Introduction
Dust emission and deposition modeling and measurements are required for the
assessment of the dust mass budget. Both emission and deposition are under-constrained in atmospheric dust models, leading to large uncertainties (Bergametti and Forêt, 2014; Schulz et al., 2012). To improve
simulations, the above authors and others suggested the establishment of
dust deposition networks in the vicinity of and away from dust source
regions, operating throughout the year. In this paper we are presenting
results from a network of dust deposition samplers located on the campus of
the King Abdullah University of Science and Technology (KAUST) along the Red
Sea coast of Saudi Arabia. This is an important dust source region (Ginoux et al., 2012; Prospero et al., 2002), the effect
of which extends thousands of kilometers downwind. To better characterize
optical, microphysical, and health effects of dust aerosols we conducted
detailed chemical, mineralogical and particle size analysis of deposition
samples collected from the air.
Importance of mineral dust
Mineral dust is the most abundant atmospheric aerosol, primarily from
suspended soils in arid and semi-arid regions on Earth (Buseck et al.,
2000; Washington and Todd, 2005; Goudie, 2006; Muhs et al., 2014), including
deserts of the Arabian Peninsula (Edgell, 2006). Dust aerosols
profoundly affect climate (Haywood and Boucher, 2000; Hsu et al., 2004;
Kumar et al., 2014), cloud properties (Twomey et al., 1984; Wang et al.,
2010; Huang et al., 2006), visibility (Kavouras et al., 2009;
Moosmüller et al., 2005), air quality (Hagen and Woodruff, 1973),
atmospheric chemistry and mineralogy (Sokolik and Toon, 1999; Kandler et
al., 2007), biogeochemical cycles in the ocean and over land (Jickells et
al., 2005; Mahowald, 2009), human health (Bennett et al., 2006; Bennion
et al., 2007; De Longueville et al., 2010; Menéndez et al., 2017), and
agriculture (Fryrear, 1981; Nihlen and Lund, 1995).
A further important implication of dust emission/deposition processes is
associated with the harvesting of the solar renewable energy in the desert
areas. Dust deposits on solar panels are known to have a severe detrimental
effect on the efficiency of photovoltaic systems (Goossens and Van
Kerschaever, 1999; Hamou et al., 2014; Mejia et al., 2014; Rao et al., 2014;
Sulaiman et al., 2014; Ilse et al., 2016), with its adverse effects
depending on mineral composition and atmospheric conditions (Supplement B).
Importance of dust mineralogy
The importance of dust mineralogy has long been recognized (Engelbrecht et
al., 2016), but only recently has the explicit transport of different
mineralogical species been implemented in climate models (Perlwitz et al.,
2015a, b; Scanza et al., 2015)
The mineralogy and chemical composition of dust generated from the Red Sea
coastal region remains uncertain. The Red Sea coastal plain is a narrow
highly heterogeneous piedmont area, and existing soil databases do not have
the spatial resolution to represent it adequately (Nickovic et al., 2012).
The specific objective of the present study is to examine mineralogical,
chemical and morphological information of deposition samples collected on
the KAUST campus. This will help to better quantify the ecological impacts,
health effects, damage to property, and optical effects of dust blown across
this area (Engelbrecht et al., 2009a, b; Weese and Abraham, 2009).
Knowledge of the mineralogy of the dust deposits will provide information on
refractive indices, which can be used to calculate dust optical properties,
providing input into radiative transfer models, and to assess the impact of
dust events on the Red Sea and adjacent coastal plain.
Previous dust studies in the region
This research complements our dust studies performed in the Arabian Peninsula
(Engelbrecht et al., 2009a; Kalenderski et al., 2013; Jish Prakash et al.,
2015, 2016) and globally (Engelbrecht et al., 2016).
The Arabian Peninsula is one of Earth's major sources of atmospheric dust,
contributing as much as 11.8 % (22–500 Mt a-1) of the total
(1877–4000 Mt a-1) global dust emissions (Tanaka
and Chiba, 2006). The Red Sea, being enveloped by the Arabian and African
deserts, is strongly impacted by windborne mineral dust. Along with profound
influence on the surface energy budget over land and the Red Sea (Kalenderski et al., 2013; Osipov et al., 2015; Brindley et al., 2015),
dust is an important source of nutrients, more so for the oligotrophic
northern Red Sea waters (Acosta et al., 2013). From preliminary
assessments it is estimated that five to six major dust storms per year impact
the Red Sea region, depositing about 6 Mt of mineral dust into the Red Sea (Jish Prakash et al., 2015). Simulations and satellite observations
suggest that the coastal dust contribution to the total deposition flux into
the Red Sea could be substantial, even during fair weather conditions (Jiang et al., 2009; Anisimov et al., 2017). Therefore, the correct
representation of the regional dust balance over the Red Sea coastal plain
is especially important. Here we specifically focus on the dust deposition
in this area, which helps to constrain the dust mass balance, as well as the
dust mineralogy and chemical composition. Dust sources impacting on the
Arabian Red Sea coastal region were shown to vary by season, coming from
local haboobs and low-level jets, delivered from the Tokar Delta of Sudan in
summer (Kalenderski and Stenchikov, 2016), and transported from the west
coast of the Arabian Peninsula (Kalenderski et al., 2013).
Minerals previously identified in continental soils from Middle East dust-generating regions include quartz, feldspars, calcite, dolomite, micas,
chlorite, kaolinite, illite, smectite, palygorskite, mixed-layer clays,
vermiculite, iron oxides, gypsum, hornblende and halite (Engelbrecht et
al., 2009b, 2016; Goudie, 2006; Jish Prakash et al., 2016;
Pye, 1987; Scheuvens and Kandler, 2014). It could be expected that similar
mineral assemblages would occur in variable proportions in the dust
deposition samples collected in the region.
Position of (a) the King Abdullah University of Science and
Technology (KAUST) campus on the Arabian Peninsula (red marker), north of the
coastal city of Jeddah, and (b) the Frisbee deposition sites
(DT1-DT4) on the KAUST campus.
![]()
Sampling and analysis
Anisimov et al. (2017) estimated that the eastern Red Sea coastal plain
emits about 5–6 Mt of dust annually. Due to its close proximity, a
significant portion of this dust is likely to be deposited into the Red Sea,
which could be comparable in amount to the estimated annual deposition rate
from remote sources during major dust events (Jish Prakash et al.,
2015). Therefore, we expect that the total dust deposition into the Red Sea
is on the order of 10 Mt a-1, but this estimate still needs to be
confirmed.
In the past few decades, wind tunnel and field tests have been performed on
different designs of deposition samplers and sand traps to compare their
efficiencies. The samplers and traps included marble dust collectors (MDCOs),
inverted Frisbees, and glass surfaces (Goossens and Rajot, 2008; Sow et
al., 2006; Goossens et al., 2000; Goossens and Offer, 2000). Most of the
experiments performed in wind tunnels failed to completely mimic the field
conditions, which resulted in an underestimation of the dust deposition,
more so for the < 10 µm size fraction (Sow et al.,
2006). Based on the field evaluations by Vallack (1995) and
suggestions by Vallack and Shillito (1998) the decision was taken
to deploy inverted Frisbee samplers with foam inserts.
Inverted Frisbee-type deposition sampler (a) on tripod and
white plastic drainage bottle. (b) View showing the foam insert in the
collection dish to help retain the deposited dust particles, as well as the
spikes with nylon thread to prevent birds from readily perching on the dish.
![]()
At each sampling site the particulate deposits were collected into a 227 mm
diameter inverted Frisbee dust deposit sampler, each with a polyester foam
insert and bird strike preventers (Hall et al., 1993; Vallack and
Chadwick, 1992, 1993; Vallack and Shillito, 1998) (Fig. 2). The purpose of
the foam insert is to enhance the particulate collection capacity of the
dust gauge (Vallack and Shillito, 1998) by better collecting and
retaining wet (from fog, dew, rain) and dry as well as fine and coarse particles,
under stable meteorological conditions, during severe dust events,
northwesterly shamal, and daily coastal winds.
For the period December 2014 to March 2015, four Frisbee samplers were
located at the New Environmental Oasis (NEO) site, about 50 m apart. The
gravimetric information from the four samplers were similar, with small
variations amongst them ascribed to the impact from local construction
activities. Due to the similarity of these gravimetric results, and to
obtain a better representation of dust deposition onto the KAUST campus, two
of the samplers (DT1 and DT2) were moved in March, the first (DT1) to a
residential area and the other (DT2) to the quay adjacent to the Coastal
& Marine Resources Core Lab (CMOR) (Table 1). (Site meta-data provided in
Supplement A.)
Locality of deposition samplers at six sites on the campus of KAUST.
Site
Latitude
Longitude
Elev.
Start
End
m a.s.l.
DT1
NEO 1
22∘18′16.12′′ N
39∘6′28.46′′ E
1
December 2014
March 2015
Res G3705
22∘18′59.06′′ N
39∘6′21.32′′ E
12
April 2015
December 2015
DT2
NEO 2
22∘18′16.84′′ N
39∘6′29.33′′ E
1
December 2014
March 2015
CMOR
22∘18′16.60′′ N
39∘6′7.91′′ E
1
April 2015
December 2015
DT3
NEO 3
22∘18′17.31′′ N
39∘6′30.51′′ E
1
December 2014
December 2015
DT4
NEO 4
22∘18′18.10′′ N
39∘6′31.52′′ E
1
December 2014
December 2015
The deposition samples were collected for intervals of a calendar month,
starting in December 2014 and ending December 2015. At the end of each
month, the samples are retrieved by flushing the dust deposit with distilled
water from the foam insert and collection dish into the downpipe and plastic
bottle. Both the insoluble particles and dissolved salts in the water
suspension are retrieved in the laboratory by a freeze-drying (sublimation)
procedure.
A total of 52 deposition samples were collected at the six sampling sites on
the KAUST campus (Fig. 1b) over a period of 13 months, largely in 2015.
Representative subsets of these samples were selected for electron microscopy (12 samples), XRD (27 samples) and chemical analysis (29 samples).
Freeze-dried sample splits were re-suspended in the laboratory onto
Teflon® filters, for elemental analysis by
XRF spectrometry using a miniaturized version of a dust
entrainment facility (Engelbrecht et al., 2016) (http://www.dri.edu/atmospheric-sciences/atms-laboratories/4185-dust-entrainment-and-characterization-facility).
With this modified system the dust sample is drawn into a vertically mounted
tubular dilution chamber, and the re-suspended dust collected onto a 47 mm
diameter Teflon® filter, for chemical analysis.
The samples re-suspended onto the Teflon® filters
were chemically analyzed for elemental content by XRF, including for Si, Ti,
Al, Fe, Mn, Ca, K, P, V, Cr, Ni, Cu, Zn, Rb, Sr, Y, Zr, and Pb (US EPA, 1999). Splits of about 2 mg from each freeze-dried sample were analyzed
for water-soluble cations of sodium (Na+), potassium (K+), calcium
(Ca2+) and magnesium (Mg2+), and anions of sulfate
(SO42-), chloride (Cl-), phosphate (PO43-), and
nitrate (NO3-), using ion chromatography (IC) (Chow and Watson, 1999).
A subset of 27 samples from the total of 52 samples, representing all months
of the year, was selected for X-ray diffraction (XRD) analysis. XRD is a
non-destructive technique particularly suited to identify and characterize
minerals such as quartz, feldspars, calcite, dolomite, clay minerals, and
iron oxides, in fine soil and dust. Dust reactivity in seawater as well as
optical properties depends on its mineralogy; e.g., carbonates and sulfates
are generally more soluble in water than silicates such as feldspars,
amphiboles, pyroxenes, or quartz. A Bruker D8®
X-ray powder diffraction system was used to analyze the mineral content of
the dust deposition samples. The diffractometer was operated at 40 kV and 40 mA, with Cu Kα radiation, scanning over a range of
4–50∘ 2θ. The Bruker Topas® software and
relative intensity ratios (RIRs) were applied for semi-quantitative XRD
analyses of the dust deposition samples (Rietveld, 1969; Chung, 1974;
Esteve et al., 1997; Caquineau et al., 1997; Sturges et al., 1989).
A likely bias in the results from applying the X-ray diffraction (XRD)
technique, together with the RIR method is widely recognized, and therefore
our methodology is considered to be semi-quantitative at best. Chung (1974) recognized that if the RIRs of all the crystalline phases in a
mineral mixture are known, the sum of all the fractions should add to 100 %. However, XRD is effective at measuring crystalline phases such as
quartz, calcite, and feldspars, and less so for partly crystalline and
amorphous phases, including some layered silicates such as clays as well as
many hydrous minerals. This could lead to an overestimation of the abundance
of the crystalline mineral species in the dust, compared to partly
crystalline and amorphous phases (Formenti et al., 2008; Kandler et al.,
2009). Other discrepancies could occur from preferred orientation of layered
silicates in the sample mounts. To minimize this effect, the dust samples
were loaded into side-mount holders.
Electron microscopy provided information on the individual particle size and
shape of micron-size particles, important for determining the optical
parameters for modeling of dust (Moosmüller et al., 2012). The scanning electron microscope (SEM)-based individual particle
analysis was performed on a subset of 12 deposition samples collected
for each month of 2015. For each sample, a portion of the deposition sample
was suspended in isopropanol and dispersed by sonication. The suspension was
vacuum filtered onto a 0.2 µm pore size polycarbonate substrate. A
section of the substrate was mounted onto a metal SEM stub with colloidal
graphite adhesive. The sample mounts were sputter-coated with carbon to
dissipate the negative charge induced on the sample by the electron beam.
The automated analysis was conducted on a Tescan MIRA 3® field emission scanning electron microscope
(FE-SEM) by rastering the electron beam over the sample while monitoring the
resultant combined backscattered electron (BE) and secondary electron (SE)
signals. Based on the grayscale levels, preset threshold values segmented
the image into particles of interest and background. The system was
configured to automatically measure the size and shape of anywhere from
5000 to 15 000 particles per sample measuring > 0.2 µm in
average diameter. A digital image was acquired of each particle, for
measurement, and stored for subsequent review. Size measurements were based
on Feret diameters obtained from the projected area of each particle, by
tracing their outer edges. This information was used to calculate the
shape-dependent particle volumes. The particles were grouped into “bins”
by their size. The field emission electron source allows for high
magnifications and sharp secondary electron images (SEIs), as well as for the
detailed study of particle size distributions.
Particle size distribution plots of 12 deposition samples collected monthly
at the KAUST campus throughout the 2015 period are shown by volume in
Appendix A and by number in Supplement C. The chemical abundance tables are
in Appendix B. The mineralogical results from XRD are described under
Sect. 5.5 and the normative mineralogy calculated from the chemistry,
presented as histogram plots in Fig. 11.
Wind (m s-1) roses for each month of 2015.
![]()
Results
Meteorology
Northwesterly shamal winds prevailed during all 12 months of 2015 (Fig. 3).
Four to five severe dust storms lasting 3 to 5 days each, contributed
to hot humid conditions during the summer months. Weaker northeasterly winds
were experienced in October and November of that year. Although the
northeasterly winds were more frequent in November, they did not reach the
maximum strength of the northwesterly winds.
Monthly averaged temperatures, humidity measurements, and
calculated dew points at KAUST during 2015.
Month
Temperature
Humidity
Dew point
Avg. ∘C
Min. ∘C
Max. ∘C
Range ∘C
Avg. %
Min. %
Max. %
N > 90 % count
Avg. ∘C
Jan
25
17
33
16
61
10
99
27
17
Feb
27
19
33
14
74
15
99
48
21
Mar
28
23
36
14
76
29
99
32
23
Apr
28
23
34
12
74
23
99
43
23
May
32
29
37
9
77
21
99
65
27
Jun
32
28
37
9
76
22
99
26
27
Jul
33
29
38
9
75
26
99
55
28
Aug
35
33
40
7
82
36
99
96
31
Sep
34
30
38
8
80
26
99
63
30
Oct
33
29
43
14
72
9
96
32
27
Nov
30
25
35
10
69
25
99
19
23
Dec
27
20
32
12
57
15
94
4
17
The first 4 months in the second half of 2015 experienced the highest
ambient temperatures (Table 2), with an average temperature of 35 ∘C
for August followed by 34 ∘C for September, bracketed by 33 ∘C for
both July and October. The highest single temperature was 43 ∘C,
recorded in October, with the coolest temperature of 17 ∘C in January
of that year. The range of temperatures was the greatest through fall,
winter and spring, with large diurnal temperature fluctuations during these
seasons. The humidity at KAUST is consistently high (Table 2), with averages
varying from 57 % for December and 61 % for January to as high as 82 % for August and 80 % for September. Dew points were calculated for each
set of hourly measurements, applying the August–Roche–Magnus approximation (Alduchov and Eskridge, 1996; August, 1828; Magnus, 1844). The highest
dew point temperatures were calculated in August (31 ∘C) and September
(30 ∘C), while the month with the greatest frequency of humidity
measurements (96) in excess of 90 % was also recorded in August (Table 2,
Fig. 4). The lowest monthly frequency (4) for humidity exceeding 90 % was
December. In 2015, there were only a few light rainfall events at KAUST, and
as such not of much importance to our measurements.
Monthly averaged minimum (blue, ▵) and maximum (red, ▵) ambient temperatures as well as dew point (green, ◊) variations for KAUST during 2015. Also shown for each month is the
frequency of hourly humidity measurements exceeding 90 % (black, ∘ ).
![]()
Monthly deposition rates (g m-2) from Frisbee samplers
(DT1-DT4) at the KAUST campus. Also shown are the monthly averages for the
four samplers.
![]()
Gravimetric analysis
With a few exceptions, the monthly gravimetric measurements from the four
samplers (DT1–DT4) are comparable (Fig. 5), changing similarly by month
and season. The deposition rates were at their lowest for December 2014
(avg. 4 g m-2), increasing steadily for 4 months to a peak value for
March 2015 (avg. 20 g m-2) before decreasing over the subsequent 4 months to a low for July (avg. 5 g m-2). The deposition rates increased
sharply for August (28 g m-2), September (23 g m-2) and October
(28 g m-2), before diminishing in November (14 g m-2) and December
(11 g m-2). The NEO terrain is close to several building construction
sites, about 400 m to the east and southeast of the installed deposition
samplers, which periodically created substantial amounts of local airborne
dust. This, together with the windy conditions, is held responsible for
elevated dust concentrations measured at the two NEO sites (DT3, DT4). The
higher deposition rate at DT3 for August, compared to DT4, is ascribed to
the fact that the former sampler is about 100 m closer to construction
material handling activities during that month. Wind-blown sea spray during
stormy conditions was responsible for elevated deposition levels of sea salt
at the CMOR (DT2) quay-side site, for the months of September and October
2015.
Source apportionment is considered to be a following step in the Red Sea
dust research program. As an approximation of source contributions, the
sampler with the lowest deposition rate can be considered to have negligible
or contain the least amount of local dust and sea salt (Fig. 5). In the
months of December 2014 and January, April, March, June, July, and December
2015, the deposition rates at the four sampling sites were similar and
considered to have similar but negligible amounts of dust from local
construction, roads, marine salt, or other particulates. In August, it is
estimated that 24–56 g m-2 month-1 (10–35 %) of the dust
captured at sites DT3 and DT4 was from local construction and motor traffic.
Similarly, it is estimated that the deposition rates of sea salt at DT2
varied from 20–21 g m-2 month-1 (51–56 %) over the months of
September and October.
Dust deposition measurements from the Middle East and other global
dust regions.
Study
Locality
Sampler type
Sampling period
Average
Range
deposition rate
deposition rate
(g m-2 month-1)
(g m-2 month-1)
(a) This study (2017)
Saudi Arabia, KAUST
Frisbee with foam insert
Dec 2014–Dec 2015
14
4–28
(b) Modaihsh and Mahjoub (2013)
Saudi Arabia, Riyadh
Dish with marbles
Jan–Mar?
42
20–140
(c) Khalaf and Al-Hashash (1983)
Kuwait, NW gulf
Polyethelene cylinders with water
Apr 1979–Mar 1980
191
10–1003
(d) Al-Awadhi (2005)
Kuwait, NE bay
PVC bucket with marbles
May 2002–Apr 2003
28
3–58
(e) Al-Awadhi and AlShuaibi (2013)
Kuwait City
PVC bucket with marbles
Mar 2011–Feb 2012
53
2–320
(f) Offer and Goossens (2001)
Israel, Negev
Marble collectors
1988–1997
17
10–25
(g) Goossens and Rajot (2008)
Niger, Banizoumbou
Frisbee with marbles, original data
8 periods in 2005
13
6–21
(h) Smith and Twiss (1965)
USA, Kansas
Cylindrical rain gauge with screens
June 1963–June 1964
6
3–14
Bearing in mind that the dust deposition samplers, sampling procedures, and conditions and sampling periods were different to those of this
study, some comparisons to similar studies in desert regions are listed in
Table 3. The deposition rates from this study, both on average
(14 g m-2 month-1) and in range
(4–28 g m-2 month-1), were found to be similar to those
previously recorded by Offer and Goossens (2001) in the Negev Desert, Israel
(average 17 g m-2 month-1, range 10–25 g m-2 month-1), and western Niger (Goossens and Rajot, 2008) (average
13 g m-2 month-1, range 6–21 g m-2 month-1). A
campaign in the Saudi Arabian capital of Riyadh (Modaihsh,
1997; Modaihsh and Mahjoub, 2013) during the dusty months of January to
March showed average monthly deposition rates of 42 g m-2 and a
range of 20–140 g m-2. The dust deposition measured in Kuwait on the
other hand, varies substantially between sites due to the contribution from
disturbed soils in lowlands during periods of northwesterly shamal winds.
Average monthly deposition rates for all four samplers (DT1-DT4) on
the KAUST campus, together with (a) monthly averaged AOD
measurements from the KAUST AERONET site, and (b) monthly averaged
visibility measurements collected from the Jeddah airport, for 2015.
![]()
Average particle size distributions and standard deviations of
(a) 12 deposition samples collected by Frisbee samplers on KAUST
campus and of (b) 13 < 38 µm sieved soil
samples from a previous study (Jish Prakash et al., 2016), both measured by SEM.
![]()
AERONET and visibility measurements
The aerosol optical depth (AOD) is one of the best observed aerosol
characteristics. It defines the aerosol radiative effect and reflects the
abundance of aerosols in the atmosphere. A CIMEL robotic sun photometer is
installed on the rooftop of the CMOR building on the campus of the KAUST and
operated by our group since 2012, as a part of the NASA AERONET, providing
aerosol optical depth (AOD) and aerosol-retrieved characteristics
(https://aeronet.gsfc.nasa.gov/). Figure 6a compares the
monthly averaged AOD at 500 nm with the dust deposition rate for 2015. In a
general sense the AOD and the deposition rates show comparable trends, both
with maxima in spring and larger maxima in fall. However, the AOD reaches a
first maximum in April, being 1 month later than that of the deposition
rate. Also, the larger second AOD maximum occurred in August while the
maximum deposition rate is broadly distributed over a 3-month period,
from August to October, with AOD in October being relatively low. The linear
correlation coefficient between the monthly deposition rates and the monthly
averaged AOD of 0.40 suggests a causal interrelationship between these two
quantities. However, for a number of reasons it is relatively low. The
photometer measures light attenuation by all aerosols along a column in the
atmosphere, while deposition rate depends on dust at ground level only, the
latter generally containing a relatively coarser dust fraction. The
low-level dust particles are partly from local dust sources while the higher
altitude dust could be transported from distal sources and chemically
transformed, i.e., aged. As was pointed out by Yu et al. (2013) the
differences between the deposition and AOD time series can in part be
attributed to modifications of the natural dust aerosol by anthropogenic
activities, including petrochemical and other large industries along the Red
Sea coast, as well as by entrainment of construction and road dust. However,
the substantial contamination of dust by anthropogenic species and sea salt
is not likely in this area, as was suggested by both observational (Osipov
et al., 2015; Brindley et al., 2015) and modeling studies (Kalenderski and
Stenchikov, 2016). To further test whether the non-dust fine aerosols (or
remotely transported fine dust) significantly contribute to the
interrelation between AOD and deposition rates, we followed two additional
approaches. We screened the AODs by low Ångström exponent value
(< 0.3), suggested to correspond to dust (Ginoux et al., 2012), and
considered the contribution to the total AOD from the spectral
deconvolution algorithm (SDA) fine-mode aerosol product. Neither technique
contributed to an increase of correlation coefficient. The discrepancy of
high AOD, low deposition rate in summer (low Ångström exponent)
and low AOD, high deposition rate in October still remained when AODs were
screened. The fine-mode AOD remains almost constant throughout the year, and
its contribution cannot explain the discrepancy.
Furthermore, a comparison between the deposition samples and the visibility
is made with measurements taken in 2015 at the Jeddah airport meteorological
station, approximately 70 km to the south of KAUST. Visibility is expressed
as the frequency of dust events with reported weather codes 06–09 or
30–35, grouped as dusty or non-dusty days, for each month (Notaro
et al., 2013; Anisimov et al., 2017), expressed as percentages. The bimodal
monthly distributions seen with the deposition rates and AERONET monitoring
are also mirrored by the visibility measurements collected at Jeddah (Fig. 6b). The linear correlation coefficient between the monthly deposition rates
and monthly averaged visibility measurements is 0.48, clearly suggesting a
causal relationship between the two variables.
Particle size distributions (PSDs)
Dust deposition rates depend on the meteorological conditions, and dust
properties such as particle size distribution, their vertical distribution,
and abundance.
Monthly measured deposition rates and assessments of < 10 and < 2.5 µm deposition rates from SEM-based particle
size measurements.
Sample no.
Month (2015)
Deposition rate
Total
< 10 µm
< 2.5 µm
g m-2 month-1
% of total
g m-2 month-1
% of total
g m-2 month-1
DT3.1_012015
January
7.34
16.5 ± 4.3
1.2 ± 0.3
4.0 ± 1.6
0.29 ± 0.12
DT3.1_022015
February
12.83
7.0 ± 3.3
0.9 ± 0.4
0.9 ± 0.4
0.12 ± 0.05
DT3.4_032015
March
15.11
12.0 ± 5.8
1.8 ± 0.9
1.6 ± 0.6
0.24 ± 0.09
DT3.4_042015
April
11.22
8.7 ± 2.8
1.0 ± 0.3
0.8 ± 0.3
0.09 ± 0.03
DT3.3_052015
May
10.51
7.3 ± 2.1
0.8 ± 0.2
0.8 ± 0.3
0.08 ± 0.03
DT3.3_062015
June
8.28
9.0 ± 2.8
0.7 ± 0.2
0.9 ± 0.3
0.07 ± 0.02
DT3.3_072015
July
5.86
9.7 ± 6.3
0.6 ± 0.4
1.1 ±, 0.7
0.06 ± 0.04
DT3.3_082015
August
43.39
6.4 ± 3.3
2.8 ± 1.4
0.6 ± 0.4
0.26 ± 0.17
DT3.3_092015
September
21.90
4.3 ± 7.1
0.9 ± 1.6
0.6 ± 1.0
0.13 ± 0.22
DT3.3_102015
October
27.39
9.2 ± 7.3
2.5 ± 2.0
0.8 ± 0.6
0.22 ± 0.16
DT3.3_112015
November
14.59
6.1 ± 1.4
0.9 ± 0.2
0.7 ± 0.2
0.10 ± 0.03
DT3.3_122015
December
9.91
7.3 ± 2.5
0.7 ± 0.2
1.1 ± 0.2
0.11 ± 0.02
Semi-quantitative XRD mineral analyses of monthly Frisbee samples
collected at the three sites DT1–DT3, for the period December 2014 to
December 2015.
![]()
Summary plots of results from SEM-based individual particle analysis for each
month of 2015, expressed by number are presented in Supplement C to this
paper. From these particle size and shape measurements, equivalent
shape-dependent volumes for the particles were calculated, the summary plots
of which are shown in Appendix A. The volume of each particle is calculated
from the measured maximum and minimum diameters, and assuming a prolate
spheroid. Also, assuming a similar average density of, for example, 2.65 g cm-3
for all minerals in the deposition samples results in similar volume and
mass distributions patterns. This was confirmed by XRD measurements and the
abundance of quartz (2.65 g cm-3), feldspar (∼ 2.65 g cm-3), micas (∼ 2.83 g cm-3), and clays (∼ 2.7–2.8 g cm-3) found in the deposition samples. The volume
distributions were applied to assess the mass percentages and deposition
rates of each size bin, e.g., the mass percentages and mass deposition rates
of particles in bins less < 10 µm in average diameter, and
similarly less than < 2.5 µm in average diameter, together with
their uncertainties (Table 4). The contribution of particles
< 10 µm to the total measured mass varies between about 4 and 17 % with an
average of 8.6 % for the 12 months. Particles less than
2.5 µm range from about 0.6 to 4 %, with an average of
1.2 % for the 12-month period. From these percentages and the total
deposition rates, average deposition rates of
1.2 ± 0.7 g m-2 month-1 for < 10 µm and
0.1 ± 0.1 g m-2 month-1 for < 2.5 µm are
estimated.
The average size distribution of the 12 deposition samples (Fig. 7a) is
compared to that of the 13 surface soils (Fig. 7b) from potential dust
source regions along the Red Sea coastal plain (Jish Prakash et al., 2016). The
deposition samples with an average diameter of 0.9 µm are much
finer than the 3.9 µm average diameter of the < 38 µm sieved soils. In addition the Frisbee sampler is biased towards the
sampling of the coarser particles, as previously documented (Bergametti and
Forêt, 2014; Goossens, 2005).
(a) Deposition sample elemental compositions, expressed as
oxides and (b) fractions normalized to unity.
![]()
Mineral analysis by XRD
XRD analysis of the 27 samples (Fig. 8) shows variable amounts of quartz
(6–38 %, avg. 22 %) and feldspars (plagioclase, K-feldspar)
(5–34 %, avg. 20 %), clays (10–18 %, avg. 13 %), micas
(6–31 %, avg. 13 %), halite (1–53 %, avg. 7 %) with lesser
amounts of gypsum (1–8 %, avg. 4 %), calcite (0–8 %, avg.
2 %), dolomite (0–7 %, avg. 3 %), hematite (0–8 %, avg.
3 %), and amphibole (and pyroxene) (0–4%, avg. 1 %).
From the XRD, four broad mineral assemblages are distinguished: the first
and major assemblage is comprised of feldspars, clays, and micas, as well as
hematite and gypsum; the second group is of quartz, the third of halite, and
the fourth of calcite.
There is an increase in the halite concentrations at sites DT1–DT3, from
about 2 % (DT1) in December 2014 to about 53 % (DT2) in July 2015
(Fig. 8). From August onwards there is an abrupt decrease in halite content
to less than 5 %, except for samples collected at the DT2 (CMOR,
quay-side) site alongside the ocean. There was a simultaneous increase in the
proportion of quartz to a maximum of 38 % in April (DT3), decreasing
to less than 25 % at all sites after July 2015. The silicate mineral
group decreased systematically from about 72 % (DT1) in December 2014 to
about 25 % (DT2) in July. Except for two samples from the DT3 site
collected in September and October 2015, the dominant minerals after July,
2015 included the silicate assemblage, with concentrations of up to 80 %.
The variation in the proportions of the four mineral assemblages, especially
the halite, is ascribed to seasonal fluctuations in wind, humidity, and
precipitation, as well as the proximity of the sea to the sampling sites.
SEM-based SEIs of individual dust particles show that the larger particles
being composed of mineral aggregates and coatings on other mineral particles.
Examples (Supplement D) include particles composed of coatings of clay
minerals on quartz and feldspar; clusters of clay minerals, calcite, gypsum,
and halite; and particles and clusters of iron oxides and clay minerals. Similar coatings and
aggregates in re-suspended soil samples are reported by Engelbrecht et
al. (2016).
(a) Ion concentrations and (b) fractions totaled
to unity.
![]()
(a) Chemical abundances combined as normative minerals, and
(b) those normalized to 100 %.
![]()
Chemistry (XRF and IC)
As expected, the chemically analyzed deposition samples contain major amounts
of SiO2 (Appendix A, Fig. 9a, b), varying between 12 and 53 % (avg.
31 %) in the sample subset, occurring as quartz, and together with
Al2O3 (avg. 4 %) and CaO (avg. 2.3 %) in plagioclase, and
K2O (avg. 0.6 %) in potassium feldspars. SiO2, together with
Al2O3, Fe2O3, TiO2, MnO, MgO, and some K2O, is
also contained in the clays, micas, and amphiboles, previously identified in
these samples by optical microscopy and XRD. Lesser amounts of CaO are
contained in gypsum and calcite and, together with MgO, in dolomite. The iron
expressed here as Fe2O3 can be contained in hematite
(Fe2O3), goethite FeO(OH), or clay minerals such as illite, each
with different solubility. It has been suggested that large fractions of iron
in soils and dusts are contained as amorphous colloidal coatings on quartz
and feldspars (Engelbrecht et al., 2016).
The water-soluble cations (Appendix A, Fig. 10a, b) account for 1–19 %
and the anions for 1–30 % of the total mass, respectively. These account
for variable amounts of halite (1–32 %), and gypsum (1–9 %), with
lesser amounts of other chlorides and carbonates. Of importance as dust-borne
nutrients likely to be deposited in the Red Sea, are the low concentrations
of both water-soluble NO3- (avg. 0.8 %), and water-soluble
PO43- (avg. 0.2 %) compared to the total P2O5 (avg.
0.3 %) in the dust deposits. The phosphorus is contained in the largely
insoluble mineral apatite (francolite), found in the sedimentary rocks
underlying large parts of the Arabian Peninsula (Notholt et al., 2005).
The sum of chemical species, including elements expressed as oxides, as well as ion
concentrations, varies from 35 to 78 %, with an average of 56 % of the
measured chemical mass. The shortfall from 100 % is attributed in part to
components not analyzed for, including H2O, OH, carbon
(CO32-, organic carbon, elemental carbon), and artifacts of debris
deposited onto the samplers.
The chemical abundances were recalculated as normative minerals
(Fig. 11a, b), comparable in composition to those identified by XRD (Fig. 8)
and optical microscopy. The relative normative mineral abundances (Fig. 11b)
show variable amounts of quartz (avg. 52.4 %) feldspar (avg. 3.9 %),
kaolinite (2.6 %), calcite (8.8 %) dolomite (0.2 %), and hematite
(8.0 %), as well as the evaporate minerals gypsum (12.1 %), halite
(12.1 %), sylvite (0.2 %), and bischofite (0.2 %). There is also,
as shown by XRD, an increase in halite content from about 7.8 % in
January to about 25.9 % in July, followed by a sharp drop to about
4.6 % in August, with greater abundances at the CMOR quayside site in
September (51.0 %) and October (31.6 %), ascribed to sea spray from
stormy conditions during those 2 months.
Elemental mass ratios for the deposition samples from this study,
compared to those of soils from the Red Sea coastal plain (Jish Prakash et
al., 2016) and TSP samples from other countries of the Middle East
(Engelbrecht et al., 2009a). The TSP filter samples were collected by
low-volume aerosol samplers without size selective inlets, for 24 h sampling
periods.
Si / Al
Ti / Al
Fe / Al
Mg / Al
Ca / Al
K / Al
Frisbee
Deposition
6.86
0.14
1.47
0.11
2.17
0.34
Saudi soils
Sieved < 38 µm
13.60
0.44
2.52
0.65
0.36
0.43
Djibouti
TSP
0.92
2.19
1.12
0.88
0.74
1.14
Afghanistan
TSP
1.05
1.25
1.00
0.94
0.69
1.96
Qatar
TSP
1.02
0.24
0.98
1.40
2.07
0.93
UAE
TSP
1.29
0.28
1.52
2.85
2.16
1.02
Iraq
TSP
1.03
0.72
0.99
1.11
1.31
1.04
Kuwait
TSP
1.07
0.65
0.99
1.25
1.23
0.94
Elemental mass ratios of the Frisbee deposition samples are compared to the
< 38 µm sieved soil samples from the Arabian Red Sea coastal
plain (Jish Prakash et al., 2016), and total suspended particulate (TSP) samples
collected at other sites in the Middle East (Engelbrecht et al., 2009a) are
compared in Table 5. The average Si / Al ratio of 6.86 of the Frisbee
deposition samplers is intermediate to the 13.60 of the Arabian Red Sea
coastal soils and the approximately unity of the Middle East samples. The
Fe / Al ratios of the sample sets show similar relationships as the
Si / Al ratios, being intermediate to the Red Sea coastal soils and four
of the five other Middle East countries, excluding the UAE, to which it is
similar. The difference is ascribed to the greater abundance of the minerals
such as quartz in the coarser sieved soil samples, and less thereof in the
finer TSP fractions. The Ca / Al ratio of 2.17 is similar to those of TSP
samples from samples of Qatar (2.07) and the UAE (2.16), ascribed to the regional
carbonate-bearing soils in all three countries. The average Ti / Al,
Mg / Al, and K / Al ratios of the Frisbee deposition samples are
substantially lower than those of the Red Sea coastal soils, which may be
related to mineralogical differences in the dust source regions. Differences
can also be ascribed to larger percentages of Al-bearing minerals such as
clays in the deposition samples from this study.
Summary and conclusions
This study provides new mineralogical, physical, and chemical information on
deposition samples collected at the KAUST campus during 2015, as well as an
assessment of the seasonal variability of the regional dust deposition rates
onto the Saudi Arabian coastal plain.
Inverted Frisbee samplers with foam inserts are found to be robust, easy to
use, and provided comparable results for the collection of wet and dry
deposits. Once a month the samples are retrieved by flushing the deposits
into plastic flasks followed by freeze drying of the slurry and recovery of
all suspended particles and dissolved salts. The average deposition rate at
KAUST for 2015 was 14 g m-2, with 4 g m-2 in December,
20 g m-2 in March, 5 g m-2 in July, and 28 g m-2 in
September and October, decreasing to 11 g m-2 the following December.
The changes are ascribed to seasonally variable meteorological conditions,
including high humidity prevailing along the Arabian Red Sea coastal plain
during the late summer and autumn months. The particle size distributions
provide an assessment of < 10 and < 2.5 µm
dust deposition rates, the former varying 0.6–2.8 g m-2 and the
latter 0.06–0.29 g m-2 per month. We suggest these deposition rates as
proxies for those of PM10 (coarse) and PM2.5 (fine), respectively.
Chemical analysis, confirmed by XRD, points to a consistent silicate mineral
fractions for the deposition samples, at all sampling sites for the entire
sampling period. The Si / Al, Fe / Al, and Ca / Al ratios of the deposition
samples fall within the range of the soil samples previously collected along
the Arabian Red Sea coastal plain as well as the TSP size fractions
collected at several sites in the Middle East. It is proposed that the dust
deposits along the Red Sea coast are a mixture of dust emissions from local
soils and soils imported from distal dust sources. Airborne mineral
concentrations are greatest at or close to dust sources, compared to dusts,
due to medium and long-range transport.
For 2015, there are marked similarities between monthly distribution patterns
of the deposition samples and AOD measured at KAUST, as well as visibility
measurements from Jeddah airport, 70 km to the south. This shows that both
the AOD and visibility measurements mirror fluctuations in dust deposition,
although it may not be justified to calculate quantitative interrelationships
without further research.
Except for the variable halite fractions and local construction dust, there
are minor variations in the mineralogical content of the dust samples
collected on the KAUST campus. To better model the dust being deposited in
the Red Sea and coastal plain, the sampling campaign should be extended to
sites beyond the KAUST campus. Such a sampling site was recently set up on
an island off the coast from KAUST. Inclusion of particle size with
mineralogical and chemical measurements provides more effective data for the
modeling community.
The deposition samplers collect all particle sizes; however, bin aerosol
models usually consider only PM10. The estimated PM10 deposition
rates are lower than the total particulate deposition rates we observed.
However, the size distribution of deposited particles shown in Fig. 7a and
Appendix A could be used to assess the contribution of PM10 in deposited
mass and reconcile models with observations. Alternatively, the calculated
particle size range in the models can be potentially be extended to cover
TSP. However, this could be computationally extensive.
As an approximation of source contributions, the sampler with the lowest
deposition rate can be considered to have negligible or the least amount of
local dust or sea salt (Fig. 5). In the months of December 2014, January,
April, March, June, July, and December 2015, the deposition rates at the four
sites were similar and considered to have no or negligible amounts of dust
from local construction, campus roads, marine salt, or other particulates.