Content uploaded by Lei-Ming Ma
Author content
All content in this area was uploaded by Lei-Ming Ma on Feb 26, 2015
Content may be subject to copyright.
Vol.20 No.4 JOURNAL OF TROPICAL METEOROLOGY December 2014
Article ID: 1006-8775(2014) 04-0297-08
NUMERICAL ANALYSIS ON THE ROLE OF TROPICAL STORM
NAMTHEUN IN THE UNUSUAL TRACKS OF THREE TROPICAL
CYCLONES IN 2010
BAI Li-na (白莉娜), MA Lei-ming (马雷鸣), ZENG Zhi-hua (曾智华), HUANG Wei (黄 伟), WANG
Dong-liang (王栋梁)
(Shanghai Typhoon Institute, Laboratory of Typhoon Forecast Technique, CMA, Shanghai, 200030 China)
Abstract: Tropical cyclones (TCs) Lionrock, Kompasu, and Namtheun were formed successively within 40
hours in 2010. Over the next several days afterwards, these TCs exhibited unusual movements which made
operational prediction difficult. Verifications are performed on the forecasts of the tracks of these TCs with
six operational models, including three global and three regional models. Results showed that the trends of
TC tracks could be reproduced by these models, whereas trajectory turning points and landfall locations
were not simulated effectively. The special track of Lionrock should be associated with its direct interaction
with Namtheun, according to a conceptual model of binary TC interaction. By contrast, the relation between
Kompasu and Namtheun satisfied the criteria for a semi-direct interaction. Numerical experiments based on
the Global and Regional Assimilation and Prediction System-Tropical Cyclone forecast Model
(GRAPES-TCM) further confirmed the effect of Namtheun on the unusual tracks of Lionrock and Kompasu.
Finally, the physical mechanism of binary TC interaction was preliminarily proposed.
Key words: binary tropical cyclone interaction; typhoon operational model; numerical experiment;
GRAPES-TCM
CLC number: P426.616 Document code: A
Received 2013-08-20; Revised 2014-07-31; Accepted 2014-10-15
Foundation item: “973” Project (2009CB421500, 2013CB430305); National Natural Science Foundation of
China (40921160381, 40705024, 40875039); Special Scientific Research Fund of Meteorological Public Welfare
of China (GYHY201006007, GYHY201006008, GYHY201006016); Shanghai Typhoon Research Foundation
(2010ST09)
Biography: BAI Li-na, Ph. D., primarily undertaking research on tropical cyclone intensity.
Corresponding author: Ma Lei-ming, e-mail: malm@mail.typhoon.gov.cn
1 INTRODUCTION
Unusual tropical cyclone (TC) tracks are often
observed when two or more TCs interact, particularly
in the western North Pacific Ocean (Neumann et al.[1];
Qin and Cao[2]). During the last week of August 2010,
three TCs were formed successively within 40 hours.
These TCs later became Severe Tropical Storm
Lionrock, Severe Typhoon Kompasu, and Tropical
Storm Namtheun. Over the next several days
afterwards, Lionrock and Namtheun made landfall in
Fujian province, whereas Kompasu turned toward the
northeast near the east coast of China, thus bringing
strong wind and rain to a wide range of coastal areas.
Forecasting the tracks of these TCs has been difficult
because of uncertainties associated with the
interactions among binary TCs.
Research on the interaction of binary vortices
dates back to Fujiwhara[3, 4] who demonstrated that
relative motion exhibits counterclockwise revolution
of one vortex around another and a tendency to attract
circulations. This phenomenon is commonly referred
to as the Fujiwhara effect in meteorological
applications. The counterclockwise rotation rate and
the tendency to attract circulations are determined by
the separation distance among TCs and the outer wind
structure of each TC (Elsberry[5]). A direct interaction
cannot occur when two TCs are separated by more
than 10 degrees of latitude (Brand[6]; Dong and
Neumann[7]; Luo and Ma[8]); however, they can
interact with each other by altering environmental
circulations (Chen and Ding[9]). In addition to these
observations, numerical studies show that binary
interaction is sensitive to initial separation distance,
intensity, and locations among TCs (DeMaria and
Chan[10]; Wang and Zhu[11]; Zhang et al.[12]).
Lander and Holland[13] performed a detailed
analysis based on data compiled for 44 years (1945 to
1988), and found that the interaction among TCs
rarely follows the classic Fujiwhara model. Instead,
they proposed a conceptual model of direct interaction,
298 Journal of Tropical Meteorology Vol.20
298
which included capturing, cyclonic orbiting, merging,
and escaping. Carr et al.[14, 15] modified the conceptual
model of Lander and Holland, and proposed a new
conceptual model with three modes based on a
seven-year compilation of western North Pacific TCs.
They further classified direct TC interaction (DTI)
into three modes: (1) one-way influence, in which the
track of the smaller TC is primarily controlled by the
circulation of the larger TC; (2) mutual interaction, in
which two TCs exhibit cyclonic rotation; and (3)
merger, in which two TCs with almost equal size get
close enough to each other such that their cyclonic
circulations overlap. Meanwhile, the semi-direct TC
interaction model has two requirements: (1) the two
TCs must be sufficiently separated such that DTI does
not occur, and (2) the two TCs must be oriented
approximately east-west and sufficiently close to the
subtropical ridge axis. Lastly, in indirect TC
interaction, the presence of an anticyclone between
two TCs is considered to be a critical factor.
The data used in the current study are described in
section 2. The evaluations performed on the track
forecasts of the three aforementioned TCs made by six
operational models are presented in section 3. In
section 4, the numerical simulation on the effect of
Namtheun on the abnormal tracks of Lionrock and
Kompasu based on a GRAPES tropical cyclone model
(TCM) is discussed. The summary and discussion are
provided in section 5.
2 DATA AND CASE
TC data are obtained from the operational reports
issued by the China Meteorological Administration
(CMA). These data include TC position and intensity
estimated every three hours. The environmental data
used in this study are based on the analysis fields of
the Global Forecast System (GFS) from the National
Centers for Environmental Prediction (NCEP), which
has a horizontal resolution of 1°×1° and 16 vertical
layers on mandatory pressure surfaces.
The tracks and intensities of Lionrock, Kompasu,
and Namtheun are shown in Figs. 1 and 2. Lionrock
strengthened into a tropical storm in the South China
Sea at 02:00 Beijing standard time (BJT) on 29
August and steadily moved northward, guided by a
cross-equatorial flow at 115°E. Then, the tropical
storm turned eastward in the morning of 30 August
and intensified into a severe tropical storm in the
afternoon of 31 August. A few hours later, Lionrock
turned to the north-northeast and degraded into a
tropical storm. In the morning of 1 September,
Lionrock shifted northwestward and intensified once
more into a severe tropical storm. Then, it made
landfall in Zhangpu, Fujian province at approximately
06:50 BJT on 2 September as a tropical storm.
Afterward, Lionrock kept moving westward and
eventually dissipated. Kompasu was formed in the
southwest of the subtropical high at 20:00 BJT on 29
August and rapidly moved toward the northwest. This
TC strengthened into a typhoon at 17:00 BJT on 30
August and into a super typhoon at 13:00 BJT on 1
September. After Kompasu moved across the axis of
the subtropical high, consistent with the steering flow
changing from southeastward to southwestward, it
turned to the north-northeast, as its intensity gradually
decreased. Kompasu made landfall in the border
between the two Koreas at 05:30 BJT on 2 September.
Then, it merged with a mid-latitude trough and
dissipated soon afterwards. The lifetime of Namtheun
was only approximately 31 hours and its intensity
remained as a tropical storm. It made landfall in
Jian’an, Fujian province at 23:50 BJT on 31 August.
Figure 1. The tracks of Lionrock, Kompasu, and Namtheun.
No.4 BAI Li-na (白莉娜), MA Lei-ming (马雷鸣) et al. 299
299
Figure 2. The intensity evolutions of Lionrock, Kompasu, and Namtheun (the numbers at the x-axis denote date and time, e.g.,
082902 is 02:00 BJT on 29 August; 082914 is 14:00 BJT on 29 August).
The unusual tracks of Lionrock and Kompasu
appeared to be associated, through binary TC
interaction, with Namtheun because these three TCs
were active simultaneously in the western North
Pacific and were close to each other. When Namtheun
was formed as a tropical depression in the South
China Sea at 08:00 BJT on 30 August, Lionrock
started turning east. After Namtheun dissipated,
Lionrock turned north-northeastward once more.
Kompasu moved to the northwest and Namtheun
moved steadily southwestward, with the distance
between them being approximately 800 km at 08:00
BJT on 31 August. After Namtheun dissipated, the
westerly component of the track of Kompasu
increased.
3 EVALUATIONS OF THE OPERATIONAL
TRACK FORECASTS
The operational track forecasts of the three TCs
made by six operational models, including three
global and three regional models, were evaluated. The
global models are the European Centre for
Medium-Range Weather Forecasts (ECMF), Global
Spectral Model of JMA (JAPN), and Global spectral
TC model of CMA at a resolution of T639L60
(TMBJ). The regional models include the (1)
Shanghai Typhoon Institute GRAPES Tropical
Cyclone Model (SGTM), which is based on the
GRAPES model and the new vortex initialization
scheme of the STI/China Meteorological
Administration (CMA); (2) Shanghai Typhoon Model
(SHTM), which is based on the MM5 model and the
bogus data assimilation (BDA) vortex scheme of the
STI/CMA; and (3) Guangzhou Tropical Model
(GZTM), which is the TCM developed by the
Guangzhou Tropical and Marine Meteorology
Institute (Ma et al.[16]).
The forecast samples of these operational models
for each TC are not the same, thus, a homogeneous
comparison should be developed based on a baseline
model, which is selected according to minimum track
error. For example, the average track error of all
samples for Lionrock is smallest from JAPN; thus,
this model is selected as a baseline model for this case.
Meanwhile, ECWMF is chosen as the baseline model
for Kompasu. The performance of these operational
models for Namtheun is not discussed in the current
evaluation because it is not sufficiently forecast by
each of the models due to its short lifetime.
3.1 Track of Lionrock
A homogeneous comparison of the performance of
the forecasts made by the six numerical models for the
track of Lionrock is shown in Table 1. The mean
absolute errors are less than 150 km at the 24-h
forecast time for all models. The track error of SHTM
is the least at the 24-h forecast, followed by those of
JAPN and ECMF. The errors are larger for the 48-h
forecast, although most of these are less than 200 km.
The performance of JAPN is significantly better than
those of the other models in this case.
The forecast tracks of the operational models for
Lionrock at different lifetimes are shown in Fig. 4.
Most models can forecast the northward track of
Lionrock within 24 hours during the early period (Fig.
3a). The forecast tracks from SHTM and GZTM are
more consistent with the observational track for
longer forecast periods. At 08:00 BJT on 30 August
(Fig. 3b), a tropical depression (later named
Namtheun) formed northeast of Lionrock. These
weather disturbances might exhibit binary interaction,
thus the uncertainty of the track of Lionrock was
increasing. Starting at this moment, Lionrock slowly
turned to the east. After 36 hours, it shifted to the
north to northeast. The global models made more
accurate track forecasts for this case compared with
the regional models. The forecast track of SGTM
moved directly northwestward, whereas those of
SHTM and GZTM were not able to forecast the
northwest turn of Lionrock after its eastward path. At
20:00 BJT on 31 August (Fig. 3c), Lionrock stopped
300 Journal of Tropical Meteorology Vol.20
300
moving eastward and started turning northwestward
as Namtheun weakened. Most operational models
were able to forecast its moving trend accurately in 24
hours. The forecast landing sites of the global models
were relatively close to that of the observation. At
08:00 BJT on 1 September (Fig. 3d), the forecast
tracks before landfall from all models are consistent
with the observational track. Namtheun moved
westward after landfall. This movement was forecast
accurately by the global models. The northward
components of the forecast tracks by the regional
models are larger than those of the observation.
Table 1. Mean errors of the track forecasts for Lionrock based on the same samples from the six numerical weather prediction
models. Models
Forecast/h Errors stat. JAPN ECMF JAPN TMBJ JAPN SGTM JAPN SHTM JAPN GZTM
sample size 4 9 8 5 3
24 errors/km 71.0 96.5 69.4 112.7 62.4 110.9 69.9 52.7 77.0 142.3
sample size 2 5 4 3 5 4
48 errors/km 68.5 117.5 91.2 196.6 96.8 238.3 111.5 185.8 91.2 137.8
Note: JAPN and GZTM do not have the same samples at the 48-h forecast time, thus, their mean errors are calculated from their own
samples.
Figure 3. The forecast tracks of various operational models for Lionrock at (a) 08:00 BJT on 29 August, (b) 08:00 BJT on 30 August,
(c) 20:00 BJT on 31 August, and (d) 08:00 BJT on 1 September.
3.2 Track of Kompasu
A similar comparison was made for Kompasu
(Table 2). The mean absolute track errors of most
models are less than 150 km in the 24-h forecast. The
most accurate model is ECMF, in which the error is
less than 90 km, followed by JAPN. In the 48-h
forecast, the track errors of most models are less than
200 km. The track error of ECMF remains the lowest,
followed by those of BJTM and JAPN.
(d)
(c)
(b)
(a)
No.4 BAI Li-na (白莉娜), MA Lei-ming (马雷鸣) et al. 301
301
Table 2. The mean errors of track forecasts of Kompasu based on the same samples from the six numerical weather prediction
models. Models
Forecast/h Errors stat. ECMF JAPN ECMF TMBJ ECMF SGTM ECMF SHTM ECMF GZTM
sample size 3 6 5 3 4
24 errors/km 44.7 74.7 70.8 212.2 73.2 132.2 85.0 153.0 73.5 120.0
sample size 1 4 3 1 3
48 errors/km 144.0 186.0 153.3 159.3 184.3 271.0 239.0 325.0 184.3 207.0
The forecast tracks of various operational models
for Kompasu at different time are shown in Fig. 4. At
08:00 BJT on 30 August (Fig. 4a), Kompasu was
moving steadily northwestward, and then, it turned to
the north after 54 hours. The turning location was at
(124.6°E, 32.3°N), which was more than 200 km east
of the forecast tracks of the operational models. The
forecast turning location of BJTM was the closest to
the observation data. At 20:00 BJT on 31 August (Fig.
4b), the forecast turning location remained at 150 km
west of the observation, although the forecast tracks
from all models shifted eastward. At 20:00 BJT on 1
September (Fig. 4c), Kompasu moved rapidly around
the subtropical high and recurved into the westerlies.
All forecast tracks are consistent with the
observational track as they include a well-defined,
strong steering flow northwest of the subtropical high.
Figure 4. The forecast tracks of various operational models for
Kompasu at (a) 08:00 BJT on 30 August, (b) 20:00 BJT on 31
August, and (c) 20:00 BJT on 1 September.
3.3 Forecast difficulties
Although the models exhibit predictive ability on
the tracks of Lionrock and Kompasu, their prediction
accuracy should be further enhanced. Global models
should be more capable of forecasting the unusual
track of Lionrock associated with Namtheun through
the effect of binary TC interaction. The forecast
turning locations for Kompasu of the six operational
models were westward from the observation, thus
making the estimated effects of rain and strong wind
on the east of China greater and earlier.
Difficulties in forecasting the tracks of Lionrock
and Kompasu include two important aspects. First, the
track of Lionrock is extremely changeable because of
its weak steering flow during its early period and the
binary TC interaction with Namtheun during its late
period. Second, the uncertainty with regard to the
recurving location of Kompasu is high because of its
rapid speed and binary TC interaction with Namtheun.
This interaction has a key role in the abnormal tracks
of Lionrock and Kompasu. The mechanism of binary
TC interaction will be discussed later based on the
conceptual model suggested by Carr et al.[14, 15] (Fig.
5), combined with NCEP/GFS 500-hPa geopotential
height analysis (Fig. 6).
At 20:00 BJT on 30 August, Namtheun moved
southwestward along the steering flow southwest of
the subtropical high, while the steering flow of
Lionrock remained weak (Fig. 6a). The distance
between Namtheun and Lionrock was approximately
751.8 km, which was close enough for a binary TC
(c)
(b)
(a)
302 Journal of Tropical Meteorology Vol.20
302
direct interaction (Fig. 5) to occur. Then, Lionrock
slowly turned to the east, and Namtheun kept moving
southwestward. By 20:00 BJT on 31 August (Fig. 6b),
the separation of Lionrock and Namtheun decreased
to 515.2 km. The relative motion of attraction and
cyclonic rotation between Lionrock and Namtheun
suggest that the two TCs may be under the direct
effect of binary interaction (Fig. 5).
(a)
(b)
Figure 5. Conceptual models of binary TC in (a) direct and (b) semi-direct interactions.
(a) (b)
Figure 6. NCEP/GFS 500-hPa geopotential height analysis at (a) 20:00 BJT on 30 August and (b) 20:00 BJT on 31 August.
Kompasu formed in the southwest of the
subtropical high, thus inducing Kompasu to move
steadily on the northwestward track during its early
period. Kompasu and Namtheun were sufficiently
separated such that their circulations did not overlap.
Thus, direct binary interaction did not occur. As
Kompasu approached the axis of the subtropical high
at 08:00 BJT on 31 August, Namtheun was almost
directly west of Kompasu. The separation distance
and the presence of the subtropical high east of
Kompasu satisfied the criteria for a semi-direct
interaction (Fig. 5). Environmental circulations
encouraged the recurvature of Kompasu, thus, the
northerly component of its track increased. A
semi-direct interaction tends to be transitory (Carr et
al.[14, 15]), thus, by 20:00 BJT on 31 August (Fig. 6b),
Kompasu turned northwest once more.
4 RESULTS OF THE NUMERICAL
SIMULATION
4.1 Model description and experimental design
Several sensitivity experiments on the intensity of
Namtheun were conducted to explore the factors that
control the interactions of binary TCs. The model
used in this study is GRAPES-TCM, which was
developed by Huang et al.[17] based on GRAPES
(Chen and Shen[18]). GRAPES-TCM has been under
real-time operational tests since 2004, and its
capability for TC simulation has been demonstrated
(Huang et al.[17]; Yu et al.[19]). The horizontal
No.4 BAI Li-na (白莉娜), MA Lei-ming (马雷鸣) et al. 303
303
resolution of GRAPES-TCM is set at 0.25°×0.25°
latitude/longitude with 31 vertical levels. The main
physics employed in the model include the
Kain–Fritsch cumulus and Medium-Range Forecast
planetary boundary layer parameterization schemes,
NCEP three-class simple-ice explicit scheme, and
Rapid Radiative Transfer Model long-wave radiation
scheme. Several key techniques on new vortex
initialization and cumulus parameterization schemes
have also been developed by STI/CMA in recent
years (Ma et al.[20]; Ma and Tan[21, 22]; Ma et al.[23, 24]).
The numerical experiments carried out in this
work are briefly introduced as follows. In the control
experiment (denoted as CTRL), the initial vortex of
Namtheun was derived from NCEP/GFS (without
bogus). In the first experiment (labeled Exp.stronger),
the initial intensity of Namtheun was enhanced
according to the cycling vortex initialization method
(Huang and Liang[25]). In the second experiment
(labeled Exp.remove), the vortex circulation of
Namtheun was removed by filter (Kurihara et al.[26]).
Each simulation was integrated for 72 hours, starting
from 08:00 BJT on 31 August. The initial analysis
fields are from NCEP/GFS.
4.2 Track and intensity verification
The modeled tracks of CTRL, Exp.stronger, and
Exp.remove, initiated at 08:00 BJT on 31 August, are
presented in Fig. 7. In CTRL, Lionrock moved slowly
eastward for the first six hours of simulation and then
it turned northeastward. At 08:00 BJT on 1 September,
Lionrock ended its northeastward track and moved
northward before shifting toward the northwest six
hours later. The simulation shows that the track of
Lionrock is reasonable, except for the estimated
turning location which was more toward the east. In
Exp.stronger (with a stronger Namtheun), Lionrock
moved eastward for the first 12 hours of simulation,
which caused the turning location to shift eastward by
0.5° compared with that in CTRL. In Exp.remove
(without the Namtheun vortex circulation), Lionrock
moved slowly and turned northeastward rapidly. The
simulation results indicate that the TC binary
interaction between Lionrock and Namtheun caused
them to exhibit cyclonic rotation, i.e., Lionrock
moved eastward and Namtheun moved southwestward.
The effect of the vortex of Namtheun on the track of
Lionrock increased when the intensity of Namtheun
was strengthened.
Figure 7. Model tracks of the three sensitivity experiments at 6-h interval for (a) Lionrock (solid) and Namtheun (dashed), (b)
Kompasu (solid) and Namtheun (dashed), initiated at 08:00 BJT on 31 August. The black curves indicate the operational tracks of
CMA; the green, red, and blue curves indicate the model tracks of the control run, Namtheun with a stronger vortex, and Namthem
with vortex removed, respectively.
The modeled tracks of Kompasu are also shown in
Fig. 7b. The northerly component of the track of
Kompasu in Exp.stronger was larger than that in CTRL
during the first 24 simulated hours. In Exp.remove, the
modeled track was moving westward more than that
in CTRL. The simulation results show that the
underestimation of the binary TC interaction between
Kompasu and Namtheun is probably the reason why
the six operational models failed to forecast the
turning location accurately.
In CTRL, the errors simulated by the models of
the initial intensities of Lionrock, Kompasu, and
Namtheun were -4, 3, and -6 hPa, respectively. The
mean error of the intensity of Lionrock during the
72-h period was 5.4 hPa. After its landfall, the
weakening trend of Lionrock in the experiment was
slower than what was observed initially. The intensity
of Kompasu in CTRL was stronger than that of the
observation, with a mean error of 7.9 hPa. Namtheun
remained as a tropical storm, which is consistent with
the observation.
(b)
(a)
304 Journal of Tropical Meteorology Vol.20
304
4.3 Steering flows
The movement of TCs is primarily controlled by
the steering flow (Dong and Neumann[27]). In Fig. 8,
the 500-hPa streamline isotach at 14:00 BJT on 31
August for Exp.stronger and Exp.remove, as well as their
association with the wind speed difference between
them and that in CTRL, are presented. In Exp.stronger
(Fig. 8a and 8c), the westerly and northerly winds
north of Lionrock were larger than those in CTRL.
The strengthened northwestward steering flow made
Lionrock move toward the east and the south more. In
Exp.remove (Fig. 8b and 8d), the situation was just the
opposite. The weakened westward steering flow
encouraged Lionrock to turn toward the northeast
rapidly.
When Namtheun was stronger, the pressure
gradient between Namtheun and the subtropical high
strengthened the westerly and southerly winds, thus
making the northerly component of the track of
Kompasu in Exp.stronger larger than that in CTRL (Fig.
8a and 8c). When the vortex of Namtheun was
removed (Fig. 8b and 8d), the westerly component of
the track of Kompasu got larger.
(a) (b)
(c) (d)
Figure 8. Wind distribution (contours) on 500 hPa at 14:00 BJT on 31 August for Exp.stronger (a, c) and Exp.enhance (b, d). The shaded
areas indicate the differences in zonal wind speed (a, b) and meridional wind speed (c, d) between the sensitivity experiments and
CTRL
4.4 Asymmetrical structure
Numerous studies found that the evolution of TC
asymmetrical structure affects the movement of TCs
(Wong and Chan[28]; Yan et al.[29]). Several scientists
have indicated that the physical mechanism of binary
TC interaction is the structural change in a TC caused
by the flow of another TC (Wang and Zhu[11]; Tian
and Shou[30]).
The asymmetrical component is calculated as
follows. First, the meteorological variable (, )
A
xy
in the Cartesian coordinate is transferred to (, )
A
r
θ
in the polar coordinate. Second, (, )
A
r
θ
is
decomposed into a symmetrical component (, )
s
A
r
θ
and an asymmetrical component (, )
a
A
r
θ
, as follows:
(, ) () (, )
sa
A
rArAr
θ
θ
=+ , (1)
where r is the radius from the TC center and
θ
is
the counterclockwise angle from the east. The
symmetrical component (, )
s
A
r
θ
in 0
rr
=
is
calculated by
00
0
(, )
() rrr r
s
A
r
Ar N
θ
≤< +Δ
=
∑
, (2)
where r
Δ
is the radial distance increment (0.25°)
and N is the number of grid points in the circular ring
00
rrr r
≤
<+Δ . Finally, the asymmetrical
component (, )
a
A
r
θ
is calculated by Eq. (1).
The 500-hPa asymmetrical geopotential heights of
Lionrock at 14:00 BJT on 31 August for Exp.stronger
and Exp.remove are shown in Fig. 9. In Exp.stronger (Fig.
9a), an NNW–SSE axis was found among vortex
doublets. The left vortex had an abnormally negative
No.4 BAI Li-na (白莉娜), MA Lei-ming (马雷鸣) et al. 305
305
geopotential height, whereas the right vortex had a
positive height. This distribution allowed for the
center southerly flow of Lionrock. In Exp.remove (Fig.
9b), the axis shifted to the NW-SE, thus indicating
that the flow among vortex doublets was
southeastward. The asymmetrical structure of
Lionrock in Exp.remove was more conducive for the
northward track than that in Exp.stronger, thus the
northerly component of the track of Lionrock in
Exp.remove was larger than in Exp.stronger.
The asymmetrical structures of Kompasu in both
the experiments also reflected the effect of Namtheun
circulation on the track of Kompasu (Fig. 9c and 9d).
When Namtheun circulation was stronger (removed),
the flow around the center of Kompasu was
southeastward (eastward), thus making the northerly
(westerly) component of the track of Kompasu larger.
Figure 9. The 500-hPa asymmetrical geopotential heights of Lionrock (a, b) and Kompasu (c, d) at 14:00 BJT on 31 August for the
(a) Exp.stronger and (b) Exp.remove. The solid (dashed) lines indicate positive (negative) values.
Yan et al.[29] and Tian and Shou[30] numerically
simulated the motion of binary TCs using a
nondivergent barotropic model with no basic flow.
The results of their simulation show that the
asymmetrical theory can explain the motional features
of binary TCs and that each binary TC can be
displaced primarily by the asymmetrical flow passing
through its center. These findings are consistent with
the effect of Namtheun on the abnormal tracks of
Lionrock and Kompasu discussed in the preceding
sections.
5 CONCLUSIONS
Three TCs, namely, Lionrock, Kompasu, and
Namtheun, were formed successively within 40 hours
in 2010. Forecasting the tracks of these TCs is
difficult because of uncertainties regarding
interactions among binary TCs. Evaluations are
performed on the operational track forecasts of the
three TCs made by six operational models, including
three global and three regional models. Although
these models exhibit ability in predicting tracks
associated with binary TC interaction, their prediction
accuracy should be enhanced further. The results
show that global models were more accurate in
forecasting the unusual track of Lionrock. The mean
absolute errors of the forecast of Lionrock’s track
were less than 150 km in the 24-h forecast for all
models and the turning position for the forecast of
(d)
(c)
(b)
(a)
306 Journal of Tropical Meteorology Vol.20
306
Kompasu was more westward by approximately 150
km from the observation.
The special track of Lionrock was associated with
a direct interaction with Namtheun, whereas the
relation between Kompasu and Namtheun satisfied the
criteria for a semi-direct interaction according to the
conceptual models of binary TC interaction (Carr et al.
[14, 15]). Lionrock and Namtheun appeared to rotate
cyclonically as they gradually approached each other
and the poleward steering flow of Kompasu was
slightly enhanced.
Numerical experiments based on GRAPES-TCM
further confirmed the effect of Namtheun on the
abnormal tracks of Lionrock and Kompasu. When
Namtheun was stronger, the larger westerly steering
flow north of Lionrock and the asymmetrical structure
of Lionrock circulation caused it to move more
toward the east. Meanwhile, the pressure gradient
between Namtheun and the subtropical high made the
westerly and southerly winds stronger, which caused
the northerly component of the track of Kompasu in
Exp.stronger to become larger than that in the CTRL.
When the Namtheun vortex was removed, the
weakened westward steering flow encouraged
Lionrock to turn rapidly to the northeast, thus, the
westerly component of the track of Kompasu became
larger.
In the current study, we discussed the effect of
binary TC interaction on the track of TCs. The effect
of this interaction on the intensity evolution of TCs
needs to be addressed in future studies. Operational
intensity forecasts of regional and global models
should be compared to gain experience in developing
typhoon models and in improving TC intensity
prediction.
REFERENCES:
[1] NEUMANN C J. Global overview. Global guide to tropical
cyclone forecasting [M]// WMO/TD-No. 560, World
Meteorological Organization, Geneva, 1993.
[2] QIN Liang-bo, CAO Xiao-gang. Analysis and comparison
of forecasting of a heavy rainfall event in Shanghai under twin
typhoons situation [J]. J. Trop. Meteorol., 2013, 29(2): 178-188
(in Chinese).
[3] FUJIWHARA S. On the growth and decay of vertical
systems [J]. Quart. J. Roy. Meteorol. Soc., 1923, 49(1): 75-104.
[4] FUJIWHARA S. Short note on the behavior of two vortices
[J]. Proc. Phys. Math. Soc. Japan, Ser. 3, 1931, 13(1): 106-110.
[5] ELSBERRY R L. Tropical cyclone motion. global
perspectives on tropical cyclones [M]// WMO/TD-No. 693,
World Meteorological Organization, Geneva, 1995.
[6] BRAND S. Interaction of binary tropical cyclones of the
western North Pacific [J]. J. Appl. Meteorol., 1970, 9(3):
433-441 (in Chinese).
[7] DONG K, NEUMANN C J. On the relative motion of
binary tropical cyclones [J]. Mon. Wea. Rev., 1983, 111(5):
945-953.
[8] LUO Zhe-xian, MA Jing-xian. Binary tropical cyclone
interactions on the south of an idealized subtropical ridge [J].
Acta. Meteorol. Sinica, 2001, 59(4): 450-458 (in Chinese).
[9] CHEN Lian-shou, DING Yi-hui. Overview on Northwest
Pacific Typhoon [M]// Science Press, Beijing, China, 1979 (in
Chinese).
[10] DEMARIA M, CHAN J C L. Comments on “A numerical
study of the interactions between two tropical cyclones” [J].
Mon. Wea. Rev., 1984, 112(8): 1643-1645.
[11] WANG Yu-qing, ZHU Yong-ti. Interactions of binary
vortices in a nondivergent barotropic model [J]. J. Trop.
Meteorol., 1989, 13(3): 208-216 (in Chinese).
[12] ZHANG Zhong-feng, LIU Qi-han, TUO Rui-fang.
Numerical study of the abnormal track of Typhoon Maggie [J].
J. Trop. Meteorol., 2006, 12(2): 174-178.
[13] LANDER M A, HOLLAND G J. On the interaction of
tropical-cyclone-scale vortices. I: Observation [J]. Quart. J.
Roy. Meteorol. Soc., 1993, 119: 1347-1361.
[14] CARR L E, BOOTHE M A, ELSBERRY R L.
Observational evidence for alternate modes of track-altering
binary tropical cyclone scenarios [J]. Mon. Wea. Rev., 1997,
125(9): 2094-2111.
[15] CARR L E, ELSBERRY R L. Objective diagnosis of
binary tropical cyclone interactions for the western North
Pacific basin [J]. Mon. Wea. Rev., 1998, 126(6): 1734-1740.
[16] MA Lei-ming, LI Jia, HUANG Wei, et al. Verification on
typhoon track forecast by operational typhoon numerical
models of China in 2007 [J]. Meteorol. Mon., 2008, 34(10):
74-80 (in Chinese).
[17] HUANG Wei, DUAN Yi-hong, XUE Ji-shan, et al.
Operational experiments and its performance analysis of the
tropical cyclone numerical model (GRAPES-TCM) [J]. Acta
Meteorol. Sinica, 2007, 65(4): 578-587 (in Chinese).
[18] CHEN De-hui, SHEN Xue-shun. Recent progress on
GRAPES research and application [J]. J. Appl. Meteorol. Sci.,
2006, 17(6): 773-777 (in Chinese).
[19] YU H, HUANG W, DUAN Y H, et al. A simulation study
on pre-landfall erratic track of typhoon Haitang (2005) [J].
Meteorol. Atmos. Phys., 2007, 1: 189-206.
[20] MA Lei-ming, LIANG Xu-dong, HUANG Wei. Progress
in research on key technologies of Shanghai Typhoon Model
[C]. 15th National Symposium of Tropical Cyclone, 2009 (in
Chinese).
[21] MA L M, TAN Z. Tropical Cyclone Initialization with
Dynamical Retrieval from a modified UWPBL model [J]. J.
Meteorol. Soc. Jpn., 2010, 88: 827-846.
[22] MA L M, TAN Z. Improving the behavior of the cumulus
parameterization for tropical cyclone prediction：Convection
trigger [J]. Atmos. Res., 2009, 92: 190-211.
[23] MA L M, CHAN J C L, DAVIDSON N E, et al.
Initialization with diabatic heating from satellite-derived
rainfall [J]. Atmos. Res., 2007, 85: 148-158.
[24] MA Lei-ming, QIN Zeng-hao, DUAN Yi-hong, et al.
Impacts of TRMM SRR assimilation on the numerical
prediction of tropical cyclone [J]. Acta Oceanol. Sinica, 2006,
25(5): 14-26 (in Chinese).
[25] HUANG Wei, LIANG Xu-dong. A cycling typhoon-like
vortex initialization scheme and its application to
No.4 BAI Li-na (白莉娜), MA Lei-ming (马雷鸣) et al. 307
307
GRAPES-TCM [J]. Acta Meteorol. Sinica, 2010, 68(3):
365-375 (in Chinese).
[26] KURIHARA Y, BENDER M A, TULEYA R E, et al.
Improvements in the GFDL hurricane prediction system [J].
Mon. Wea. Rev., 1995, 123(9): 2791-2801.
[27] DONG K, NEUMANN C J. The relationship between
tropical cyclone motion and environment geostrophic flow [J].
Mon. Wea. Rev, 1986, 114(1): 115-122.
[28] WONG M L M, CHAN J C L. Tropical cyclone motion in
response to land surface friction [J]. J. Atmos. Sci., 2006, 63(4):
1324-1337.
[29] YAN Li-jun, TIAN Yong-xiang, HUANG Xian-xiang. A
numerical study of the interaction of twin typhoons [J]. J. Trop.
Meteorol., 2004, 20(2): 167-175 (in Chinese).
[30] TIAN Yong-xian, Shou Shao-wen. Investigation on
interaction of binary tropical cyclones [J]. Acta Meteorol.
Sinica, 1998, 56(5): 584-593 (in Chinese).
Citation: BAI Li-na, MA Lei-ming, ZENG Zhi-hua, et al. Numerical analysis on the role of tropical storm Namtheun in the unusual
tracks of three tropical cyclones in 2010. J. Trop. Meteorol., 2014, 20(4): 297-307.
