Analysis of conveyor belts in winter Mediterranean cyclones

Abstract

The relevance of the midlatitude conveyor belt model to Mediterranean cyclones (MCs) is examined using data from two winters.
Eight MCs, which exhibit typical midlatitude cyclone structure, were scrutinized and their conveyor belts were examined. The
analysis was based on satellite imagery, isentropic wind maps, vertical cross-sections of potential and equivalent potential
temperatures, and air back-trajectories. The conveyor belts found in the studied MCs were similar to the common features of
midlatitude cyclones, except for three aspects. First, the warm conveyor belt was not associated with massive organized cloudiness
in five of the eight cyclones since it consisted of dry air originated from the Saharan desert. Second, the anticyclonic branch
of the cold conveyor belt was not found in half of the MCs. Third, the dry air intrusion originated north of the cyclone and
extended southward around it, unlike its common midlatitudinal northwest–southeast orientation. This is consistent with the
relatively small baroclinic vertical-westward tilt of the cyclones analyzed.

Full text

Page 1

ORIGINAL PAPER
Analysis of conveyor belts in winter Mediterranean cyclones
B. Ziv & H. Saaroni & M. Romem & E. Heifetz &
N. Harnik & A. Baharad
Received: 14 January 2009 /Accepted: 3 May 2009 /Published online: 2 June 2009
# Springer-Verlag 2009
Abstract The relevance of the midlatitude conveyor belt
model to Mediterranean cyclones (MCs) is examined using
data from two winters. Eight MCs, which exhibit typical
midlatitude cyclone structure, were scrutinized and their
conveyor belts were examined. The analysis was based on
satellite imagery, isentropic wind maps, vertical cross-
sections of potential and equivalent potential temperatures,
and air back-trajectories. The conveyor belts found in the
studied MCs were similar to the common features of
midlatitude cyclones, except for three aspects. First, the
warm conveyor belt was not associated with massive
organized cloudiness in five of the eight cyclones since it
consisted of dry air originated from the Saharan desert.
Second, the anticyclonic branch of the cold conveyor belt
was not found in half of the MCs. Third, the dry air
intrusion originated north of the cyclone and extended
southward around it, unlike its common midlatitudinal
northwest–southeast orientation. This is consistent with the
relatively small baroclinic vertical-westward tilt of the
cyclones analyzed.
1 Introduction
The conveyor belt model (CBM) describes air motion in the
frame of reference of the cyclone track and provides a good
framework for understanding the precipitation and cloud
fields accompanying midlatitude cyclones during their
development (see review articles by Browning 1990 and
Carlson 1991). The CBM was developed for and applied to
midlatitude cyclones over the US and the main global storm
tracks. A somewhat weaker secondary storm track runs
along the Mediterranean Basin (MB) and is especially
active in the winter season (Hoskins and Hodges 2002).
Studies have shown that Mediterranean storms exhibit
some characteristics typical of midlatitude storms (e.g.,
Alpert and Ziv 1989; Alpert et al. 1990, Tsidulko and
Alpert 2001) with some differences stemming from the
proximity of the Mediterranean to the arid subtropics and
an influence of stronger storms over Europe which typically
exist alongside the Mediterranean cyclones (MCs).
1.1 The conveyor belt model
The CBM has evolved over many decades of study (starting
from Shaw 1903; Kutzbach 1979; Cohen 1993; Wernli
1995; Gall and Shapiro 2000; Schultz 2001 for reviews of
the history of the CBM development) before being
integrated into the familiar conceptual CBM by Carlson
(1980) in an investigation of a winter storm in Central
United States in 1977. Carlson’s CBM (which was
extended further by Browning 1990, 1997, 1999) describes
three airstreams affecting the structure of clouds and
precipitation—the warm conveyor belt (WCB), the cold
conveyor belt (CCB), and the dry air intrusion (DAI). They
are identified on isentropic streamline maps of the wind
field relative to the cyclone velocity. The horizontal
Theor Appl Climatol (2010) 99:441–455
DOI 10.1007/s00704-009-0150-9
B. Ziv (*)
The Open University of Israel,
P.O. Box 808, Ra’anana 43107, Israel
e-mail: baruchz@openu.ac.il
H. Saaroni:M. Romem
Department of Geography and the Human Environment,
Tel Aviv University,
Tel Aviv 69978, Israel
E. Heifetz:N. Harnik:A. Baharad
Department of Geophysics and Planetary Sciences,
Tel Aviv University,
Tel Aviv 69978, Israel

Page 2

projection of the conveyor belts is shown schematically in
Fig. 1.
The WCB is composed of warm air with high moisture
content. This belt originates in the subtropics at the lower
levels and ascends moving northward, in an anticyclonic
curvature, up to ~6,000 m (the midtroposphere; Eckhardt et
al. 2004) to the right of the cyclone center. The WCB is
about 2,000 km long and 3 km deep. It is represented in the
satellite imagery by a thick cloud band curved cyclonically
in its southernmost part and anticyclonically at the jet level
in its northern part, yielding an “S” shape. Since its
associated cloudiness is normally precipitating, the WCB
is considered as the main rainfall contributor of the cyclone
(Eckhardt et al. 2004). Kurz (1988), Wernli (1995, 1997)
Bader et al. (1995), and Browning and Roberts (1996)
found that, during the early stages of cyclogenesis, if the
upper-level flow is open, i.e., no closed systems exist, the
WCB rises to the upper troposphere and turns anticycloni-
cally downstream of the cyclone. Later, during the
occlusion stage, if the upper-level flow is characterized by
a closed cyclone, some of the WCB may turn cyclonically
around the cyclone center (known as the trowal airstream
of Martin 1999), and in some cases, lead to the formation of
a comma-shaped cloud pattern.
The CCB transports cold air that originates from the
lower levels, ~1,500 m, northeast of the cyclone center.
The air flows westward relative to the cyclone, beneath
the WCB, and while passing north of the cyclone center,
it splits into two branches, cyclonic and anticyclonic
(Carlson 1987; Browning 1990; Carlson 1991). The
cyclonic branch encircles the cyclone counterclockwise
while descending slightly, whereas the anticyclonic branch
turns sharply clockwise, northeastward, and ascends until it
merges with the northeastern part of the WCB. Schultz
(2001) argued that the anticyclonic branch is very thin.
Though the cold air mass in this belt is originally dry, it
may become wet due to evaporation from the rain
associated with the WCB while it flows below it (Fig. 1).
The latent heat release within the upper layer of the CCB
tends to increase the potential vorticity (PV) that it advects
westward, toward the lower-tropospheric cyclone center.
The clouds associated with the cyclonic branch of the CCB
are mostly medium and low and produce smaller amounts
of rain than in the WCB. Since the precipitation generated
within the WCB falls through the CCB, the temperature and
humidity of the CCB determines the type and amount of
precipitation reaching the surface (ranging from virga, via
sleet and freezing rain, and up to an abrupt change from
rain to heavy snow; Kain et al. 2000).
The DAI originates from the northwest sector of the
cyclone, at the upper troposphere and lower polar strato-
sphere, and descends on the west side of the upper-level
trough, west of the cyclone center (Fig. 1). This air mass is
dry and stable with high PV, of ~5 potential vorticity units
(PVU) typically. The DAI encircles the cyclone cyclonical-
ly through its southwest part, behind the cold front. While
descending into the subtropical troposphere, it attains a
“hammerhead” shape. The DAI plays a significant role in
baroclinic cyclogenesis and in the mesoscale convective
destabilization that affects clouds development (Carr and
Millard 1985; Young et al. 1987; Durran and Weber 1988;
Gurka et al. 1995; Browning and Golding 1995; Browning
1997).
Most of the literature on CBM explored extratropical
cyclones in Europe, North America and the Atlantic, or the
Pacific Oceans. In this study, we focus on MCs and
examine the relevancy of the CBM to their structure and
resulting weather conditions.
1.2 Mediterranean cyclones
The MB is highly populated by cyclones, especially in the
winter season (HMSO 1962; Reiter 1975; Radinovic 1987;
Campins et al. 2000; Romem et al. 2007). The MCs are
considered to be of midlatitude frontal type (in spite of
being located south of the global midlatitude climatic belt)
with a cyclogenetic contribution of orography (Tafferner
and Egger 1990; Stein and Alpert 1993). The annual
number of MCs varies along the MB, reaching its
maximum near the Gulf of Genoa, ~80 (Campins et al.
2000). Their horizontal size differs between mesoscale
(Alpert et al. 1999) and synoptic-scale (Campins et al.
2000; Trigo et al. 2002). The MCs tend to propagate
eastward along the northern coast of the Mediterranean
until reaching the Levant region where they weaken or even
dissipate (Kahana et al. 2002). The main MCs’ tracks for
January are shown in Fig. 2.
The CBM has been examined for the MCs only partly.
Eckhardt et al. (2004), in their climatological study of
Fig. 1 Schematic structure of the conveyor belts
442B. Ziv et al.

Page 3

WCBs based on a Lagrangian approach, found a weak
signal in the Mediterranean, implying that strong ascent of
air masses also occurs in this area of the globe. Other two
studies point at the relevance of conveyor belts for this
region. One describes a dust plume, “which was transported
northward more than 2400 km” and its western boundary
coincided with a cold front (Alpert and Ganor 1994). This
dust plume reflects, presumably, a WCB which extended
from the Sahara northward over the Mediterranean Sea. A
second study dealt with the role of high-PVadvection in the
genesis of a Genoa Low (Tsidulko and Alpert 2001), which
resembles the existence of a DAI.
This paper studies the analogy of the MCs to the typical
midlatitude cyclones by examining to what extent the CBM
holds for the Mediterranean and, correspondingly, whether
the main precipitation and cloud features arise due to
similar processes. It analyses the 3-D flow structure of
MCs, identifies the conveyor belts, and compares their
structure with the classical CBM. Section 2 specifies the
data and methodology used. Section 3 presents the results
for the analyzed MCs and Section 4 summarizes the results
and discusses their dynamic and some of their climatolog-
ical implications.
2 Data and methodology
We examined the MCs during two winters (December–
March of 2001/2002 and 2002/2003). The data base is the
six hourly National Centers for Environmental Prediction/
National Center for Atmospheric Research (NCEP/NCAR)
reanalysis (Kalnay et al. 1996; Kistler et al. 2001) with a
2.5×2.5° spatial resolution. The eight cyclones selected
were well-developed, with a central vorticity of >2×
10−5s−1, with a diameter of >1,500 km, and with no
additional cyclone in the range of 2,000 km. Each of them
existed at least 36 h. The characteristics of the selected
cyclones are listed in Table 1. Their typical diameter and
speed generally agree with that evaluated by Alpert and Ziv
(1989), being >1,500 km and 5 m s−1, respectively, and the
average value of their central vorticity (3.5×10−5s−1) is
close to those found by Alpert and Neeman (1992) and
Trigo et al. (2002).
The identification of the conveyor belts is based on a
“semi-Eulerian” approach, i.e., along streamlines of the
relative wind field (the wind velocity with respect to the
cyclone velocity) on isentropic surfaces, following Browning
(1990). The cyclone velocity was determined by tracking its
center along 12-h intervals around the pertinent time. Since
the conveyor belts were found in various heights throughout
the troposphere, the use of the surface cyclone alone did not
seem to be proper for tracking it. We, therefore, averaged the
flow over the 1,000- to 300-hPa range. In this averaging,
closed cyclones were not always found (but open troughs),
so we used the relative vorticity maximum to represent the
cyclone center.
The isentropic surfaces better approximate the actual air
movement than the pressure surfaces do. But, since
condensation exists along conveyor belts, with its implied
latent heat release, the isentropic surfaces may be poor
approximation where condensation takes place, so that air
changes its potential temperature (θ). In that case, the air
conserves only its equivalent potential temperature (θe),
implying that it moves along θesurfaces. However, the use
of θe surfaces for analyzing conveyor belts was found
impractical because, in several cases, θedid not increase
Fig. 2 Winter MCs tracks, rep-
resented by the long-term mean
SLP for January, based on
NCEP/NCAR reanalysis data
(Kalnay et al. 1996; Kistler et al.
2001). The thick arrows repre-
sent the main tracks, as inferred
from the individual tracks for
January 1983–1988 (after Alpert
et al. 1990). Their thickness is
proportional to the track
population
Analysis of conveyor belts in winter Mediterranean cyclones 443

Page 4

with height but remained constant. In order to assess the
validity of using isentropic sections, we examined the
Lagrangian θ change of individual air parcels traveling
along the conveyor belts. The largest changes were found in
the WCBs, being <3°K within 12 h. This suggests that, for
our practical needs, isentropic sections are sufficient.
The locations of the WCB and the CCB were determined
subjectively along bands of streamlines which coincided
with a tongue of warm or cold air, searching for features
that meet the “classical” description (e.g., Browning 1990).
A “tongue” is represented by a trough or a ridge in the
geometry of the isentropic surface along the WCB and the
CCB, respectively. The WCBs were searched along the
southerly flow separating between the cyclone to the west
and the anticyclonic circulation to the east. The CCBs were
searched, starting from the cold reservoir north of the
cyclone and encircling the cyclone from the west. The
choice of the representative isentropic level for each
conveyor belt was done by inspecting the isentropes (in
2°K increments) ranging from 298 to 316°K for the WCB
and from 284 to 302°K for the CCB. The actual level for
each type of conveyor belt differed from one case to
another. The conveyor belts found in the isentropic maps
were also searched for in the IR METEOSAT satellite
imagery, attempting to identify cloud bands that coincide
with them. The images used were of 6- to 8-km resolution
and could not enable us to differentiate between convective
and stratified clouds confidently. For some of the cases,
MODIS images of 2-km resolution (http://www.nasa.gov/
about/highlights/HP_Privacy.html#links) were available
and were used to augment the analysis.
The DAI was identified in the isentropic level that best
represents the tropopause, i.e., in which positive PV
anomalies exceed 2 PVU. This threshold follows Browning
(1990), Wernli (1997), and Wernli and Davies (1997). It
was depicted along anomalously high-PV bands connected
to the high-PV reservoir to the north, parallel to the
streamlines. In order to further assess the stratospheric
source of the DAI, a relative humidity (RH) map was
derived. Pattern of low values that coexist with a high-PV
anomaly is very suggestive of a stratospheric source.
To better elucidate the thermodynamic characteristics of
the air masses within the conveyor belts, we analyzed them
also by vertical cross-sections of θe, calculated following
Bolton (1980). Since θeis conserved during condensation,
layers that are subjected to moist convection would have
zero vertical θe gradient, combined with high RH. The
difference, θe−θ, is a measure for specific humidity, as it is
higher in moist regions. Following the above, θecan be
considered as an efficient tracer and conveyor belts are
expected to appear as slanted layers of relatively constant
high or low θe(when the air mass transported is moist or
dry relative to its surrounding, respectively).
In order to identify thesources of the air masses within the
conveyor belts, we derived back-trajectories of air parcels
located at significant reference points on the conveyor belts,
using the National Oceanic and Atmospheric Administration
HYSPLIT4 (Hybrid Single-Particle Lagrangian Integrated
Trajectory) Model (1997).
3 Results
The main features of the studied cyclones and the
methodology applied are exemplified by a case study
presented in the following subsection. The CBM structure
in this case is characteristic of the other MC cases
examined.
3.1 A case study: 10–12 January 2003
This cyclone was formed on 9 January 2003 18 UTC in
Alger, at 30° N, 2° W, and existed for 72 h. The cyclone
moved east–northeastward at an average speed of 13 m s−1,
Table 1 General features of the eight studied MCs
Dates of existence 500-hPa systemDiameter (km)Min SLP (hPa)Average speed (m s−1)Max vorticity at 295°K (10−5s−1)
16–18 Jan 2002
10–13 Mar 2002
22–23 Mar 2003
6–8 Jan 2003
10–12 Jan 2003
17–20 Jan 2003
26–28 Jan 2003
31 Jan–2 Feb 2003
Average
Cutoff low
Cutoff low
Open trough
Open trough
Cutoff low
Open trough
Cutoff low
Open trough
1,800
1,900
2,000
1,700
1,900
2,150
2,200
1,600
1,900
1,017
1,005
997
995
994
1,013
1,004
1,000
1,003
5.8
7.5
9.6
9.6
12.6
6.6
6.5
8.6
8.4
4
5
4
5
5
4
4.5
6
4.5
The cyclone diameter was defined with respect to the region of positive relative vorticity at the 295°K isentropic level
444B. Ziv et al.

Page 5

until it reached the eastern part of the Black Sea (44° N, 40°
E). Its minimum sea-level pressure (SLP) was 994 hPa and
the maximum relative vorticity at the 295°K isentropic
level was 5×10−5s−1. The SLP map and the cyclone
trajectory are shown in Fig. 3. The detailed analysis was
done for 11 January 2003 06 UTC, when the cyclone
attained its maximum intensity. The 850-hPa geopotential
height and temperature fields (Fig. 4) demonstrate the
frontal structure of the cyclone. The 850-hPa thermal field
indicates that the fronts (denoted in Fig. 4) are located
slightly east of the cyclone center, implying that the cyclone
has just reached its occlusion phase. The cloud structure,
seen in the high-resolution satellite image (which was
available only for 11:55 UTC; Fig. 5), reflects the warm
and the cold conveyor belts (to be described below) and the
northern part of the cold front, along ~17°E.
The WCB (Fig. 6, thick arrow) was best identified on the
303°K isentropic surface. It was ~2,000 km long, extending
from southern Libya (23° N, 19° E), southeast of the
cyclone center, toward the Aegean Sea (40° N, 24° E) and
climbing from 1.5 to 4.5 km. Similar ascent was noted by
air-trajectory analysis. A band of cirrus clouds, of ~500 km
width, coinciding with the WCB is clearly seen in the
satellite image (Fig. 5). The clouds are thin in their southern
Fig. 3 SLP for January 11,
2003, 06 UTC and the trajecto-
ry of the cyclone
Fig. 4 The 850-hPa geopoten-
tial height and the temperature
fields for January 11, 2003, 06
UTC together with the notation
of the cold and warm fronts
Analysis of conveyor belts in winter Mediterranean cyclones445