;;7510A5-?5:9(/519/1>&=1>/:?? :77131:2A5-?5:9
)41C?=-?=:;5/-7)=-9>5?5:9:2)=:;5/-7D/7:91>:=1/->?)41C?=-?=:;5/-7)=-9>5?5:9:2)=:;5/-7D/7:91>:=1/->?
4-771931>@==19?*901=>?-90593-90@?@=15=1/?5:9>4-771931>@==19?*901=>?-90593-90@?@=15=1/?5:9>
(-=-4 :91>
*95A1=>5?D:2#@95/4
&-?=5/6-==
$-A-7&:>?3=-0@-?1(/4::7
58.=-4-8
#1?1:=:7:35/-7(1=A5/1:2-9-0-
"-9/1:>-=?
)41*95A1=>5?D:27.-9D(?-?1*95A1=>5?D:2$1B,:=6
&1?1= :BD1=
-9-05-9@==5/-9119?=1
(1191C?;-312:=-005?5:9-7-@?4:=>
:77:B?45>-90-005?5:9-7B:=6>-?4??;>/:88:9>1=-@10@;=81?1:=:7:3D
&-=?:2?41?8:>;41=5/(/519/1>:88:9>-90?41#1?1:=:7:3D:88:9>
(/4:7-=7D:88:9>5?-?5:9(/4:7-=7D:88:9>5?-?5:9
:91>(-==&.=-4-8 :>-=?":BD1=& A-9> "-971D-9>?=@8$
-=?'"-7-@=1??1(59/7-5=#'(85?4'!)4:=9/=:2?)41C?=-?=:;5/-7)=-9>5?5:9:2
)=:;5/-7D/7:91>:=1/->?4-771931>@==19?*901=>?-90593-90@?@=15=1/?5:9>
+1-?41=-90
:=1/->?593
4??;>0:5:=3))%)%
G:;D=534?81=5/-9#1?1:=:7:35/-7(:/51?D#(&1=85>>5:9?:@>1K3@=1>?-.71>-90.=5121C/1=;?>
2=:8?45>B:=659>/519?5K/-9010@/-?5:9-7B:=6>5>41=1.D3=-9?10;=:A5010?4-??41>:@=/15>-/69:B7103109D
@>1:28-?1=5-759?45>B:=6?4-?5>01?1=85910?:.1I2-5=@>1J@901=(1/?5:9:2?41*(:;D=534?/?
(1;?18.1=&-31:=?4-?>-?5>K1>?41/:905?5:9>>;1/5K1059(1/?5:9:2?41*(:;D=534?/?*(
F->=1A5>10.D&"0:1>9:?=1<@5=1?41#(H>;1=85>>5:9'1;@.75/-?5:9>D>?18-?5/=1;=:0@/?5:9
;:>?59359171/?=:95/2:=8>@/4->:9-B1.>5?1:=59->1-=/4-.710-?-.->1:=:?41=@>1>:2?45>8-?1=5-71C/1;?
->1C18;?10.D?41-.:A1>?-?1819?=1<@5=1>B=5??19;1=85>>5:9:=-75/19>12=:8?41#(005?5:9-701?-57>-=1
;=:A501059?41#(:;D=534?&:75/D-A-57-.71:9?41#(+1.>5?17:/-?10-?4??;BBB-81?>:/:=3:=2=:8
?41#(-?:=/:;D=534?>-81?>:/:=3
)45>=?5/715>.=:@34??:D:@2:=2=11-90:;19-//1>>.D?41:77131:2A5-?5:9-?(/4:7-=7D:88:9>?4->
.119-//1;?102:=59/7@>5:959;;7510A5-?5:9(/519/1>&=1>/:??.D-9-@?4:=5E10-08595>?=-?:=:2(/4:7-=7D
:88:9>:=8:=1592:=8-?5:9;71->1/:9?-/?/:88:9>1=-@10@
@?4:=>@?4:=>
(-=-4 :91>&-?=5/6-== 58.=-4-8"-9/1:>-=?&1?1= :BD1= 1995"A-9>1.:=-4
-971D-==D$-9>?=@8':.1=?-=?=-9/:5>"-7-@=1??1#-=6'(59/7-5=':31=!(85?4-904=5>
)4:=9/=:2?
)45>-=?5/715>-A-57-.71-?(/4:7-=7D:88:9>4??;>/:88:9>1=-@10@;=81?1:=:7:3D
1052 V
OLUME
18WEATHER AND FORECASTING
q 2003 American Meteorological Society
The Extratropical Transition of Tropical Cyclones: Forecast Challenges, Current
Understanding, and Future Directions
S
ARAH
C. J
ONES
,
a
P
ATRICK
A. H
ARR
,
b
J
IM
A
BRAHAM
,
c
L
ANCE
F. B
OSART
,
d
P
ETER
J. B
OWYER
,
e
J
ENNI
L. E
VANS
,
f
D
EBORAH
E. H
ANLEY
,
g
B
ARRY
N. H
ANSTRUM
,
h
R
OBERT
E. H
ART
,
f
F
RANC¸OIS
L
ALAURETTE
,
i
M
ARK
R. S
INCLAIR
,
j
R
OGER
K. S
MITH
,
a
AND
C
HRIS
T
HORNCROFT
d
a
Meteorological Institute, University of Munich, Munich, Germany
b
Department of Meteorology, Naval Postgraduate School, Monterey, California
c
Meteorological Service of Canada, Dorval, Quebec, Canada
d
Department of Earth and Atmospheric Sciences, The University of Albany, State University of New York, Albany, New York
e
Canadian Hurricane Centre, Dartmouth, Nova Scotia, Canada
f
Department of Meteorology, The Pennsylvania State University, University Park, Pennsylvania
g
Center for Ocean–Atmosphere Prediction Studies, The Florida State University, Tallahassee, Florida
h
Bureau of Meteorology, Perth, Western Australia, Australia
i
European Centre for Medium-Range Weather Forecasts, Reading, United Kingdom
j
Embry-Riddle Aeronautical University, Prescott, Arizona
(Manuscript received 17 September 2002; in final form 25 March 2003)
ABSTRACT
A significant number of tropical cyclones move into the midlatitudes and transform into extratropical cyclones.
This process is generally referred to as extratropical transition (ET). During ET a cyclone frequently produces
intense rainfall and strong winds and has increased forward motion, so that such systems pose a serious threat
to land and maritime activities. Changes in the structure of a system as it evolves from a tropical to an extratropical
cyclone during ET necessitate changes in forecast strategies. In this paper a brief climatology of ET is given
and the challenges associated with forecasting extratropical transition are described in terms of the forecast
variables (track, intensity, surface winds, precipitation) and their impacts (flooding, bush fires, ocean response).
The problems associated with the numerical prediction of ET are discussed. A comprehensive review of the
current understanding of the processes involved in ET is presented. Classifications of extratropical transition
are described and potential vorticity thinking is presented as an aid to understanding ET. Further sections discuss
the interaction between a tropical cyclone and the midlatitude environment, the role of latent heat release,
convection and the underlying surface in ET, the structural changes due to frontogenesis, the mechanisms
responsible for precipitation, and the energy budget during ET. Finally, a summary of the future directions for
research into ET is given.
1. Introduction
The threat to life and property from tropical cyclones
is well recognized. Even in countries that are not directly
affected by these storms, considerable media coverage
is given to significant tropical cyclone events. Often,
the level of interest diminishes when the tropical cy-
clone moves to higher latitudes and begins to lose its
tropical cyclone characteristics such that official tropical
cyclone warnings are no longer posted or the storm is
thought to be no longer a threat to life or property.
Corresponding author address: Dr. Sarah C. Jones, Meteorolo-
gisches Institut, Theresienstr. 37, 80333 Mu¨nchen, Germany.
E-mail: sarah@meteo.physik.uni-muenchen.de
However, a decaying tropical cyclone often evolves into
a fast-moving and occasionally rapidly developing ex-
tratropical cyclone that produces intense rainfall, very
large waves, and even hurricane-force winds. This ex-
tratropical transition (ET) of a decaying tropical cyclone
may result in a storm that continues to pose a serious
threat to land and maritime activities by extending trop-
ical cyclone–like conditions over a larger area and to
latitudes that do not typically experience such events—
essentially, bringing the strong winds typical of major
winter storms to midlatitude locations during a summer
or autumn season. Extratropical transition poses a sig-
nificant challenge to the forecaster. There is a high de-
gree of uncertainty associated with predicting the timing
D
ECEMBER
2003 1053JONES ET AL.
F
IG
. 1. Tracks of all tropical cyclones that underwent extratropical
transition during 1970–99. (a) Western North Pacific. Tracks of trop-
ical cyclones defined to be extratropical in JTWC best-track data. (b)
Southwest Pacific [data as in (a)] and southeast Indian Ocean [tracks
of tropical cyclones that accelerated toward the southeast under the
influence of a midlatitude frontal system and maintained gales into
midlatitudes, the so-called captured cyclones in Foley and Hanstrum
(1994); best-track data taken from http://www.australiasevereweather.
com/cyclones/history.htm]. (c) North Atlantic. Tracks of tropical cy-
clones defined to be extratropical in National Hurricane Center best-
track data.
of the increased translation speed, the potential for rapid
intensification, and the severity of the weather associ-
ated with ET, especially given the fact that the quality
of numerical forecasts is not yet reliable for ET.
a. A brief climatology
Extratropical transition of a tropical cyclone occurs
in nearly every ocean basin that experiences tropical
cyclones (Fig. 1), with the number of ET events fol-
lowing a distribution in time similar to that of the total
number of tropical cyclone occurrences (Fig. 2). The
largest number of ET events occur in the western North
Pacific (Fig. 2b) while the North Atlantic basin contains
the largest percentage of tropical cyclones that undergo
ET (Fig. 2a), with 45% of all tropical cyclones under-
going ET in the 30-yr period shown. In the eastern North
Pacific the synoptic conditions associated with the pres-
ence of a strong subtropical ridge are not conducive to
ET.
In the southwest Pacific, ET has a significant impact
on Australia and New Zealand, being triggered by the
approach of a midlatitude trough from the west (Sinclair
2002). However, because of the time frame shown, and
due to the lack of a comprehensive definition of ET (as
discussed later in this section), Fig. 1b underestimates
the threat posed by ET to eastern Australia (J. Callaghan
2002, personal communication). Over the southeast In-
dian Ocean, relatively few tropical cyclones undergo
ET (Figs. 1b and 2d). There, ET occurs most frequently
when a large-amplitude cold front approaches within a
distance of around 1700 km or less of a tropical cyclone.
This is most likely to occur in the late summer or au-
tumn. Such events result in significant impacts occurring
every 20 yr or so (Foley and Hanstrum 1994).
Hart and Evans (2001) show that ET in the Atlantic
occurs at lower latitudes in the early and late hurricane
season and at higher latitudes during the peak of the
season. The highest percentage of ET events occurs in
September and October. The increased probability that
Atlantic ET will occur during these months can be ex-
plained by comparing the geographical location of the
areas that support tropical and extratropical develop-
ment (see Hart and Evans 2001 for details of the cal-
culation). In September and October the distance that a
tropical cyclone must travel from the region that sup-
ports tropical development to the region that supports
extratropical development is shorter than in other
months, implying that a tropical cyclone is more likely
to reach the region of extratropical development and
thus more likely to undergo ET. Hart and Evans (2001)
showed also that most of the storms that intensify after
ET form in the deep Tropics and that a large number
of these are Cape Verde cyclones.
Tropical cyclones that have undergone ET have been
tracked across the Atlantic or the Pacific (e.g., Thorn-
croft and Jones 2000). Such systems may reintensify
many days after ET and bring strong winds and heavy
rain to the eastern side of the ocean basin [e.g., Hur-
ricane Floyd in 1993 (Rabier et al. 1996); Hurricane
Lili in 1996 (Browning et al. 1998)].
More detailed climatologies of ET are available for
the west coast of Australia (Foley and Hanstrum 1994),
the western North Pacific (Klein et al. 2000), the North
Atlantic (Hart and Evans 2001), and the southwest Pa-
cific (Sinclair 2002).
1054 V
OLUME
18WEATHER AND FORECASTING
F
IG
. 2. Monthly total number of tropical cyclones (open bars) in each basin during 1970–99 and the number of tropical cyclones that
underwent extratropical transition (shaded bars). Data as in Fig. 1.
b. Characteristics of extratropical transition
Extratropical transition is, as its name suggests, a
gradual process in which a tropical cyclone loses trop-
ical characteristics and becomes more extratropical in
nature. As a tropical cyclone moves poleward it expe-
riences changes in its environment (Schnadt et al. 1998).
These changes may include increased baroclinity and
vertical shear, meridional humidity gradients, decreased
sea surface temperature (SST) or strong SST gradients
(e.g., those associated with the Gulf Stream), and an
increased Coriolis parameter. The tropical cyclone may
come into proximity with an upper-level trough or a
mature extratropical system. If the tropical cyclone
makes landfall, it will experience increased surface drag,
a reduction of surface fluxes of latent and sensible heat,
and it may encounter orography.
When a tropical cyclone begins to interact with the
midlatitude baroclinic environment, the characteristics
of the cyclone change dramatically (Palme´n 1958; Mur-
amatsu 1985; Foley and Hanstrum 1994; Klein et al.
2000). In satellite imagery the inner core of the tropical
cyclone loses its symmetric appearance and gradually
takes on the appearance of an extratropical cyclone. The
nearly axisymmetric wind and precipitation distribu-
tions that are concentrated about the circulation center
of the tropical cyclone evolve to broad asymmetric dis-
tributions and expand greatly in area. Although the ex-
panding cloud field associated with a poleward-moving
tropical cyclone includes large amounts of high clouds
due to the tropical cyclone outflow into the midlatitude
westerlies, regions of significant precipitation are typ-
ically embedded in the large cloud shield.
Movement of a decaying tropical cyclone into the
midlatitude westerlies results in an increased translation
speed (Fig. 3), which contributes to the asymmetric dis-
tributions of severe weather elements. Over the ocean,
high wind speeds and large translation speeds contribute
to the generation of large ocean surface waves and swell
(Bigio 1996; Bowyer 2000).
The southwest Pacific basin is unique in that tropical
cyclones have an average eastward component of mo-
tion throughout most of their lives (Fig. 3). This is be-
cause they interact with the midlatitude westerlies early
in their life cycle: these may extend to 158S during the
Southern Hemisphere tropical cyclone season. As a con-
sequence, tropical cyclones start acquiring the asym-
metries characteristic of the onset of ET (section 3a)
around 208S, closer to the equator than in any other
ocean basin. On average, ET is complete by 308S (Sin-
clair 2002). In contrast, Northern Hemisphere storms
may preserve tropical characteristics as far north as
508N.
![FIG. 16. Horizontal plot on the 200-hPa surface for the ET composite containing cases with a total decrease in central pressure of more than 10 hPa. (a), (c), (e) Vectors of the total wind [reference arrow indicated at the bottom left of (e)], and Ertel potential vorticity (increment of 0.5 3 1026 K m2 s21 kg21; values greater than 1.5 3 1026 K m2 s21 kg21 shaded as indicated) at times t0 2 12 h, t0, and t0 1 12 h, respectively. (b), (d), (f ) Total wind speed (values greater than 15 m s21 shaded as indicated), velocity potential (solid lines, contour interval 6 3 105 m2 s21), and divergent wind [reference arrow indicated at the bottom left of (f )] at times t0 2 12 h, t0, and t0 1 12 h, respectively. Here, t0 is the 12-hourly observation time immediately before the central sea level pressure begins to fall. Asterisk denotes the location of the composite tropical cyclone center, and the increment in latitude and longitude is 108. [Taken from Hanley (1999).]](/figures/fig-16-horizontal-plot-on-the-200-hpa-surface-for-the-et-30q6ew3d.webp)
![FIG. 18. Forecast surface latent heat fluxes in W m22 (shaded) and analyzed mean sea level pressure in hPa (contours) at 0000 UTC on the dates shown for the extratropical transitions of (a)–(c) Hurricane Iris in 1995, where dark shading indicates strong upward fluxes, and (d) Hurricane Felix in 1995, where dark shading indicates strong downward fluxes. Data from ECMWF ERA-40 reanalysis (available online at http://www.ecmwf.int/research/era). Surface fluxes are a 6-h average centered on 0000 UTC from a forecast initialized at 1200 UTC on the previous day. [Adapted from Thorncroft and Jones (2000).]](/figures/fig-18-forecast-surface-latent-heat-fluxes-in-w-m22-shaded-2sx6r5xh.webp)
![FIG. 21. (a) Track of Tropical Cyclone Bola (1988) based on 6-h positions with day numbers listed at the 0000 UTC position. (b) Rainfall (mm) over the North Island of New Zealand associated with the passage of Bola. (c) Isotachs (shaded with a contour interval of 20 kt) and heights (dynamical meters) at 200 h Pa at 0000 UTC 7 Mar 1988. (d) Streamlines at 850 hPa, moisture convergence (shaded in units of 1028 s21), and region of 850-hPa total frontogenesis defined by a value of 3 3 10210 K m21 s21. [Adapted from Sinclair (1993).]](/figures/fig-21-a-track-of-tropical-cyclone-bola-1988-based-on-6-h-12ua7fhz.webp)