The Opposite Effects of Inner and Outer Sea Surface Temperature
on Tropical Cyclone Intensity
Yuan Sun1, Zhong Zhong1,2, Lan Yi3, Yao Ha1, Yimei Sun4
1. College of Meteorology and Oceanography, PLA University of Science
and Technology, Nanjing, China
2. Jiangsu Collaborative Innovation Center for Climate Change and School of
Atmospheric Sciences, Nanjing University, Nanjing, China
3. National Climate Center, Beijing, China
4. Chinese Satellite Maritime Tracking and Control Department, Jiangyin, China
Published in the Journal of Geophysical Research-Atmospheres
on March 11, 2014
Corresponding author: Zhong Zhong
College of Meteorology and Oceanography,
PLA University of Science and Technology,
No. 60 Shuanglong Road, Nanjing, 211101, China.
Email: zhong_zhong@yeah.net
1
1
Abstract
2
A suite of semi-idealized numerical experiments are conducted to investigate the
3
sensitivity of tropical cyclone (TC) intensity to changes of sea surface temperature
4
(SST) over different radial extents. It is found that the increase of inner SST within
5
the range 1.5–2.0 times the radius of maximum wind (RMW), defined as the effective
6
radius (ER), contributes greatly to the increase of TC intensity and the reduction of
7
TC inner-core size, whereas the increase of outer SST (defined as SST outside the
8
ER) reduces TC intensity and increases TC inner-core size. Further analysis suggests
9
that, the effects of SST inside and outside the ER on TC intensity rely on the factors
10
that influence the TC development.
11
As the SST increases inside the ER, more surface enthalpy flux enters TC eyewall
12
and less enters the outer spiral rainbands. This will decrease the RMW, leading to a
13
smaller eyewall radius where strong latent heating is released. As a result, the central
14
pressure of the TC deepens with stronger radial pressure gradient. Meanwhile,
15
difference between SST and upper-tropospheric temperature increases. All above
16
factors contribute to TC intensification as the inner SST increases. The opposite
17
happen as the SST increases outside the ER. How TC intensity responds to the change
18
of entire SST depends on the competitive and opposite effects of inner and outer SST.
19
Moreover, understanding the mechanisms is vital to the forecast of variations in TC
20
intensity and inner-core size when a TC comes across an ocean cold or warm pool.
2
21
1. Introduction
22
Over the tropical oceans, low-level sensible and latent heat fluxes are important
23
energy sources for tropical cyclones (TCs). Therefore, the sea surface temperature
24
(SST) is a vital factor that determines TC genesis and intensification [Emanuel, 1986;
25
Rotunno and Emanuel, 1987; Holland, 1997; Persing and Montgomery, 2005; Bell
26
and Montgomery, 2008]. Placing the energetics of a TC in the framework of a Carnot
27
cycle, Emanuel [1995] demonstrated that high SST could lead to a strong TC with
28
high potential intensity.
29
Further studies based on numerical models attempt to find a radius within which
30
the surface enthalpy flux (the sum of sensible heat and latent heat fluxes) plays a vital
31
role in determining the intensity of the TC [Xu and Wang, 2010; Miyamoto and
32
Takemi, 2010; Sun et al., 2013a]. Although the size of this radius is controversial,
33
results of above studies have shown that the impact of the surface enthalpy flux and
34
SST outside this radius on TC intensity is substantially different from that inside the
35
radius. This radius is about 7-8 times the radius of the maximum wind (RMW) in the
36
axisymmetric model of Miyamoto and Takemi [2010]. Farther outward, the surface
37
enthalpy flux doesn’t affect the TC intensity. As suggested by Xu and Wang [2010]
38
(hereafter XW10), this radius is about 2-2.5 times the RMW and the surface enthalpy
39
fluxes outside this radius are crucial to the growth of the TC inner-core size but
40
reduce the TC intensity. Similar to results of XW10, Sun et al. [2013a] demonstrated
41
that the inner SST within a radius 2-3 times the RMW contributes greatly to the
42
increase in TC intensity while the outer SST outside this radius reduces TC intensity.
3
43
More important, this radius may be useful to estimate when a TC is influenced by a
44
sudden change in fluxes while it is approaching a warm or cold eddy.
45
Previous studies implied that the high sensitivity of TC intensity to the inner SST
46
can be attributed to the negative effect of the outer SST on the TC intensity
47
[Miyamoto and Takemi, 2010; XW10; Sun et al., 2013a]. These studies are more
48
focused on the different effects of inner and outer surface enthalpy fluxes on TC
49
intensity, but less involved in exploration of the possible mechanisms. One in-depth
50
research topic on this issue is to figure out the mechanisms that favor or suppress the
51
TC development as the inner and outer SST changes. Sun et al. [2013] have noted the
52
importance of the inner and outer SST on TC intensification based on energy transfer
53
between the air and sea and revealed its importance in TC intensification [e.g.,
54
Emanuel, 1991; Holland, 1997; Sun et al., 2013a]. Unfortunately they underestimate
55
or ignore the role of TC eyewall structure on determining TC intensity as the SST
56
changes, i.e., the SST can affect the TC intensity by influencing the TC eyewall
57
structure. The former impact emphasizes the change of the surface enthalpy fluxes,
58
while the latter impact focuses on the change in the eyewall structure. The eyewall of
59
a TC is a ring with deep convection near the RMW. The TC obtains most of its energy
60
from heat fluxes transferred from the ocean surface to the atmosphere, which is
61
released later as latent heat due to moist convection in the eyewall [Wang and Wu,
62
2004]. The eyewall structure is closely related to the TC intensity (e.g., minimum
63
surface level pressure and maximum wind speed) and influenced by the underlying
64
SST [e.g., Chen et al., 2010; Lee and Chen, 2012]. There is a strong dependence of
4
65
the simulated storm size on the surface enthalpy fluxes outside the eyewall. The
66
results are confirmed by further study of XW10. Nevertheless, both Wang and Xu
67
[2010] and XW10 did not address the relation between changes in storm intensity and
68
that in storm structure (e.g., RMW, radial pressure gradient, etc.). According to the
69
sensitivity experiments in Fierro et al. [2009] and Sun et al. [2013b], the simulated
70
storms with smaller RMW are often accompanied by larger radial pressure gradients
71
near the eyewall. This larger radial pressure gradient is favorable for TC
72
intensification. However, previous studies have not explicitly revealed whether a
73
storm over an ocean warm pool exhibits structural features that favor TC
74
intensification. Thereby, the impact of SST on TC intensity in view of storm structure
75
still remains a hypothesis that needs further studies.
76
It should be mentioned that the reasonable design of sensitivity experiments is a
77
prerequisite to get reliable results on the response of TC intensity to SST change. In
78
their sensitivity experiments of Miyamoto and Takemi [2010] and XW10, the surface
79
enthalpy fluxes outside certain radii are eliminated to determine the radius inside
80
which the enthalpy flux is of great impact on TC intensity. However, although the
81
underlying SST determines TC intensity by feeding the surface enthalpy flux to the
82
TC, the change of SST may not be consistent with the change of the surface enthalpy
83
flux induced by SST change. To better understand the TC intensity change especially
84
when a TC is approaching an ocean warm or cold pool, it is more appropriate to
85
conduct a series of sensitivity experiments on SST change rather than on the surface
86
enthalpy flux change. In the ten sensitivity experiments of Sun et al. [2013a], the
5
87
change of SST is modified with a weighting function that is maximized in the TC
88
center. SST changes linearly outwards until the change becomes zero at a prescribed
89
radius and remains zero further outwards. Due to the non-uniform change of SST
90
within the certain radius, the design of their weighting function is somewhat
91
inappropriate and inefficient to estimate the difference in SST anomalies and their
92
impacts on TC intensity between the sensitivity experiments. In this study, we
93
carefully design the sensitivity experiments to overcome the weaknesses of previous
94
studies.
95
Although the importance of the local SST and the surface enthalpy flux to TC
96
intensity has been addressed, questions remain as to 1) how will TC activity (e.g., the
97
intensity, inner-core size and structure of TC) respond to the SST change over
98
different radial extents with respect to its center, and 2) what are the possible reasons
99
for the different impacts of SSTs over different radial extent on TC intensity from the
100
viewpoint of the surface enthalpy flux and eyewall structure. While the emphasis of
101
other studies [e.g., Miyamoto and Takemi, 2010; XW10; Sun et al., 2013a] was mainly
102
on the impact of SST on TC intensity in view of the surface enthalpy flux, our study
103
give a more comprehensive view that also includes the impact of SST in view of
104
storm structure by examining the near-eyewall structure and upper-level temperature.
105
To investigate the impact of SST on TC activity, a suite of sensitivity experiments are
106
designed for the case of Typhoon Shanshan (2006). The Weather Research and
107
Forecasting (WRF) model forced by artificially changed SSTs is used to simulate
108
Shanshan. The paper is organized as follows. Section 2 describes the model
6
109
configuration and experiment designs. Section 3 discusses the sensitivity of TC
110
intensity and inner-core size to SST anomalies over various radial extents. The
111
possible reasons for the change in TC intensity are analyzed in section 4. Conclusions
112
and discussions are given in the final section.
113
2. Model configuration and experiment designs
114
TC Shanshan formed over the western Pacific (16.7°N, 134.9°E) at 1200 UTC 10
115
September 2006 and moved westward as it intensified. At 1800 UTC 14 September, it
116
reached eastern Taiwan (20.7°N, 124.6°E) and intensified into an extremely intense
117
typhoon. After turning to the northeast, Shanshan made landfall on Japan at about
118
1000 UTC on 17 September. The movement of Shanshan followed a typical recurving
119
track with typical intensity and duration, which are illustrated by the best-track data
120
obtained
121
(http://www.usno.navy.mil/NOOC/nmfc~ph/RSS/jtwc/best_tracks/wpindex.html).
from
the
Joint
Typhoon
Warning
Center
(JTWC)
122
The model used in this study is the Advanced Research Weather Research and
123
Forecasting (WRF-ARW) model. The model domain is triply-nested with two-way
124
interactive nesting. The inner domains, which are designed to define Shanshan,
125
automatically move to follow the position of model storm via automatic vortex-
126
following algorithm [Skamarock et al., 2008]. The initial and lateral boundary
127
conditions are extracted from the 1°×1° National Centers for Environmental
128
Prediction
129
(http://dss.ucar.edu/dsszone/ds083.2). The TC Bogus scheme in the WRF model is
130
adopted to generate the initial field [Skamarock et al., 2008]. In this study, daily SST
(NCEP)
final
analysis
data
7
(FNL)
at
6-h
intervals
131
is updated using the Tropical Rainfall Measuring Mission (TRMM) Microwave
132
Imager (TMI) level-1 standard product. It is a gridded data with 0.25° × 0.25°
133
resolution provided by the Earth Observation Research Center (EORC)/National
134
Space Development Agency (NASDA) (http://www.eorc.nasda.go.jp/TRMM).
135
The model integration starts at 0000 UTC 14 September 2006 and ends at 1200
136
UTC on 16 September 2006, which covers the intensification and the steady-state
137
period of Shanshan (hereafter CTRL). The Yonsei University non-local-K planetary
138
boundary layer scheme [Hong et al., 2006] and Monin-Obukhov surface layer scheme
139
[Dyer and Hicks, 1970; Paulson, 1970; Webb, 1970; Beljaars, 1994] are used in this
140
study. Some other important physical schemes include Lin microphysical scheme [Lin
141
et al., 1983], and Betts-Miller-Janjić convective parameterization scheme [Betts,
142
1986; Betts and Miller, 1986; Janjić, 1994; Janjić, 2000]. Details of the model
143
configuration and simulation results can be found in Sun et al. [2012]. The triply-
144
nested model is run with grid spacing of 45, 15, and 5km, respectively and the model
145
output is saved at 5-min interval. Statistical analysis of the simulation results in the
146
inner-most domain (D3) shows that the average track error is about 55 km, minimum
147
surface level pressure (MSLP) error is 4.7 hPa and the maximum wind speed error is
148
4.4 m s-1. Figures 1a and 1b show the triply-nested model domains and the tracks of
149
Shanshan at 6-h intervals from the observation and CTRL, respectively.
150
In addition to the CTRL, 21 sensitivity experiments are performed to investigate
151
the response of TC intensity to the changes of SST over different radial extents. In
152
these experiments, the underlying SST is either reduced or increased within a
8
153
specified radius of the TC center in all domains (D1, D2 and D3). In particular, the
154
SST is modified with a weighting function that is maximized within a prescribed
155
radius of the TC center and decreases linearly outward to zero at the prescribed radius
156
plus 15 km, and remains zero farther outward (This extra 15-km-smoothing is not
157
necessary and not performed in D1). Note that, since the TC is a moving system, the
158
SST field is modified on the basis of TRMM/TMI SST data in each time-step to
159
ensure the center of the changed SST region is always consistent with the TC center in
160
our sensitivity experiments. The extents and radii of some sensitivity experiments are
161
shown in Fig. 1c.
162
The effects of underlying SST on intensity, inner-core size and structure of the TC
163
are investigated through comparing results of the sensitivity experiments. Table 1 lists
164
the sensitivity experiments and their simulated maximum intensities and RMWs. It
165
should be pointed out that, such artificially moving SST anomalies in our sensitivity
166
experiments may not exist in reality. In addition to SST, ocean heat content and mixed
167
layer depth could also be important to storm intensity and structure due to their
168
influence on the strength of the storm-induced SST cooling. Since it is the SST in the
169
WRF model that directly influences the energy transfer between the ocean and
170
atmosphere, results of our experiments with the imposed SST changes described
171
above will help answering the question that, within what radial extent of a storm is the
172
underlying SST relevant for determining TC intensity and how does TC intensity
173
change as the TC approaches an ocean cold or warm pool? In addition, in all our
174
sensitivity experiments, the simulated TCs follow a similar track (figure omitted), and
9
175
thus prevent the differences in TC structure to be due to different environmental fields
176
between the experiments. This allows us to examine the sensitivity of simulated TC
177
intensity and structure to the underlying SST in an environment basically similar to
178
that of Typhoon Shanshan.
179
3. Sensitivity to SST anomalies over different radial extent
180
3.1. Storm intensity
181
To demonstrate the sensitivity of TC intensity to SST anomalies over different
182
radial extents, we compare in Figs. 2 and 3 the temporal evolutions of MSLP and
183
maximum surface wind speed simulated by the sensitivity experiments. Note that the
184
evolutions of TC intensity for the sensitivity experiments with 1°C SST change are
185
not shown because the change of TC intensity is gradual as the magnitude of SST
186
change varies from 1°C to 2°C. Since the RMW is about 55 km in CTRL (Table 1),
187
E45 can be considered as a sensitivity experiment for the SST change in the TC eye
188
region. In contrast, in E90 the SST change occurs in both the eye and eyewall regions.
189
E180 and E270 can be regarded as the experiments with SST changes in the inner and
190
outer spiral rainbands, respectively.
191
Comparing E45-2 with CTRL, it can be seen that, the contribution of the SST in
192
the eye region to the storm intensity is relatively small. This is consistent with results
193
of XW10 when the surface enthalpy flux is changed in the eye region in one of their
194
experiments. In E45+2, however, the increase of SST and thus the surface enthalpy
195
flux in the eye can reduce RMW (Table 1). Similar results are found in Table 1 of
196
XW10. The region with a radius of 45 km contains not only the eye, but also part of
10
197
the eyewall in E45+2, suggesting that the increase of the surface enthalpy flux inside
198
the eyewall contributes to the significant increase of TC intensity and the additional
199
reduction of RMW compared with that of CTRL (Figs. 2b, 3b and Table 1).
200
Therefore, there may exist a positive feedback between the increase of the surface
201
enthalpy flux inside the eyewall and the reduction of RMW in E45+2. The smaller
202
increase of SST in E45+1, however, cannot trigger this positive feedback mechanism,
203
and thus results in a negligible change of RMW and intensification of TC (figure
204
omitted).
205
As shown in Figs. 2 and 3, E180-2 and E90+2 are the experiments with the most
206
notable decrease and increase of TC intensity, respectively. Compared with the
207
CTRL, the storm minimum MSLP is increased by 30.1 hPa in E180-2 and decreased
208
by 31.3 hPa in E90+2. The maximum wind speed is reduced by 7.7 m s-1 for a 2°C
209
decrease of inner SST in E180-2 and increased by 8.5 m s-1 for a 2°C increase in
210
E90+2 (Table 1). Based on above results, we define an effective radius (ER) of about
211
1.5–2.0 times the RMW (i.e., not the RMW in CTRL, but the RMW in each
212
experiment), within which the SST is effective for the maintenance and intensification
213
of the TC. SST within the ER is defined as “inner SST” in this study while outside the
214
ER it is named “outer SST”. The size of the ER is basically consistent with that
215
suggested by XW10 and Sun et al. [2013a], but different from that of Miyamoto and
216
Takemi [2010], who indicated that the effective radius of enthalpy flux is about 7-8
217
times the RMW.
11
218
The experiments with SST being changed farther outside of the eyewall, e.g.,
219
E270 and Eall, can be viewed as the outer SST experiments in contrast to the inner
220
SST experiments (e.g., E90+2 and E180-2). In other words, E270 and Eall are the
221
sensitivity experiments with outer SST being changed if we consider the inner SST
222
experiments (e.g., E90+2 and E180-2) as references. As mentioned above, the outer
223
SST is defined as the SST outside the ER. It can be seen that the decrease of SST in
224
E270-2 does not cause such a great magnitude decrease as in E180-2 (Figs. 2 and 3).
225
Similarly, the increase of SST in E270+2 doesn’t cause the same magnitude increase
226
of TC intensity as in E90+2. Results of E270-2 versus E180-2 and E270+2 versus
227
E90+2 suggest that a decrease of outer SST suppresses the weakening of the
228
simulated storm, while an increase of outer SST suppresses the intensification of the
229
simulated storm. The same conclusion can be obtained by comparing Eall-2 with
230
E180-2 and Eall+2 with E90+2. Above results are largely attributed to the fact that the
231
decrease of outer SST weakens the activity of outer spiral rainbands while the
232
increase of outer SST reinforces it. As suggested by Wang [2009] and Hill and
233
Lackmann [2009], the TC intensity will reduce as convection in the outer spiral
234
rainbands intensifies and the opposite is true when the convection weakens there.
235
Thereby, the outer SST is responsible for the less change in TC intensity in the entire-
236
SST experiments (e.g., Eall-2 and Eall+2) than that in the inner SST experiments
237
(e.g., E180-2 and E90+2), namely, TC intensity is relatively insensitive to the entire
238
SST change.
12
239
The TC intensity change is not only related to the extent and magnitude of SST
240
changes but also associated with the mean SST itself. It is found that 30°C may be a
241
critical temperature for the development of a TC, since the TC experiences rapid
242
intensification when the SST is between 27°C and 30°C but slows down as the SST is
243
above 30°C [Chan et al., 2001]. SST near the TC in CTRL is basically between 29°C
244
and 32°C. Comparing Eall+2 with the CTRL, the TC intensity is not increased
245
notably (except for the last 6 hours in Eall+2 in Fig. 2b), following the increase of the
246
entire SST. The outer SST is related to the mean SST and may be responsible for the
247
insensitivity of TC intensity to the entire SST increase under warm SST conditions,
248
for the difference of TC intensity between Eall-2 and E180-2 is much smaller than
249
that between Eall+2 and E90+2 (Figs. 2 and 3). It is concluded that the outer SST
250
contributes little to storm intensification when the mean SST is relatively cold (see the
251
difference between Eall-2 and E180-2 in Figs. 2 and 3), whereas it plays a critical role
252
in reducing TC intensity when the mean SST is larger than 30°C (see the difference
253
between Eall+2 and E90+2 in Figs. 2 and 3).
254
The above results imply that the TC may not become much stronger as the entire
255
SST increases since the opposite effect of SST inside and outside the ER cancels each
256
other, resulting in little changes in TC intensity. Whether TCs will become stronger as
257
the entire SST increases under the global warming condition in future, however,
258
remains a controversial issue and requires further investigation.
259
3.2. Storm inner-core size
13
260
To provide a detailed description of the eyewall radius, the evolution of the
261
overall inner-core size of the simulated storms is defined as the azimuthal mean RMW
262
at 10 m (Fig. 4). Since the eyewall of the simulated storm is asymmetric from time to
263
time [Wang, 2007], we only use the azimuthal mean as a proxy of the location of the
264
overall eyewall. As is shown, the SST changes over different radial extents play
265
radically different roles in determining the storm inner-core size.
266
Due to the decrease of SST in the eye region, the simulated RMW in E45 is
267
notably increased as compared to that in the CTRL. The opposite is true as SST
268
increases in the eye region. However, the decrease of inner SST within the ER (such
269
as in E180-2) is responsible for the greatest increase in the inner-core size of TC after
270
12-h model spin-up time. Similarly, the increase of inner SST within the ER (such as
271
in E90+2) reduces the TC inner-core size. This is somewhat different from the
272
idealized model results of XW10, which ignored the impact of the surface enthalpy
273
flux near the eyewall on the TC inner-core size. On the contrary, it is found that the
274
decrease of outer SST is responsible for the sharp decrease of the storm inner-core
275
size while the increase of outer SST results in a sharp increase of it. This is clearly
276
reflected in the difference between results of Eall-2 and E180-2 and that between
277
results of Eall+2 and E90+2 (Fig. 4). The strong impact of outer SST on TC inner-
278
core size indicates that TC may become much larger as the entire SST increases.
279
Interestingly, there is a notable negative correlation between TC intensity and its
280
inner-core size. Table 1 shows the differences between the sensitivity experiments and
281
CTRL. These differences demonstrate various impacts of SST change over different
14
282
radial extents on TC intensity and its inner-core size. Based on the statistical analysis
283
of observations of many storms, Stern and Nolan [2009] suggested that the size of the
284
RMW is directly proportional to the outward slope of the RMW with height, but there
285
is no distinct relationship between the size of the RMW and TC intensity. However,
286
they also emphasized that for a given simulation, the RMW is approximately
287
inversely proportional to TC intensity. Their results are consistent with the findings of
288
Emanuel [1986], who suggested that for the simulation of a single storm, the storm
289
intensity appears to be closely related to slope of the RMW and thus its size, although
290
the relationship is nonlinear. Results in our study reveal a strong relationship between
291
the RMW and TC intensity, despite the fact that the underlying SST is varied. Details
292
will be further discussed in section 4.2. According to this negative correlation, the
293
dramatic decrease of RMW in Eall+2 may be responsible for the increased
294
intensification of TC after 0300 UTC 16 September 2006. It is noteworthy that the
295
storm intensity in Eall+2 is still much weaker than that in E90+2 even after its
296
increased intensification.
297
4. Possible reasons for the change of TC intensity
298
4.1. Impact of SST in view of the surface enthalpy flux
299
4.1.1. Area-integrated surface enthalpy flux
300
Previous studies suggested that a TC intensifies and maintains its strength
301
against surface frictional dissipation by extracting energy from the underlying oceans.
302
Thus, energy exchange at the air-sea interface is the key to the intensity change of a
303
TC [Malkus and Riehl, 1960; Black and Holland, 1995]. Figure 5 shows the temporal
15
304
evolution of area-integrated kinetic energy at 10 m and the surface enthalpy flux
305
within a 200-km radius in some representative experiments. Owing to the effect of
306
warm SST and other favorable conditions, the area-integrated surface enthalpy flux
307
and kinetic energy increase from 0600 UTC 14 September to 0300 UTC 16
308
September. Note that a significant reduction in the surface enthalpy flux and kinetic
309
energy occurs at about 0300 UTC 16 September 2006 in these sensitivity experiments
310
except for Eall-2, when the simulated TC passed over a cold pool near Miyako Island
311
during that period.
312
In our experiments, though the TC intensity in E90+2 is much stronger than that
313
in Eall+2 (Figs. 2), the integrated surface enthalpy flux and kinetic energy in E90+2
314
are notably less than those in Eall+2 (Fig.5). In contrast, the TC intensity in E180-2 is
315
much weaker than that in Eall-2 (Figs. 3), but the integrated surface enthalpy flux and
316
kinetic energy in E180-2 are just slightly higher than those in Eall-2 (Fig. 5).
317
Apparently differences in the integrated surface enthalpy flux might explain the
318
difference in storm intensity between sensitivity experiments and CTRL, but cannot
319
explain the difference between the inner SST experiments (e.g., E180-2 and E90+2)
320
and the entire SST experiments (e.g., Eall-2 and Eall+2). There must have other
321
factors that promote or suppress storm intensification in the entire SST experiments.
322
We will discuss this issue in section 4.1.2.
323
4.1.2. Different effects of the surface enthalpy flux inside and outside the ER
324
The wind-induced surface heat exchange, which describes a positive feedback
325
between the increase in the surface enthalpy flux and the surface wind speed in the
16
326
near-core region of a TC, is regarded as the dominant process that controls the rapid
327
intensification of a TC [Emanuel, 1986; Rotunno and Emanuel, 1987]. Hence the
328
surface enthalpy flux may play a more important role in determining TC intensity in
329
the near-core region than in other region. As mentioned above, the SST and thus the
330
surface enthalpy flux within the ER contributes greatly to the storm intensification,
331
while the surface enthalpy flux outside the ER actually reduces the storm intensity,
332
suggesting that the effects of SST anomalies over different radial extents on TC
333
intensification are substantially different.
334
Figure 6 illustrates the distributions of the surface enthalpy flux difference and
335
reflectivity difference between two specific experiments (i.e., E180-2 and E90+2) and
336
CTRL, averaged over the mature stage (from 1800 UTC 15 September to 0600 UTC
337
16 September). It clearly shows that the surface enthalpy flux near the eyewall
338
reduces considerably with the decrease of inner SST, leading to weaker convection
339
near the eyewall in terms of simulated reflectivity and thus weaker TC intensity.
340
Meanwhile, the decrease of inner SST inside the ER also induces an increase in
341
convective activity of outer spiral rainbands. The opposite happen when the inner SST
342
is increased. As proposed by Houze et al. [2006] and Houze et al. [2007], there is a
343
competition between effects of convection in outer spiral bands and that near the
344
eyewall. This competition plays a vital role in determining the change of TC intensity.
345
Wang [2009] has suggested that active convection in outer spiral rainbands can reduce
346
the storm intensity. Thus, the increase in inner SST contributes greatly to TC
347
intensification by enhancing convection near the eyewall and weakening the
17
348
convection in outer spiral rainbands. The decrease in inner SST has the opposite
349
effects.
350
To investigate the impact of the outer SST on TC intensity, we compare the
351
distributions of the surface enthalpy flux and reflectivity between the inner SST
352
experiments (e.g., E180-2 and E90+2) and the entire SST experiments (e.g., Eall-2
353
and Eall+2, Fig. 7). The difference between E180-2 and Eall-2 can be regarded as the
354
result of the increased outer SST, while the difference between E90+2 and Eall+2
355
reflects the impact of the decreased outer SST. It is found that the increase in outer
356
SST not only induces more surface enthalpy flux outside the ER and thus stronger
357
outer spiral rainbands, but also reduces the surface enthalpy flux inside the ER and
358
thus the activity of convection near the eyewall (Fig. 7). Above results are completely
359
reversed when the outer SST decreases. Impacts of the increase (decrease) in outer
360
SST are similar to that of the decrease (increase) in inner SST (Figs. 6 and 7). As
361
suggested by Sun et al. [2013a], due to the impact of the warm outer SST, the air-sea
362
temperature and humidity difference and thus the surface enthalpy flux decrease
363
gradually as the surface inflow approaches the TC center, resulting in a decrease in
364
the surface enthalpy flux near the eyewall. This may also be used to explain the
365
changes in surface enthalpy flux inside the ER, which decreases when the outer SST
366
increases but increases when the outer SST decreases. Moreover, the impact of outer
367
SST on the surface enthalpy flux and thus TC intensity under entire warmer SST
368
conditions is much stronger than that under entire colder SST conditions (Figs. 2 and
369
7). Therefore, in contrast with the impact of the inner SST on TC intensity, the outer
18
370
SST reduces TC intensity by enhancing convection in the outer spiral rainbands and
371
weakening the convection near the eyewall, especially under entire warmer SST
372
conditions (e.g., in Eall+2). In addition, it is important to point out that the change of
373
SST is not consistent with the change of surface enthalpy flux induced by SST change
374
as shown in Fig. 7. The increase of outer SST can not only increase the surface
375
enthalpy flux outside the ER, but also decrease the surface enthalpy flux inside the
376
ER. The opposite is true when the outer SST decreases (Figs. 7a and 7c). This fact
377
further proves that it is more appropriate to conduct a series of reasonable sensitivity
378
experiments with SST change rather than with the surface enthalpy flux change to
379
investigate the mechanism for changes in TC intensity.
380
Changes in the surface enthalpy flux distribution have a significant impact on the
381
storm inner-core size by influencing the activity of outer spiral rainbands. Comparing
382
E180-2 with Eall-2, the warmer outer SST in E180-2 results in larger amount of
383
simulated surface enthalpy flux released outside the ER, which is favorable for the
384
development of outer spiral rainbands (Figs. 7a and 7b). The pressure outside the
385
RMW and hence the pressure gradient across the RMW decreases subsequently.
386
These changes help weaken the storm but increase the storm inner-core size. All
387
above changes and their impact on the storm intensity and the inner-core size are
388
reversed when the outer SST decreases, which shows the impact of colder outer SST
389
in E90+2 compared to that in Eall+2 (Figs. 7c and 7d). Note that the effect of the
390
inner SST decrease (increase) is similar to that of outer SST increase (decrease)
391
outside the ER. Above results clearly indicate that the inner and outer SST changes
19
392
can either increase or decrease the activity of outer spiral bands, leading to changes in
393
both the storm intensity and the storm inner-core size (Table 1).
394
The effect of the entire SST change on TC intensity and inner-core size can be
395
regarded as the sum of the effects of inner and outer SST. As the entire SST increases,
396
the effects of the inner SST that help intensify the TC are stronger than the effects of
397
the outer SST that suppress the TC intensification. Meanwhile, the effects of the inner
398
SST that reduce the inner-core size are weaker than the effects of the outer SST that
399
increase the inner-core size. As a result, the TC becomes stronger and larger as the
400
entire SST increases. The opposite happen when the entire SST decreases. However,
401
the positive effect of the entire SST on TC intensification is small under warm area-
402
averaged SST condition. This is consistent with the findings of Sun et al. [2013a].
403
4.2. Impact of SST in view of the thermodynamic structure
404
4.2.1. Latent heating
405
Many recent studies suggested that energy exchange between the ocean and
406
atmosphere affects TC structure [e.g., Chen et al., 2010; Lee and Chen, 2012]. In the
407
previous section, we have shown a negative correlation between TC intensity and
408
inner-core size when imposing SST changes at various radial extents. Stern and Nolan
409
[2009] found a linear relationship between the size of RMW (which basically
410
corresponds to the radius of eyewall) and the outward slope of eyewall based on both
411
observational data analysis and theoretical deduction. They argued that the
412
relationship between the eyewall slope and TC intensity may be attributed to the
413
impact of the radial pressure gradient. The ascending air parcels in the eyewall
20
414
normally experience a reduction of the inward-directed pressure gradient force,
415
leading to an outward centrifugal displacement with increasing height, i.e. the eyewall
416
slope [Gentry and Lackmann, 2010]. Most importantly, a less upright eyewall often
417
comes out with a weaker radial gradient of tangential and vertical velocities and
418
pressure, which would eventually suppress storm intensification. In contrast, a more
419
upright eyewall is often accompanied with a stronger radial gradient of vertical
420
velocity and pressure, which would promote storm intensification [Fierro et al., 2009;
421
Sun et al., 2013b]. Thereby, the TC intensity is related to the TC eyewall slope and
422
thus the TC inner-core size.
423
In this study, the change of RMW may be an important index to estimate the
424
change of TC intensity as SST varies in the sensitivity experiments. As mentioned
425
above, due to the difference in activity of outer spiral rainbands caused by SST
426
difference, the simulated storms in the sensitivity experiments exhibit a large
427
difference in inner-core size. The storm inner-core size determines the radius at which
428
strong latent heating occurs, and thus contributes to storm intensification.
429
Latent heating is the direct forcing that drives the eyewall convection and the
430
secondary circulation of TC, which makes the storm to maintain its strength and
431
develop further. In this study, the latent heating is found to be sensitive to the change
432
of SST and its radial extent. This is mainly because the SST-related surface enthalpy
433
flux is favorable for active convection in the storm, which will increase the latent heat
434
release [Wang, 2009]. As suggested by Eliassen [1951] and Holland and Merrill
435
[1984], one of the fundamental conditions for TC development is that heating occurs
21
436
in the region of high inertial stability. The heating in the high-inertial-stability region
437
inside the RMW is efficient at generating a localized temperature tendency, and this
438
efficiency increases dramatically with storm intensity; whereas the heating in the low-
439
inertial-stability region outside the RMW is inefficient at generating a warm core. In
440
other words, the vortex intensification rate depends on how much of the heating is
441
occurring inside the RMW. Moreover, Vigh and Schubert [2009] proposed that
442
heating in the outer region of about 2-3 time (80 km: 30 km) the RMW from TC
443
center could still spin up the local tangential wind and radially constrained the
444
circulation response, though the heating is occurring outside of the RMW.
445
Figure 8 shows the azimuthal- and time-averaged cross sections of the model-
446
simulated latent heating at the TC mature stage. As expected, due to more energy
447
exchange at the air-sea interface (i.e., the surface enthalpy flux), all the warmer-SST
448
experiments except for Eall+2 produce larger magnitudes of latent heating near the
449
eyewall. The larger latent heating results in a larger central pressure fall and thus
450
larger radial pressure gradient. However, compared with results in the entire warmer-
451
SST experiment (i.e., Eall+2), the simulated latent heating in the inner warmer-SST
452
experiments are concentrated in much smaller areas close to the eye due to the smaller
453
RMW and eyewall slope. Similar results are found when we compare simulations of
454
the entire colder-SST (i.e., Eall-2) experiment with that of the inner colder-SST
455
experiments. Hack and Schubert [1986] have shown that the smaller the radius at
456
which latent heating occurs, the more significant contribution it brings to the central
457
pressure fall and radial pressure gradient near the eyewall. Hence the simulated latent
22
458
heating at the smaller radius in the inner warmer-SST experiments and the entire
459
colder-SST experiment contributes to a deeper central pressure fall and larger pressure
460
gradient, and eventually leads to either a larger increase or a smaller decrease in TC
461
intensity (Figs. 2 and 3). Conversely, the larger radius in the inner colder-SST
462
experiments and entire warmer-SST experiment is responsible for the larger decrease
463
or smaller increase in TC intensity (Figs. 2 and 3).
464
4.2.2. Upper-tropospheric temperature
465
Impacts of SST on TC activity have been addressed in previous studies with a
466
focus on whether the changes of SST can modify the tropospheric temperature profile
467
in the tropics [Vecchi and Soden, 2007; Vecchi et al., 2008; Ramsay and Sobel, 2011].
468
The large-scale SST can affect TCs via its influence in upper-tropospheric
469
temperature, which has to maintain small horizontal gradients and thus must adjust to
470
stay consistent with the SST changes in the tropics. Moreover, the upper-tropospheric
471
temperature responds much less to the local SST than the large-scale SST [Vecchi and
472
Soden, 2007; Vecchi et al., 2008]. In this study, the impact of inner SST change on
473
upper-tropospheric temperature is similar to that of local SST change, while the
474
impact of entire SST change is similar to that of large-scale SST change.
475
Figure 9 shows the temporal evolution of the difference in the domain-averaged
476
(average over the entire D3) upper-tropospheric temperature at 15-km height
477
(approximately the height of outflow layer) between the sensitivity experiments and
478
CTRL. Comparing with that of CTRL, as the extent of SST anomalies increases in the
479
sensitivity experiments, the change of upper-tropospheric temperature will increase
23
480
notably. Apparently the change of upper-tropospheric temperature is sensitive to the
481
extent of SST change. Compared with the experiments with the change of entire SST,
482
the defined inner SST change contributes less to the change in upper-tropospheric
483
temperature. Thereby, the increase of inner SST causes a much less increase of upper-
484
tropospheric temperature than that induced by the increase of the entire SST, which
485
results in a larger difference between sea surface air temperature and upper-
486
tropospheric temperature and thus more unstable atmospheric condition. This unstable
487
condition favors the intensification of the TC. The opposite happen as the inner SST
488
decreases. In contrast, analysis of the difference in upper-tropospheric temperature
489
between the inner-SST experiments (e.g., E90+2 or E180-2) and entire-SST
490
experiments (e.g., Eall+2 or Eall-2) shows that the outer SST plays a significant role
491
in determining upper-tropospheric temperature (Fig. 9). Increase in outer SST makes
492
the local stratification more stable and thus suppresses the intensification of TC. Note
493
that, different from the concerns in the previous studies [e.g., Vecchi and Soden, 2007;
494
Vecchi et al., 2008; Ramsay and Sobel, 2011], our concern is on “transient” response
495
of upper-tropospheric temperature to the SST on storm scale, which depends on the
496
strength of convective activity. Convective activity could affect upper-tropospheric
497
temperature by influencing not only vertical flux of temperature but also condensation
498
in the anvil clouds [Sun et al., 2013c]. This is a complex issue, which will be
499
discussed in detail in our further work.
500
This study reveals the different impacts of inner and outer SST changes on upper-
501
tropospheric temperature and explores the physical mechanisms of TC intensification.
24
502
Due to the strong convection in TC region, however, the involved mechanisms
503
discussed here may not be applicable to other tropical regions. We will address this
504
issue in our future study.
505
4.3. Validation of the opposite effects of inner and outer SST
506
To verify and validate the effects of inner and outer SST on TC intensity, a suite
507
of semi-idealized sensitivity experiments are conducted. While the imposed SST
508
anomalies are unrealistic, results of these experiments will help us understand the
509
physical and dynamic mechanisms of the SST impact on TC intensity, which is
510
important for successful forecast of TC intensity and inner-core size especially when a
511
TC comes across an ocean cold or warm pool. Previous studies suggested that TC
512
intensity undergoes a significant change as the TC passes over an ocean warm or cold
513
pool [e.g., Shay et al., 2000; Lin et al., 2005; Wu et al., 2007]. Results of the present
514
study indicate that TC intensity may start to change even before the TC passes over
515
the ocean warm or cold pool.
516
To validate our conclusions on the opposite effects of inner and outer SST, we
517
have conducted two additional sensitivity experiments to investigate the response of
518
TC intensity to a fixed change of SST. The fixed SST anomalies is either reduced or
519
increased by 2oC within a radius of 180 km in the sensitivity experiments (EF180).
520
The center of the fixed anomalies is consistent with the TC center at 0000 UTC 16
521
September 2006 when the TC is approximately steady. Figure 10 shows the temporal
522
evolution of MSLP and RMW for the fixed SST anomalies experiments (i.e., EF180-2
523
and EF180+2) and CTRL, and the differences between the fixed SST anomalies
25
524
experiments and CTRL. As a TC approaches a cold or warm pool, the impact of the
525
pool on TC intensity is related to the distance from the TC center to the pool. Before
526
1800 UTC 14 September, the TC center is about 350 km away from the fixed ocean
527
cold pool in EF180-2. Comparing EF180-2 and CTRL, due to the long distance
528
between the cold pool and the TC center, the impact of the SST anomalies on TC
529
intensity is very slight and can be neglected. As the TC approaches the fixed cold pool
530
but the cold pool is still outside the ER of the TC (approximately from 1800 UTC 14
531
to 1800 UTC 15 September), the impact of the cold pool on TC intensity becomes
532
much stronger. Consistent with the aforementioned impact of the outer SST, the fixed
533
cold pool outside the ER increases the TC intensity, but reduces the TC inner-core
534
size (Figs. 10a and 10c). However, as the TC further approaches the fixed cold pool
535
and the fixed cold pool is within the ER of the TC (after 1800 UTC 15 September),
536
the cold pool reduces the TC intensity (Fig. 10a), but increases the TC inner-core size
537
(Fig. 10c). Apparently the impact of the cold pool within the ER of the TC is
538
consistent with the aforementioned impact of the inner SST.
539
On the other hand, the impact of the fixed ocean warm pool on the TC intensity is
540
notably weaker than that of the fixed ocean cold pool (Figs. 10a and 10b). This is
541
consistent with the results of Chan et al. [2001], which suggested that the sensitivity
542
of TC intensity to SST may decrease under higher SST condition (e.g., the SST is
543
above 30°C). Despite its relatively small impact on the TC intensity, the fixed ocean
544
warm pool contributes greatly to the increase of the TC inner-core size. As the TC
545
moves closer to the fixed ocean warm pool but the warm pool is still outside the ER of
26
546
the TC (from 0000 UTC 15 to 1800 UTC 15 September), the TC inner-core size
547
increases notably due to the impact of the fixed ocean warm pool, which is consistent
548
with the aforementioned the impact of outer SST. However, as the TC gets closer to
549
the fixed ocean warm pool and the fixed warm pool is within the ER of the TC (after
550
1800 UTC 15 September), the TC inner-core size does not decrease notably as
551
expected. This may be attributed to the large radius of the ocean warm pool (i.e., 180
552
km), which covers not only the region inside the ER but also the region outside the
553
ER. Thus, the impact of the increase of inner SST on the TC inner-core size is offset
554
by that of the increase of outer SST, resulting in a steady TC inner-core size after
555
0000 UTC 16 September.
556
5. Conclusions and discussions
557
In this study, the sensitivity of TC intensity to SST change over different radial
558
extents has been investigated through a suite of sensitivity experiments using the
559
high-resolution WRF model. Note that this is a case study and the simulated TCs
560
follow a similar track in all our experiments. This is important because it allows us to
561
examine the sensitivity of simulated TC structures to SST change in a relatively
562
consistent environment. The results indicate that the SST in the TC eye region
563
contributes little to the storm intensity. Instead, the inner SST within the range 1.5-2.0
564
times the RMW, defined as the ER, plays a significant role in TC intensification and
565
can reduce the inner-core size of the TC. In contrast, the outer SST (defined as SST
566
outside the ER) contributes greatly to the growth of TC inner-core size, but reduces
567
the TC intensity. Changes in the inner SST have stronger impact on TC intensification
27
568
than changes in the entire SST, whose impact is weakened by the negative effect of
569
outer SST on TC intensification.
570
The mechanisms for the different effects of SST over different radial extents on
571
TC intensification are discussed systematically from perspectives of the surface
572
enthalpy flux and TC structures. Different from many previous studies that focus on
573
the impact study in terms of the surface enthalpy flux, we pay more attention to the
574
impact of SST on the storm structure. Our present study based on both the surface
575
enthalpy flux and TC structure analysis suggests that the change of SST over different
576
radial extents has substantially different impacts on TC intensity. There exists a
577
competition between the effects of SST change inside and outside the ER on TC
578
intensity. The possible mechanisms responsible for the different effects of SST
579
change over different radial extents on TC intensity are identified through
580
comprehensive diagnostic analysis. The results are schematically summarized in Fig.
581
11. As the inner SST increases within the ER, more surface enthalpy flux enters the
582
eyewall and contributes directly to TC intensification. Meanwhile, the impact of SST
583
on storm structure is also significant. As a result of the radial distribution of the
584
surface enthalpy flux caused by increased SST within the ER, the activity of outer
585
spiral rainbands is weakened. Such changes not only promote storm intensity but also
586
decrease the storm inner-core size. The reduced storm inner-core size leads to a
587
smaller radius of eyewall where strong latent heating is released. As a result, the
588
central pressure of TC deepens with stronger radial pressure gradient, leading to a
589
larger increase in TC intensity. In contrast, the opposite happen as the outer SST
28
590
increases outside the ER. Further analysis shows that the upper-tropospheric
591
temperature responds less to the inner SST change than to the outer SST change. The
592
increase of inner SST causes a less increase of upper-tropospheric temperature than
593
that induced by the increase of outer SST, resulting in more unstable atmospheric
594
condition which is favorable for the intensification of TC.
595
In this study, the positive effects of the inner SST increase within the ER are
596
strong enough to overcome the negative effects of the outer SST increase and result in
597
a stronger TC. Due to the negative effects of the outer SST, however, the storm
598
intensity simulated by the experiments with SST increase over the entire domain is
599
notably weaker than that in the warmer inner SST experiments. Similarly, the
600
opposite effects of inner SST and outer SST increase can also be used to explain the
601
response of TC intensity to the decrease of the SST inside and outside the ER. How
602
TC intensity responds to the change of entire SST depends on the competitive effects
603
of SST changes inside and outside the ER.
604
In addition, the competition between the effects of SST changes inside and outside
605
the ER and thus the change of TC intensity in the entire SST experiments is related to
606
the area-averaged SST. The TC intensification caused by the entire SST increase
607
under warmer SST conditions is notably weaker than that under colder SST
608
conditions. While several previous studies have suggested that TCs will become much
609
stronger when the entire SST increases, results of our study indicate that this is not
610
always the case. One possible reason for the discrepancy is because the negative
611
effect of outer SST on TC intensity is largely underestimated in previous studies.
29
612
While response of TC intensity to the increase of entire SST in the context of global
613
warming still remains controversial, results of this study will help shed light on this
614
issue and reduce the uncertainties in prediction of future change scenario.
615
Furthermore, understanding the mechanisms for the opposite effects of inner and
616
outer SST is also vital to the forecast of variations in TC intensity and inner-core size
617
when a TC comes across an ocean cold or warm pool. As the TC approaches a cold or
618
warm pool, the impact of the pool on TC intensity is related to the distance from the
619
TC center to the pool. As the ocean cold pool is outside the ER from TC center,
620
similar to the impact of inner SST anomaly, the ocean cold pool increases the TC
621
intensity but reduces the TC inner-core size; while as the cold pool is within the ER
622
from the TC center, similar to the impact of outer SST anomaly, it reduces the TC
623
intensity but increases the TC inner-core size. The opposite phenomena occur when
624
the TC comes across an ocean warm pool.
625
Acknowledgement: This work is supported by the R&D Special Fund for Public
626
Welfare Industry (Meteorology) (GYHY201306025), the National Key Basic
627
Research Program of China (2013CB956203), and the National Natural Science
628
Foundation of China (41175090), and the Jiangsu Collaborative Innovation Center for
629
Climate Change.
30
630
References
631
Beljaars, A. C. M. (1994), The parameterization of surface fluxes in large-scale
632
models under free convection, Quart. J. Roy. Meteorol. Soc., 121, 255-270,
633
doi:10.1002/qj.49712152203.
634
Bell, M. M., and M. T. Montgomery (2008), Observed structure, evolution, and
635
potential intensity of category 5 Hurricane Isabel from 12 to 14 September, Mon.
636
Weather Rev., 136, 2023–2046, doi:10.1175/2007MWR1858.1.
637
Betts, A. K. (1986), A new convective adjustment scheme. Part I: Observational and
638
theoretical
639
doi:10.1002/qj.49711247307.
basis,
Quart.
J.
Roy.
Meteorol.
Soc.,
112,
677–692,
640
Betts, A. K., and M. J. Miller (1986), A new convective adjustment scheme. Part II:
641
Single column tests using GATE wave, BOMEX, ATEX and arctic air-mass data
642
sets, Quart. J. Roy. Meteorol. Soc., 112, 693–709, doi:10.1002/qj.49711247308.
643
Black, P. G., and G. J. Holland (1995), The boundary layer of Tropical Cyclone Kerry
644
(1979),
645
0493(1995)123<2007:TBLOTC>2.0.CO;2.
Mon.
Weather
Rev.,
123,
2007–2028,
doi:10.1175/1520-
646
Chan, J. C. L., Y. Duan, and L. K. Shay (2001), Tropical cyclone intensity change
647
from a simple ocean–atmosphere coupled model, J. Atmos. Sci., 58, 154–172,
648
doi:10.1175/1520-0469(2001)058<0154:TCICFA>2.0.CO;2.
649
Chen, S., T. J. Campbell, H. Jin, S. Gaberšek, R. M. Hodur, and P. Martin (2010),
650
Effect of two-way air–sea coupling in high and low wind speed regimes, Mon.
651
Weather Rev., 138, 3579–3602, doi:10.1175/2009MWR3119.1.
652
653
Dyer, A. J., and B. B. Hicks (1970), Flux-gradient relationships in the constant flux
layer, Quart. J. Roy. Meteorol. Soc., 96, 715-721, doi:10.1002/qj.49709641012.
31
654
655
Eliassen, A. (1951), Slow thermally or frictionally controlled meridional circulation in
a circular vortex, Astrophys. Norv., 5,19-60.
656
Emanuel, K. A. (1986), An air-sea interaction theory for tropical cyclone: Part I:
657
Steady state maintenance, J. Atmos. Sci., 43, 585–604, doi:10.1175/1520-
658
0469(1986)043<0585:AASITF>2.0.CO;2.
659
660
Emanuel, K. A. (1991), The theory of hurricanes, Annual Review of Fluid Mechanics,
23, 179–196, doi:10.1146/annurev.fl.23.010191.001143.
661
Emanuel, K. A. (1995), Sensitivity of tropical cyclones to surface exchange
662
coefficients and a revised steady-state model incorporating eye dynamics, J.
663
Atmos.
664
0469(1995)052<3969:SOTCTS>2.0.CO;2.
Sci.,
52,
3969–3976,
doi:10.1175/1520-
665
Fierro, A. O., R. F. Rogers, and F. D. Marks (2009), The Impact of Horizontal Grid
666
Spacing on the Microphysical and Kinematic Structures of Strong Tropical
667
Cyclones Simulated with the WRF-ARW Model, Mon. Weather Rev., 137, 3717–
668
3743, doi:10.1175/2009MWR2946.1.
669
Gentry, M. S., and G. M. Lackmann (2010), Sensitivity of simulated tropical cyclone
670
structure and intensity to horizontal resolution, Mon. Weather Rev., 138, 688–
671
704, doi:10.1175/2009MWR2976.1.
672
Hack, J. J., and W. H. Schuert (1986), Nonlinear response of atmospheric vortices to
673
heating by organized cumulus convection, J. Atmos. Sci., 43, 1559–1573,
674
doi:10.1175/1520-0469(1986)043<1559:NROAVT>2.0.CO;2.
675
Hill, K. A., and G. M. Lackmann (2009), Influence of environmental humidity on
676
tropical
677
doi:10.1175/2009MWR2679.1.
cyclone
size,
Mon.
Weather
Rev.,
137,
3294–3315,
32
678
Holland, G. J. (1997), The maximum potential intensity of tropical cyclones, J. Atmos.
679
Sci.,
680
0469(1997)054<2519:TMPIOT>2.0.CO;2.
681
54,
2519–2541,
doi:10.1175/1520-
Holland, G. J., and R. T. Merrill (1984), On the dynamics of tropical cyclone
682
structural
683
doi:10.1002/qj.49711046510.
changes,
Quart.
J.
Roy.
Meteorol.
Soc.,
110,
723–745,
684
Hong, S. Y., Y. Noh, and J. Dudhia (2006), A new vertical diffusion package with an
685
explicit treatment of entrainment processes, Mon. Weather Rev., 134, 2318-2341,
686
doi:10.1175/MWR3199.1.
687
Houze, R. A., Jr., S. S. Chen, W.-C. Lee, et al. (2006), The hurricane rainband and
688
intensity change experiment: Observations and modeling of Hurricanes Katrina,
689
Ophelia, and Rita, Bull. Amer. Meteor. Soc., 87, 1503-1521, doi:10.1175/BAMS-
690
87-11-1503.
691
Houze, R. A., Jr., S. S. Chen, B. F. Smull, et al. (2007), Hurricane intensity and
692
eyewall replacement, Science, 315, 1235-1239, doi:10.1126/science-1135650.
693
Janjić, Z. I. (1994), The step-mountain eta coordinate model: Further developments of
694
the convection, viscous sublayer and turbulence closure schemes, Mon. Weather
695
Rev.,
696
0493(1994)122<0927:TSMECM>2.0.CO;2.
122,
927–945,
doi:10.1175/1520-
697
Janjić, Z. I. (2000), Comments on “Development and evaluation of a convection
698
scheme for use in climate models”, J. Atmos. Sci., 57, 3686, doi:10.1175/1520-
699
0469(2000)057<3686:CODAEO>2.0.CO;2.
700
Lee, C.-Y., and S. S. Chen (2012), Symmetric and asymmetric structures of hurricane
701
boundary layer in coupled atmosphere–wave–ocean models and observations, J.
702
Atmos. Sci., 69, 3576–3594, doi:10.1175/JAS-D-12-046.1.
33
703
Lin, I-I., C.-C. Wu, K. A. Emanuel, I-H. Lee, C.-R. Wu, and I.-F. Pun (2005), The
704
interaction of Supertyphoon Maemi (2003) with a warm ocean eddy, Mon.
705
Weather Rev., 133, 2635–2649, doi:10.1175/MWR3005.1
706
Lin, Y.-L., R. D. Farley, and H. D. Orville (1983), Bulk parameterization of the snow
707
field in a cloud model. J. Climate Appl. Meteorol., 22, 1065-1092,
708
doi:10.1175/1520-0450(1983)022<1065:BPOTSF>2.0.CO;2.
709
Malkus, J. S., and H. Riehl (1960), On the dynamics and energy transformations in
710
steady-state
711
3490.1960.tb01279.x.
hurricane,
Tellus,
12,
1–20,
doi:10.1111/j.2153-
712
Miyamoto, Y., and T. Takemi (2010), An effective radius of the sea surface enthalpy
713
flux for the maintenance of a tropical cyclone, Atmos. Sci. Lett., 11, 278–282,
714
doi:10.1002/asl.292.
715
Paulson, C. A. (1970), The mathematical representation of wind speed and
716
temperature profiles in the unstable atmospheric surface layer, J. Appl. Meteorol.,
717
9, 857-861, doi:10.1175/1520-0450(1970)009<0857:TMROWS>2.0.CO;2.
718
Persing, J., and M. T. Montgomery (2005), Is environmental CAPE important in the
719
determination of maximum possible hurricane intensity?, J. Atmos. Sci., 62, 542–
720
550, doi:10.1175/JAS-3370.1.
721
Ramsay, H. A., and A. H. Sobel (2011), Effects of relative and absolute sea surface
722
temperature on tropical cyclone potential intensity using a single-column model,
723
J. Climate, 24, 183–193, doi:10.1175/2010JCLI3690.1.
724
Rotunno, R., and K. A. Emanuel (1987), An air–sea interaction theory for tropical
725
cyclones. Part II: Evolutionary study using a nonhydrostatic axisymmetric
726
numerical
727
0469(1987)044<0542:AAITFT>2.0.CO;2.
model,
J.
Atmos.
Sci.,
44,
542–561,
doi:10.1175/1520-
34
728
Shay, L. K., G. J. Goni, and P. G. Black (2000), Effects of a warm oceanic feature on
729
Hurricane Opal, Mon. Weather Rev., 128, 1366–1383, doi:10.1175/1520-
730
0493(2000)128<1366:EOAWOF>2.0.CO;2
731
Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, M. G. Duda, X.
732
Y. Huang, W. Wang, and J. G. Powers, 2008: A Description of the Advanced
733
Research WRF Version 3, NCAR Tech. Note NCAR/TN-475+STR, 113 pp.
734
[Available online at http://www.mmm.ucar.edu/people/skamarock/.]
735
Stern, D. P., and D. S. Nolan (2009), Reexamining the vertical structure of tangential
736
winds in tropical cyclones: Observations and theory, J. Atmos. Sci., 66, 3579–
737
3600, doi:10.1175/2009JAS2916.1.
738
Sun, Y., Z. Zhong, and Y. Wang (2012), Characteristics of asymmetric flow of
739
tropical cyclone Shanshan (2006) during its turning and intensification period,
740
Acta Meteorol. Sinica, 26, 147-162, doi:10.1007/s13351-012-0202-4.
741
Sun, Y., Z. Zhong, Y. Ha, Y. Wang, and X. D. Wang (2013a), The dynamic and
742
thermodynamic perspectives of relative and absolute sea surface temperature on
743
tropical cyclone intensity, Acta Meteorol. Sinica, 27, 40-49, doi:10.1007/s13351-
744
013-0105-z.
745
Sun, Y., Y. Lan, Z. Zhong, Y. Hu, and Y. Ha (2013b), Dependence of model
746
convergence on horizontal resolution and convective parameterization in
747
simulations of a tropical cyclone at grey-zone resolutions, J. Geophys. Res., 118,
748
7715–7732, doi:10.1002/jgrd.50606.
749
Sun, Y., Z. Zhong, W. Lu, and Y. Hu (2013c), Why are tropical cyclone tracks over
750
the western North Pacific sensitive to the cumulus parameterization scheme in
751
regional climate modeling?, A case study for Megi (2010), Mon. Wea. Rev.,
752
doi:10.1175/MWR-D-13-00232.1, in press.
35
753
Vecchi, G. A., and B. J. Soden (2007), Effect of remote sea surface temperature
754
change on tropical cyclone potential intensity, Nature, 450, 1066-1070,
755
doi:10.1038/nature06423.
756
757
Vecchi, G. A., K. L. Swanson, and B. J. Soden (2008), Whither hurricane activity?,
Science, 322, 687-689, doi:10.1126/science.1164396.
758
Vigh, J. L., and W. H. Schubert (2009), Rapid development of the tropical cyclone
759
warm core, Mon. Weather Rev., 66, 3335-3350, doi:10.1175/2009JAS3092.1.
760
Wang, Y. (2007), A multiply nested, movable mesh, fully compressible,
761
nonhydrostatic tropical cyclone model—TCM4: Model description and
762
development of asymmetries without explicit asymmetric forcing, Meteorol.
763
Atmos. Phys., 97, 93-116, doi:10.1007/s00703-006-0246-z.
764
765
Wang, Y. (2009), How do outer spiral rainbands affect tropical cyclone structure and
intensity?, J. Atmos. Sci., 66, 1250–1273, doi:10.1175/2008JAS2737.1.
766
Wang, Y., and C.-C. Wu (2004), Current understanding of tropical cyclone structure
767
and intensity changes – a review, Meteorol. Atmos. Phys., 87, 257-278,
768
doi:10.1007/s00703-003-0055-6.
769
Wang, Y., and J. Xu (2010), Energy production, frictional dissipation, and maximum
770
intensity of a numerically simulated tropical cyclone, J. Atmos. Sci., 67, 97-116,
771
doi:10.1175/2009JAS3143.1.
772
Webb, E. K. (1970), Profile relationships: The log-linear range, and extension to
773
strong
774
doi:10.1002/qj.49709640708.
775
776
stability,
Quart.
J.
Roy.
Meteorol.
Soc.,
96,
67-90,
Wu, C.-C., C.-Y. Lee, and I-I. Lin (2007), The effect of the ocean eddy on tropical
cyclone intensity, J. Atmos. Sci., 64, 3562–3578, doi:10.1175/JAS4051.1
36
777
Xu, J., and Y. Wang (2010), Sensitivity of tropical cyclone inner-core size and
778
intensity to the radial distribution of surface entropy flux, J. Atmos. Sci., 67,
779
1831–1852, doi:10.1175/2010JAS3387.1.
780
37
781
Figure Captions
782
Figure 1. (a) Triply-nested model domains; (b) the observed and simulated track of
783
Shanshan (2006) at 6-h intervals (b); and (c) the schematic of extents and
784
radii of SST change in sensitivity experiments.
785
Figure 2. The temporal evolution of MSLP in sensitivity experiments.
786
Figure 3. The Temporal evolution of maximum wind speed at 10 m in sensitivity
787
experiments.
788
Figure 4. The temporal evolution of RMW at 10 m in sensitivity experiments.
789
Figure 5. (a) The temporal evolutions of area-integrated kinetic energy at 10 m; (b)
790
area-integrated the surface enthalpy flux (calculated within a radius of 200
791
km in some representative experiments).
792
Figure 6. Distributions of the surface enthalpy flux difference (W m-2) and max-
793
reflectivity difference (dBZ) between results of sensitive experiments and
794
CTRL. (a) Surface enthalpy flux difference between inner SST
795
experiments E180-2 and CTRL; (b) Max-reflectivity difference between
796
inner SST experiments E180-2 and CTRL; (c) Surface enthalpy flux
797
difference between E90+2 and CTRL; (d) Max-reflectivity difference
798
between E90+2 and CTRL. (All calculation is done over the mature stage
799
from 1800UTC on 15 September to 0600UTC on 16 September).
800
Figure 7. As in Fig. 6, but for the difference between E180-2 and Eall-2 (a, b), and
801
the difference between E90+2 and Eall+2 (c, d).
802
Figure 8. Azimuthal- and time-averaged cross sections of the model-simulated latent
803
heat rate (°C h-1; shaded) and radial pressure gradient (10-2 Pa m-1;
804
contoured) in the TC mature stage.
38
805
Figure 9. The temporal evolution of the difference in the averaged upper-tropospheric
806
temperature (°C) at 15 km in the entire moving domain (D3) between
807
sensitivity experiments and CTRL.
808
Figure 10. Temporal evolutions of the MSLP and RMW for the fixed SST anomalies
809
experiments (i.e., EF180-2 and EF180+2) and CTRL, and the differences
810
between the fixed SST anomalies experiments and CTRL.
811
Figure 11. Schematic diagram summarizing the possible mechanisms responsible for
812
different effects of SST change over different radial extent on TC
813
intensity. The thin dashed (thin solid) line indicates the impact of SST on
814
TC intensity in view of the surface enthalpy flux (storm structure).
815
39
816
Table 1. Summary of the control experiment and sensitivity experiments. Variable r
817
is the radial distance from the storm center; ΔSST is the artificial
818
modification of SST. MSLPmin and Vmax are the lifetime minimum MSLP
819
and lifetime maximum wind speed at 10 m for each simulation, respectively.
820
RMWave is the RMW at the 10 m averaged in the last 36-h simulation (from
821
0000 UTC on 15 September to 1200 UTC on 16 September) when the
822
difference between sensitivity experiment and CTRL is significant.
823
Exp.
Modification to the underlying SST
MSLPmin(hPa)
Vmax(m s-1)
RMWave (km)
CTRL
Control experiment
920.6
53.3
55.0
E45-2
ΔSST = -2 for r≤30 km, and linearly increases to 0℃ at r=45 km
923.4
51.6
62.4
E90-2
ΔSST = -2 for r≤75 km, and linearly increases to 0℃ at r=90 km
939.6
48.8
82.1
E180-2
ΔSST = -2 for r≤165 km, and linearly increases to 0℃ at r=180 km
950.7
45.6
108.8
E270-2
ΔSST = -2 for r≤255 km, and linearly increases to 0℃ at r=270 km
934.0
45.7
38.7
Eall-2
ΔSST = -2℃ for all the simulation domain
941.3
45.9
45.0
E45+2
ΔSST = 2 for r≤30 km, and linearly decreases to 0℃ at r=45 km
900.3
58.9
47.8
E90+2
ΔSST = 2 for r≤75 km, and linearly decreases to 0℃ at r=90 km
889.3
61.8
47.1
E180+2
ΔSST = 2 for r≤165 km, and linearly decreases to 0℃ at r=180 km
907.9
59.2
60.8
E270+2
ΔSST = 2 for r≤255 km, and linearly decreases to 0℃ at r=270 km
915.6
54.6
83.1
Eall+2
ΔSST = 2℃ for all the simulation domain
905.6
52.8
82.8
824
40
825
826
827
828
829
830
831
Figure 1. (a) Triply-nested model domains; (b) the observed and simulated track of
832
Shanshan (2006) at 6-h intervals (b); and (c) the schematic of extents and radii of SST
833
change in sensitivity experiments.
834
835
836
837
838
839
840
841
842
843
844
41
845
846
847
848
Figure 2. Temporal evolutions of MSLP in the sensitivity experiments.
849
850
851
852
853
854
855
42
856
857
858
859
860
861
862
Figure 3. Temporal evolutions of maximum wind speed at 10 m in the sensitivity
863
experiments.
864
865
866
867
868
869
43
870
871
872
873
874
875
876
877
Figure 4. Temporal evolutions of RMW at 10 m in the sensitivity experiments.
878
879
880
881
882
44
883
884
885
886
887
888
Figure 5. (a) Temporal evolution of area-integrated kinetic energy at 10 m; (b) area-
889
integrated the surface enthalpy flux (calculated within a radius of 200 km in some
890
representative experiments).
891
892
45
893
894
895
Figure 6. Distributions of the surface enthalpy flux difference (W m-2) and max-
896
reflectivity difference (dBZ) between results of sensitive experiments and CTRL. (a)
897
Surface enthalpy flux difference between inner SST experiments E180-2 and CTRL;
898
(b) Max-reflectivity difference between inner SST experiments E180-2 and CTRL; (c)
899
Surface enthalpy flux difference between E90+2 and CTRL; (d) Max-reflectivity
900
difference between E90+2 and CTRL. (All calculation is done over the mature stage
901
from 1800UTC on 15 September to 0600UTC on 16 September).
902
903
904
46
905
906
907
908
909
910
Figure 7. As in Fig. 6, but for the difference between E180-2 and Eall-2 (a, b), and
911
the difference between E90+2 and Eall+2 (c, d).
912
913
47
914
915
916
Figure 8. Azimuthal- and time-averaged cross sections of the model-simulated latent
917
heat rate (°C h-1; shaded) and radial pressure gradient (10-2 Pa m-1; contoured) in the
918
TC mature stage.
919
920
921
48
922
923
924
925
Figure 9. Temporal evolutions of the difference in the averaged upper-tropospheric
926
temperature (°C) at 15 km in the entire moving domain (D3) between sensitivity
927
experiments and CTRL.
928
929
49
930
931
932
Figure 10. Temporal evolutions of the MSLP and RMW for the fixed SST anomalies
933
experiments (i.e., EF180-2 and EF180+2) and CTRL, and the differences between the
934
fixed SST anomalies experiments and CTRL.
935
50
936
(a) Positive effect of inner SST on TC intensity
(b) Negative effect of outer SST on TC intensity
Increase of SST
outside ER
Increase of SST
within ER
More SEF input
into eyewall
Less SEF input
into outer spiral
rainbands
Less increase of
upper-tropospheric
temperature
937
938
More SEF input
into outer spiral
rainbands
More increase of
upper-tropospheric
temperature
Weaker convections
near the eyewall
and stronger outer
spiral rainbands
Stronger convections
near the eyewall
and weaker outer
spiral rainbands
Smaller storm innercore size
Less SEF input
into eyewall
More unstable
atmospheric
condition
Larger storm innercore size
Smaller radius at which
latent heating occurs
Larger radius at which
latent heating occurs
Larger radial gradients
near eyewall and central
pressure fall
Smaller radial gradients
near eyewall and central
pressure fall
Stronger TC intensity
Less unstable
atmospheric
condition
Weaker TC intensity
939
Figure 11. Schematic diagram summarizing the possible mechanisms responsible for
940
different effects of SST change over different radial extent on TC intensity. The thin
941
dashed (thin solid) line indicates the impact of SST on TC intensity in view of the
942
surface enthalpy flux (storm structure).
51