1. religion and spirituality
2. judaism

INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. (in press)
Published online in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/joc.1587
Suspended dust over southeastern Mediterranean and its
relation to atmospheric circulations
Uri Dayan,a * Baruch Ziv,b Tova Shooba and Yehouda Enzela,c
a
c
Department of Geography, The Hebrew University of Jerusalem, Israel
b The Open University of Israel, Raanana, Israel
Institute of Earth Sciences, The Hebrew University of Jerusalem, Israel
ABSTRACT: The Middle East deserts are often subjected to dust, which reduces horizontal visibility to 5 km, and
sometimes even to <1 km. The present study examines the annual and inter-annual occurrences of dust events based on
37 years of visibility observations from Hazerim (near Beer Sheba) correlated with PM10 dust concentration. The visibility
data was converted to PM10 dust concentration, using concurrent data for three years. We then analyse the linkage between
dust and synoptic- to global-scale weather systems.
The monthly data indicate that the dust season starts in October and ends in May, with a maximum in March. More
than 89% of the total annual dust is accumulated between December and May, the ‘high dust season’. The annual totals
vary as much as an order of magnitude from year to year.
The synoptic system that produces the majority of the dust over the northern Negev is the Cyprus Low, contributing
2/3 of both the total yearly dust yield and of the number of dust observations. This suggests that a positive relationship
exists between the dust in the Negev and rainfall in north Israel, both of which are generated by Cyprus Lows. Indeed,
a significant (at 0.05 level) correlation of +0.30 was found between the two. Correlation maps evidence that in dustrich years the cyclonic activity over the Mediterranean is abnormally high and in poor-dust years it is low. A highly
significant negative correlation (−0.66) was found between the dust yield and the intensity of the North Atlantic oscillation
(NAO), which modulates the cyclonic activity over Europe and the northern Mediterranean. This may also imply that
periods in which more dust accumulated as loess in the northern Negev may indicate the existence of negative NAO
phase, and concurrently, warmer conditions over the Sahara, colder conditions over Europe and enhanced rainfall over the
Mediterranean Basin. Copyright  2007 Royal Meteorological Society
KEY WORDS
PM10 ; visibility; synoptic systems; NAO; dust; Israel; paleoclimate
Received 14 December 2006; Revised 28 May 2007; Accepted 3 June 2007
1.
Introduction
The Negev Desert in southern Israel (Figure 1) is located
between the Saharan and the Arabian deserts. The Negev
is frequently subjected to suspended dust, reaching at
times the degree of a dust storm (e.g. Dayan et al., 1991).
Ganor and Foner (2001), using direct dust measurements
from Tel Aviv and Jerusalem, showed that the total
suspended particulates (TSP) in Israel consists of a
background constant value of 80–100 µg m−3 , which
originates from local sources, on top of which dust is
transported in pulses during spring and autumn.
The Sahara is known as the largest source of Aeolian
soil dust (Goudie and Middleton, 2001), imparting almost
half of the dust supplied to the world’s oceans (Schutz
et al., 1981; D’Almeida, 1987; Goudie and Middleton,
2001). Dust storms, moving from the Sahar Desert to
the eastern Mediterranean, occur between October and
May, but mostly from December to April (Lunson,
* Correspondence to: Uri Dayan, Department of Geography, The
Hebrew University of Jerusalem, 91905, Israel.
E-mail: [email protected]
Copyright  2007 Royal Meteorological Society
1950; Katsnelson, 1970; Ganor, 1975). The North African
origin of the dust over the study region was also shown
by Israelevich et al. (2002) using total Ozone mapping
spectrometer (TOMS) data.
Dust-fall has beneficial effects on the soil as a source of
nutrients. In addition, suspended dust plays an important
role in neutralizing acid rain in the region (Mamane
et al., 1987; Ganor and Foner, 2001). The amount of
dust deposited as primary and reworked loess over the
northwest Negev in the Quaternary period has reached a
depth of 13.5 m (Ganor and Foner, 2001).
Synoptically, the majority of dust storms, which occur
above the Mediterranean over the year, are caused by
an increased surface pressure gradient. They are usually
associated with the passage of either a cold or a warm
low-pressure system. Saharan depressions develop when
a polar air mass flows behind dry desert air. When this
happens, there is a corresponding induced flow of continental tropical air northwards, transporting dust with
it (Dayan et al., 1991). The frequency of dust mobilization over North Africa and of dust intrusions over
the Mediterranean Sea are, therefore, strongly related to
U. DAYAN ET AL.
Figure 1. Study area. The asterisk in the middle denotes the location
of Beer Sheba (Israel).
the climatology of depressions affecting North Africa
(Moulin et al., 1998). Katsnelson (1970) attributed dust
events over Israel to thermal lows (including Sharav
cyclones and Red Sea trough) and mentioned dynamic
cold cyclones (Cyprus Lows) as a possible additional
factor.
Moulin et al. (1998), using METEOSAT imagery to
explain the general pattern of the dust transport over
the Mediterranean basin (MB), stated that in its eastern
part most of the dust transport occurs between January
and July. They found that the pulsative nature of dust
intrusions into the region indicate that they originate
from transient disturbances, i.e. cyclonic events, such as
the Sharav cyclone. Dayan et al. (1991) differentiated
between dust reaching the region either from the west
and southwest (i.e. the Sahara) or from east of the
Negev (i.e. from Arabia). They showed that the Saharan
dust transport that reaches the eastern Mediterranean is
usually intensive and of longer duration (2–4 days) in
comparison with the dust transported from the Arabian
Desert.
We question why the Cyprus Low has not been
mentioned as a major synoptic-scale system responsible
for dust transport to the study region, since it seems
to supply the necessary conditions that can cause dust
mobilization and transport. Ganor and Foner (2001) and
Moulin et al. (1998) showed that the mean annual TSP
begins rising as early as January, which is the month
in which Cyprus Lows are most frequent (Alpert et al.,
1990), while other systems, regarded as producing dust,
are less frequent. Both studies also mentioned the fact
that dust is removed by precipitations, and therefore the
residence time of particles in the atmosphere is much
shorter when it rains, as is the case during the Cyprus
Lows, and so reduces their signature as a major dust
contributor. This issue urged us to further elaborate the
relative contribution of the various synoptic systems in
dust production over the Negev Desert.
Once the synoptic-scale circulations that are responsible for mobilization, transport and deposition of dust in
Copyright  2007 Royal Meteorological Society
the northern Negev are determined, one can deduce from
the deciphering of loess sediments the past occurrence of
the pertinent synoptic system over the region.
Here, we explore atmospheric circulations controlling
suspended dust over the southeastern Mediterranean. For
this end we approached the problem somewhat differently
from previous researches in that the data is taken from
the northern Negev, where most of loess deposition has
occurred, and the analysis is based on a long record
of 37 years. The data is the 3-h visibility observations,
representing dust concentrations.
First, visibility is converted to dust concentration (Section 2). Section 3 specifies the annual and inter-annual
variations of dust concentrations. Section 4 analyses the
synoptic systems with which the dust is associated and
quantifies the relative contribution of each of these systems to the observed suspended dust. Then, as dust processes are shown to be related to the spatial scale of
the Sahara, Europe and the entire MB, we analyse the
relations between the inter-annual variation of the dust
concentrations in the study area and large-scale oscillations over the northern hemisphere. Finally (Section 5),
we propose a way in which dust-rich and dust-poor past
episodes can provide information regarding past climatic
conditions over Africa, Europe and especially the eastern
Mediterranean.
2.
Methods and databases
The visibility data was recorded as 3 hourly observations
from the station located at the airport of Hazerim (near
Beer Sheba, 31.25 ° N, 34.8 ° E) in the northern Negev
(Figure 1). Since these observations serve for aviation
activity they are reliable and uninterrupted. Moreover,
this station has records beginning from as early as
1967. To minimize possible inaccuracies in subjectively
estimated visibility as reported by observers during the
night hours, only daylight visibility records (06–15 UTC)
were used. This limit was imposed for all of the months
to eliminate a possible bias as an outcome of differences
in the day-length. Two criteria were applied to select dust
observations:
• Visibility ≤5 km
• No weather phenomena, other than haze or dust,
responsible for reduction in visibility, were reported.
To quantify the dust, we converted the visibility to PM10 (small particles with a diameter ≤10 ×
10−6 m). The conversion function was developed, using
2001–2003 visibility data, together with co-located measurements of PM10 dust concentration collected by the
Israeli Ministry of Environment. There is no unique and
common function for the estimation of PM10 from visibility. Various regression equations have been proposed that
relate the two variables (Chepil and Woodruff, 1957; Patterson and Gillette, 1977; Tews, 1996). One of the main
reasons is that the visibility is strongly influenced by dust
Int. J. Climatol. (in press)
DOI: 10.1002/joc
ATMOSPHERIC CIRCULATIONS GOVERNING DUST IN THE SE MEDITERRANEAN
particle detailed size distribution (El Fandy, 1953). We,
therefore, derived a function that best fit our data that
include 74 individual observations, in which the visibility was ≤5000 m. The relation which yielded the highest
correlation between the visibility and PM10 concentration
(R = 0.72, p = 0.005), is:
y = −505 Ln(x) + 2264,
(1)
where y is the PM10 concentration, in µg m−3 , and
x is the visibility, in 100 m units. The function and
the raw data are presented in Figure 2 and specified in
Table I. This function indicates that even the threshold
visibility of 5 km, implying PM10 concentration of 287 ±
169 µg m−3 , reflects dust concentrations that exceed
three times the background value (Ganor and Foner,
2001). Even if the uncertainty value is subtracted, the
resulting PM10 concentration of 118 µg m−3 , is still
higher than their upper background value (100 µg m−3 ).
Therefore, the 5 km threshold can be considered as
reflecting the pulses associated with dust outbreaks.
The evaluation of the amount of dust is done using
two indicators. One, is the number of ‘dust observations’,
i.e. the number of 3-hourly observations in which the
visibility was ≤5 km. The second is ‘accumulated dust’,
i.e. the summation of the dust concentrations, derived
from the 3-h visibility observations (when ≤5 km) by
Equation (1). The accumulated dust can be regarded as
a proxy for the regional dust yield, due to the exclusion
of night observations, and therefore, is treated here as an
index, without using physical units.
The relation between dust and the regional synoptic
patterns was analysed, based on the NCEP/NCAR reanalysis data set (www.cdc.noaa.gov/cdc/reanalysis/), with
2.5° × 2.5° resolution (for more details, see Kalnay et al.,
1996 and Kistler et al., 2001).
The synoptic systems dominating the study region for
each dust observation were determined subjectively by
three trained forecasters, using sea level pressure (SLP)
maps for the domain 0° –50 ° E, 10° –40° N, as one of the
following types: Cyprus Low, Sharav cyclone, Red Sea
trough, Persian trough, High over Israel, and High to the
east of the region Dayan et al. (1991).
The preference of subjective classification over objective is based on Yarnal (1993), who stated that despite
the positive features of objective automated classification,
it has an important drawback, i.e. that no matter how
much the parameters are fine tuned, there will always be
a ‘black box’ element. On the other hand, ‘At every step
of the manual classification process, the investigator . . .
manipulates the classification to match his understanding
of the physical world and to produce the type of results
required by the research design’ (Yarnal, 1993).
The monthly numerical indices representing the largescale oscillations and global teleconnections used in
this study are taken from: http://www.cdc.noaa.gov/
ClimateIndices/corr.html and are based on the analysis
of Branston and Livezey (1987).
3.
Figure 2. The regression curve between visibility (in 100 m units)
and dust concentration (of PM10 , in µg m−3 ). The error bars denote
the standard error between the calculated and observed individual
concentrations for each visibility group.
Table I. Calculated PM10 concentrations, according to Equation (1), for selected visibility ranges, together with standard
errors.
Visibility
(km)
1
2
3
4
5
Calculated PM10
concentration
(µg m−3 )
Standard error
(SE, µg m−3 )
Relative
SE
(%)
1100
750
545
399
287
385
412
250
255
169
35
55
46
64
59
Copyright  2007 Royal Meteorological Society
Annual and inter-annual variability of dust
Figure 3 shows the average annual distribution of number
of dust observations and the respective accumulated dust.
The uncertainty in the calculated PM10 concentration,
being ∼50%, reduces the significance of the accumulated
dust, but the similarity between its annual course and that
of the number of dust observations makes it acceptable.
The annual course is dominated by a gradual increase
from the fall season through the end of the winter
(March), when it reaches the yearly maximum and then
drops sharply from April to June. In the 5 months,
December–April, the monthly average accumulated dust
exceeds 50% of the maximum monthly average value,
so they can be regarded as the ‘high dust season’. This
general behaviour is in agreement with Ganor and Foner
(2001). In the summer months (June–September) there
are almost no dust events. Accordingly, the ‘dust-year’
was set to start in August and end the following July.
Hereafter, a dust-year will be denoted by both years
included, e.g. 1967–1968 refers to the year spanning
from August 1967 to July 1968. The segmentation of
the time series to dust-years reduced the number of years
contained in the sample to 36.
Figure 4 shows the time series of the annual accumulated dust during the study period. The first 10 years
(1968–1979) are characterized by high amplitude annual
Int. J. Climatol. (in press)
DOI: 10.1002/joc
U. DAYAN ET AL.
Figure 3. Monthly distribution of average accumulated dust, represented by columns, and number of observations (line) in Beer Sheba,
averaged over the years 1967–2003.
variations in comparison with the rest of the record. There
are two distinct peaks in the accumulated dust: one in the
late 60s (with the absolute maximum in 1968–1969) and
the other in the late 70s. Note that extreme years tend to
group, especially in the first 15 years (e.g. two successive
rich-dust years in 1967–1969 and four poor-dust years in
1972–1976). The limited record (36 dust-years) and the
fluctuations observed in the accumulated dust (between
3161 in 1990–1991 and 41 058 in 1968–1969) make any
analysis of long-term trend for this time-series meaningless. It is noteworthy that the uncertainty in the yearly
dust yield (∼1/2, see error bars in Figure 4) cannot mask
the inter-annual variation (factor of ∼10), and therefore,
has no adverse effect on the validity of the results.
4.
4.1.
Relation to synoptic and large-scale systems
The contributions of individual synoptic types
Figure 5 and Table II present the contribution of the
various synoptic systems to the suspended dust over the
study region. The contribution for each synoptic type is
estimated by both the accumulated dust and the number of
dust observations. The most prominent feature, reflected
by both indicators, is the dominant contribution of the
Figure 4. Inter-annual variation of yearly (from the preceding August
to the current July) accumulated dust in Beer Sheba.
Cyprus Low, being 4 times larger than that of the second
one, the Sharav cyclone. This striking finding contradicts
the widely accepted idea that the Sharav cyclone is
the main contributor to the suspended dust over this
region. The annual occurrence distribution of the two
most effective synoptic types explains why the yearly
maximum is in the late winter and early spring, when
these two systems are most frequent (Alpert et al., 2004).
For 4.1% of the dust observations the synoptic types were
difficult to classify, or coincided with transition between
different distinct synoptic types (termed ‘undefined’ in
Table II).
The average intensity of dust concentration for each
synoptic type was calculated by dividing the accumulated
dust by the number of dust observations. The type which
was found to be most effective is the Cyprus Low, producing an average PM10 concentration of 676 µg m−3 ,
while the less effective is the ‘high to the east’ type,
yielding 469 µg m−3 on the average. It is noteworthy,
that in addition to the superiority of the Cyprus Low over
the Sharav cyclone in its associated dust observations, its
average intensity also exceeds that of the latter, by 13%.
4.2. Regional patterns for extreme dust-years
To identify the most prominent features in the regional
circulation patterns for extremely dusty years, we
Figure 5. Dust distribution among the synoptic systems, demonstrated by the number of dust observations and accumulated dust (the y axis
refers to both representations), based on the period 1967–2003.
Copyright  2007 Royal Meteorological Society
Int. J. Climatol. (in press)
DOI: 10.1002/joc
ATMOSPHERIC CIRCULATIONS GOVERNING DUST IN THE SE MEDITERRANEAN
Table II. Distribution of dust among the synoptic systems represented by total number of dust observations and accumulated
PM10 (in 1000 µg m−3 units) along the study period.
Synoptic
system
Cyprus Low
Sharav cyclone
Red Sea trough
Persian trough
High over Israel
High to the east
Undefined
No. of
observations
Accumulated
dust
472
109
100
17
48
18
33
319
65
62
11
31
8
24
extracted composite maps, one for the four dust-rich
and other for the four dust-poor years, i.e. top and
lowest ∼10% of the years. The advantage of adopting an
extreme value approach is that it provides insights into
(and sometimes resolve) the relation existing between
environmental phenomena and the state of the causative
large-scale atmospheric circulations (e.g., Enzel et al.,
1989; Kahana et al., 2002; Xoplaki et al., 2003; Dayan
and Lamb, 2005).
The rich-dust years were 1967–1968, 1968–1969,
1976–1977 and 1977–1978, and the poor-dust years
were 1989–1990, 1990–1991, 1996–1997 and
1998–1999. The season chosen for this analysis is
December–May, in which 89% of the average yearly dust
is obtained, according to both number of dust observations and accumulated dust (Figure 3). The most prominent feature in the composite SLP for the dust-rich years
(Figure 6(a)) is a pair of low-pressure centres with central pressure <1014 hPa, one over the Gulf of Genoa
and the second over southern Turkey. In contrast, the
four dust-poor years (Figure 6(b)) were characterized
by a belt of high pressure extending across southern
Europe, and a relatively shallow and dispersed low pressure (∼1016 hPa) along the MB. These features are further emphasized by the concurrent anomaly composites
(Figure 7(a) and (b)). In the dust-rich years, the SLP
over Europe is anomalously low by more than −2 hPa
over northwestern Europe, and by −1.5 hPa over Turkey.
The SLP average anomaly for the four dust-poor years
exceeds +2 hPa over Central and Western Europe and
over Turkey. The North Atlantic experiences anomalies
of opposite signs with respect to Europe, with even higher
amplitudes, i.e. +5 hPa and −3 hPa for the dust-rich and
dust-poor years, respectively.
Another difference between the dust-rich and dust-poor
years is the well-pronounced contrast between the Azores
High and the Icelandic Low, being more emphasized in
the poor dust years. As the pressure difference between
these two regions measures the intensity of the North
Atlantic oscillation (NAO), we suggest that the NAO
plays an important role in the inter-annual variability of
the suspended dust over the northern Negev.
4.3. Correlation maps
In order to search for inter-relation between the dust in the
Negev and remote climatic conditions, correlation maps
were generated between the seasonally accumulated dust
and the respective averages of atmospheric fields in the
lower levels, i.e. SLP and 850 hPa temperature, for which
the correlation patterns were most distinct (Figure 8(a)
and (b), respectively). Since the correlations are based
on 36 years, all correlations exceeding 0.28 and 0.39
are significant at the 0.05 and 0.01 levels, respectively. A
region with positive (negative) correlation indicates that
dust concentrations over the study region is associated
with a positive (negative) anomaly in the relevant field
over the pertinent location.
The most prominent feature is a negative correlation
region covering the entire MB and west North Africa,
Figure 6. Composite SLP (hPa) from December to May for (a) the four rich-dust years (1967–1968, 1968–1969, 1976–1977, and 1977–1978)
and (b) the low dust years (1989–1990, 1990–1991, 1996–1997, and 1998–1999).
Copyright  2007 Royal Meteorological Society
Int. J. Climatol. (in press)
DOI: 10.1002/joc
U. DAYAN ET AL.
with an extreme correlation of −0.7 over the eastern Mediterranean. A positive correlation region extends
from the North Atlantic toward northeastern Europe, with
a maximum of > + 0.55 over Iceland (Figure 8(a)). In
the 850 hPa temperature (Figure 8(b)) a most significant positive correlation centre (+0.6) exists over the
Sahara, reflecting its tendency for warmer temperatures
during dust-rich years, accompanied by lower temperatures over Europe (R exceeds −0.4 to the west of France
and over western Russia). Similar features were found in
correlation maps based on the number of dust observations.
The SLP correlation and composite maps indicate that
the dust yield is larger in years with intensive cyclonic
activity, which agrees with the dominance of Cyprus
Lows for dust in the study region. The temperature correlation map indicates that in rich-dust years the MB is
subjected to enhanced temperature gradient, leading to
baroclinic conditions, which may explain the increase
in cyclonic activity. Furthermore, the higher temperature in the lower levels over the Sahara in rich-dust
years suggests that the region is subjected to abnormally
enhanced instability, making it more prone to dust mobilization.
4.4. Correlation between dust in the Negev and the
NAO
As mentioned above, the gradient in the SLP correlation
(Figure 8(a)) and the difference in the pressure gradient across the British Isles, between the dust-poor and
dust-rich years (compare Figure 6(a) and (b)), indicates
that the NAO plays a role in determining dust concentrations over the study area. More specifically, the more
intense the NAO the less dust is recorded over the study
area, implying that a negative correlation exists between
the two. Since the NAO is most indicative in the months
Figure 7. As in Figure 6, but for SLP anomaly.
Figure 8. Correlation between the yearly accumulated PM10 and (a) SLP and (b) 850 hPa temperature.
Copyright  2007 Royal Meteorological Society
Int. J. Climatol. (in press)
DOI: 10.1002/joc
ATMOSPHERIC CIRCULATIONS GOVERNING DUST IN THE SE MEDITERRANEAN
December–March (Hurrel, 1995), these month were used
for analysing this relation. Figure 9 demonstrates the
expected inverse relation. A significant negative correlation (R = −0.66; p = 0.005) was obtained, explaining 43% of the variance of dust concentrations over the
Negev.
The relation between the NAO and the cyclonic
activity over the MB was formulated by Hurrel (1995),
who stated that high NAO is associated with a shifting
of cyclone tracks toward Scandinavia, at the expense of
the Mediterranean, and vice versa. This implies that when
the NAO is in its positive phase, the main dust carriers to
the Eastern Mediterranean, the Mediterranean cyclones,
become less frequent and of lower intensity; then the
region experiences less dust during NAO positive phases.
The coincident negative correlation between the NAO
and dust in our study region and the SLP in the eastern
Mediterranean is consistent with the positive correlation
found by Ben-Gai et al. (2001) between the SLP in Israel
and the NAO.
4.5. Correlation between dust in the Negev and
rainfall in northern Israel
The enhanced cyclonic activity found over the MB in
rich-dust years, and the contributions of the Mediterranean cyclones to rainfall in northern Israel, suggest
that a positive relation exists between rainfall in northern
Israel and the dust in the Negev, i.e. that most rainy years
are characterized by high dust yield. This relationship is
demonstrated by Figure 10, showing a tendency of rainy
years to coincide with rich-dust years. However, several
rainy years show the opposite. For instance, 1982–1983
and 1991–1992, which belong to the upper 20% quintile
of rainfall (Ziv et al., 2006), were poor with dust. This
results in a positive and significant correlation (R = 0.30;
p = 0.05), though weak, between the rainfall in northern
Israel and the dust concentrations in the Negev.
To investigate whether different types of rainy years
exist, we compiled two composite maps of SLP-anomaly,
for four rainy and dust-rich years and for four rainy
and dust-poor years. The results (Figure 11) indicate the
Figure 9. Annual variation of accumulated dust in Beer Sheba for October–May (dashed) and the NAO index averaged over December–March
(solid).
Figure 10. Annual variation of accumulated dust in Beer Sheba for October–May (dashed) and the seasonal rainfall in the northern half of Israel
(solid).
Copyright  2007 Royal Meteorological Society
Int. J. Climatol. (in press)
DOI: 10.1002/joc
U. DAYAN ET AL.
Figure 11. Composite SLP anomaly in hPa units for (a) four rainy winters which were rich with dust (1967–1968, 1968–1969, 1979–1980 and
1980–1981) and (b) four rainy winters which were low dust years (1973–1974, 1982–1983, 1987–1988 and 1991–1992).
existence of a distinct pattern for each group. For the
rainy/dust-rich years, a pronounced negative anomaly,
corresponding to enhanced cyclonic activity, covers the
eastern and central Mediterranean. For the rainy/dustpoor years, the negative anomaly is confined to the
eastern Mediterranean only, whereas positive anomaly
covers the rest of the basin and Europe. The rainy/dustrich type is consistent with the general situation for richdust years, in which cyclones sweep the whole MB. These
cyclones produce strong southwesterly winds ahead of
their cold fronts to the south of their tracks, i.e. over
North Africa and Sinai, and so transport dust from North
African sources eastward, to the Negev. Ganor and Foner
(2001) did not identify the Cyprus Low as the major
synoptic system for dust transport since their sites in
which they collected dust were located over central Israel
and not over the Negev. The latter type, rainy/poordust years, is exceptional in the sense that the SLP
anomaly gradient along eastern North Africa reflects
a northwesterly flow there that deviates dust plumes,
whenever produced, from being transported to Sinai and
Israel. In addition, the existence of a negative pressure
anomaly in the eastern Mediterranean only reflects the
tendency of cyclones to be formed there rather than to
approach from the west, as typifying the rainy/dust-rich
years.
5.
Discussion and summary
The monthly and annual characteristics of the lowerlevel suspended dust over south Israel were analysed,
together with its causative synoptic- to hemispheric-scale
conditions. Our study is based on day-time 3-hourly
visibility observations from Beer Sheba for 1967–2003.
These data are represented by number of observations in
which the visibility was ≤5 km and the concentration of
PM10 . The concentration was extracted from the visibility
through an empirical logarithmic function; derived from
Copyright  2007 Royal Meteorological Society
3 years of concurrent PM10 concentration and visibility
observations. The dust-year extends from August through
the next July, with 89% of the yearly accumulated dust
obtained in December–May, and the maximum in March.
The seasonality found here is similar to that found by
Lunson, 1950; Katsnelson, 1970 and Ganor, 1975. The
total annual dust yield varied in the study period between
3161 and 41 058, with a median of 11 538.
Classification of the synoptic systems dominating the
region in observed dust-days indicates that ∼60% of the
dust yield was generated by Cyprus Lows, while the
Sharav cyclones, which had been regarded as the main
contributor, and the Red Sea trough, contributed ∼12%
each.
Composite maps for years extremely rich and poor
with dust, together with correlation maps, show that
dust in the southeast Mediterranean is associated with
an intensification of cyclonic activity all over the MB.
This seems to be related to the enhanced baroclinity over
that region due to warming of the Sahara and cooling
of Europe, as reflected in the 850 hPa temperature
correlation map.
The NAO seems to play a crucial role for dust conditions over the Negev. Its negative phase results in colder
winters over northern Europe and a more frequent passage of winter storms over the MB, which enhances dust
over the Negev. We obtained a negative correlation, of
−0.66, between them, which is, apparently, inconsistent
with Moulin et al. (1997), who found positive correlation between the NAO and dust over the entire MB.
This disagreement stems from two reasons: (1) our study
region (Figure 1) occupies only a small portion of their
research domain and (2) the timing of the ‘high dust season’. We identified the winter as the ‘high season’, which
was poorly documented by Moulin et al. (1997) due to
frequent cloud masking in this season.
The close relation between the NAO and the dust yield
in the Negev Desert implies that the thick loess sequences
Int. J. Climatol. (in press)
DOI: 10.1002/joc
ATMOSPHERIC CIRCULATIONS GOVERNING DUST IN THE SE MEDITERRANEAN
in the Negev may corresponded to past variations in the
NAO and are in temporal agreement with other records
in Europe and the Mediterranean controlled or modulated
by NAO.
Our findings can also shed light directly on the
regional paleoclimates. During the late Pleistocene, the
southern Levant experienced the following conditions:
(1) high stands of Lake Lisan, the precursor of the Dead
Sea (Bartov et al., 2003), indicating very wet conditions in northern Israel, northern Jordan, Lebanon, and
Syria, and (2) intensive dust (fine sand and silt) deposition that formed the thick loess deposits in the northern Negev (e.g., Magaritz, 1986). This intensive dust
deposition affected also areas farther north in Israel
where late Pleistocene fine silts and clays were able
to accumulate as Mediterranean soils. This finer dust
deposition occurred when the Sahara was dry in contrast with early to middle Holocene times, when dust
deposition in the Negev and central Israel was minimal (Gerson et al., 1985) as the Sahara was relatively
wet as in other interglacials (e.g., Frumkin and Stein,
2004) and the Dead Sea level was lower than that of
Lake Lisan, indicating much less rainfall in northern
Israel. These regional patterns are in full agreement with
the finding of this study but on a significantly larger
timescale.
Appendix A: the Synoptic Systems Contributing the
Bulk of the Dust.
Figure A1 exemplifies the three synoptic systems which
are the main dust contributors for the study area. These
are, in a descending order of importance, the Cyprus Low,
the Sharav cyclone and the Red-Sea trough.
Acknowledgements
We thank the Department of Climate, the Israel Meteorological Service, for the visibility data and the Israeli
Ministry of Environment for the measurements of dust
concentration. This study was also supported by the
Israeli Science foundation (ISF, grant No. 764/06). Special thanks are due to Michal Kidron from the Cartographic Laboratory of the Department of Geography at
the Hebrew University of Jerusalem for her assistance in
preparation of the figures.
Figure A1. Sea level pressure (hPa) for three selected cases in which dust was observed in: (a). Cyprus Low on 30 Dec 1969, (b). Sharav cyclone
on 31 Mar 1969 and (c). Red Sea trough on 14 Oct 1978.
Copyright  2007 Royal Meteorological Society
Int. J. Climatol. (in press)
DOI: 10.1002/joc
U. DAYAN ET AL.
References
Alpert P, Neeman BU, Shay-El Y. 1990. Climatological analysis of
mediterranean cyclones using ECMWF data. Tellus 42A: 65–77.
Alpert P, Osetinsky I, Ziv B, Shafir H. 2004. Semi-objective classification for daily synoptic systems: application to the Eastern Mediterranean climate change. International Journal of Climatology 24(8):
1001–1011.
Bartov Y, Goldstein S, Stein L, Enzel Y. 2003. Catastrophic arid
events in the East Mediterranean linked with the North Atlantic
Heinrich events. Geology 31: 439–442.
Ben-Gai T, Bitan A, Manes A, Alpert P, Kushnir Y. 2001. Temperature and surface pressure anomalies in Israel and the North Atlantic
Oscillation. Theoretical and Applied Climatology 69: 171–177.
Branston AG, Livezey RE. 1987. Classification, seasonality and
persistence of low-frequency atmospheric circulation patterns.
Monthly Weather Review 115: 1083–1126.
Chepil W, Woodruff N. 1957. Sedimentary characteristics of dust
storms. II. Visibility and dust concentration. American Journal of
Science 255: 104–114.
D’Almeida GA. 1987. Desert aerosol characteristics and effects
on climate. In Paleoclimatology and Paleometeorology: Modern
and Past Patterns of Global Atmospheric Transport. Lienen M,
Sarnthein M (eds). Kluwer Academic Publishers; 311–338.
Dayan U, Lamb D. 2005. Global and synoptic-scale weather patterns
controlling wet atmospheric deposition over central Europe.
Atmospheric Environment 39: 521–533.
Dayan U, Heffter J, Miller J, Gutman G. 1991. Dust intrusion events
into the mediterranean basin. Journal of Applied Meteorology 30:
1185–1199.
El Fandy FG. 1953. On the physics of dusty atmosphere. Journal of
the Royal Meteorological Society 79: 284–287.
Enzel Y, Cayan DR, Anderson RY, Wells SG. 1989. Atmospheric
circulation during Holocene lake stands in the Mojave Desert:
evidence of regional climate change. Nature 341: 44–47.
Frumkin A, Stein M. 2004. The Sahara-East Mediterranean dust and
climate connection revealed by strontium and uranium isotopes in
a Jerusalem speleothem. Earth and Planetary Science Letters 217:
451–464.
Ganor E. 1975. Atmospheric dust in Israel – sedimentological and
meteorological analysis of dust deposition, PhD thesis, The Hebrew
University of Jerusalem; 224.
Ganor E, Foner HA. 2001. Mineral dust concentrations, deposition
fluxes and deposition velocities in dust episodes over Israel. Journal
of Geophysical Research 106(D16): 18,431, (2000JD900535).
Gerson R, Grossman S, Amit R. 1985. A procedure for evaluation of
dust potential and variability in desert terrains. In Proceedings, Dust
Environment Workshop, II, US Army Corps of Engineers, Vicksburg,
Mississippi; 84.
Goudie AS, Middleton NJ. 2001. Saharan dust storms: nature and
consequences. Earth Science Reviews 56: 179–204.
Hurrel JW. 1995. Decadal trends in the North Atlantic Oscillation
and relations to regional temperature and precipitation. Science 269:
676–679.
Israelevich PL, Levin Z, Joseph JH, Ganor E. 2002. Desert aerosol
transport in the Mediterranean region as inferred from the TOMS
Copyright  2007 Royal Meteorological Society
aerosol index. Journal of Geophysical Research 107(D21): 4572,
Doi:10.1029/2001JD002011.
Kahana R, Ziv B, Enzel Y, Dayan U. 2002. Synoptic climatology of
major floods in the Negev desert, Israel. International Journal of
Climatology 22: 867–882.
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L,
Iredell M, Saha S, White G, Woollen J, Zhu Y, Chelliah M,
Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C,
Wang J, Leetmaa A, Reynolds R, Jenne R, Joseph D. 1996. The
NCEP/NCAR 40-Year reanalysis project. Bulletin of the American
Meteorological Society 77: 437–471.
Katsnelson J. 1970. The frequency of dust storms at Beer Sheba. Israel
Journal of Earth Sciences 19: 69–76.
Kistler R, Kalnay E, Collins W, Saha S, White G, Woollen J,
Chelliah M, Ebisuzaki W, Kanamitsu M, Kousky V, van den
Dool H, Jenne R, Fiorino M. 2001. The NCEP/NCAR 50-year
reanalysis: monthly means CD-ROM and documentation. Bulletin
of the American Meteorological Society 82: 247–267.
Lunson E. 1950. Sand storms on the northern coasts of Libya and
Egypt. Great Britain Meteorological Office, Professional Notes, 102,
1–12.
Magaritz M. 1986. Environmental changes recorded in the upper Pleistocene along the desert boundary, southern Israel. Palaeogeography
Palaeoclimatology Palaeoecology 53: 213–229.
Mamane Y, Dayan U, Miller JM. 1987. Contribution of Alkaline
and acidic sources to precipitation in Israel. Science of the Total
Environment 61: 15–22.
Moulin C, Lambert CE, Dulac F, Dayan U. 1997. Control of
atmospheric export of dust from North Africa by the North Atlantic
oscillation. Nature 387: 691–694.
Moulin C, Lambert CE, Dayan U, Masson V, Ramonet M, Bousquet P, Legrand M, Balkanski YJ, Guelle W, Marticorena B, Bergametti G, Dulac F. 1998. Satellite climatology of African dust transport in the Mediterranean atmosphere. Journal of Geophysical
Research-Atmospheres 103(D11): 13137–13144.
Patterson EM, Gillette DA. 1977. Measurements of visibility vs. mass
concentration for airborne soil particles. Atmospheric Environment
11: 193–196.
Schutz L, Jaenicke R, Petrick H. 1981. Saharan dust transport over the
North Atlantic Ocean. In Desert Dust: Origin, Characteristics and
Effects on Man. Pewe TL. (ed.). Geological Society of America;
87–100.
Tews EK. 1996. Wind erosion rates from meteorological records in
eastern Australia 1960–92. Thesis, Griffith University, Australia.
Xoplaki E, Gonzalez–Rouco FJ, Luterbacher J, Wanner H. 2003.
Mediterranean summer air temperature variability and its connection
to the large-scale atmospheric circulation and SSTs. Climate
Dynamics 20: 723–839.
Yarnal B. 1993. Synoptic Climatology in Environmental Analysis: a
Primer. Bellhaven Press, Boca Raton, FL; 195.
Ziv B, Dayan U, Kushnir Y, Roth C, Enzel Y. 2006. Regional and
global atmospheric patterns governing rainfall in the southern
Levant. International journal of Climatology 26: 55–73.
Int. J. Climatol. (in press)
DOI: 10.1002/joc