3950 MONTHLY WEATHER REVIEW VOLUME 141

Effects of Asymmetric SST Distribution on Straight-Moving Typhoon Ewiniar (2006) and Recurving Typhoon Maemi (2003)

YUMI CHOI,KYUNG-SOOK YUN, AND KYUNG-JA HA Division of Earth Environmental System, Pusan National University, Busan, South

KWANG-YUL KIM School of Earth and Environmental Sciences, National University, Seoul,

SOON-JO YOON Water Resources Operations Center, Korea Water Resources Corporation, Daejeon, South Korea

JOHNNY C. L. CHAN Guy Carpenter Asia-Pacific Climate Impact Centre, School of Energy and Environment, City University of Hong Kong, Hong Kong,

(Manuscript received 26 July 2012, in final form 12 June 2013)

ABSTRACT

The effects of asymmetric sea surface temperature (SST) distribution on the tropical cyclone (TC) motion around East Asia have been examined using the Weather Research and Forecasting Model for the straight- moving Typhoon Ewiniar (2006) and recurving Typhoon Maemi (2003). The SST–TC motion relationships associated with the two different TCs and the physical mechanism of recurvature are investigated in the context of the potential vorticity tendency framework. A zonally asymmetric SST distribution alters the TC translating direction and speed, which is ascribable to the interaction between a TC and the environmental current associated with asymmetric SST forcing. A north–south SST gradient has an insignificant role in the TC motion. It is noted that the straight-moving (i.e., northward moving) TC deflects toward the region of warmer SST when SST is zonally asymmetric. A contribution of the horizontal advection including asym- metric flow induced by asymmetric forcing is dominant for the deflection. The recurving TC reveals north- eastward acceleration and deceleration after the recurvature point in the western warming (WW) and eastern warming (EW) experiments, respectively. When it comes to a strong southerly vertical wind shear under the recurvature condition, diabatic heating can be a significant physical process associated with the downward motion over the region of upshear right. The enhanced (reduced) southwesterly flow effectively produces the acceleration (deceleration) of northeastward movement in WW (EW) after recurvature.

1. Introduction as the beta drift. This mechanism, which depends on vortex structure and latitude, refers to the northwest- During the past several decades, much effort has been ward TC motion in the Northern Hemisphere on a beta made to understand the tropical cyclone (TC) motion. plane in the absence of an environmental steering flow The complexity of the TC motion derives from a wide (Holland 1983; Chan and Williams 1987; Fiorino and variety of external, internal, and interactive dynamical Elsberry 1989); this motion is determined by the beta- forcing (Wang et al. 1998; Chan 2005). The most fun- induced secondary steering flow over the vortex cen- damental process of the interactive dynamics is known ter, referred to as the ventilation flow. The secondary steering flow can be modulated by the internal dynamic factors and the interaction between a TC and external Corresponding author address: Prof. Kyung-Ja Ha, Depart- ment of Atmospheric Sciences, Pusan National University, Busan forcing (Wang et al. 1998; Chan 2005). 609-735, South Korea. Sea surface temperature (SST) is an important factor E-mail: [email protected] affecting internal and external TC dynamics. The role of

DOI: 10.1175/MWR-D-12-00207.1

Ó 2013 American Meteorological Society Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3951

SST in the TC genesis and intensity has been widely diabatic heating (DH) (Wang and Holland 1996a,b,c). investigated (Chang 1979; Tuleya and Kurihara 1982; A potential vorticity tendency (PVT) diagnostic ap- Schade 2000; Michaels et al. 2006). In addition, ther- proach is useful in understanding the physical mecha- modynamical influences of SST also modulate the TC nisms of baroclinic and diabatic TC motion (Wu and motion. Wu et al. (2005) investigated how the TC- Wang 2000, 2001; Chan et al. 2002; Wong and Chan induced SST anomaly affects the TC motion. They de- 2006). These studies suggested that TCs tend to move signed symmetric and asymmetric SST anomalies with into the region of the maximum wavenumber-1 (WN1) respect to a TC center to focus on the impacts of air–sea PVT. We, therefore, adopt the PVT approach to un- interaction on the TC motion. A large-scale asymmetric derstand the role of asymmetric SST distributions, SST distribution may also affect the TC motion in a which are responsible for various TC motions in re- different way. Chang and Madala (1980) showed the alistic environments. influence of various SST distributions with a mean flow The objective of this study is to investigate the effects on the behavior of a translating TC. They showed that of asymmetric SST distributions upon straight-moving TCs tend to move into regions of warmer SST and a fa- and recurving TC motions in real environments using vorable condition for TC deflection is established when the PVT approach. Experiments with asymmetric SST the SST gradient is perpendicular to the mean flow. The distributions are performed with the Weather Research SST distributions affect the TC motion by altering total and Forecasting Model (WRF), version 3.2. Section 2 surface friction and heat flux exchange. Yun et al. (2012) describes the model experiments, real TC cases, and studied a northeastward-moving TC to demonstrate that methodology used in this study. How an asymmetric SST the TC motion is sensitive to the SST magnitude and distribution affects the straight-moving and recurving TC gradient. They suggested that an eastward SST increase motions is examined in sections 3 and 4, respectively. produces a greater eastward TC deflection than does a Section 5 contains our major findings and discussion. meridional SST gradient or variation of SST magnitude. While most studies used idealized numerical experi- 2. Model experiment and methodology ments to understand the dynamics of the TC motion, real a. Model experiment TCs experience further complicated environments which result in intricate interactions. Although an attempt has WRF version 3.2 is used to simulate Super Typhoons been made to understand the effects of SST gradient on Ewiniar (2006) and Maemi (2003); details of these TCs Typhoon Maemi (2003) in a realistic environment (Yun are included in the following subsection. The model et al. 2012), a single northeastward-moving TC case may domain consists of 240 3 240 grid points with a uniform be insufficient to verify and generalize the results. It is horizontal resolution of 12 km and 27 vertical levels with well known that TCs exhibit a wide variety of motions a top at 50 hPa. Each experiment is conducted as a 72-h according to such environmental conditions as baro- integration with a 60-s time step. The Kain–Fritsch clinicity, vertical wind shear, SST gradients, and the scheme (Kain and Fritsch 1993) is used for cumulus position of the subtropical ridge. For example, as a TC parameterization. Details of the model physics are docu- moves to a higher latitude, the energy source of the cy- mented in Yun et al. (2012). The National Centers for clones comes gradually more from baroclinic processes Environmental Prediction (NCEP)Final(FNL)Opera- than latent heat release (Jones et al. 2003). Further, tional Global Analysis 6-hourly data with a 1.0831.08 Holland and Wang (1995) showed that potential for resolution are used as initial and boundary conditions. The recurvature depends on the initial location of a TC best track dataset is obtained from the Joint Typhoon relative to an idealized subtropical ridge with a mid- Warning Center (JTWC). latitude westerly trough. The environmental wind field A simple bogussing scheme used in the fifth-generation observed in the northwest region of a TC is also im- Pennsylvania State University–National Center for portant for determining the TC recurvature (Hodanish Atmospheric Research Mesoscale Model (MM5; Davis and Gray 1993). A comparison between straight-moving and Low-Nam 2001) is employed to insert a bogus and recurving TCs, therefore, is critical for understand- vortex into a background field for improving simulation ing the physical processes that contribute to various TC performance. Briefly, the bogussing procedure consists motions in nature. In the present study, Super Typhoons of identifying an initial TC vortex as compared with the Ewiniar (2006) and Maemi (2003) are chosen respectively best track data, and removing the vortex from the first as a representative of straight-moving and recurving TCs. guess field. Finally, the bogus vortex generated by using Baroclinic TC motion is controlled by various phys- the simple Rankine vortex (Davis and Low-Nam 2001) ical processes such as asymmetric flow within the vor- replaces the removed vortex in the initial guess field. tex core region, vertical shear, vortex structure, and To generate the vortex, which is an improvement over

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3952 MONTHLY WEATHER REVIEW VOLUME 141

TABLE 1. An experimental design of the control run and runs with Fig. 2, Typhoon Maemi is a typical recurving TC de- asymmetric SST distributions. veloped in September. Westward-moving TC is nor- Expt Change in SST condition mally referred to as straight-moving TC. In this study, however, northward-moving Typhoon Ewiniar is de- CTL (control run) Initial condition of SST obtained from FNL data fined as the straight-moving TC to differentiate the two SW (southern warming) SST increased from north to south different TCs in terms of their recurvature. Accordingly, WW (western warming) SST increased from east to west these storms are selected as respective representatives EW (eastern warming) SST increased from west to east of straight-moving and recurving TCs for this study. The western North Pacific subtropical high is a crucial com- ponent in the East Asian weather system (Yun et al. the initial NCEP vortex, we choose 300 km (400 km) as 2008; Lee et al. 2013). Previous studies investigated in- the radius of Typhoon Ewiniar (Typhoon Maemi) with teractions between the synoptic-scale circulations (west- a radius of maximum wind of 27 780 m (46 300 m) and erly trough and subtropical ridge) and TCs with respect 2 2 amaximumwindspeedof60ms 1 (45 m s 1)basedon to the recurvature process (Hodanish and Gray 1993; the observation records. Holland and Wang 1995). Eastward-retreating sub- A baseline control (CTL) experiment (Figs. 3a and 7a) tropical ridge and an approaching westerly trough are is set with an initial SST condition obtained from the FNL dominant environmental features during the recurvature data. SST distributions for the experiments are either (Li and Chan 1999). A comparison of the geopotential zonal or meridional (as shown in Figs. 3b–d and 7b–d). heights at 500 hPa between Typhoon Ewiniar (Fig. 1) and For example, SST increases southward to study the Typhoon Maemi (Fig. 2) demonstrates that the straight- effects of meridional asymmetry of SST on the TC mo- moving and recurving TC motions mainly result from tion in our southern warming (SW) experiment. Other different large-scale circulations. Thus, responses and experiments in this study are labeled as western warming their physical mechanisms are expected to vary for vari- (WW) and eastern warming (EW; see Table 1). The SST ous asymmetric SST distributions between the two cases. 2 gradient is approximately 18C(480km) 1. To minimize Yun et al. (2012) focused on the motion of Typhoon the discrepancy of the initial TC intensity among the Maemi for the period of 0000 UTC 11 September– experiments, SST at the initial location of a TC is set to 0000 UTC 13 September, during which it showed a be similar to that of CTL. The TC center is determined straight northeastward motion. To focus on the recur- by the minimum sea level pressure. Experimental designs vature process, however, we investigate the following are identical for both the typhoons. three consecutive 12-h periods during the 72-h inte- gration interval: before (from 1200 UTC 10 September b. Case selection: Typhoon Ewiniar (2006) to 0000 UTC 11 September), during (from 0000 UTC and Typhoon Maemi (2003) 11 September to 1200 UTC 11 September), and after (from A major problem in TC forecasting is the deter- 1200 UTC 11 September to 0000 UTC 12 September) mination of the recurvature. Different TC motions recurvature. In comparison, the following two consecu- could be expected, depending on the SST distribution tive 12-h periods are investigated for Typhoon Ewiniar— and environmental flow. A comparison of straight- 1200UTC8July–0000UTC9Julyand0000UTC moving and recurving TCs is thus useful for under- 9 July–1200 UTC 9 July—during the 72-h integration standing the physical processes contributing to diverse interval. The period of 1200 UTC 9 July–0000 UTC TC motions in nature. To determine the influence of 10 July is not considered in this study to exclude the SST distribution on TC passing through the western land influence; TCs could drift toward the land asso- North Pacific and affecting the Korean Peninsula, two ciated with roughness length over land (Wong and TCs are considered in this study: Typhoon Ewiniar Chan 2006). To remove temporal fluctuations and fo- (2006), which destroyed more than 600 homes through cus on the tendency of the TC motion, a 12- or 24-h severe flooding and claimed 62 lives in Korea; and composite analysis is employed. Typhoon Maemi (2003), the strongest typhoon to hit the c. Potential vorticity tendency framework Korean peninsula in nearly 100 years of the recorded history (Ye 2004). The PVT diagnostic approach is useful in under- Typhoon Ewiniar exhibits a straight northward path standing the physical mechanisms of baroclinic and dia- during the 3-day integration in the present study (Fig. 1). batic TC motion. Previous research has demonstrated The JTWC defines the tropical cyclone recurvature as that TCs tend to migrate to the region of maximum WN1 ‘‘the turning of a tropical cyclone from an initial path PVT (Wu and Wang 2000, 2001; Chan et al. 2002; Wong west and poleward to east and poleward.’’ As shown in and Chan 2006; Yun et al. 2012).

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3953

FIG. 1. The best track of straight-moving Typhoon Ewiniar (black solid line) obtained from JTWC and 500-hPa geopotential height (gray solid line, m) from FNL data at 0000 UTC 9 Jul 2006. The 3-day integration is conducted from 0000 UTC 7 Jul to 0000 UTC 10 Jul 2006.

In this study, vertically averaged PVT is used because friction is assumed negligible since the vertical average environmental flow can vary significantly with height. above the boundary layer (1.0 $ s $ 0.9; ;1010–914 hPa) According to Wong and Chan (2006), the PVT analysis is used. Wu and Wang (2000) suggested that the contri- is conducted in the lower layer (LL: 0.9 $ s $ 0.55; butions of various physical processes could be determined ;914–578 hPa) and upper layer (UL: 0.55 $ s $ 0.25; by investigating their effects on the WN1 component of ;578–290 hPa). A baroclinic vortex moves approxi- PVT. Adopting this point of view, a tendency of the TC mately with the environmental flow in the low- to mid- motion can be attributed to the most dominant term if the tropospheric layer (Holland and Wang 1995; Wang and most dominant physical process for PVT is determined. Holland 1996b,c; Wang et al. 1998). As demonstrated We follow this concept to examine the physical mecha- by Yun et al. (2012), the WN1 component of the PVT nisms of both straight-moving and recurving TCs. Total based on a LL average is fairly consistent with the move- PVT is primarily controlled by three components: hori- ment of a TC. However, changes in environmental flow zontal advection (HA), vertical advection (VA), and DH. associated with the recurvature are more evident in upper- The HA term includes beta-induced circulation, environ- tropospheric flow than in lower-tropospheric flow (George mental steering flow, and heating-induced steering flow. A and Gray 1977; Hodanish and Gray 1993; Li and Chan dominant term is chosen among the three by considering 1999). In addition, asymmetric flow which affects TC both the magnitude and location of maximum total PVT. motion depends on large-scale circulation (Chan and Cheung 1998). Therefore, different vertical averages 3. Effects of asymmetric SST distribution for are used for the straight-moving typhoon Ewiniar (LL) a straight-moving case (Ewiniar 2006) and recurving typhoon Maemi (UL). The PVT arises from the horizontal and vertical PV To understand the effects of asymmetric SST dis- advections, DH, and friction (the appendix). In this study, tribution under a less-complicated environment, a

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3954 MONTHLY WEATHER REVIEW VOLUME 141

FIG. 2. As in Fig. 1, but for recurving Typhoon Maemi and 500-hPa geopotential height (m) at 1800 UTC 10 Sep 2003 (before recurvature). The 3-day integration is conducted from 0000 UTC 9 Sep to 0000 UTC 12 Sep 2003. straight-moving TC is first investigated. As shown in present study, produced a larger eastward deflection for Fig. 3a, the track of the TC in CTL is similar to the best anortheastward-movingTC,owingtothesouthwest- track in Fig. 1, which shows a northward TC trans- ward tilt of the vortex axis and the resulting vertical lation. During the mature stage (from 0000 UTC 7 July southeasterly wind shear. However, a significant track to 0000 UTC 10 July), the TC track in the SW experi- change was realized only in an eastward SST increase. ment (Fig. 3b) is similar to that of CTL, which has a This discrepancy may arise from dominant baroclinic similar SST distribution to SW but a steeper SST gra- processes in higher latitudes. Jones et al. (2003) sug- dient. Asymmetric SST distributions in the zonal di- gested that the main energy source for the cyclone is rection shift the TC westward or eastward (Figs. 3c,d). baroclinic processes rather than latent heat release It is noted that a northward-moving TC deflects toward during a TC translation to higher latitudes. Thus, a the region of warmer SST. Larger deflections occur in northeastward-moving TC in higher latitude is mainly WW and EW than in SW, which indicates that SST affected by the vertical wind shear, in contrast to gradient perpendicular to TC translation offers more Ewiniar in the present study. favorable conditions for TC deflection. This result is To understand the physics of the SST distribution- consistent with earlier studies (Chang and Madala 1980; induced TC motion, we compare both the PVT analysis Yun et al. 2012). Chang and Madala (1980) suggested that and the asymmetric flow in LL among WW, CTL, and TCs tend to move toward the regions of warmer SST EW, which reveal northwestward, northward, and north- when the SST gradient is perpendicular to the idealized eastward motion, respectively. As shown in Fig. 4, TC easterly current (i.e., a westward-moving TC) due to the heading direction is largely consistent with the location asymmetries associated with heat exchange, enhanced of maximum total PVT averaged over a 24-h period. evaporation, and friction. Yun et al. (2012) showed that The location of maximum HA corresponds reasonably an eastward SST increase, which is the same as EW in the to that of the total PVT (Figs. 4a–f). For example, the

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3955

FIG. 3. The simulated tracks from (a) CTL (black solid line), (b) SW, (c) WW, and (d) EW (gray solid lines) of Typhoon Ewiniar. The initial forecast time is at 0000 UTC 7 Jul 2006 (0 h). The SST distributions (gray dotted line) are asymmetric in zonal and meridional directions for the each experiment. The contour interval is 1.0 (8C). maximum total PVT in EW is located northeastward (i.e., ventilation flow) and the advection of asymmetric and agrees well with the TC heading direction. The lo- PV by symmetric flow (Chan et al. 2002). The steering cation and magnitude of the maximum HA are more flow defined over the inner core includes environmental consistent with those of the total PVT than VA or DH steering flow and secondary steering flow generated by (Figs. 4c,f,i,l). The VA and DH terms, which have max- internal dynamics and interaction between a TC and ima in opposite locations, tend to cancel each other. We, external forcing (Wang et al. 1998). In the present study, therefore, infer that the contribution of HA to a TC the effect of steering flow is investigated by examining motion dominates. The contribution of HA becomes asymmetric flow induced by both large-scale circulation more significant as a TC moves to higher latitudes in all and asymmetric SST forcing. An asymmetric flow was the experiments (figure not shown). Chan et al. (2002) obtained by subtracting a symmetric wind field from found that the main contributor to the total PVT is HA a total wind field (Wong and Chan 2006; Yun et al. for straight-moving TCs. 2012). The HA term consists of both the advection of sym- Southwesterly asymmetric flow over the TC core re- metric PV by asymmetric flow, which includes envi- gion in EW (Figs. 5e,f) effectively provides a better ronmental steering flow and beta-induced circulation condition for eastward deflection compared to CTL

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3956 MONTHLY WEATHER REVIEW VOLUME 141

FIG. 4. Wavenumber-1 components of (a)–(c) total potential vorticity tendency, (d)–(f) horizontal advection, (g)–(i) vertical advection, and (j)–(l) diabatic heating, which are 24-h time composites of lower-level (0.9 $ s $ 0.55) averages during 36–60 h in (left) WW, (middle) CTL, and (right) EW of Typhoon Ewiniar. Positive values are shaded. The contour interval is 3.0 in (a)–(c) and (j)–(l), 4.0 in (d)–(f), 1.0 in 2 2 2 (g), 0.5 in (h), and 2.0 in (i) [106 potential vorticity unit (PVU, where 1 PVU 5 10 6 Km2 kg 1 s 1)].

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3957

FIG. 5. The 12-h averages of lower-level asymmetric flow in the TC core region over two consecutive periods (36–48 and 48–60 h) in (a),(b) WW, (c),(d) CTL, and (e),(f) EW of Typhoon Ewiniar. The wind speed is shaded according to the scale at the bottom of the figure.

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3958 MONTHLY WEATHER REVIEW VOLUME 141

(Figs. 5c,d). In WW (Figs. 5a,b), a westward deflection begins to recurve at 48 h (0000 UTC 11 September). occurs related to the southerly or southeasterly over the During recurvature (48–60 h), the TC moves northward TC center. These results can be inferred from the PVT before showing a northeastward translation after re- analysis, which reveals that the TC motion can be de- curvature (60–72 h). The point of initial recurvature is termined by the location of maximum HA. In addition, defined as the beginning of recurvature (Hodanish and these results also support previous studies in that the Gray 1993). In the present study, the recurvature point environmental steering is well estimated by the PVT is 0000 UTC 11 September. A comparison between CTL approach (Wu and Wang 2000; Yun et al. 2012). The (Fig. 7a) and SW (Fig. 7b) reveals an insignificant dif- horizontal advection of PV implicitly includes the in- ference in the TC motion, which is analogous to the fluence of the heating-induced asymmetric flows (Wang straight-moving TC. This result indicates that a north– and Holland 1996b; Wu and Wang 2001). Therefore, the south SST gradient has an insignificant role in the TC location of maximum HA faithfully depicts the TC motion. Unlike the straight-moving TC, WW and EW motion with various SST distributions. reveal different TC translation speeds (Figs. 7c,d) after A comparison of PVT and asymmetric flow over the the recurvature point. A TC tends to accelerate after TC center between CTL and SW exhibits little differ- a sudden track change (i.e., recurvature) over the western ence despite the difference in SST gradient (Figs. 3a,b). North Pacific under the steering flow associated with the As shown in Figs. 6a,b, total PVT and HA in SW are subtropical high (Wu et al. 2011). Compared with the TC almost the same as those in CTL in magnitude and lo- motion in CTL, WW and EW show northeastward ac- cation (Figs. 4b,e). The VA and DH terms in SW (Figs. celeration and deceleration, respectively. This result dif- 6c,d) are comparable in magnitude with those in CTL fers from that of Yun et al. (2012), in which a west–east (Figs. 4h,k), although the maximum is rotated more zonal SST gradient offered a favorable condition for a counterclockwise; this does not make a much difference large eastward deflection. Therefore, relative configura- in the TC motion between CTL and SW since VA and tion of TC heading direction and SST distribution could DH tend to cancel each other. The southerly asymmetric play a significant role in determining the TC deflecting flow over the TC center in SW is shown, which is similar direction as well as the translation speed in a zonally to that of CTL (Figs. 5c,d and 6e,f). This seems to imply asymmetric SST distribution. that a north–south SST gradient plays an insignificant To understand the physics of the SST distribution- role in the TC motion. induced TC motion under the recurvature environment, Although the location of maximum HA does not we analyze the PVT in UL before (36–48 h), during precisely align with the TC heading direction, the former (48–60 h), and after (60–72 h) recurvature. The HA term agrees approximately with the tendency of a TC motion. after recurvature agrees well with the total PVT in The magnitude of both VA and DH in EW is greater magnitude and location in all the experiments (Figs. 8c,f than that in WW (Figs. 4g,i,j,l). This result may imply and 9c,f). Although the region of maximum total PVT that a warmer SST over the eastern side of the ocean after recurvature coincides with the northeastward- presents a more favorable condition for TC intensi- moving TC direction of movement, those before and fication than does a warmer SST over the western side. during recurvature do not align with the northwestward- Chang and Madala (1980) suggested that TC intensi- and northward-moving TC, but are located to the east of fication in the Northern Hemisphere is more likely to the TC core in WW and EW (Figs. 8a–c and 9a–c); this is occur when warmer SSTs appear to the right of the TC consistent with CTL and SW (figure not shown). Chan heading direction than to the left. This result is partially (1984) reported that the maximum relative vorticity caused by the enhanced DH associated with relatively tendency of a recurving TC rotates to its future di- strong wind on the right-hand side of the TC heading rection before recurvature. In this study, the location of direction. The effects of asymmetric SST distribution on maximum total PVT before and during recurvature the recurving TC motion are explained in the following corresponds to that after recurvature. The location of section. maximum total PVT and HA cannot depict the TC heading direction before and during recurvature since complex nonlinear interaction can occur between the 4. SST–TC motion relationship under the TC and environmental flow such as a subtropical ridge recurvature environment (Maemi 2003) and westerly trough; Chan and Cheung (1998) demon- A similar approach is used to study Typhoon Maemi. strated that interaction between environment and TC The simulated TC track in CTL (Fig. 7a) shows a fair circulation developed wavenumber-2 flow about 36 h agreement with the best track data (Fig. 2). The TC prior to recurvature, and wavenumber-1 flow becomes moves northwestward before recurvature (36–48 h) and dominant again after recurvature.

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3959

FIG. 6. Wavenumber-1 components of (a) total potential vorticity tendency, (b) horizontal advection, (c) ver- tical advection, and (d) diabatic heating, which are 24-h time composite of lower-level (0.9 $ s $ 0.55) averages during 36–60 h in SW of Typhoon Ewiniar. Positive values are shaded. The contour interval is 3.0 in (a) and (d), 4.0 in (b), and 0.5 in (c) (106 PVU). (e),(f) The 12-h averages of lower-level asymmetric flow in the TC core region over two consecutive periods (36–48 and 48–60 h). The wind speed is shaded according to the scale at the bottom of the figure.

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3960 MONTHLY WEATHER REVIEW VOLUME 141

FIG. 7. The simulated tracks from (a) CTL (black solid line), (b) SW, (c) WW, and (d) EW (gray solid lines) of Typhoon Maemi. The initial forecast time is at 0000 UTC 9 Sep 2003 (0 h). The SST distributions (gray dotted line) are asymmetric in zonal and meridional directions for each experiment. The contour interval is 1.0 (8C).

Despite the fact that the region of maximum total to a weak zonal vertical wind shear over the TC center PVT reveals the future TC moving direction before and in EW (Fig. 10). DH is determined by the vertical and during recurvature, it is meaningful to investigate the horizontal variations of heating, as well as vertical wind dominant term constituting total PVT; relative impor- shear (Wu and Wang 2000; Chan et al. 2002). Vertical tance of various physical processes can be determined by tilt of a TC induced by external forcing alters advection investigating their contributions to the WN1 component of PV through a vertical coupling. Wang and Holland of PVT (Wu and Wang 2000). Based on this point of (1996c) determined that the vertical interaction between view, we examine the dominant term contributing to upper- and lower-level PV associated with DH and the total PVT in order to understand the most dominant development of convective asymmetries within the TC physical process associated with the two different TC core region can modulate the TC motion. It is noted that motions in the zonally asymmetric SST distributions. westerly and southerly vertical wind shears are domi- Note that HA is a major contributor to total PVT in WW nant in WW and EW, respectively (Fig. 10). As shown in (Fig. 8), whereas DH is a dominant term in EW (Fig. 9) Fig. 11, the location of maximum WN1 component of before and during recurvature. In connection with DH, vertical wind in UL indicates upward and downward a strong southerly vertical wind shear is seen compared motions over the region of downshear left and upshear

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3961

FIG. 8. The 12-h time composite of the wavenumber-1 components of (a)–(c) total potential vorticity tendency and (d)–(f) a dominant term mainly contributing to total potential vorticity tendency in upper level (0.55 $ s $ 0.25) over three consecutive periods divided into (left) before, (middle) during, and (right) after recurvature in WW of Typhoon Maemi. Positive values are shaded. The contour interval is 10.0 (106 PVU). right, respectively. This result is consistent with previous to the downshear left (upshear right) of the TC center studies (Wang and Holland 1996c; Bender 1997; Frank due to the TC’s response to imbalances caused by the and Ritchie 2001; Corbosiero and Molinari 2002), which vertical wind shear. The potential temperature anomaly reveal that an upward (downward) motion is enhanced associated with the downward motion can affect mainly

FIG. 9. The 12-h time composite of wavenumber-1 components of (a)–(c) total potential vorticity tendency and (d)–(f) a dominant term mainly contributing to total potential vorticity tendency in upper level (0.55 $ s $ 0.25) over three consecutive periods divided into (left) before, (middle) during, and (right) after recurvature in EW of Typhoon Maemi. Positive values are shaded. The contour interval is 2.0 in (a), 3.0 in (b), and 5.0 in (c)–(f) (106 PVU).

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3962 MONTHLY WEATHER REVIEW VOLUME 141

FIG. 10. The 12-h averaged (a)–(c) zonal and (d)–(f) meridional winds averaged within 100 km from the TC center over three consecutive periods (before, during, and after recurvature) in CTL, SW, WW, and EW of Typhoon Maemi. the location and magnitude of maximum DH. Jones As shown in Fig. 2, the TC shows a rapid northward (1995) explained that warm (cold) anomaly develops track change, and then its northeastward translation in the descent (ascent) region through vertical advec- accelerates. Holland and Wang (1995) suggested that tion when PV is tilted by vertical shear, which shows as a TC approaches a subtropical ridge, an anticyclonic a physical relationship between vertical circulation gyre develops poleward of the ridge axis and is advected and potential temperature anomaly. Thus, the southerly eastward by the westerly flow. As shown in Fig. 12, vertical wind shear under the recurvature condition the asymmetric flow over the TC core region changes may play a significant role in the vertical circulation from southeasterly to southerly, and then southwesterly and the resulting potential temperature anomalies, throughout the recurvature process. These asymmetric which affect the location and magnitude of maximum flows correlate well with the recurving TC motion, which DH (Figs. 9d,e). is characterized as a northwestward movement (before

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3963

FIG. 11. The 12-h time composite of wavenumber-1 components of vertical wind in upper level (0.55 $ s $ 0.25) over three consecutive periods divided into (left) before, (middle) during, and (right) after recurvature in (a)–(c) WW and (d)–(f) EW of Typhoon Maemi. 2 Positive values are shaded. The contour interval is 0.5 in (b),(c) and 0.3 in (a),(d),(e),(f) (m s 1). recurvature), a turn toward north (during recurvature), et al. 2011). Because of the steering flow associated with the and an acceleration of northeastward movement (after subtropical high, the TC showed a sudden track change and recurvature). northeastward acceleration. Large-scale circulations at Comparing the fast-moving TC in WW with the slow- 500 hPa in WW and EW are shown in Fig. 13. A relatively moving TC in EW, a significant difference is revealed weak steering flow is seen in EW compared to WW because in the magnitude of asymmetric flow after recurvature oftheinterferencebyacyclonicvortexovertheregionof (Figs. 12e,f). Wu and Wang (2001) demonstrated that the subtropical high. The counterclockwise flow of the cyclone advection of symmetric PV by heating-induced asymmet- offsets the southerly or southwesterly steering flow, which ric flow could affect the TC motion. Thus, a nonlinear in- provides a favorable condition for recurvature and accel- teraction between heating-induced asymmetric flow and eration. In other words, the environmental steering effect environmental steering flow can affect the TC motion does not sufficiently influence the TC motion in EW, through the resulting horizontal advection of PV. The particularly after recurvature. The TC therefore shows southwesterly flow becomes a strong and deep layer in the a slow motion compared to other experiments. low- to upper troposphere after recurvature (Figs. 10c,f). This steering flow effectively produces an acceleration of 5. Summary and discussion northeastward movement in WW, whereas the southwest- erly flow in EW does not strengthen after recurvature (Figs. The effects of asymmetric SST distributions on the 10c,f), which is an unfavorable condition for acceleration. tracks of the straight-moving Typhoon Ewiniar (2006) The asymmetric flow over the TC core in UL is enhanced and recurving Typhoon Maemi (2003) have been ex- throughout the recurvature process, particularly after re- amined through the PVT analysis with WRF. Different curvature, in WW (Figs. 12a,c,e). The enhanced synoptic- TC motions could be expected, depending on both the scale wind strongly shifts Typhoon Maemi northward, configuration of TC heading direction and asymmetric placing the TC under the southwesterly steering flow (Wu SST distribution and the interaction with environmental

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3964 MONTHLY WEATHER REVIEW VOLUME 141

FIG. 12. The 12-h averaged asymmetric flow in upper level over three consecutive periods (before, during, and after recurvature) in (left) WW and (right) EW of Typhoon Maemi. The wind speed is shaded according to the scale at the bottom of the figure.

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3965

21 FIG. 13. Large-scale flow (m s ) at 500 hPa in (a) WW and (b) EW of Typhoon Maemi after recurvature (1800 UTC 11 Sep 2003). The gray contour indicates geopotential height (m).

flows. A comparison of the straight-moving and recurving southwesterly asymmetric flow over the TC core region TCs is thus useful for understanding the physical pro- in WW and EW effectively provides a larger deflection cesses contributing to various TC motions in nature. compared to CTL, respectively. Unlike the straight- Baroclinic vortices move approximately with the envi- moving TC, the recurving TC reveals northeastward ronmental flow in the low- to midtropospheric layer acceleration (deceleration) after the recurvature point (Holland and Wang 1995; Wang and Holland 1996b,c; in WW (EW). The location of maximum total PVT be- Wang et al. 1998). Changes in environmental flow asso- fore and during recurvature reveals the future TC moving ciated with the recurvature are more related to the upper- direction, whereas it agrees well with the northeastward- tropospheric flow than the lower-tropospheric flow moving TC direction after recurvature. While HA is (George and Gray 1977; Hodanish and Gray 1993; Li a major contributor to total PVT in WW, DH is a domi- and Chan 1999). Different vertical averages are thus nant term in EW before and during recurvature. In con- used to investigate the TC motion for the straight-moving nection with DH, a strong southerly vertical wind shear is Typhoon Ewiniar in LL and recurving Typhoon Maemi shown compared to a weak zonally vertical wind shear in UL. over the TC center in EW. The southerly vertical wind Zonally asymmetric SST distributions result in dif- shear under the recurvature condition may play a signifi- ferent TC translating directions and speeds depending cant role in the vertical circulation and the resulting po- on the interaction between the TC and environmental tential temperature anomalies, which affect the location current associated with asymmetric SST distribution. A and magnitude of maximum DH. The enhanced south- north–south SST gradient has an insignificant role in the westerly flow effectively produces an acceleration of TC motion. It is noted in the zonally asymmetric SST northeastward movement in WW after recurvature, distribution that the straight-moving (i.e., northward whereas the environmental steering effect does not suf- moving) TC is deflected toward the region of warmer ficiently affect the TC motion in EW, particularly after SST. A larger deflection occurs in the zonal direction recurvature. A configuration of TC heading direction than in the meridional direction. This result indicates and SST distribution, therefore, may be crucial in that a SST gradient perpendicular to the TC translating determining not only the TC deflecting direction but direction offers a more favorable condition for a TC also the translation speed when SST distribution is deflection toward the region of warmer SST (Chang and zonally asymmetric. The main results are summarized Madala 1980; Yun et al. 2012). A contribution of HA in Table 2. including asymmetric flows induced by asymmetric forc- Even though we suggest a possible role of vertical ing dominates the deflection. The location and magnitude wind shear in the location and magnitude of maximum of maximum HA are generally consistent with those of DH under the recurvature condition, the physical process total PVT compared to VA or DH. Southeasterly and can be much more complicated depending on the vertical

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC 3966 MONTHLY WEATHER REVIEW VOLUME 141

TABLE 2. Summary of the results from runs with zonally asymmetric SST distributions for the straight-moving Typhoon Ewiniar and recurving Typhoon Maemi. Mean distance from the CTL and mean translation speed are averages during the 3-day integration.

Straight-moving Typhoon Ewiniar Recurving Typhoon Maemi WW EW WW EW Mean distance from 115.3 (westward 114.7 (eastward Mean translation 4.8 3.9 2 CTL (km) deflection) deflection) speed (m s 1) Dominant term Horizontal Horizontal Dominant term Horizontal Diabatic heating contributing to total advection advection contributing to total advection PVT in LL PVT in UL Asymmetric flow in LL Southeasterly Southwesterly Vertical wind shear Westerly Southerly Large-scale circulation Enhanced Reduced steering flow steering flow

structure and intensity of TC and its environment (Wang demonstrated that a baroclinic TC migrates to the re- and Holland 1996c; Chan 2005). Although a lot of effort has gion in which the azimuthal wavenumber-1 (WN1) been put into understanding the role of vertical wind shear component of PVT reaches a maximum. On the basis of in the TC dynamics over the last several decades (Jones this finding, we identified the contributions of various 1995; DeMaria 1996; Wang and Holland 1996c; Wang and physical processes to the TC motion. Chan et al. (2002) Wu 2004), how it influences the TC dynamics is not thor- validated the PVT framework by analyzing different oughly understood. More study is needed in this regard. observational datasets. Observations from satellites, aircraft reconnaissance, As shown by Wu and Wang (2000), the potential etc., have been utilized by many of the recent data assim- vorticity P can be written in sigma coordinates as ilation efforts to improve the performance of TC forecasts (Zhang and Pu 2010; Singh et al. 2011; Liu and Xie 2012; g ›u ›u ›u ›y ›u Uhlhorn and Nolan 2012). Therefore, modeling of more P 52 (z 1 f ) 1 2 , p ›s ›s ›y ›s ›x realistic TC cases should be carried out in future em- s ploying data assimilation to relax uncertainties coming z 1 u from the idealized SST distribution. Such simulations where ps,( f ), and are surface pressure, absolute will show more realistic performance of the model and vorticity, and potential temperature, respectively; and reveal relative importance of bogussing scheme and SST distribution in determining the TC track. ›P 2C Á $P 5 1 , s ›t Acknowledgments. This work was supported by a GRL grant of the National Research Foundation (NRF) fun- where C, Ps, and P1 are the TC motion vector, sym- ded by the Korean Government (MEST 2011-0021927). metric PV, and WN1 components of PV, respectively. This research was also a part of the project titled ‘‘Con- These expressions indicate that TC dominated by sym- struction of Ocean Research Stations and their Applica- metric circulation tends to move to the region with a tion Studies’’ funded by the Ministry of Land, Transport maximum WN1 component of PVT. Thus, the TC mo- and Maritime Affairs, South Korea. The work of JCLC tion is determined by defining the location of the maxi- was supported by the Research Grants Council of the mum WN1 component of PVT. Hong Kong Special Administrative Region Grant CityU The PVT equation is given by 100210. The authors thank Prof. Bin Wang for his in- sightful comments. Valuable comments and suggestions ›P ›P ›P ›P from the anonymous reviewers improved this manuscript. 52u 2 y 2 s_ ›t ›x ›y ›s 2 3 APPENDIX ›u_ › ›u_ ›y ›u_ 1 g 42 z 1 2 u 1 5 ( f ) ›s ›s › ›s › , ps y x Potential Vorticity Tendency Approach Wu and Wang (2000) provided a general dynamic where s_ is vertical velocity in the sigma coordinates and framework for the study of baroclinic TC motion with u_ is rate of change of potential temperature. The fol- the potential vorticity tendency (PVT) approach. They lowing three terms contribute to total PVT when friction

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC NOVEMBER 2013 C H O I E T A L . 3967 is negligible: horizontal advection (HA), vertical ad- monsoon precipitation and the Indian Ocean precipitation in the vection (VA), and diabatic heating (DH). For further boreal summertime. Asia-Pac.J.Atmos.Sci.,49 (2), 171–182. details, we refer to Wu and Wang (2000). Li, Y. S., and J. C. L. Chan, 1999: Momentum transports associated with tropical cyclone recurvature. Mon. Wea. Rev., 127, 1021– 1037. REFERENCES Liu, B., and L. Xie, 2012: A scale-selective data assimilation ap- proach to improving tropical cyclone track and intensity Bender, M. A., 1997: The effect of relative flow on the asymmetric forecasts in a limited-area model: A case study of Hurricane structure in the interior of hurricanes. J. Atmos. Sci., 54, 703–724. Felix (2007). Wea. Forecasting, 27, 124–140. Chan, J. C. L., 1984: An observational study of the physical pro- Michaels, P. J., P. C. Knappenberger, and R. E. Davis, 2006: Sea- cesses responsible for tropical cyclone motion. J. Atmos. Sci., surface temperatures and tropical cyclones in the Atlantic basin. 41, 1036–1048. Geophys. Res. Lett., 33, L09708, doi:10.1029/2006GL025757. ——, 2005: The physics of tropical cyclone motion. Annu. Rev. Schade, L. R., 2000: Tropical cyclone intensity and sea surface Fluid Mech., 37, 99–128. temperature. J. Atmos. Sci., 57, 3122–3130. ——, and R. T. Williams, 1987: Analytical and numerical studies of Singh, R., C. M. Kishtawal, P. K. Pal, and P. C. Joshi, 2011: As- the beta-effect in tropical cyclone motion. Part I: Zero mean similation of the multisatellite data into the WRF model for flow. J. Atmos. Sci., 44, 1257–1265. track and intensity simulation of the Indian Ocean tropical ——, and K. K. W. Cheung, 1998: Characteristics of the asymmetric cyclones. Meteor. Atmos. Phys., 111, 103–119. circulation associated with tropical cyclone motion. Meteor. Tuleya, R. E., and Y. Kurihara, 1982: A note on the sea surface Atmos. Phys., 65, 183–196. temperature sensitivity of a numerical model of tropical storm ——, F. M. F. Ko, and Y. M. Lei, 2002: Relationship between po- genesis. Mon. Wea. Rev., 110, 2063–2069. tential vorticity tendency and tropical cyclone motion. J. At- Uhlhorn, E. W., and D. S. Nolan, 2012: Observational under- mos. Sci., 59, 1317–1336. sampling in tropical cyclones and implications for estimated Chang, S. W., 1979: The response of an axisymmetric model trop- intensity. Mon. Wea. Rev., 140, 825–840. ical cyclone to local variations of sea surface temperature. Wang, B., R. L. Elsberry, Y. Wang, and L. Wu, 1998: Dynamics in Mon. Wea. Rev., 107, 662–666. tropical cyclone motion: A review. Chin. J. Atmos. Sci., 22, ——, and R. V. Madala, 1980: Numerical simulation of the in- 416–434. fluence of sea surface temperature on translating tropical cy- Wang, Y., and G. J. Holland, 1996a: The beta drift of baroclinic clones. J. Atmos. Sci., 37, 2617–2630. vortices. Part I: Adiabatic vortices. J. Atmos. Sci., 53, 411–427. Corbosiero K. L., and J. Molinari, 2002: The effects of vertical wind ——, and ——, 1996b: The beta drift of baroclinic vortices. Part II: shear on the distribution of convection in tropical cyclones. Diabatic vortices. J. Atmos. Sci., 53, 3737–3756. Mon. Wea. Rev., 130, 2110–2123. ——, and ——, 1996c: Tropical cyclone motion and evolution in Davis, C. A., and S. Low-Nam, 2001: The NCAR-AFWA tropical vertical shear. J. Atmos. Sci., 53, 3313–3332. cyclone bogussing scheme. NCAR Tech. Note, 13 pp. ——, and C.-C. Wu, 2004: Current understanding of tropical cy- DeMaria, M., 1996: The effect of vertical shear on tropical cyclone clone structure and intensity changes—A review. Meteor. intensity change. J. Atmos. Sci., 53, 2076–2087. Atmos. Phys., 87, 257–278. Fiorino, M., and R. L. Elsberry, 1989: Some aspects of vortex Wong, M. L. M., and J. C. L. Chan, 2006: Tropical cyclone motion in structure related to tropical cyclone motion. J. Atmos. Sci., 46, response to land surface friction. J. Atmos. Sci., 63, 1324–1337. 975–990. Wu, L., and B. Wang, 2000: A potential vorticity tendency di- Frank, W. M., and E. A. Ritchie, 2001: Effects of vertical wind agnostic approach for tropical cyclone motion. Mon. Wea. shear on the intensity and structure of numerically simulated Rev., 128, 1899–1911. hurricanes. Mon. Wea. Rev., 129, 2249–2269. ——, and ——, 2001: Effects of convective heating on movement George, J. E., and W. M . Gray, 1977: Tropical cyclone recurvature and vertical coupling of tropical cyclones: A numerical study. and nonrecurvature as related to surrounding wind-height J. Atmos. Sci., 58, 3639–3649. fields. J. Appl. Meteor., 16, 34–42. ——, ——, and S. A. Braun, 2005: Impacts of air–sea interaction on Hodanish, S., and W. M. Gray, 1993: An observational analysis of tropical cyclone track and intensity. Mon. Wea. Rev., 133, tropical cyclone recurvature. Mon. Wea. Rev., 121, 2665–2689. 3299–3314. Holland, G. J., 1983: Tropical cyclone motion: Environmental in- ——, H. Zong, and J. Liang, 2011: Observational analysis of sudden teraction plus a beta effect. J. Atmos. Sci., 40, 328–342. tropical cyclone track changes in the vicinity of the East China ——, and Y. Wang, 1995: Baroclinic dynamics of simulated tropical Sea. J. Atmos. Sci., 68, 3012–3031. cyclone recurvature. J. Atmos. Sci., 52, 410–426. Ye, Q., 2004: Typhoon Rusa and Super Typhoon Maemi in Korea. Jones, S. C., 1995: The evolution of vortices in vertical shear. I: NCAR/ESIG, 35 pp. Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, Yun, K.-S., K.-H. Seo, and K.-J. Ha, 2008: Relationship between 821–851. ENSO and northward propagating intraseasonal oscillation in ——, and Coauthors, 2003: The extratropical transition of tropical the East Asian summer monsoon system. J. Geophys. Res., cyclones: Forecast challenges, current understanding, and fu- 113, D14120, doi:10.1029/2008JD009901. ture directions. Wea. Forecasting, 18, 1052–1092. ——, J. C. L. Chan, and K.-J. Ha, 2012: Effects of SST magnitude Kain, J. S., and J. M. Fritsch, 1993: Convective parameterization for and gradient on typhoon tracks around East Asia: A case study mesoscale models: The Kain–Fritsch scheme. The Represen- for Typhoon Maemi (2003). Atmos. Res., 109–110, 36–51. tation of Cumulus Convection in Numerical Models, Meteor. Zhang, L., and Z. Pu, 2010: An observing system simulation ex- Monogr., No. 24, Amer. Meteor. Soc., 165–170. periment (OSSE) to assess the impact of Doppler wind lidar Lee, S.-S., Y.-W. Seo, K.-J. Ha, and J.-G. Jhun, 2013: Impact of (DWL) measurements on the numerical simulation of a tropical the western North Pacific subtropical high on the East Asian cyclone. Adv. Meteor., 2010, 1–14, doi:10.1155/2010/743863.

Unauthenticated | Downloaded 10/03/21 11:47 PM UTC