Transport of dust and anthropogenic aerosols across Alexandria, Egypt

The flow of pollutants from Europe and desert dust to Europe from the Sahara desert both affects the air quality of the coastal regions of Egypt. As such, measurements from both ground and satellite observations assume great importance to ascertain the conditions and flow affecting the Nile Delta and the large city of Alexandria. We note that special weather conditions prevailing in the Mediterranean Sea result in a westerly wind flow pattern during spring and from North to South during the summer. Such flow patterns transport dust-loaded and polluted air masses from the Sahara desert and Europe, respectively, through Alexandria, and the Nile Delta in Egypt. We have carried out measurements acquired with a groundbased portable sun photometer (Microtops II) and the satellite-borne TERRA/Moderate Resolution Imaging Spectroradiometer (MODIS) sensor during the periods of October 1999–August 2001 and July 2002–September 2003. These measurements show a seasonal variability in aerosol optical depth (AOD) following these flow patterns. Maximum aerosol loadings accompanied by total precipitable water vapor (W) enhancements are observed during the spring and summer seasons. Pronounced changes have been observed in theÅngstr̈om exponent ( α) derived from groundbased measurements over Alexandria (31.14 ◦ N, 29.59 E) during both dust and pollution periods. We have followed up the observations with a 3-day back-trajectories model to trace the probable sources and pathways of the air masses causing the observed aerosol loadings. We have also used other NASA model outputs to estimate the sea salt, dust, sulfates and black carbon AOD spatial distributions during different seasons. Our results reveal the probable source regions Correspondence to: H. El-Askary (elaskary@chapman.edu) of these aerosol types, showing agreement with the trajectory andÅngstr̈om exponent analysis results. It is confirmed that Alexandria is subjected to different atmospheric conditions involving dust, pollution, mixed aerosols and clean sky.


Introduction
Egypt is influenced by the regional scale trade wind system caused by differential heating between the Mediterranean Sea and the land regions of North Africa (bare and dry) and Southern Europe (partial vegetation coverage).This system is enhanced during the warm period of the year by the strong gradient between the Atlantic anticyclone activity and monsoon activity over the Indian Ocean and Middle East.The resulting circulation patterns over Egypt are winds from North that enhance the sea breezes along the North African coastline.This flow regime penetrates deep inside Egypt and assists in cleaning the air in the urban locations within the greater Delta region, such as Cairo, Alexandria, and adjoining areas (Kallos et al., 1998(Kallos et al., , 2007)).The boundary layer height changes from a few hundred meters at night to 2-4 km during daytime, thus also assisting in pumping local air pollutants away from their sources (Kallos et al., 1998).During fall, this pattern is weakened significantly and local circulations start to develop, while stagnation, local boundary layer formation along the coast, and shallow boundary layer formations in general, are the characteristics that lead to poor pollutant dispersion conditions.Also, during fall, biomass burning in the greater Nile Delta becomes a major contributor to local air quality degradation due to poor ventilation processes (e.g.light winds and shallow boundary layer).
The objective of the current study is to investigate the regional pattern of pollution and dust aerosols exchange from and to Europe.Dust is predominantly emitted from Sahara desert, which is the major source of dust in this region.The current study investigates a multi-year remotely sensed aerosol data acquired from both MODIS sensor and ground Microtops II sun photometer, over Alexandria, Egypt, which is centrally situated within the region of interest.Alexandria is the second largest city in Egypt, and houses more than 60% of the Egyptian national industries.It lies at the junction of several weather systems, including the Mediterranean and the Saharan ones.As a result, it is characterized by a complex synoptic meteorology, which follows an annual cycle (Moulin et al., 1998).

Instrumentation, data and methodology
In this research, ground-based aerosol remote sensing was conducted using a portable Microtops II sun photometer manufactured by Solar Light Co., USA.The Sun photometer measures solar radiance (W sr −1 m −2 ) in five spectral wave bands (340, 440, 675, 870, and 936 nm) from which the spectral aerosol optical depth (AOD) is derived.Since the 936 nm wavelength is greatly affected by water-vapor absorption, it is used to derive the total water-vapor column amount, W .The derived AOD and W data were subsequently adjusted following calibration processes performed as described by Ichoku et al. (2002).The absolute accuracy of the calibration-adjusted AOD measurements is expected to be ±0.02 over the mean, with precision 4 times better than that of the corresponding values before adjustment (Sabbah et al., 2001).
Microtops II sun photometer measurements were conducted for several years, on clear days when no cloud patch was even close to the sun, to avoid cloud contamination.For better quality control, measurements were taken in triplets.One or several such sets of measurements were obtained per day.Approximately once per year, the sun photometer was returned to the National Aeronautics and Space Administration (NASA), Goddard Space Flight Center (GSFC), Greenbelt, Maryland, USA, and calibrated against a reference instrument, which is the master automatic tracking sun photometer/sky radiometer (CIMEL Electronique 318A) belonging to AERONET (Holben et al., 1998).The calibration enabled the determination of calibration coefficients used for the adjustment of all data acquired between two successive calibrations.Overall, the data used in this work were obtained from 19 October 1999 to 20 August 2001, and from 28 July 2002 to 4 September 2003.
The spectral AOD at a given wavelength λ (τ λ ) represents the extinction of radiation at wavelength λ as a result of the presence of atmospheric aerosols.AOD measurements in two adjacent wavelengths can be used to calculate the corresponding Ångström exponent (α) given in Eq. ( 1), which represents the slope of the wavelength dependence of the AOD in logarithmic coordinates ( Ångström, 1929).
In the solar spectrum, α is a good indicator of the size range of the atmospheric particles responsible for the AOD: α > 1 when fine mode (submicron) aerosols are dominant, while α < 1 for aerosols dominated by coarse or supermicron particles.In fact, King et al. (1978), Nakajima et al. (1986), Kaufman (1993), Eck et al. (1999), O'Neill et al. (2001a, b, 2003), Schuster et al. (2006) and Gobbi et al. (2007) discuss how the spectral variation of the Ångström exponent can provide further information about the aerosol-size distribution.Many satellite-based and ground-based measurements provide retrievals of spectral AOD, but no direct sizedistribution retrievals.Therefore, analysis of α is important in the interpretation of these data and in providing further information on particle size.

Aerosol (pollution/dust) discrimination using Ångström exponent
A study of the aerosol parameters during major dust storm events (2001)(2002)(2003)(2004)(2005) over the Indo-Gangetic (IG) Plains, India (Prasad and Singh, 2007) shows a large increase in AOD from 0.4-0.6 to more than 1, and a corresponding sharp decline in the Ångström exponent (to the 0-0.4 range) with the arrival of dust storms compared to non-dusty days.A sharp decrease in the Ångström exponent to low and even negative values was observed over Delhi and Hyderabad during dust storm events (Badarinath et al., 2007;Singh et al., 2005) and also during Saharan dust events (Kaskaoutis et al., 2008;Hamonou et al., 1999).An enhanced level of W (>3 cm, in near infra-red, clear column) has been found to be associated with higher aerosol loading during these dust storms.Aerosols over the Northern coast of Egypt can have both local and distant origins depending on the season.Aerosols from local sources may either be from natural causes, for instance desert dust particles directly transported from the Western desert regions or maritime particles produced over the Mediterranean; or anthropogenic emissions from the Alexandria industrialized areas and smoke particles from seasonal biomass fires over the Greater Delta region (e.g.Barnaba and Gobbi, 2004).This area is also exposed to longrange transport of not only the Saharan dust, but also anthropogenic pollutants (mainly from Europe) (Luria et al., 1996;Millan et al., 1997;andKallos et al., 1998, 2007).According to their findings, long-range transported anthropogenic aerosol concentrations are very high during the warm period of the year due to the prevailing trade wind patterns, photochemical activity, and the absence of precipitation.Desert dust transport is a major contributor during the spring and the transitional seasons.Sea-salt particles contribute to aerosol concentrations over the coastal areas, mainly during strong wind weather conditions (synoptic scale component).Sea salt becomes more important during winter due to the passage of synoptic systems and during summer due to trade winds from the north and associated wave breaking activity along the coastline.Knowledge of these seasonal patterns can be combined with the aerosol Ångström exponent α analysis to achieve an approximate differentiation of the aerosols' origins.In this study, α was determined from AOD measurements at 675 nm and 440 nm wavelengths as given in Eq. ( 2) following Ichoku et al. (2002).
where, α 675/440 is the Ångström exponent derived using only the AOD values (τ a675 and τ a440 ) measured at the 675 and 440 nm channels.These two wavelengths have been used because the computed α 675/440 can be used to interpolate τ a550 , which can then be plotted together with equivalent MODIS values.Also, the closeness of this wavelength pair allows us to avoid the curvature in spectral dependence described by Eck et al. (1999).High α values indicate a higher abundance of smaller aerosol particles, while low α ones reflect coarser particle sizes.The coarse-mode aerosols are mainly from sea salt and mineral dust, whereas the mainly anthropogenic submicron aerosols are mostly carbonaceous and sulfate aerosols generated from biomass burning and urban/industrial activities in addition to those from gas-to-particle conversion mechanisms (Ramachandran, 2004).Therefore, either an increase in the number of smaller particles or a decrease in the number of larger particles can cause an increase in the value of α.
In that context, we classified the months into different categories (clean/dust/pollution/mixed) according to the number of days that fall into each.This categorization is based on the AOD τ a675 values and the Ångström exponent α 675/440 (Sabbah et al., 2001).Cases with τ a675 ≤ 0.06 were classified as "clean" conditions.For those with τ a675 > 0.06, further classification was based on the values of the Ångström exponent, being categorized as "dust" for α 675/440 < 0.25, "pollution" for α 675/440 > 1.0, and "mixed" otherwise (Table 1).

Inferring aerosol seasonality
A time-series analysis of the frequency of occurrence of the four pre-defined aerosol categories, together with monthly During the first data period, the months with the highest frequency of days with dust events were April 2001 (52.2%),May 2001 (50.0%),April 2000 (34.5%), and March 2001 (32.0%).However, during the second time period, the highest dust percentages were in April 2003 (29.2%),January 2003 (24.1%),February 2003 (19.0%), and September 2002 (17.2%).It is clear that April is the month with the most frequent dust event occurrences in Alexandria (khamaseen storms), and is probably the peak month for transport of Saharan dust to the Near East and Eastern Europe.In terms of the monthly mean AOD (Fig. 3 Table 1), which is significantly high for this period of the year.The presence of dust produces a noticeable dip in the Ångström exponent value, but also causes an enhancement in the monthly mean AOD value, which is showing as a peak in the time-series (Fig. 2).This unusual dust outbreak during the month of July in the Southeastern Mediterranean region in 2000 was also reported by Alpert et al. (2002), who suggested that a strong dry convection that may have developed over the Saharan source regions may have lifted the dust to a high altitude (say above 700 hPa), thereby allowing the upper-level southwesterly winds to transport the dust northeastward over the Southeastern Mediterranean region.This observation agrees with the findings reported by Kaskaoutis et al. (2008).It is also noteworthy that the synoptic systems prevailing during this period and responsible for dust transport are the "North African Depressions" that move along the coast of North Africa towards Middle East that have as a result the transport of dust from Sahara towards the Mediterranean.
Since smoke is also included producing high α-values, an attempt was made to determine the influence of smoke from the biomass burning within the Greater Nile Delta region on these pollution peaks.Fire pixel counts were obtained from MODIS on the Terra satellite, and the total counts per month were calculated and plotted (Fig. 3) for the study period.Based on these data, peak biomass burning was found to occur in September (2001 and2002) or October (2003 and 2004).Nevertheless, it could be inferred that fine-particles were the dominant aerosols from local biomass burning in the Egyptian Nile delta peaks in September or October, and may be partly responsible for the peak pollution found in September-October 2000 (El-Askary et al., 2006, 2008).Since there is virtually no biomass burning in the locality at other times of the year, it is probable that the rest of the peaks in the pollution category observed from the monthly frequency data may have been due to long-range transport from the Near East and Southeastery Europe.Hence, high AOD values during April-May are attributed to large dust loading in the atmosphere, while FMF shows high values during the summer and fall seasons due to the urban/industrial and bio-fuel emissions.It is noteworthy that MODIS AOD values seem to be higher than the ground observations in the summer-fall months for reasons relating to the complexity of satellite retrieval of mixed aerosols over coastal semi-arid land surfaces (Remer et al., 2005).

Model analysis of aerosol transport
For better understanding of the aerosol-transport patterns in the study region, we have used the NOAA Air Resources Laboratory (ARL) HYSPLIT 4 model (Draxler and Hess, 1998;Draxler, 1999) to generate air-mass back trajectories over Alexandria for selected days, in order to investigate the probable sources of the different aerosol types.Also, Goddard Chemistry Aerosol Radiation and Transport (GO-CART) model (Chin et al., 2000) outputs were obtained to determine the possible sources and spatial distribution of sea salt, dust, sulfates and black-carbon AOD over Alexandria and the surrounding delta region during the four seasons of 2003.Table 2 shows the daily average AOD measurements at 675 nm, α 675/440 , W , the aerosol type, for the selected days representing the four scenarios addressed, namely, dust/pollution/mixed/clean.Some of these days were selected for the back trajectory analysis during 2003, where the GOCART outputs are displayed.
Air-mass back trajectories are typically calculated from gridded meteorological data derived from various sources and archived at regular time intervals each day.A complete description of input data, methodology, equations involved, and sources of error for the air mass trajectory calculations is presented in Draxler and Hess (1997).Figure 4 shows 3day air-mass back trajectories at three vertical levels ending over Alexandria for the four aerosol categories: dust, pollution, clean, and mixed for highlighted days during 2003 taken from Table 2.The wind directions observed during the dusty days were found to be variable.However, the major dust wind directions were W, S, and SW when the air masses come from the Sahara or North Africa during spring.We observe that when the wind directions are from N and NW, although the origin of the wind may be in Europe, the trajectory passes (typically) over North Africa where the airmass becomes loaded with dust before reaching Alexandria.Therefore, the air-mass back-trajectory allows for the visualization of a sizeable segment of the air-mass path, giving an indication at where the dust may have been collected from.Dust situations with elevated water-vapor content are often cases where an air mass collects dust from Sahara before or after it collects some moisture over the Mediterranean, after which part of it is blown back by headwinds toward Alexandria.During the pollution days, (summer and fall seasons), the air masses come predominantly from different regions of Eastern, Central or Western Europe mainly due to the wellknown convection patterns and the associated prevailing meteorology (Kallos et al., 1998).In other words, elevated W values indicate that the air-mass containing the aerosol probably passed over the Mediterranean to collect moisture before reaching Alexandria.In particular, W is higher during the summer and fall because that is when come predominantly from Europe, gathering moisture over the Mediterranean.Hence, while passing over the Mediterranean toward North Africa, there is a high chance for cloud nucleation and indirect radiative forcing due to the associated sea breeze (El-Askary and Kafatos, 2008).During the days with mixed aerosols, the air masses come from two sources, one that brings pollution from Europe and the other that brings dust from the Sahara and North Africa.The wind directions during the mixed days are from all directions; however, having Westerly flow usually Alexandria is receiving aerosols from the central and western Mediterranean.These different directions reflect the convergence of air masses from diverse source areas.During the clean days, the air masses mainly come from the Atlantic Ocean via the Mediterranean.Therefore, the wind directions during the clean days typically fall between N and W directions.Our outputs have been further corroborated by GOCART model outputs exploring what appears to be significantly high sea salt, dust, sulfates and black carbon concentrations over the Nile Delta region (Fig. 5).peaks in the Ångström exponent and corresponding increase in the water-vapor column.Biomass burning and local pollution are most prominent during the fall and winter seasons, as confirmed by the fire pixel counts and the black-carbon concentrations observed from other model outputs.In summary, it is clear that the city of Alexandria and the Greater Nile Delta region are exposed to the impacts of aerosols from various sources throughout the year.This is becoming a serious health hazard because of the persistence of high pollution throughout the year or for prolonged time periods.In addition, the sea-breeze component of the atmospheric circulation brought by the prevailing European trade winds tends to inhibit the development of the land breeze component, thereby allowing the deposition of pollutants from the sky above the city at night, especially during the fall season where the inversion layer is observed to become lower as discussed by El-Askary and Kafatos (2008).This study has demonstrated that long-term measurements of aerosol optical depth (AOD) and precipitable water vapor (W ) from the ground and from satellite over Alexandria can capture the general transport signal of aerosols from Europe and the Sahara.

Fig. 1 .
Fig. 1.Number of days with measurements and occurrences for different aerosol categories and their corresponding Ångström exponent over Alexandria for two time periods from October 1999 to August 2001 and from August 2002 to August 2003.
), the peak months during these dust seasons were April 2000, May 2001, and April 2003.Incidentally, the April 2000 and May 2001 dust peaks coincide with W peaks, which may have provided favorable conditions for cloud seeding and nucleation, with the possible enhancement of indirect radiative forcing (El-Askary et al., 2003).El-Askary et al. (2006, 2008) observed a slowly increasing water-vapor column from March through May in 2000, 2001 and 2003.However, they found that the highest water-vapor column value observed during spring is still lower than the values found during September through November.July generally represents a dust free month, but in July 2000 the dust occurrence percentage was 25.8% (see Fig. 1 and

Fig. 2 .
Fig. 2. Microtops II monthly mean AOD at (550 nm), precipitable water vapor and Ångström exponent (α675/440) with corresponding monthly means of MODIS-AOD and FMF over Alexandria for two time periods from October 1999 to August 2001 and from August 2002 to August 2003.

Fig. 3 .
Fig. 3. Fire pixel counts in the Greater Nile Delta region as observed by MODIS on Terra, whose average local solar transit time over that location is 8:50 a.m.

Fig. 5 .
Fig. 5. GOCART model outputs of (a) sea salt in winter 2003, (b) dust in spring 2003, (c) sulfates in summer 2003, and (d) black carbon AOD in fall 2003 respectively showing potential sources and relative concentrations of these aerosol types over Alexandria and the Nile Delta region.

Table 1 .
Percentages of occurrence for different aerosol categories and their corresponding Ångström exponent.

Table 2 .
Characterization of the aerosol types over Alexandria, Egypt, for selected dates.Back ward trajectories' are shown for dates in bold (Fig.5).Shaded cells represent great variations associated with the specified atmospheric scenario in the highlighted parameters.