Tropical cyclones (TCs) always develop from synoptic-scale disturbances. While early studies suggested that the presence of synoptic-scale disturbances may enhance large-scale conditions for TC formation, recent studies argued that TC-precursor disturbances can establish a rotation-dominant area, which can play a crucial role in organizing convective activity and converting convective heating to rotational energy for storm-scale intensification. To demonstrate the synoptic-scale influence of TC-precursor disturbances, 91 TC formation events within the monsoon trough over the western North Pacific during 2000–10 were examined by separating TC-precursor disturbances from the low-frequency background. The composite analysis shows that the synoptic disturbances indeed enhance the mid- and low-level relative vorticity and convergence, but contribute little to reducing vertical wind shear.
The dynamic composite that is conducted with respect to disturbance centers indicates that TC-precursor disturbances within the monsoon trough establish a rotation-dominant region with a radius of less than 550 km. The cyclonic rotation increases with time 72 h prior to TC formation and nearly all air particles keep recirculating in the core area with a radius of about 220 km. Analysis of a specific case suggests that vorticity increase occurs through the merger of mesoscale convective systems in the rotation-dominant area. The enhancing rotation in the core area may efficiently convert diabatic heating to kinetic energy for TC formation. Thus, it is suggested that the important role of TC-precursor disturbances in TC formation is the establishment of a limited, rotation-dominant area.
It has long been known that tropical cyclones (TCs) always develop from a preexistent low-level cyclonic disturbance, an area of organized but weak synoptic-scale (~700 km) winds (Riehl 1948; Ramage 1959; Zehr 1992; Gray 1998). Such a synoptic-scale disturbance can maintain its identity for one day or more with multiple localized areas of heavy rain and deep convection (Gray 1998). Although necessary large-scale conditions for TC formation (e.g., warm SST, large low-level vorticity, weak vertical wind shear, and high humidity in the low- to midtroposphere) have been well established (Ramage 1959; Gray 1968, 1998), the influences of preexistent synoptic-scale disturbances on TC formation have not been fully understood.
Examination of satellite imagery indicates that only a small fraction of tropical disturbances can lead to TC formation (Hennon et al. 2013). Using composited rawinsonde data, Zehr (1976), McBride and Zehr (1981), and Lee (1989) were among the first to examine the differences in low-level vorticity, vertical motion, temperature, and moisture between pretyphoon cloud clusters and the nondeveloping cloud clusters over the western North Pacific. While there are larger low-level vorticity and stronger vertical motion in developing cloud clusters, no obvious differences can be found in temperature and moisture between developing and nondeveloping disturbances. In agreement with previous studies, Fu et al. (2012) recently estimated the relative importance of TC formation factors and also found that dynamic factors are more important to TC formation in the western North Pacific.
Many studies have shown that TC formation in the western North Pacific is modulated by tropical intraseasonal oscillations, among which the Madden–Julian oscillation (MJO) is the dominant mode of intraseasonal convective variability in the tropics (Madden and Julian 1971; Liebmann et al. 1994). Once the monsoon trough is established, the shear flows and confluent zone between the monsoon westerly and easterly trades are favorable spots for TC formation (Holland 1995; Ritchie and Holland 1999). The arrival of the MJO convective phase further enhances the monsoon trough and local eddy activity (Maloney and Hartmann 2001) as well as amplifies and contracts various incoming equatorial waves, leading to the formation of TCs (Aiyyer and Molinari 2003; Frank and Roundy 2006; Schreck et al. 2011, 2012). Because of the presence of low-frequency modulations, the differences between developing and nondeveloping tropical disturbances, which were identified in the early observational analysis (Zehr 1976; McBride and Zehr 1981; Lee 1989), can be contaminated by the low-frequency circulation. That is, it is difficult to distinguish whether the contribution of the dynamic factors result from tropical disturbances and/or its large-scale environment.
A few recent studies suggested that TC-precursor synoptic-scale disturbances can play crucial roles in organizing convective activity that is essential to TC formation and in converting convective heating to rotational energy for the storm-scale intensification (Dunkerton et al. 2009; Wang et al. 2009; Montgomery et al. 2010a; Wang et al. 2012; Tory et al. 2013). Dunkerton et al. (2009) found that TC formation in Atlantic African easterly waves prefers to occur in Kelvin cat’s-eye near the intersection of the wave critical latitude and trough axis. While air is repeatedly moistened by convection and protected from dry air intrusion, they argued that the synoptic-scale easterly wave or wave’s pouch also provides an environment favorable for vorticity aggregation leading to TC formation. The marsupial theory provides an effective way to identify regions of TC formation with strong cyclonic vorticity, weak strain, and high saturation (Wang et al. 2009; Montgomery et al. 2010b; Wang et al. 2012). Using a normalized Okubo–Weiss parameter, Tory et al. (2013) found that the flow near the center of the pouch is in near-solid-body rotation and emphasized the importance of low-deformation vorticity in the marsupial pouch. The enhanced rotation can substantially reduce the Rossby radius of deformation and effectively prevent diabatic heating being diffused away by gravity waves (Schubert et al. 1980; Schubert and Hack 1982; Ooyama 1982; Simpson et al. 1997).
In the western North Pacific, synoptic-scale tropical disturbances are usually embedded in the monsoon trough and are often associated with northwest-propagating wave trains, which are characterized by alternating regions of cyclonic and anticyclonic circulations along the monsoon trough axis. The development of synoptic-scale wave trains are attributed to a transition of mixed Rossby–gravity (MRG) waves to off-equatorial tropical-depression-type (TD type) waves (Takayabu and Nitta 1993; Liebmann and Hendon 1990; Dickinson and Molinari 2002; Aiyyer and Molinari 2003; Chen and Huang 2009), scale contraction and energy accumulation of equatorial waves (Tai and Ogura 1987; Holland 1995; Kuo et al. 2001; Sobel and Maloney 2000; Hartmann and Maloney 2001; Maloney and Hartmann 2001; Maloney and Dickinson 2003; Tam and Li 2006; Gall and Frank 2010), instability of summer mean flows in the presence of a convection–frictional convergence feedback (Li 2006), and Rossby wave energy dispersion of a preexisting TC (Holland 1995; Li et al. 2003; Li and Fu 2006; Li et al. 2006). Montgomery et al. (2010a; Montgomery and Smith 2012) demonstrated that the wave pouch of the easterly wave critical layer in the western North Pacific provided a protective region for TC formation, and Wang et al. (2012) found that the marsupial theory for tropical cyclogenesis in easterly waves also worked for tropical cyclogenesis associated with northwestward-propagating disturbances in the western North Pacific. This suggests that the northwestward-propagating disturbances may have the similar influence on tropical cyclogenesis as the easterly waves have in the Atlantic. The marsupial theory provides a way for finding the favorable or possible areas for tropical cyclogenesis. But, more attention should be paid to why these areas are favorable for tropical cyclogenesis.
The objective of this study is twofold. First, the influence of synoptic-scale tropical disturbances on TC formation is examined through a composite analysis of the May–October TC formation events within the monsoon trough over the western North Pacific during 2000–10. We obtain synoptic-scale disturbances using a time filter while the monsoon trough in our analysis is treated as the low-frequency background. Second, following Tory et al. (2013), the role of tropical disturbances in establishing a closed, rotational area for TC formation is investigated by calculating the Okubo–Weiss (OW) parameter and air parcel trajectories. The rest of the paper is organized as follows. The data and methodology are described in section 2. The structures of developing disturbances and their contributions to large-scale conditions for TC formation are presented in sections 3 and 4. The role of tropical disturbances in establishing a closed, rotational area for TC formation is discussed in section 5, followed by the conclusions and discussion in section 6.
2. Data and methodology
The primary data used in this study are the National Centers for Environmental Prediction (NCEP) Final (FNL) Operational Global Analysis data with 1° × 1° latitude–longitude grids at 6-h intervals. The zonal and meridional wind components, temperature, and relative humidity from 1000 to 100 hPa are used in the study. The hurricane satellite (HURSAT) B1 geostationary data (version 5) from the National Climatic Data Center (NCDC) (Knapp and Kossin 2007) is used to examine the convective activity in the tropical disturbance associated with Typhoon Bilis (2006). The dataset consists of raw satellite observations from the International Satellite Cloud Climatology Project (ISCCP) B1 data centered on the historical TCs. The data include infrared window, visible, and other channels and are provided in 3-h intervals and gridded to 8-km resolution. The TC formation information is obtained from the Joint Typhoon Warning Center (JTWC) best track data. The TC formation events in May–October within the monsoon trough during the period 2000–10 are discussed in this study. It is acknowledged that the 11-yr data period was rather anomalous in that during many of the years easterly trade winds were stronger than normal, which altered the typical monsoon trough environment over the region.
b. Identification of TC formation events within monsoon troughs
In this study, the monsoon trough is treated as the background relative to synoptic-scale flows such as tropical disturbances and TCs. The TCs and synoptic-scale wave trains in the monsoon trough can lead to changes in the structure and position of the monsoon trough (Harr and Wu 2011). To reduce the bias of TC circulation, following Hsu et al. (2008) and Wu et al. (2013), the TC circulation in the FNL data is first removed with the approach proposed by Kurihara et al. (1993, 1995). Then a low-pass Lanczos filter with a 10-day cutoff period is applied to the 850-hPa wind fields to obtain the low-frequency flows (Duchon 1979). The identification of monsoon troughs is based on the low-frequency 850-hPa winds.
The monsoon trough (MT) can be divided into two parts: a monsoon shear line region and a confluence zone (Ritchie and Holland 1999). The two parts reflect the meridional and zonal changes of the zonal winds, respectively. The monsoon trough axis that is identified with the line of zero-zonal wind and positive relative vorticity is used to recognize the changes of the zonal winds. The successive westerly and easterly winds on the both sides of the trough axis are examined to determine the monsoon trough. To determine the eastern boundary of the monsoon trough, the negative zonal stretching deformation in the previously identified area is checked. If a TC first reaches the maximum sustained wind of 34 kt (17.5 m s−1) within this area, it will be counted as a TC formation event within the monsoon trough. Readers are referred to Zong and Wu (2015) for details.
Two typical types of MTs are identified in this study. The southeast–northwest (SE–NW)-oriented MTs in our study are predominant, accounting for 80.2%. The angle between the axis and the equator is 16.1° on average, with the 90% axis angles less than 30°. As Lander (1996) showed, there are reverse-oriented [southwest–northeast (SW–NE) oriented] MTs in the western North Pacific. Additionally, the monsoon flow pattern sometimes evolved into a monsoon gyre pattern that behaves as a nearly closed cyclonic gyre with a diameter of 2500 km (Lander 1994, 1996). Although TC formation in SW–NE-oriented MTs is important, in this study we focus only on the SE–NW-oriented MTs.
c. Identification of tropical disturbances associated with TC formation
A 3–8-day bandpass filter is used to extract synoptic-scale tropical disturbances from the FNL data (e.g., Lau and Lau 1990; Fu et al. 2007; Peng et al. 2012). Based on the cases of TC formation within monsoon troughs, the disturbances associated with TC formation are traced back to 72 h from the JTWC formation time. The maxima of the synoptic-scale relative vorticity at 850 hPa are used to track tropical disturbances while the TC center at the TC formation time or the previously determined disturbance center is used as the first guess. Considering that the movement of the tropical disturbance is confined to several hundreds of kilometers within 6 h (Kerns et al. 2008), a radius of 550 km is used to search for the disturbance center. The grid point with the maximum synoptic-scale relative vorticity at 850 hPa is taken as the position of the disturbance center.
As an example, Fig. 1 shows the identified disturbance centers associated with the formation of Typhoon Bilis (2006), as well as the 3–8-day bandpass-filtered 850-hPa wind and relative vorticity fields at −72 h (1200 UTC 6 July) from the TC formation time. The disturbance center is identified at 6-h intervals, but for clarity we only show the 24-h disturbance centers in Fig. 1. With the JTWC formation location at 1200 UTC 9 July, the disturbance center at 0600 UTC 9 July was determined based on the synoptic-scale relative vorticity. Then the disturbance center at 0000 UTC 9 July was found by taking the previous center as the first guess. In this study, we trace the centers of tropical disturbances back to 72 h from the TC formation time. Note that, as shown in Fig. 1, the disturbance center is not necessarily the circulation center of the tropical disturbance.
3. The composite structure of TC-precursor disturbances
A total of 91 TC formation events are found within the monsoon trough in May–October during 2000–10. Figure 2 shows the 10-day low-pass-filtered 850-hPa wind field composited with respect to the confluence zone center for the 91 TC formation events and the tropical disturbance centers at −72 and −48 h from the TC formation time. Note that this composite is conducted based on the monsoon trough confluence zone center, which is different from the following composites conducted based on the disturbance center. The confluence zone center is defined as the most southward and eastward point with zero zonal wind and positive relative vorticity. The positions of the disturbance centers are plotted with respect to the confluence zone center. Also, considering that the orientation of the monsoon troughs varies significantly, the monsoon trough orientation is rotated to the mean orientation of 91 troughs around the confluence zone center, in order to reduce the uncertainty in our composite analysis. The trough axis was identified first by the zero contour of zonal wind in the monsoon trough and then fitted with a straight line. The trough orientation is measured by the angle between the axis and the equator. Finally the monsoon trough axis is rotated to the mean angle of the 91 cases. Although the rotation of the MT axis may turn the western part of the trough in the midlatitude into the tropic, this effect should have little effect in the eastern portion of the MT where most TC formation events are identified. Readers are referred to Zong and Wu (2015) for details.
As shown in Fig. 2, most of the TC-precursor disturbances occur in the shear region of the monsoon trough, especially near the eastern end of the shear line. Compared to the centers at −72 h (Fig. 2a), the disturbance centers tend to move toward the region of maximum relative vorticity, which can be seen from the contours of the occurrence frequency (Fig. 2b). In the next section, we will show that the tendency toward the maximum relative vorticity affects the low-frequency relative vorticity composited with respect to disturbance centers.
Figure 3 gives the composite vertical profiles of the 91 disturbances in the meridional wind component and relative vorticity. At −72 h (Fig. 3a), strong northerly winds occur to the west of the disturbance center, and the positive relative vorticity extends up to 200 hPa. Both of the meridional wind and relative vorticity have maximum around 850 hPa. The 850-hPa maximum wind occurs about 250 km away from the disturbance center. At −24 h (Fig. 3b), compared to that at −72 h, the relative vorticity is nearly doubled, covering more than 1000 km at the lower levels and extending up to 200 hPa. The meridional wind also intensifies, in particular on the eastern side of the tropical disturbance.
4. The contribution of tropical disturbances
Previous studies indicated that tropical disturbances mainly contribute to TC formation through dynamic factors (Zehr 1976; McBride and Zehr 1981; Lee 1989; Fu et al. 2012), and thus we focus on the relative vorticity and vertical wind shear associated with tropical disturbances. As shown in Fig. 3, the strong positive relative vorticity and winds are mostly confined within a radius of 550 km from the disturbance center. For this reason, the synoptic-scale relative vorticity and wind profiles are averaged on the area with a radius of 550 km from the disturbance center and then composited with respect to the disturbance center.
Figure 4 shows the vertical profiles of synoptic-scale relative vorticity and divergence at −72, −48, and −24 h. For comparison, the corresponding vertical profiles of low-frequency relative vorticity and divergence are also plotted in this figure. The synoptic-scale relative vorticity remains positive up to 200 hPa and reaches its maximum around 850 hPa. There is little change with height in the synoptic-scale relative vorticity between 900 and 600 hPa. As the formation time approaches, the synoptic-scale relative vorticity increases with time. The low-frequency relative vorticity also reaches its maximum around 850 hPa, decreasing with height and being negative at the upper levels. Note that the low-frequency relative vorticity is much stronger than the synoptic-scale one at lower levels and also substantially increases with time. Further inspection of Fig. 2 indicates that the increase of the low-frequency relative vorticity with time is due to the tendency of tropical disturbances toward the maximum low-frequency relative vorticity. The synoptic-scale divergence profile shows a deep layer of convergence in the middle and lower level and divergence above 400 hPa, which resembles the convective divergence profile of the mesoscale scale systems (Fig. 4b) (Mapes and Houze 1995). The low-frequency divergence profile also takes the form of a convective divergence profile. The increasing relative vorticity is associated with the low-level convergence, implying the importance of the stretching effect. The upper-level divergence suggests outflow at the upper levels. Note that the divergence in the upper troposphere mainly comes from the low-frequency flow as the synoptic-scale divergence is much weaker than the low-frequency one.
In addition, vertical wind shear is one of the important factors influencing TC formation. Figure 5 shows the zonal and meridional wind components at −72, −48, and −24 h. As in the case of relative vorticity, the corresponding profiles of low-frequency wind components are also plotted in Fig. 5. An important feature is that the vertical wind shear on the synoptic scale is negligibly small while the vertical wind shear results mainly from the low-frequency zonal wind between 900 and 350 hPa, where the low-frequency zonal wind decreases in a quasi-linear manner. We conclude that synoptic disturbances contribute little in vertical wind shear.
5. Establishment of the rotation-dominant area
The marsupial theory emphasizes the near-solid-body rotation (low deformation) of the flow in the synoptic-scale easterly wave or wave pouch (Dunkerton et al. 2009; Wang et al. 2009; Montgomery et al. 2010a; Wang et al. 2012; Tory et al. 2013). In this section, we show that synoptic-scale disturbances in the western North Pacific monsoon trough play an important role in establishing such a closed, rotational area for TC formation by calculating the OW parameter and air parcel trajectories. The OW parameter has been used to quantify the rotation of the flow associated with TC formation (Dunkerton et al. 2009; Tory et al. 2013; Wang and Hankes 2014).
where ζ is relative vorticity and s1 and s2 are the stretching deformation and shearing deformation, respectively. Positive (negative) values of the OW parameter indicate that the rotation (deformation) is dominant in the flow. Figure 6 shows the composited 850-hPa OW parameter of the synoptic-scale and low-frequency flows at −72, −48, and −24 h. The positive values of the synoptic-scale OW parameter near the disturbance center substantially increases as the formation time approaches. The positive values are confined to an area with a radius of less than 550 km around the disturbance center. The rotation-dominant flow is mainly in a smaller region with a radius of about 220 km. The OW parameter associated with the low-frequency flow is also positive near the disturbance center. Despite the dominant influence of the synoptic-scale flow, the low-frequency flow also plays a role in establishing the rotational area for TC formation, especially at −24 h.
Tory et al. (2013) showed that 95% of TCs that formed in tropical waves are associated with low-deformation flow at the 850- and 500-hPa levels. Here we averaged the OW parameter within a radius of 220 km from the disturbance center, the area with enhanced OW parameter (Fig. 6). Figure 7 shows the vertical profiles of the OW parameter on the synoptic and low-frequency scales. An important feature is that the OW parameter of the synoptic-scale flow is much larger than that of the low-frequency flow at low- to midlevels. When the TC formation time approaches, the OW parameter increases with time on both of the synoptic and low-frequency scales, in good agreement with Tory et al. (2013).
As suggested in previous studies, the enhanced OW parameter near the disturbance center also establishes a closed area in which air is repeatedly moistened by convection and protected from dry air intrusion (Dunkerton et al. 2009; Tory et al. 2013; Wang and Hankes 2014). To demonstrate this, the trajectories of air parcels around the disturbance center are calculated using the composited wind fields:
where xi and yi represent the position of air parcels at the time i; u and υ are the Eulerian velocity field as a function of space and time, respectively; and Ud and Vd are the averaged translation speeds of the disturbances. The translation speed of the disturbance center is subtracted from the composited wind fields. The trajectories are calculated by using the second-order Runge–Kutta method. The linear interpolation and bilinear interpolation methods are used to obtain the wind velocity between the temporal or spatial grid points, respectively.
The trajectories are seeded around the disturbance center and calculated for 66 h forward in time, terminating 6 h prior to TC formation, as shown in Fig. 8. Among the air parcels within a radius of less than 550 km from the disturbance center (Fig. 8a), 75.3% of the air parcels take a circular track toward the disturbance center. In particular, nearly all of the air particles recycle in the core area or the region of the enhanced OW parameter within a radius of 220 km from the disturbance center. It is suggested that the enhanced OW parameter identifies a region of enhanced rotation that is favorable for TC formation.
In the rotation-dominant core region, as suggested in the marsupial theory, vorticity aggregation occurs through the merger of mesoscale convective systems and diabatic heating is efficiently converted to kinetic energy for system-scale intensification. Figure 9 gives an example of the MCS activity in the synoptic-scale disturbance associated with the formation of Typhoon Bilis (2006). Bilis formed in a cyclonic circulation of a wave train in the monsoon trough (Fig. 10). Both the synoptic-scale and low-frequency flows provided a rotation-dominant environment around the disturbance associated with the formation of Bilis.
As indicated by the low brightness temperature, mesoscale convective systems can be identified in the rotation-dominant region with positive values of the OW parameter. Relatively small mesoscale convective systems were scattered within the disturbance (Figs. 9a–c), 36–24 h prior to the formation of Bilis (0000, 0600, and 1200 UTC 8 July 2006). The horizontal size of the mesoscale convective system became larger during the last 24 h as the formation time approached (Figs. 9d–f). We noticed that the convective activity weakened 12 and 6 h prior to the formation of Bilis (Figs. 9e and 9f). That is, the convection tended to weaken at 0900 local time. Mapes and Houze (1993) found that the convection in the organized cloud clusters in the western Pacific warm pool is strongest before dawn and decreased in the morning. It seems that the weakening of the convective activity is probably related to the diurnal cycle. This specific case indicates that the rotation-dominant core region is favorable for the merger of mesoscale convective systems.
Tropical cyclones always develop from a preexistent low-level cyclonic disturbance (Riehl 1948; Ramage 1959; Zehr 1992; Gray 1998), which may enhance large-scale conditions for TC formation (Zehr 1976; McBride and Zehr 1981; Lee 1989; Fu et al. 2012). However, recent studies suggested that TC-precursor disturbances can establish a rotation-dominant area, in which vorticity aggregation occurs through the merger of mesoscale convective systems and convective heating is efficiently converted to kinetic energy for storm-scale intensification (Dunkerton et al. 2009; Wang et al. 2009; Montgomery et al. 2010a; Wang et al. 2012; Tory et al. 2013). To demonstrate the role of TC-precursor disturbances in establishing a rotation-dominant area, we examined 91 TC formation events within the monsoon trough over the western North Pacific during 2000–10.
Our composite analysis with respect to the disturbance center indicates that tropical disturbances indeed enhance the mid- and low-level relative vorticity and convergence, but contribute little to reducing vertical wind shear. In this study, the TC-precursor disturbances were separated from their low-frequency background using a 3–8-day bandpass filter. This is different from previous observational analysis (e.g., Zehr 1976; McBride and Zehr 1981; Lee 1989), in which the synoptic-scale disturbances were defined mainly on their horizontal scale. In fact, the synoptic-scale disturbances in these studies contained the low-frequency background circulation. In the present study, the synoptic-scale disturbances are defined on both time and horizontal scales. Our analysis shows that the synoptic-scale relative vorticity and divergence at mid- and low levels are weaker than the low-frequency ones.
We argue that the important role of TC-precursor disturbances in TC formation is the establishment of a limited, rotation-dominant area. As shown by Tory et al. (2013), the rotation-dominant area is characterized with positive values of the OW parameter with a radius of less than 550 km. Calculation of the trajectories of air particles indicates that nearly all air particles keep recirculating in the core area with a radius of about 220 km, where the cyclonic rotation increases with time 72 h prior to TC formation.
Gray (1998) pointed out that tropical disturbances cannot intensify over the entire region instantaneously, but it is more efficient for the disturbance to intensify in a limited area. Our results show that such a rotation-dominant area is established by TC-precursor disturbances. The enhanced rotation may reduce the local Rossby deformation radius, leading to efficient conversion from the convective heating to the kinetic energy for the system-scale intensification (Schubert et al. 1980; Schubert and Hack 1982; Ooyama 1982). Moreover, as illustrated in the formation case of Typhoon Bilis, vorticity aggregation occurs through the merger of mesoscale convective systems in the rotation-dominant area, which has been suggested to be important for TC formation (Simpson et al. 1997). Further investigation is needed to understand the merger of mesoscale convective systems through observational analysis and high-resolution numerical simulation.
This research was jointly supported by the National Basic Research Program of China (2013CB430103 and 2015CB452803), the National Natural Science Foundation of China (Grant 41275093), and the project of the specially appointed professorship of Jiangsu Province.