of 0.44 (0.27).

392 After that, mean AOT440 (AOTfine) began to decrease and achieved its winter values in

PT
393 December (AOT440 ~ 0.18, AOTfine ~ 0.10). Monthly mean of α440-870 denoted steady decrease

RI
394 from 1.40 to 0.94 between January and April. Then α440-870, raised about 40 %, reaching value

395 of 1.30 in June whilst α440-870 did not pointed out remarkable change from June to December.

SC
396 The seasonal cycle of water vapor, on the other hand, was similar to that of observed for

397 AOTfine. Monthly mean water vapor exhibited substantial increase from January (1.0 cm) to

398
U
August (3.2 cm) and later its monthly values drastically dropped to 1.1 cm towards end of the
AN
399 year.
M
D
TE
C EP
AC

400
17
 ACCEPTED MANUSCRIPT
401 Figure 5. Mean monthly values of AOT440 (a), AOT870 (b), coarse mode AOT (c) and fine
402 mode AOT (d) at 500 nm, α440-870 (e), water vapor (f), total rain (h) and temperature (i). The
403 bars represent standard deviations.
404 The AOT versus α440-870 gives qualitative information about the aerosol burden and

405 particle size and hence it is a useful tool to distinguish antropic from natural aerosols (Eck et

406 al., 2001; Holben et al., 2001; Smirnov et al., 2002; Fotiadi et al., 2006). Seasonal scatter

PT
407 diagrams for AOT440 versus α440-870 together with the frequency of occurrence of these two

408 parameters are depicted in Figure 6. As can be deduced from scatter plots, winter (DJF) and

RI
409 spring (MAM) showed a strong trend of increase in AOT440 values as α440-870 decreases,

SC
410 though, summer (JJA) and fall (SON) exhibited a very weak trend of reduction in AOT440

411 values with decreasing α440-870. Seasonal histograms for AOT440 and α440-870 also denoted

U
412 distinct difference. The frequency of occurrence histogram of AOT440 showed gradual shift
AN
413 toward higher values from winter to summer and it decreased from summer to fall. In winter,

414 spring and fall, the AOT440 probability distributions exhibited peak values of 0.1-0.2 (51 %)
M

415 0.2-0.3 (35 %) and 0.1-0.2 (32 %), respectively, whereas AOT440 illustrated peak value of 0.3-

0.4 (25 %) in summer. The AOT440 frequency distributions clearly showed that the
D

416

417 atmosphere over the Eastern Mediterranean was more turbid in summer than those observed
TE

418 for the remaining seasons. As stated before, this might be result of higher gas to particle
EP

419 conversion, sluggish airflow and lack of rain. The frequency of occurrences histogram for

420 α440-870 displayed relatively narrow peak from 1.1 to 1.6 in summer (77 %) and fall (56 %) on
C

421 the other hand, only few percent had values less than 0.5 (summer: 1.2 %, fall: 3.5 %),
AC

422 showing that fine particles were dominant. Unlike summer and fall, the α440-870 probability

423 distribution was rather broader (wide range particle size) in spring, 16 % being lower than 0.5.

424 About 50 % of α440-870 < 0.5 values had AOT440 larger than 0.4, implying the transport of

425 mineral dust from North Africa and Middle East deserts (Koçak et al., 2004). The remaining

426 episodes might be attributed to weak dust events and/or sea salt. Furthermore, more than 16 %

427 of the α440-870 in winter, spring and fall ranged from 0.5 to 1.0, suggesting mixed mineral dust

18
 ACCEPTED MANUSCRIPT
428 and anthropic particles. Therefore, it might be argued that the aerosol population was

429 comparatively more diverse in spring compared to winter, summer and fall.

PT
RI
SC
430

U
AN
M
D
TE

431
C EP
AC

432
433 Figure 6. Seasonal scatter plot of AOT440 versus α440-870 (a-b) and frequency of occurrence of
434 AOT440 and α440-870 for winter (c-d), spring (e-f), summer (g-h), and fall (i-j).
435

19
 ACCEPTED MANUSCRIPT

PT
RI
436

U SC
AN
M

437
D

438 Figure 6. continue ...

439
TE

440 Correlations between variables illustrate the degree to which they together. In order to
EP

441 clarify monthly relationship between AOT, α440-870, water vapor and local meteorological

442 parameters, namely temperature and rain were used. The individual monthly means from
C

443 2000 to 2014 were applied to carry out correlation analysis when the number of data were at
AC

444 least 15 days per month, remaining 98 out of 132 months. The significance level of the

445 correlation coefficient is considerably influenced by the number of the sample. Therefore,

446 even a small correlation coefficient can be considered as statistically significant in

447 populations with a high number of samples. Subsequently, Three different terms will be

448 utilized during the interpretation of correlation coefficients: (a) weak correlation (r = 0-0.3),

449 (b) moderate correlation (r = 0.3-0.6) and (c) strong correlation (r = 0.6-1.0).

20
 ACCEPTED MANUSCRIPT
450 The correlation coefficients between aforementioned variables are given in Table 2.

451 The obtained correlation coefficients indicated that water vapor, temperature and total rain

452 amount significantly modified the seasonality of aerosol physics in the atmosphere over the

453 Erdemli. Rain was moderately correlated with water vapor (r = -0.37), denoting the removal

454 of the water from the atmosphere column in the form of precipitation. Except for AOTcoarse,

PT
455 there were negative correlation between rain amount and AOT870, AOT440 and AOTfine.

RI
456 However, the correlation coefficients were only moderate for AOT440 (r = -0.30) and AOTfine

457 (r = -0.42). In addition, AOT440 (r = 0.63) and AOTfine (r = 0.70) showed strong correlation

SC
458 coefficients with temperature. Therefore, during winter month’s AOT440 and AOTfine

459 decreased not only due to the removal of particles from atmospheric compartment but also

460
U
because of less efficient gas to particle conversion (e.g. cloudy days, inferior incoming solar
AN
461 influx). Influence of rain on AOT870 (r = -0.12) and AOTcoarse (r = 0.11) was not important and
M

462 hence this might be ascribed to the mechanical formation of coarse particles and their source

463 strength. For example, atmospheric sea salt production especially increases in rainy winter
D

464 months owing to more active and frequent storms (Marks, 1990; O'Dowd, C. D. and M. H.
TE

465 Smith, 1993). Furthermore, in spite of considerable rain, the mineral dust transport from

466 desert sources to the receptor site has primarily been determined in spring (March, April and
EP

467 May) when the prevailing conditions are more favorable for the Saharan cyclones (Kubilay et

468 al., 2000). Angstrom Exponent indicated strong negative correlation with AOTcoarse (-0.77)
C
AC

469 although it showed moderate negative correlation coefficient with AOT870 (-0.49).

470 Considering these relationship, it might be note that the seasonality of α440-870 was mainly

471 controlled by the existence of coarse particles.

472 Table 2. Matrix correlation for individual monthly (n = 98) AOT (440 nm), AOT (870 nm),
473 fine as well as coarse AOT, α (440/870), water vapor (WV), temperature (T) and total rain
474 (R). Bold and italic characters illustrate strong and moderate correlation coefficients,
475 respectively.
476
AOT870 AOT440 AOTcoarse AOTfine α440-870 WV T R

21
 ACCEPTED MANUSCRIPT
AOT870 1.00
AOT440 0.80 1.00
AOTcoarse 0.85 0.38 1.00
AOTfine 0.49 0.91 -0.03 1.00
α440-870 -0.49 0.02 -0.77 0.52 1.00
WV 0.38 0.76 0.01 0.87 0.27 1.00
T 0.31 0.63 0.02 0.70 0.22 0.90 1.00
R -0.12 -0.30 0.11 -0.42 -0.11 -0.37 -0.41 1.00
477

PT
478 3.4. Influence of Air Flow on Aerosol Optical Properties

479 The origin of air masses influence the aerosol optical thickness and angstrom exponent

RI
480 (Smirnov et al., 1995; Smirnov et al., 1996; Eck et al., 2008; Giles et al., 2011). To assess the

SC
481 influence of air flow on aerosol optical properties, the categorization of trajectories were

482 utilized. The Kruskal-Wallis (K-W) test was applied to test for the presence of significant

U
483 differences in aerosol optical properties namely aerosol optical thickness at 440, 675, 870 as
AN
484 well as 1020 nm, angstrom exponent and fine fraction (%) classified according to air masses.

485 Table 3 and Figure 7 illustrate statistical summary of the AOT, α440-870 and FF (%) for each air
M

486 flow sector.

D

487 Table 3. Arithmetic means and standard deviations for AOT, AE and fine mode fraction as a
488 function of air flow for 1 km (a) and 4 km (b).
TE

a) WE EE TR MID SAH MED

(n=327) (n=1028) (n=771) (n=147) (n=194) (n=244)
Mean±Std Mean±Std Mean±Std Mean±Std Mean±Std Mean±Std
AOT440 0.27 ± 0.15 0.28 ± 0.16 0.34 ± 0.18 0.36 ± 0.25 0.35 ± 0.20 0.29 ± 0.15
EP

AOT675 0.17 ± 0.11 0.15 ± 0.09 0.20 ± 0.11 0.28 ± 0.24 0.26 ± 0.19 0.17 ± 0.09
AOT870 0.13 ± 0.09 0.11 ± 0.07 0.15 ± 0.09 0.25 ± 0.24 0.23 ± 0.18 0.13 ± 0.07
AOT1020 0.12 ± 0.09 0.09 ± 0.06 0.12 ± 0.08 0.23 ± 0.24 0.21 ± 0.17 0.11 ± 0.06
α440-870 1.1 ± 0.4 1.3 ± 0.3 1.3 ± 0.3 0.8 ± 0.4 0.8 ± 0.4 1.3 ± 0.3
C

FF 0.62 ± 0.17 0.73 ± 0.11 0.71 ± 0.14 0.49 ± 0.19 0.47 ± 0.19 0.68 ± 0.13
489 FF, std denotes fine mode fraction at 500 nm and standard deviation, respectively.
AC

b) WE EE TR MID SAH
(n=1508) (n=325) (n=273) (n=81) (n=511)
Mean±Std Mean±Std Mean±Std Mean±Std Mean±Std
AOT440 0.30 ± 0.16 0.21 ± 0.11 0.33 ± 0.17 0.46 ± 0.26 0.37 ± 0.20
AOT675 0.17 ± 0.11 0.11 ± 0.06 0.18 ± 0.10 0.30 ± 0.18 0.26 ± 0.17
AOT870 0.13 ± 0.09 0.08 ± 0.04 0.13 ± 0.08 0.24 ± 0.15 0.22 ± 0.16
AOT1020 0.11 ± 0.08 0.07 ± 0.03 0.11 ± 0.07 0.21 ± 0.14 0.20 ± 0.15
α440-870 1.3 ± 0.3 1.4 ± 0.3 1.4 ± 0.3 1.1 ± 0.4 0.9 ± 0.4
FF 0.70 ± 0.14 0.73 ± 0.12 0.74 ± 0.12 0.62 ± 0.17 0.55 ± 0.20
490 FF, std denotes fine mode fraction at 500 nm and standard deviation, respectively.
491

22
 ACCEPTED MANUSCRIPT
492 Air flow reaching at altitude of 1 km: Table 3.a and Figure 7 illustrate statistical summary of

493 the AOT, α440-870 and FF (%) for each air flow sector at 1 km. Application of Kruskal-Wallis

494 test exhibited that aerosol optical thicknesses at 440 nm for MID, SAH and TR were not

495 statistically different. Likewise, AOT for MED, EE and WE was found to be comparable and

496 they were statistically not different. AOT for MID, SAH and TR was at least 15 % larger than

PT
497 those observed for MED, EE and WE. On the other hand, mean AOT for air flow showed

RI
498 dramatic difference with increasing wavelength. For MID and SAH air flow, AOT exhibited

499 slight decrease from 440 nm to 1020 nm, indicating dominance of coarse particle. Angstrom

SC
500 exponent and the percent contribution of fine fraction in these air masses were also support

501 the above statement, former and later being around 0.75 and 47 %, respectively. Conversely,

502
U
AOT for MED, EE, WE and TR denoted almost 65 % decrease between 440 nm and 1020 nm
AN
503 since aerosols arising from these air flow sectors were mainly associated with fine mode (α440-
M

504 870 > 1.1 and FF > 62 %, see Table 3.a and Figure 7). Results from K-W test demonstrated

505 statistically significant difference for AOT between MID, SAH and MED, EE, WE as well as
D

506 TR at 675 nm, 870 nm and 1020 nm, first two sectors being approximately 2 times higher
TE

507 than the remaining air flows. Furthermore, angstrom exponent and the percent contribution of

508 fine fraction were remarkably lower for MID and SAH compared to air flow originated from
EP

509 MED, EE, WE and TR.

510 Air flow reaching at altitude of 4 km: The results from K-W test for 4 km air flow was less
C
AC

511 dramatic relative to 1 km, suggesting influence of different air masses in the atmospheric

512 column at the same time. For example, the air flow reaching at 4 km altitude from SAH sector

513 exhibited diverse air masses transport at 1 km. Correspondingly, 18, 5, 17, 21, 27 and 12 % of

514 trajectories at 1 km altitude were associated with air flow from EE, MED, MID, SAH, TR and

515 WE when air masses arriving at 4 km altitude from SAH. Figure 8.a and b show AOT at 1020

516 nm and angstrom exponent according to air masses transport from EE, MED, MID, SAH, TR

23
 ACCEPTED MANUSCRIPT
517 and WE for SAH air flow at 4 km altitude. As expected, K-W test exhibited dramatic

518 difference for air masses categories. For AOT at 1020 nm, air masses back trajectories arising

519 from SAH and MID was remarkably larger (at least 30 %) than those observed for the

520 remaining air flow sectors. Angstrom exponent of MID and SAH was also statistically

521 different than the other sectors, the former group being less than 0.62 and the latter group

PT
522 being higher than 0.86. Therefore, it might be suggested that the impact of mineral dust

RI
523 transport was less obvious in the atmospheric column when the air flows originated from

524 industrialized and populated areas due to the dilution effect of anthropic aerosols.

U SC
AN
M
D
TE
C EP
AC

24
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP

525
C

526 Figure 7. Variability in aerosol optical thickness (a), fine fraction and angstrom exponent (b)
527 according to the clusters at 1 km.
AC

528
529

25
 ACCEPTED MANUSCRIPT

PT
RI
U SC
530
AN
531
M
D
TE
C EP
AC

532
533 Figure 8. Bow-whisker plot of AOT at 1020 nm (a) and angstrom exponent (b) according to
534 air masses transport from EE, MED, MID, SAH, TR and WE at 1 km when air masses
535 arriving at 4 km altitude from SAH.SE refers to standard error.
536

537

538 3.5. Hygroscopic Nature of Aerosols and their Potential Source

26
 ACCEPTED MANUSCRIPT
539 In order to investigate relationship between water vapor and AOT, value of α440-870 has

540 generally been classified into groups namely α440-870 > 0.70 (or 0.75) and α440-870 < 0.70 (or

541 0.75) since the threshold for α440-870 tentatively differentiate anthropogenic dominated aerosol

542 population from natural (Smirnov et al., 2002; Eck et al., 2008). It should be noted that the

543 application of such an arbitrary threshold might not fully separate aerosol type from the one

PT
544 another, yet this type of application is inevitable because it provides simplicity to explore

RI
545 hygroscopic growth or hygroscopic nature of aerosol type. Nevertheless, in this study the

546 relationships between these two variables were investigated for three different α classes

SC
547 specifically: (i) α440-870 < 0.5, (ii) 0.5 < α440-870 < 1.0 and (iii) 1.0 < α440-870 < 2.0. The first

548 group was dominated by coarse particles, accounting of 75 % of AOT440 whilst the second

549
U
group was equally influenced by coarse (~ 51 %) and fine (~ 49 %) particles. In contrast, the
AN
550 third class was dominated by fine particles, elucidating 80 % of the observed AOT440. The
M

551 relationship between AOT440 and water vapor is shown in Figure 9.The logarithmic fit was

552 found to explain much more of the variance between AOT440 and water vapor compared to
D

553 linear fit. In general, AOT440 increased with increasing water vapor, however; the degree of
TE

554 relationship between AOT440 and water vapor for each group was different. The correlation

555 coefficient between two variables was found to increase with enhancing α. The relationship
EP

556 for the first group was weak (R2 = 0.13, p < 0.0002, Figure 9.a), suggesting that the particles

557 in this group exhibited the least hydroscopic growth compared to the remaining groups. The
C

relationship for the second group was moderate (R2 = 0.36, p < 0.0002, Figure 9.b), denoting
AC

558

559 moderate hydroscopic growth. As it is well known, the sea salt has a high hygroscopicity

560 (Petters and Kreidenweis, 2007) with a small α440-870. Thus, the first group might be

561 influenced by sea spray. The sea salt is produced by bubble bursting which commences at a

562 wind speed of about 3-4 m s-1 (O’Dowd et al., 1997) and therefore its production is greatly

563 related to the local wind speed. In order to assess the impact of sea salt on these groups the

27
 ACCEPTED MANUSCRIPT
564 relationship of wind speed with water vapor and AOT was investigated when the local wind

565 originated from sea. The regression analysis did not reveal any statistically significant

566 relationship between wind speed and AOT (first group: r = -0.1; for second group: r = -0.3) as

567 well as water vapor (first group: r = -0.1; for second group: r = -0.02). It is seem that the

568 aerosols in the first group was mainly dominated by mineral dust with a low hygroscopicity

PT
569 (Koehler et al., 2009). Moderate hygroscopic property of the second group might be ascribed

RI
570 to mixing of different type aerosols such as mineral dust and anthropogenic sulfates, nitrates

571 and/or organic acids (Koçak et al., 2007, 2012). If the first and second group were originated

SC
572 from distinct aerosol populations, then the former would show less interaction with

573 anthropogenic sulfates, nitrates and/or organic acids relative to later. The ratio of nssSO42-

574
U
/nssCa2+ is a useful tool to investigate such an interaction. The ratio of 0.4 ± 0.1 for nssSO42-
AN
575 /nssCa2+ has been suggested for non-enriched cases, in other words, gypsum mainly arising

from dust (Putaut et al., 2004). Regarding, the nssSO42-/nssCa2+ ratio presented by Koçak et
M

576

577 al. (2012) for Erdemli in October 2007 and April 2008, the two individual cases may at least
D

578 provide further information about the chief aerosol populations of the first and second groups.
TE

579 On 14 April 2008, AOT440 and α440-870 were respectively 0.68 and 0.14 whereas AOT440 and

580 α440-870 were 0.58 and 0.96 on 11 October 2007, respectively. Correspondingly nssSO42-
EP

581 /nssCa2+ ratios (Al concentrations) for 14 April 2008 and 11 October 2007 were reported

582 around 0.5 (5.3 µg m-3) and 1.5 (2.6 µg m-3) during these dust events (Koçak et al., 2012),
C
AC

583 implying that the former event was less interacted with man-made particles than that of

584 observed for later. Consequently, it might be argued that the hygroscopic capacity of the first

585 group was not remarkably modified during long range transport of dust particles into the

586 region.

587 The coefficient of determination (R2 = 0.60, p < 10-5, Figure 9.c) for the last group was

588 strong with a value of 0.60, stating that aerosols associated with this class exhibit a large

28
 ACCEPTED MANUSCRIPT
589 hydroscopic growth. As stated above, the third category was chiefly associated with

590 hygroscopic fine particles, which may include ammonium sulfate and ammonium nitrate

591 (Tang and Munkelwitz, 1994). Simirnov et al. (2002) have shown similar relationship

592 between AOT500 and water vapor for coarse (R2 = 0.2, α440-870 <0.7) and fine (R2 = 0.67, α440-

593 870 > 0.7) mode at Bahrain/Arabian Gulf. The increasing trend of AOT500 as augmenting water

PT
594 vapor has been reported for a low humid site in Hamim/Arabian Gulf, nonetheless, the

relationship between these two variables was found to be independent of aerosol type (R2 ~

RI
595

596 0.30 for both coarse and fine mode, Eck et al., 2008) .

U SC
AN
M
D
TE

597
C EP
AC

598
599 Figure 9. Aerosol optical thickness versus water vapor with regression analysis for the values
600 of α440-870 < 0.5 (a), 0.5 < α440-870 < 1.0 (b), and 1.0 < α440-870 < 2.0 (c).

29
 ACCEPTED MANUSCRIPT
601

602 In addition to α440-870, OMI-AI was also distinctly different for the three groups (Man-

603 Whitney U-test p < 0.002 for 1st and 2nd groups, p < 10-5 for 1st and 3rd as well as 2nd and 3rd

604 groups, see Figure S12). Correspondingly, arithmetic mean values of OMI-AI were 1.4, 1.1

605 and 0.8 for the first, second and third groups (AOT440 > 0.4). Therefore, regarding the

PT
606 difference in the aerosol types, the first, second and third classes were grouped into three

RI
607 categories namely: (a) Non-hygroscopic dominated, (b) moderately hygroscopic and (c)

608 hygroscopic dominated.

SC
609 Results of PSCF analysis for non-hygroscopic, moderately hygroscopic and

610 hygroscopic aerosols are presented in the maps with a color scale (Figure 10). Regarding

611
U
PSCF maps for each group the following observations may be made:
AN
612 (a) Non-hygroscopic dominated (α440-870 < 0.5, Figure 10a-b; n =80) Aerosols: As expected,
M

613 the 80 % of such particles were characterized in spring due to the frequent dust intrusions in

614 to the region. The PSCF map for 1 km level indicates the presence of intermediate and strong
D

615 PSCF valued grid cells in North Africa and the Middle East between coordinates 22˚N−34˚N
TE

616 and 8˚E−48˚E (Figure 10a). Particularly, the strong potential probabilities were located at

617 regions covering patches of Southern Tunisia, Western Libya, Libya/Chad border, southern
EP

618 Egypt in North Africa and areas extending from Iraq to eastern Saudi Arabia in the Middle

619 East. On the other hand, the map for 4 km level only shows the major source regions located
C
AC

620 on North Africa (Figure 10b). It could be interpreted that mineral dust transported to the

621 region at high altitudes, which was characterized by low angstrom exponent, originated

622 mainly from North Africa. The intermediate and strong PSCF valued cells for 4 km level took

623 place between 16˚N-30˚N and 10˚W-30˚E. The map identifies northeastern Mauritania and

624 part of north Mali, southern Algerian Northwestern Niger as strong potential source locations

625 (Figure 10b). Identified source regions in North African and Middle East are in

30
 ACCEPTED MANUSCRIPT
626 correspondence with previous studies (Washigngton et al., 2003, Ginoux et al., 2012,

627 Gherboudj et al., 2017). For instance, Gherboudj et al. (2017) highlighted Chott el-Jerid

628 (Southern Tunisia), Aljafra plain/ Nafusa mountains slopes (Tunisia/Libya), Tibesti

629 mountains slopes (Libya-Chad border), river drainage basin of the Aiir (southern Algerian),

630 Erg of Bilma (Northwestern Niger) and Nubian desert (Sudan/Egypt) as important dust

PT
631 emission source areas, which was related to the source regions identified in this study.

RI
632 (b) Moderately hygroscopic (0.5 <α440-870 <1.0, Figure 10c-d; n = 110) Aerosols: These

633 particles were almost equally identified in spring (31 %), summer (32 %) and fall (28 %). The

SC
634 PSCF map for 1 km level identifies similar region as the map of 1 km level for α440-870 < 0.5.

635 However, a wide area of Western Europe from Spain to Italy and other small areas in France

636
U
and Germany appeared to be potential source locations (Figure 10c). These results show that
AN
637 air masses belonging to this group was influenced by both anthropogenic and natural aerosols
M

638 before reaching to Erdemli (Koçak et al., 2007). As for the PSCF map for 4 km level, region

639 stretching from Northeastern Sudan to Syria were also identified as possible source locations
D

640 compared to first group’s 4 km level PSCF map (Figure 10d).

TE

641 (c) Hygroscopic dominated (1.0 < α440-870, Figure 10 e-f; n = 484) Aerosols: About 80 % of

642 the hygroscopic aerosols were observed in summer. The PSCF map for 1.0 < α440-870 doesn’t
EP

643 clearly show any significant source regions. This may mainly point to the local sources

644 formed under summer conditions. Similar PSCF patterns have been demonstrated for
C
AC

645 secondary aerosols measured in Erdemli (Koçak et al., 2009).

31
 ACCEPTED MANUSCRIPT

PT
RI
646

U SC
AN
M
D

647
TE

648
EP
C
AC

649
650

651
652

32
 ACCEPTED MANUSCRIPT
653 Figure 10. Distribution of PSCF values at 1km and 4km for the values of α440-870 < 0.5 (a-b),
654 0.5<α440-870 < 1.0 (c-d), 1.0<α440-870 (e-f). Explanation of abbreviation: Spain (Sp), France (Fr),
655 Germany (G), Poland (P), Ukraine (U), Turkey (Tr), Syria (S), Iraq (I), Saudi Arabia (S.A),
656 Egypt (E), Libya (L),Tunisia (T), Algeria (A), Morocco (M), Mauritania (Mau), Mali (Ma),
657 Niger (N), Chad (C) and Sudan (Su).
658

659 4. Summary and Conclusion

PT
660 The present study has produced a unique and extensive dataset of temporal aerosol optical and

661 physical properties in the atmosphere over the rural site located at the coast of the Eastern

RI
662 Mediterranean. From these findings the following summary may be made:

SC
663 Aerosol Optical Thickness (AOT) exhibited log-normal distribution, therefore; AOT is

664 best characterized by it geometric mean concentration, being 15 % less than that of arithmetic

U
665 mean at 440 nm. The arithmetic mean AOT decreased 2.5 times between 440 nm (0.30) and
AN
666 1020 nm (0.12), demonstrating the dominance of fine particles. Angstrom Exponent (α =

667 1.22), particle size distribution (bi-modal distribution, majority being between 0.1 µm and 0.3
M

668 µm) and fine mode fraction (68 %) contribution also showed that the aerosol burden was

chiefly associated with fine particles.

D

669

670 The lowest AOT values was mainly found in winter during rain or just the day after
TE

671 rain since wet deposition removes aerosol particles from the atmospheric compartment
EP

672 efficiently. Enhanced AOT values at 440 nm was observed in spring with low α440-870 owing

673 primarily to sporadic mineral dust intrusions originated from North Africa and the Middle
C

674 East. High AOT at 440 nm and α440-870 were principally characterized in summer because of
AC

675 high gas-to particle conversion, sluggish air masses and absence of rain.

676 Seasonal frequency distribution for AOT at 440 nm exhibited that summer (25 %

677 being between 0.3 and 0.4) was more turbid than those of the remaining seasons. 50 % of the

678 α440-870 < 0.5 in spring was found to be associated with AOT values higher than 0.4, indicating

679 impact of dust transported from deserts located at North Africa and the Middle East. The

680 seasonality of the aerosol optical and physical properties was remarkably modified by water

33
 ACCEPTED MANUSCRIPT
681 vapor, temperature and rain events. AOT440 and AOTfine showed statistically significant

682 negative and positive correlation with rain and temperature, respectively.

683 Results of air flow categorization for 1 km altitude exhibited that elevated AOT with

684 relatively lower α (0.7) and fine mode fraction (47 %) were observed when the air masses

685 originated from North Africa (SAH) and the Middle East (MID). Mean AOT also showed

PT
686 slight decrease between 440 nm and 1020 nm for SAH and MID air flows, denoting

RI
687 importance of coarse mode particles, whereas, this decrease was 65 % for air flow from

688 Mediterranean Sea, Eastern Europe, Western Europe and Turkey with comparatively higher α

SC
689 (> 1.1) and fine mode fraction (> 62 %).

690 Regarding the relationship between AOT and Water Vapor (WV) as a function of α,

691
U
three classes of aerosol were identified: (a) non-hygroscopic dominated, (b) moderately
AN
692 hygroscopic and (c) hygroscopic dominated. Potential source contribution function (PSCF)
M

693 maps exhibited that non-hygroscopic particles were associated with mineral dust dominated

694 aerosol population, originating from North Africa (Southern Tunisia, Western Libya,
D

695 Libya/Chad border, southern Egypt northeastern Mauritania, north Mali, southern Algeria,
TE

696 Northwestern Niger) and the Middle East (extending from Iraq to eastern Saudi Arabia).

697 Moderately hygroscopic particles, on the other hand, were a mixture of mineral dust and
EP

698 anthropic aerosols due mainly to arrival of the dust after passing through industrialized and

699 populated sites (such as Spain, France and Italy). PSCF maps did not show any particular
C
AC

700 potential source area for hygroscopic particles, exhibiting nearly mesoscale formation of

701 secondary aerosols under the prevailing summer conditions.

702 Aerosols play an essential role in precipitation acting as ice and/or cloud condensation

703 nuclei. Most of the non-hygroscopic cases were observed in spring while majority of the

704 hygroscopic cases were characterized in summer. On the other hand, moderately hygroscopic

705 events were almost exclusively uniformly found in spring, summer and fall. Taking into

34
 ACCEPTED MANUSCRIPT
706 account the origin, composition, air masses history and aging processes, there is a clear need

707 to explore the influence of such particles on the hydrological cycle in the Eastern

708 Mediterranean.

709 Acknowledge

710 This study was supported by METU project (BAP-08-11-2014-021). We acknowledge the

PT
711 GSFC/NASA AERONET group for production of sun photometer data sets used in this
712 research effort. Satellite aerosol optical thickness and aerosol index values from satellite
713 instruments used in this study were produced with the Giovanni online data system,

RI
714 developed and maintained by the NASA GES DISC. We also acknowledge the MODIS and
715 OMI mission scientists and associated NASA personnel for the production of the data used in
716 this research effort.

SC
717
718 References

U
723 Barnaba, F., Gobbi, G.P., 2004. Aerosol seasonal variability over the Mediterranean region
724 and relative impact of maritime, continental and Saharan dust particles over the basin from
AN
725 MODIS data in the year 2001. Atmos. Chem. Phys. 4, 2367–2391.
726
727 Cadle, S.H., Dasch, J. M., 1988. Wintertime concentrations and sinks of atmospheric
728 particulate carbon at a rural location in northern Michigan. Atmos. Environ. 22, 1373–1381.
M

729
730 Charlson, R.J., Schwartz, S.E., Hales, J.M., Cess, R.D., Coakley, J.A., Hansen, J.E.,
731 Hofmann, D.J., 1992. Climate forcing by anthropogenic aerosols. Science 255, 423–430.
D

732
733 Chen, Y., Mills, S., Street, J., Golan, D., Post, A., Jacobson, M., Paytan, A., 2007. Estimates
TE

734 of atmospheric dry deposition and associated input of nutrients to Gulf of Aqaba seawater. J.
735 Geophys. Res. 112, D04309.
736
Dayan, U., Ricaud, P., Zbinden, R., and Dulac, F., 2017. Atmospheric pollution
EP

737
738 concentrations over the Eastern Mediterranean during summer – A review, Atmos. Chem.
739 Phys. Discuss. 17, 13233-13263, https://doi.org/10.5194/acp-17-13233-2017.
740
C

741 Draxler, R.R., 1999. HYSPLIT4 user's guide. NOAA Tech. Memo. ERL ARL-230, NOAA
742 Air Resources Laboratory, Silver Spring, MD.
AC

743
744 Dubovik, O., King, M.D., 2000. A flexible inversion algorithm for retrieval of aerosol optical
745 properties from Sun and sky radiance measurements. J. Geophys. Res. 105, 20673-20696.
746
747 Dubovik, O., Holben, B., Eck, T.F., Smirnov, A., Kaufman, Y.J., King, M.D., Tanre, D.,
748 Slutsker, I., 2002. Variability of absorption and optical properties of key aerosol types
749 observed in worldwide locations. J. Atmos. Sci. 59, 590–608.
750
751 Dubovik, O., Sinyuk, A., Lapyonok, T., Holben B.N., Mishchenko, M., Yang, P., Eck, T.F.,
752 Volten, H., Muñoz, O., Veihelmann, B., van der Zande, W.J., Leon, J.F., Sorokin, M.,
753 Slutsker, I., 2006. Application of spheroid models to account for aerosol particle
754 nonsphericity in remonte sensing of desert dust. J. Geophys. Res. 111, D11208.
35
 ACCEPTED MANUSCRIPT
755
756 Eck, T.F., Holben, B.N., Reid, J.S., Dubovik, O., Smirnov, A., O'Neill, N.T., Slutsker, I.,
757 Kinne, S., 1999. Wavelength dependence of the optical depth of biomass burning, urban and
758 desert dust aerosols. J. Geophys. Res. 104, 31333-31350.
759
760 Eck, T.F., Holben, B.N., Ward, D.E, Dubovik, O., Reid, J.S., Smirnov, A., Mukelabai, M.M.,
761 Hsu, N.C., O’Neill, N.T., Slutsker, I., 2001. Characterization of the optical properties of
762 biomass burning aerosols in Zambia during the 1997 ZIBBEE Field Campaign. J. Geophys.
763 Res. 106, 3425 – 3448.

PT
764
765 Eck, T.F., Holben, B.N., Dubovik, O., Smirnov, A., Goloub, P., Chen, H.B., Chatenet, B.,
766 Gomes, L., Zhang, X.Y., Tsay, S.C., Ji, Q., Giles, D., Slutsker, I., 2005. Columnar aerosol

RI
767 optical properties at AERONET sites in central eastern Asia and aerosol transport to the
768 tropical mid‐Pacific. J. Geophys. Res. 110, D06202.
769

SC
770 Eck, T.F., Holben, B.N., Reid J.S., Sinyuk, A., Dubovik, O., Smirnov, A., Giles, D., O'Neill,
771 N.T., Tsay, S.C., Ji, Q., Al Mandoos, A., Ramzan Khan M., Reid, E.A., Schafer, J.S.,
772 Sorokine, M., Newcomb, W., Slutsker, I., 2008. Spatial and temporal variability of column
773 integrated aerosol optical properties in the southern Arabian Gulf and United Arab Emirates

U
774 in summer. J. Geophys. Res. 113, D01204.
775
AN
776 Eck, T.F., Holben, B.N., Sinyuk, A., Pinker, R.T, Goloub, P., Chen, H., Chatenet, B., Li, Z.,
777 Singh, R.P., Tripathi, S.N., Reid, J.S., Giles, D.M., Dubovik, O., O'Neill, N.T., Smirnov, A.,
778 Wang, P., Xia, X., 2010. Climatological aspects of the optical properties of fine/coarse mode
M

779 aerosol mixtures. J. Geophys. Res. 115, D19205.

780
781 Fotiadi, A., Hatzianastassiou, N., Drakakis, E., Matsoukas, C., Pavlakis, K. G.,
D

782 Hatzidimitriou, D., Gerasopoulos, E., Mihalopoulos, N., and Vardavas, I., 2006. Aerosol
783 physical and optical properties in the Eastern Mediterranean Basin, Crete, from Aerosol
784 Robotic Network data. Atmos. Chem. Phys. 6, 5399-5413, https://doi.org/10.5194/acp-6-
TE

785 5399-2006.
786
787 Gama, C., Tchepel, O., Baldasano, J.M., Basart, S., Ferreira, J., Pio, C., Cardoso, J., Borrego,
EP

788 C., 2015. Seasonal patterns of Saharan dust over Cape Verde – a combined approach using
789 observations and modelling. Tellus B: Chemical and Physical Meteorology 67:1, 24410,
790 https://doi.org/10.3402/tellusb.v67.24410.
791
C

792 Gerasopoulos E., Koulouri E., Kalivitis N., Kouvarakis G., Saarikoski S., Mäkelä T., Hillamo
AC

793 R., Mihalopoulos N., 2007. Size-segregated mass distributions of aerosols over Eastern
794 Mediterranean: Seasonal variability and comparison with AERONET columnar size
795 distributions. Atmos. Chem. Phys. 7, 2551-2561.
796
797 Gherboudj, I., Beegum, N., Ghedira, H., 2016. Identifying natural dust source regions over the
798 Middle-East and North-Africa: Estimation of dust emission potential. Earth Sci. Rev. 165,
799 342-355.
800
801 Giles, D.M., Holben, B.N., Tripathi, S.N., Eck, T.F., Newcomb, W.W., Slutsker, I., Russell
802 R. Dickerson, R.R., Thompson, A.M., Mattoo, S., Wang, S.H., Singh, R.P., Sinyuk, A.,
803 Schafer, J.S., 2011. Aerosol properties over the Indo‐Gangetic Plain: A mesoscale perspective

36
 ACCEPTED MANUSCRIPT
804 from the TIGERZ experiment. J. Geophys. Res. 116, D18203,
805 https://doi.org/10.1029/2011JD015809.
806
807 Ginoux, P., Prospero, J.M., Gill, T.E., Hsu, N.C., Zhao, M., 2012. Global scale attribution of
808 anthropogenic and natural dust sources and their emission rates based on MODIS deep blue
809 aerosol products. Rev. Geophys. 50, 3005.
810
811 Haywood, J.M., Shine, K.P., 1995. The effect of anthropogenic sulfate and soot aerosol on the
812 clear sky planetary radiation budget. Geophys. Res. Lett. 22, 603–606.

PT
813
814 Haywood, J.M., Shine, K.P., 1997. Multi-spectral calculations of the radiative forcing of
815 tropospheric sulphate and soot aerosols using a column model. Quart. J. Roy. Meteor. Soc.

RI
816 123, 543, 1907-1930.
817
818 Haywood, J.M., Osborne, S.R., Francis, P.N., Keil, A., Formenti, P., Andrea, M.O., Kaye,

SC
819 P.H., 2003. The mean physical and optical properties of regional haze dominated by biomass
820 burning aerosol measured from the C-130 aircraft during SAFARI 2000. J. Geophys. Res.
821 108(D13), 8473, https://doi.org/10.1029/2002JD002226.
822

U
823 Holben, B.N., Eck, T.F., Fraser, R.S., 1991. Temporal and spatial variability of aerosol optical
824 depth in the Sahel region in relation to vegetation remote sensing. Int. J. Remote Sens. 12,
AN
825 1147–1164.
826
827 Holben B.N., Eck, T.F., Slutsker, I., Tanre, D., Buis, J.P., Setzer, A., Vermote, E., Reagan,
M

828 J.A., Kaufman, Y., Nakajima, T., Lavenue, F., Jankowiak, I., Smirnov, A., 1998. AERONET
829 - A federated instrument network and data archive for aerosol characterization, Rem. Sens.
830 Environ. 66, 1-16.
D

831
832 Holben, B.N., Tanre, D. Smirnov, A., Eck, T.F., Slutsker, I., Abuhassan, N., Newcomb,
833 W.W., Schafer, J., Chatenet, B,. Lavenue, F., Kaufman, Y., Vande Castle, J., Setzer, A.,
TE

834 Markham, B., Clark, D., Frouin, R., Halthore, R., Karnieli, A., O’Neill, N.T., Pietras, C.,
835 Pinker, R.T., Voss, K., Zibordi, G., 2001. An emerging ground-based aerosol climatology:
836 Aerosol Optical Depth from AERONET, J. Geophys. Res. 106, 12 067-12 097.
EP

837
838 Holben, B.N., Eck, T.F., Slutsker, I., Smirnov, A., Sinyuk, A., Schafer, J., Giles, D., Dubovik,
839 O., 2006. Aeronet’s Version 2.0 quality assurance criteria, Proc. SPIE 6408, Remote Sensing
840 of the Atmosphere and Clouds, 64080Q, https://doi.org/10.1117/12.706524.
C

841
AC

842 Huang, J., Lin, B., Minnis, P., Wang, T., Wang, X., Hu, Y., Yi, Y., Ayers, J.K., 2006.
843 Satellite-based assessment of possible dust aerosols semidirect effect on cloud water path over
844 East Asia. Geophys. Res. Lett. 33, L19802, https://doi.org/10.1029/2006GL026561.
845
846 Husar, R., Prospero, J., Stowe, L.L., 1997. Characterization of tropospheric aerosols over the
847 oceans with the NOAA AVHRR optical thickness operational product. J. Geophys. Res. 102,
848 16889-16909, https://doi.org/10.1029/96JD04009.
849
850 IPCC (Intergovernmental Panel on Climate Change), 2013. The Physical Science Basis,
851 Working Group I Contribution to the IPCC 5th Assessment Report – Changes to the
852 Underlying Scientific/Technical Assessment. Cambridge Univ. Press, New York.
853

37
 ACCEPTED MANUSCRIPT
854 Kaufman, Y., Gitelson, A., Karnieli, A., Ganor, E., Fraser, R., Nakajima, T., Mattoo, S.,
855 Holben, B.N., 1994. Size distribution and scattering phase function of aerosol particles
856 retrieved from sky brightness measurements. J. Geophys. Res. 99 (D5), 10341-10356.
857
858 Kaufman, Y.J., Tanre, D., Boucher, O., 2002. A satellite view of aerosols in the climate
859 system. Nature 419(6903), 215-223.
860
861 Koçak, M., Kubilay, N., Mihalopoulos, N., 2004. Ionic composition of lower tropospheric
862 aerosols at a Northeastern Mediterranean site: implications regarding sources and long-range

PT
863 transport. Atmos. Environ. 38, 2067-2077.
864
865 Koçak, M., Kubilay, N., Herut, B., Nimmo, M., 2005. Dry atmospheric fluxes of trace metals

RI
866 (Al, Fe, Mn, Pb, Cd, Zn, Cu) over the Levantine Basin: A refined assessment. Atmos.
867 Environ. 39, 7330-7341.
868

SC
869 Koçak, M., Mihalopoulos, N., Kubilay, N., 2007. Chemical composition of the fine and
870 coarse fraction of aerosols in the Northeastern Mediterranean. Atmos. Environ. 41, 7351-
871 7368.
872

U
873 Koçak, M., Mihalopoulos, N., Kubilay, N., 2009. Origin and source regions of PM10 in the
874 eastern Mediterranean atmosphere. Atmos. Res. 92, 464-474.
AN
875
876 Koçak, M., Theodosi, C., Zarmpas, P., Séguret, M.J.M., Herut, B., Kallos, G., Mihalopoulos,
877 N., Kubilay, N., Nimmo M., 2012. Influence of mineral dust transport on the chemical
M

878 composition and physical properties of the Eastern Mediterranean aerosol. Atmos. Environ.
879 57, 266-277.
880
Koehler, K.A., Kreidenweis, S.M., DeMott, P.J., Petters, M.D., Prenni, A.J., Carrico, C.M.,
D

881
882 2009. Hygroscopicity and cloud droplet activation of mineral dust aerosol. Geophys. Res.
883 Lett. 36, L08805.
TE

884
885 Kubilay, N., Saydam, C., 1995. Trace elements in atmospheric particulates over the Eastern
886 Mediterranean: concentration, sources and temporal variability. Atmos. Environ. 29, 2289–
EP

887 2300.
888
889 Kubilay N., Nickovic S., Moulin C., Dulac F., 2000. An illustration of the transport and
890 deposition of mineral dust onto the eastern Mediterranean. Atmos. Environ. 34, 1293-1303.
C

891
AC

892 Kubilay, N., Çokacar, T., Oğuz, T., 2003. Optical properties of mineral dust outbreaks over
893 the northeastern Mediterranean. J. Geophys. Res. 108(D21), 4666,
894 doi:10.1029/2003JD003798.
895
896 Levin, Z., Teller, A., Ganor, E., and Yin, Y., 2005. On the interactions of mineral dust, sea
897 salt particles and clouds – Measurements and modeling study from the MEIDEX campaign. J.
898 Geophys. Res. 110, D20202.
899
900 Lionello, P., Malanotte-Rizzoli, P., Boscolo, R., Alpert, P., Artale, V., Li, L., Luterbacher, J.,
901 May, W., Trigo, R.M., Tsimplis, M., Ulbrich, U., Xoplaki, E., 2006. The Mediterranean
902 Climate: An overview of the main characteristics and issues, in: Lionello, P., Malanotte-

38
 ACCEPTED MANUSCRIPT
903 Rizzoli, P., Boscolo, R. (eds.), Mediterranean Climate Variability. Elsevier Amsterdam, pp.
904 1–26.
905
906 Mallet, M., Dubovik, O., Nabat, P., Dulac, F., Kahn, R., Sciare, J., Paronis, D., and Léon, J.
907 F., 2013. Absorption properties of Mediterranean aerosols obtained from multi-year ground-
908 based remote sensing observations. Atmos. Chem. Phys. 13, 9195–9210. 2013.
909
910 Marks, R., 1990. Preliminary investigations on the influence of rain on the production,
911 concentration, and vertical distribution of sea salt aerosol. J. Geophys. Res. 95, 22299-22304.

PT
912
913 Mehta M., Singh R., Singh A., Singh N., Anshumali, 2016. Recent global aerosol optical
914 depth variations and trends—a comparative study using MODIS and MISR level 3 datasets.

RI
915 Remote. Sens. Environ. 181, 137–150, https://doi.org/10.1016/j.rse.2016.04.004.
916
917 Meloni, D., di Sarra A., Pace, G., Monteleone, F., 2006. Aerosol optical properties at

SC
918 Lampedusa (Central Mediterranean). 2. Determination of single scattering albedo at two
919 wavelengths for different aerosol types. Atmos. Chem. Phys., 6, 715-727,
920 https://doi.org/10.5194/acp-6-715-2006.
921

U
922 Mihalopoulos, N., Stephanou, E., Kanakidou, M., Pilitsidis, S., Bousquet, P., 1997.
923 Tropospheric aerosol ionic composition in the E. Mediterranean region. Tellus 49B, 1–13.
AN
924
925 Moulin, C., Lambert, E., Dayan, U.,Masson, V., Ramonet, M., Bousquet, P., Legrand, M.,
926 Balkanski, Y.J., Guelle, W., Marticorena, B., Bergametti, G., Dulac, F., 1998. Satellite
M

927 climatology of African dust transport in the Mediterranean atmosphere. J. Geophys. Res. 103,
928 13137–13144.
929
O'Dowd, C.D., Smith, M.H., 1993. Physicochemical properties of aerosol over the northeast
D

930
931 Atlantic: Evidence for wind-speed related submicron sea-salt aerosol production. Geophys.
932 Res. 98, 1137-1149.
TE

933
934 O’Dowd, C.D., Smith, M.H., Consterdine, I.E., Lowe, J.A., 1997. Marine aerosol, sea salt,
935 and the marine sulfur cycle: a short review. Atmos. Environ. 31, 73–80.
EP

936
937 O’Neill, N.T., Ignatov, A., Holben, B.N., Eck, T.F., 2000. The lognormal distribution as a
938 reference for reporting aerosol optical depth statistics: Empirical tests using multi-year, multi-
939 site AERONET sunphotometer data. Geophys. Res. Lett. 27, 3333-3336.
C

940
AC

941 O'Neill, N.T., Eck, T.F., Holben, B.N., Smirnov, A., Dubovik, O., Royer, A., 2001. Bimodal
942 size distribution influences on the variation of Angstrom derivatives in spectral and optical
943 depth space. J. Geophys. Res., 106, 9787-9806.
944
945 O'Neill, N.T., Eck, T.F., Smirnov, A., Holben, B.N, Thulasiraman, S., 2003. Spectral
946 discrimination of coarse and fine mode optical depth. J. Geophys. Res. 108, D17, 4559-4573,
947 https://doi.org/10.1029/2002JD002975.
948
949 Pace, G., di Sarra, A., Meloni, D., Piacentino, S., Chamard, P., 2006. Aerosol optical
950 properties at Lampedusa (Central Mediterranean). 1. Influence of transport and identification
951 of different aerosol types. Atmos. Chem. Phys. 6, 697-713, https://doi.org/10.5194/acp-6-697-
952 2006.

39
 ACCEPTED MANUSCRIPT
953
954 Patterson, E.M., Gillete, D.A., and Stockton, B.H., 1977. Complex index of refraction
955 between 300 and 700 nm for Saharan aerosol. J. Geophys. Res. 82, 3153-3160.
956
957 Petters, M.D. and Kreidenweis, S.M, 2007. A single parameter representation of hygroscopic
958 growth and cloud condensation nucleus activity. Atmos. Chem. Phys., 7, 1961–1971.
959
960 Polissar, A.V., Hopke, P.K., Harris, J.M., 2001. Source regions for atmospheric aerosol
961 measured at Barrow, Alaska. Environ. Sci. Technol. 35, 4214-4226.

PT
962
963 Prospero, J.M., Ginoux, P., Torres, O., Nicholson, S.E., and Gill, T.E., 2002. Environmental
964 characterization of global sources of atmospheric soil dust identified with the nimbus 7 total

RI
965 ozone mapping spectrometer (TOMS) absorbing aerosol product. Rev. Geophys. 40, 1002,
966 doi:10.1029/2000RG000095
967

SC
968 Rosenfeld, D., Lohmann, U., Raga, G.B., O’Dowd, C.D., Kulmala, M., Fuzzi, S., Reissell, A.,
969 Andreae, M.O., 2008. Flood or drought: How do aerosols affect precipitation?, Science, 321,
970 1309–1313, https://doi.org/10.1126/science.1160606.
971

U
972 Santese, M., De Tomasi, F., and Perrone, M.R., 2008. Advection patterns and aerosol optical
973 and microphysical properties by AERONET over south-east Italy in the central
AN
974 Mediterranean. Atmos. Chem. Phys. 8, 1881-1896, https://doi.org/10.5194/acp-8-1881-2008.
975
976 Schafer, J.S., Eck, T.F., Holben, B.N., Artaxo, P., Duarte, A.F., 2008. Characterization of the
M

977 optical properties of atmospheric aerosols in Amazônia from long-term AERONET

978 monitoring (1993-1995 and 1999-2006). J. Geophys. Res. 113, D04204,
979 doi:10.1029/2007JD009319.
D

980
981 Sciare, J., Oikonomou, K., Favez, O., Liakakou, E., Markaki, Z., Cachier, H., Mihalopoulos,
982 N., 2008. Long-term measurements of carbonaceous aerosols in the Eastern Mediterranean:
TE

983 evidence of long-range transport of biomass burning. Atmos. Chem. Phys. 8, 5551-5563,
984 doi:10.5194/acp-8-5551-2008.
985
EP

986 Smirnov, A., Villevalde, Y., O’Neill, N.T., Royer, A., Tarussov, A., 1995. Aerosol optical
987 depth over the oceans: Analysis in terms of synoptic air mass types. J. Geophys. Res. 100,
988 16639–16650.
989
C

994 Smirnov, A.V., O’Neill, N.T., Royer, A., Tarussov, A., McArthur, B., 1996. Aerosol optical
AC

995 depth over Canada and the link with synoptic air masstypes. J. Geophys. Res. 101, 19299–
996 19318.
997
998 Smirnov, A., Holben, B.N., Dubovik, O., O'Neill, N.T., Eck, T.F., Westphal, D.L., Goroch,
999 A.K., C. Pietras, C., Slutsker, I., 2002. Atmospheric Aerosol Optical Properties in the Persian
1000 Gulf. J. Atmos. Sci. 59, 620–634.
1001
1002 Stein, A.F., Draxler, R.R, Rolph, G.D., Stunder, B.J.B., Cohen, M.D., Ngan, F., 2015.
1003 NOAA's HYSPLIT atmospheric transport and dispersion modeling system. Bull. Amer.
1004 Meteor. Soc. 96, 2059-2077.
1005

40
 ACCEPTED MANUSCRIPT
1006 Tang, I.N. and Munkelwitz, H.R., 1994. Water activities, densities, and refractive indices of
1007 aqueous sulfates and sodium nitrate droplets of atmospheric importance. J. Geophys. Res. 99,
1008 18801-18808.
1009
1010 Tegen, I., Lacis, A.A., 1996. Modeling of particle size distribution and its influence on the
1011 radiative properties of mineral dust aerosol. J. Geophys. Res. 101, 19237-19244,
1012 https://doi.org/10.1029/95JD03610.
1013
1014 Toledano, C., Cachorro, V.E., de Frutos, A.M., Torres, B., Berjon, A., Sorribas, M., Stone, R.

PT
1015 S., 2009. Airmass Classification and Analysis of Aerosol Types at El Arenosillo (Spain). J.
1016 Appl. Meteorol. Clim. 48(5), 962–981.
1017

RI
1021 Vasconcelos, L., Kahl, J., Liu, D., Macias, E., White,W., 1996. A tracer calibration of back
1022 trajectory analysis at the Grand Canyon. J. Geophys. Res. 101 (D14), 19329–19335.
1023

SC
1024 Washington, R., Todd, M., Middleton, N.J., Andrew, S., Goudie, A.S., 2003. Dust-storm
1025 source areas determined by the Total Ozone Monitoring Spectrometer and surface
1026 observations. Ann. Assoc. Am. Geogr. 93 (2), 297–313.
1027

U
1028 Xia, X.A., Li, Z.Q., Holben, B., Wang, P.C., Eck, T., Chen, H.B., Cribb, M., Zhao, Y.X.,
1029 2007. Aerosol optical properties and radiative effects in the Yangtze Delta region of China. J.
AN
1030 Geophys. Res. 112, D22S12, https://doi.org/10.1029/2007JD008859.
1031
1032 Xin, J., Wang, Y., Li, Z., Wang, P., Hao, W.M., Nordgren, B.L., Wang, S., Liu, G., Wang, L.,
M

1033 Wen, T., Sun, Y., Hu, B., 2007. Aerosol optical depth (AOD) and Angstrom exponent of
1034 aerosols observed by the Chinese Sun Hazemeter Network from August 2004 to September
1035 2005. J. Geophys. Res. 112, D05203, https://doi.org/10.1029/2006JD007075.
D

1036
1037 Yoon, J., Hoyningen-Huene, W., Kokhanovsky, A.A, Vountas, M., Burrows, P.J., 2012.
1038 Trend analysis of aerosol optical thickness and Angstrom exponent derived from the global
TE

1039 AERONET spectral observations. Atmos. Meas. Tech. 5, 1271-1299. doi:10.5194/amt-5-

1040 1271-2012.
1041
EP

1042
1043
1044
1045
C

1046
AC

1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058

41
 ACCEPTED MANUSCRIPT
1059
1060
1061
1062
1063
1064
1065
1066
1067

PT
1068
1069
1070

RI
1071
1072
1073

SC
1074
1075
1076
1077

U
1078
1079
AN
1080
1081
1082
M

1083
1084 Supplement:
1085
D
TE
C EP
AC

1086

42
 ACCEPTED MANUSCRIPT
1087 Figure S1. Classification of 5-day back trajectories arriving at Erdemli. Airflow sectors are
1088 presented as WE (Western Europe), EE (Eastern Europe), TUR (Turkey), SAH (Saharan),
1089 MID (Middle East) and MED (Mediterranean Sea).
1090
1091

PT
RI
U SC
AN
1092
M

1093 Figure S2. Frequency histogram of the log-transferred columnar AOT440 and its expected
1094 distribution at Erdemli.
1095
D
TE
C EP
AC

1096
1097 Figure S3. Locations of AERONET sites in Mediterranean, Middle East, Africa, Asia, and
1098 America.
1099

1100

43
 ACCEPTED MANUSCRIPT

PT
RI
U SC
1101
AN
1102 Figure S4. Variability in Aerosol Optical Thickness (orange bar) and Angstrom Exponent
1103 (blue line) during dust dominated (07 March 2009), mixed (14 May 2010) and pollution
M

1104 dominated (14 July 2013) events at Erdemli.

1105
D

1106
TE
C EP
AC

1107
1108 Figure S5. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m
1109 (black star), 3000m (black square) and 4000m (black triangle) on 1st of April 2014 (a) and 2nd
1110 of April 2014 (b) with corresponding Aerosol Index (AI) from OMI (Ozone Mapping
1111 Instrument) for Erdemli.

44
 ACCEPTED MANUSCRIPT

PT
RI
SC
1112
1113 Figure S6. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m

U
1114 (black star), 3000m (black square) and 4000m (black triangle) on 7 March 2009 with
1115 corresponding Aerosol Index (AI) from OMI (Ozone Mapping Instrument) for Erdemli.
AN
1116

1117
M
D
TE
EP

1118
C
AC

1119
1120 Figure S7. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m
1121 (black star), 3000m (black square) and 4000m (black triangle) on 25th of February 2007 (a),

45
 ACCEPTED MANUSCRIPT
1122 26th of February 2007 (b), and 27th of February 2007 (c) with corresponding Aerosol Index
1123 (AI) from OMI (Ozone Mapping Instrument) for Erdemli.
1124

PT
RI
U SC
1125
1126
AN
1127 Figure S8. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m
1128 (black star), 3000m (black square) and 4000m (black triangle) on 14 May 2010 with
1129 corresponding Aerosol Index (AI) from OMI (Ozone Mapping Instrument) for Erdemli.
M
D
TE
EP

1130
C
AC

1131
1132 Figure S9. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m
1133 (black star), 3000m (black square) and 4000m (black triangle) on 25th of August 2014 (a), 26th
46
 ACCEPTED MANUSCRIPT
1134 of August 2014 (b) and 27th of August 2014 (c) with corresponding AOD from MODIS Terra
1135 for Erdemli

PT
RI
U SC
1136
AN
1137 Figure S10. Vertical movement of five day back trajectories of air masses reaching at 1000m
1138 (black circle), 2000m (black star), 3000m (black square) and 4000m (black triangle) on 25th of
August 2014 (a), 26th of August 2014 (b) and 27th of August 2014 (c).
M

1139
D
TE
C EP
AC

1140
1141 Figure S11. Five day back trajectories of air masses reaching at 1000m (black circle), 2000m
1142 (black star), 3000m (black square) and 4000m (black triangle) on 14 July 2013 with
1143 corresponding Aerosol Index (AI) from OMI (Ozone Mapping Instrument) for Erdemli.
1144

47
 ACCEPTED MANUSCRIPT

PT
RI
SC
1145

1146 Figure S12. Bow-whisker plot of OMI-AI for first (non-hygroscopic dominated), moderately

U
1147 (moderately hygroscopic) and third (hygroscopic dominated) groups.
1148
AN
1149
M

1150
D
TE
C EP
AC

48
 ACCEPTED MANUSCRIPT
Highlights

• Daily and seasonal variability in AOT and α440-870 over the Eastern Mediterranean.

• Impact of air flow on aerosol optical properties over the Eastern Mediterranean.

• Hygroscopic growth of aerosol types and their potential source.

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:

PT
RI
U SC
AN
M
D
TE
C EP
AC