HTML
--> --> -->With aircraft observing data, Harr et al. (1996) proposed that the structure of a mesoscale convective system (MCS) is beneficial to the transformation of the monsoon depression to a TC by amplifying the vorticity of the monsoon depression. Bister and Emanuel (1997) and Ritchie and Holland (1997) investigated the formation of Hurricane Guillermo (1991) and Typhoon Irving (1992), respectively, and both found that a midlevel mesocyclone in the stratiform precipitation region preceded the development of the near-surface cyclone. Simpson et al. (1997) argued that the formation of TC Oliver might have been induced by the interactions between MCSs and associated mesoscale vortices at the mid-levels. Based on a large amount of statistical observational data of the early stage of TC formation, Zehr (1992) and Gray (1998) suggested that tropical cyclogenesis is a two-stage process, with the MV establishing in the first stage. Recent observational work has continued to indicate that TC formation is preceded by a moist, mid-level mesoscale vortex (Komaromi, 2013; Zawislak and Zipser, 2014). Idealized numerical simulations confirm the observational studies. In idealized numerical simulations with a pre-existing, weak tropical disturbance in a quiescent atmosphere, the genesis of a TC only occurs if the vortex in the mid-troposphere has intensified (Nolan, 2007; Ge et al., 2013). With a cloud model initialized by a radiative?convective equilibrium state perturbed by random signals with amplitude of 0.25 K, the simulation by Davis (2015) exhibited the development of rotating, coherent mid-tropospheric structures under the condition without large-scale forcing and an initial vortex, revealing that a TC formed about 6.5 days later after the emergence of the mid-level vortex.
As a significant precursor of TC formation, the roles of an MV in tropical cyclogenesis have been discussed in detail in the literature. Simpson et al. (1997) and Ritchie and Holland (1997) found that a strengthened MV with increased spatial scale can result from the concentration of mid-level vortices, which can lead to the growth of the near-surface cyclonic circulation and thus favor the intensification of a tropical disturbance. Bister and Emanuel (1997) hypothesized that the spin-up of the surface vortex can happen under the effect of the downward transport of mid-level cyclonic vorticity by the mesoscale downdraft within the stratiform precipitation. Gray (1998) argued that the convectively initiated mid-level cyclonic vortex can extend downwards to the surface in the form of a much smaller and more concentrated cyclonic vortex, which can be the embryo of the future storm.
Different from the above-mentioned works focusing on the downward extension of the mid-level vorticity, there are many other studies in the literature that argue the MV is closely linked to vortical hot towers (VHTs), which are critical to the intensification of the surface vortex in tropical cyclogenesis (e.g., Hendricks et al., 2004; Montgomery et al., 2006). Montgomery et al. (2006) suggested that the positive vorticity-rich environment provided by an MV favors VHT development. Dunkerton et al. (2009) argued that, because of the approximately closed Lagrangian recirculation, the mid-level cyclonic circulation associated with the easterly wave can protect the TC embryo inside from the hostile environment, such as dry air intrusion, strain/shear deformation and so on, thus providing a critical point for VHT concentration. This perspective was advocated by Wang et al. (2010a, b) and Montgomery et al. (2010), who showed that the region occupied by an MV is characterized by strong rotation as well as weak deformation and vertical wind shear, favoring further VHT aggregation and persistent convection. In addition, Wang (2012) indicated that the closed cyclonic region of the MV in the central area of pre-Hurricane Felix (2007) was beneficial for VHT aggregation and tropical cyclogenesis because of the abundant moisture and short incubation time scale.
With a focus on the thermodynamic structure of the MV, Raymond et al. (2011, 2014) argued that the cold anomaly below the MV favors the outbreak of surface-based convection and then the spin-up of TCs in that the downdrafts induced by evaporative cooling are inhibited. However, Davis (2015) argued that, during TC genesis, the downdrafts intensify/weaken along with the strengthening/weakening of the upward motion, and the bottom-heavy mass flux profile within an MV can primarily be ascribed to the intense updrafts rather than the weakened downdrafts in the lower troposphere. In addition, the simulation by Davis (2015) highlighted that the enhanced low?mid-level stratification corresponding to the remarkable mid-level cyclonic vorticity was unfavorable for deep convective bursts. His work suggested that, prior to genesis, strong and large surface cold pools induced by strengthened downdrafts would interplay with the vertical wind shear in organizing the low-level updrafts that contribute to the strengthening of the surface cyclonic circulation. It is worth mentioning that, although the MV has been widely accepted as a precursor in tropical cyclogenesis, the idealized experiments performed by Kilroy et al. (2017), in which the ice microphysical processes were absent, indicated that the mid-level vortex may not be necessary for the occurrence of TC formation.
Although the relationships between MVs and TC formation have been discussed in detail since Bosart and Sanders (1981), the evolution of the MV in the pre-genesis stage of a TC has not been discussed comprehensively owing to the lack of continuous observations. With the aim to further understand the tropical cyclogenesis associated with the MV, the current study examines the formation of super Typhoon Megi (2010) on the basis of a cloud-resolving simulation. Similar to Davis (2015), this work focuses on the MV evolution in TC genesis, but for a real-world TC. Megi was the strongest typhoon over the globe in 2010, with a minimum sea-level pressure of 895 hPa and maximum wind of 72 m s?1 (best-track data of the Japan Meteorological Agency; JMA). It originated from several positive vorticity anomalies (PVAs) initially scattered in the western tropical Paci?c Ocean and was designated as a tropical depression by the JMA at 0000 UTC 13 October. During Megi’s lifetime, rapid intensification, sudden track changes, and rapid weakening occurred sequentially, which induced large prediction errors in the operational forecasting. As a result, Megi drew the attention of many studies (e.g., Kieu et al., 2012; Qian et al., 2013; Shi et al., 2014; Wang and Wang, 2013, 2014). The current study focuses on its formation, which has not been fully explored in the literature.
The rest of this paper is organized as follows: The numerical simulation and verification are briefly introduced in section 2. Section 3 presents the model-derived evolution of the MV and associated convection during Megi’s formation. Section 4 focuses on the influence of the MV on the formation of Megi. Section 5 gives some concluding remarks.
The model-simulated and GFS-derived track and area-mean vorticity of the pre-Megi disturbance①, as well as the environmental vertical wind shear between 850 and 200 hPa following the disturbance, are shown in Fig. 1. The observational data indicate that the formation of Megi occurred in the environment with moderate vertical wind shear (~7.5 m s?1; Fig. 1c). At the beginning, the pre-Megi disturbance moved westwards, and then took a northwest route and a southwest route sequentially (Fig. 1a), while the vorticity experienced fluctuating variations so that the disturbance developed into a TC at 0000 UTC 13 October 2010 according to the JMA best track (Fig. 1b). In general, the above-mentioned characteristics of the environment, track and intensity of the pre-Megi disturbance were properly described by the numerical model. The model-simulated vertical wind shear was larger than that derived from GFS, but the difference was less than ~2 m s?1. The deviation of the model-simulated track from that derived from GFS was smaller than 150 km. The model-derived vorticity was a bit larger than that derived from GFS. This difference can be partially ascribed to the fact the horizontal resolution of the model data was higher than the observation. In addition to the environment, track and intensity, the key mesoscale processes in the formation of Megi, such as the first episode of vigorous deep convection starting from 0600 UTC 8 October, the vorticity aggregation after 0600 UTC 9 October, and convection reinvigoration after 0000 UTC 10 October, were also represented pretty well in the numerical simulation (Fig. 2). We acknowledge that the spectral nudging may have a role in the model performance, but it was only used in the first 24 h in the innermost domain. Overall, the good performance of the numerical model in the simulation of Megi’s formation provided four-dimensional and dynamically consistent data for our investigation of the formation of Megi.
Figure1. (a) GFS-derived (black circles) and model-derived (red circles) track of the disturbance, overlaid by the simulated 1000 hPa wind (vectors), and sea level pressure (blue lines) at 0000 UTC 13 October. The values of the contours are 1006.5, 1007, 1007.5, 1008, 1009, 1010, and 1011 hPa. (b) Area-mean vorticity (units: 10?5 s?1) at 850 hPa within 200-km radius of the disturbance center. The vertical dashed line denotes the genesis time (0000 UTC 13 October, from the JMA). (c) Area-mean vertical wind shear (units: m s?1) between 850 hPa and 200 hPa within a 20° × 20° area centered at the disturbance center.
Figure2. (a) Time?longitude section of GMS/MTSAT-derived blackbody temperature (shading; units: °C) and GFS-analysis-derived 850-hPa winds averaged between 8°N and 16°N (vectors) in the period from 0600 UTC 8 to 0600 UTC 13 October. (b) As in (a) but for the summation of positive vorticity (units: 10?5 s?1) between 8° and 16°N derived from the GFS analysis data at 850 hPa. (c) As in (b) but for 600 hPa. (d?f) As in (a?c) except for (d) the model-derived column-maximum reflectivity (units: dBZ) and 850-hPa winds, (e) 850-hPa and (f) 600-hPa vorticity (units: 10?5 s?1).3.1. Overview of the genesis process of Megi
In consistent with Gray (1998), Figs. 2a and d show that the pre-Megi disturbance experienced two episodes of deep convection separated by an intermediate one-day inactive period of little convection. The first episode of massive intense convection began from the model start to around 0600 UTC 9 October, which was associated with the passage of a tropical Kelvin wave (Fang and Zhang, 2016), while the second episode of deep convection started from about 0600 UTC 10 October and continued towards the end. Figures 2b-c and 2e-f further indicate that the distinct increase of lower-tropospheric vorticity mainly occurred in the second episode of deep convection, prior to which, i.e., in the interval between the first and second deep convection episode, the mid-tropospheric vorticity enhanced significantly. According to the variations of the convection and low- and mid-level vorticity described above, and for the convenience of ensuing discussions, the genesis process of Megi is roughly partitioned into three phases (with 0600 UTC 9 October and 0600 UTC 10 October subjectively chosen as the dividing points): (1) massive convective burst associated with the tropical Kelvin wave; (2) decaying convection and mid-level vorticity enhancement; and (3) deep convection reinvigoration, low-level vortex intensi?cation, and formation of Megi. Such a divide is consistent with that in Fang and Zhang (2016).Figure 3② displays the convection and low- and mid-tropospheric vorticity in the pre-Megi disturbance during the formation of Megi with time intervals of 12 h. At the beginning of the model integration (0600 UTC 8 October; Fig. 3a), a small PVA and two PVAs were presented in the lower and middle troposphere, respectively. After a short time of model spin-up, there was a convection burst in the pre-Megi disturbance and the low-level PVA enlarged while one more PVA appeared in the middle troposphere (stage 1; Fig. 3b). In the following 12 h, the low-level PVA shrank a little and its vorticity increased considerably, while the mid-level PVAs merged into a mesoscale vorticity anomaly (Fig. 3c). Afterwards, the deep convection and the low-level vorticity decayed gradually in the pre-Megi disturbance, while the mid-level mesoscale vorticity anomaly kept strengthening and developed into a well-defined mid-level mesoscale vortex, i.e., the MV, at about 1800 UTC 9 October (stage 2; Fig. 3d). As the convection became active again in the pre-Megi disturbance after 0600 UTC 10 October (stage 3; Fig. 3e), the low-level vorticity began to enhance significantly, while the MV weakened slightly (Figs. 3e and f). Meanwhile, the MV began to shift eastwards relative to the low-level vorticity under the impact of the westerly shear. The westerly shear also exerted some influences on the convection, as they tended to occur more frequently in the downshear region (Figs. 3e-g). In the following period of Megi’s genesis process (Figs. 3f-l), both the low-level PVA and MV persistently occupied the center area of the pre-Megi disturbance and exhibited fluctuating intensification. The fluctuation of the intensity of the MV can also be detected in Fig. 2c. After 0000 UTC 13 October, i.e., Megi’s genesis time, the low-level mesoscale vortex and MV began to couple together to form a monopole structure, though there still existed little displacement between the low-level vortex and MV due to the vertical wind shear (Fig. 3l).
Figure3. The re-gridded relative vorticity at 1000 hPa (red contours; units: 10?5 s?1) and 600 hPa (blue contours; units: 10?5 s?1), the 1000?600-hPa-averaged vertical velocity (shading; units: m s?1), as well as the 1000-hPa winds (vectors) in the region centered at pre-Megi’s center with side length of ~1125 km. The contour intervals are 2 × 10?5 s?1 and dashed contours designate negative values. The dashed circles center at pre-Megi’s center with radius of 2°. The vector in the corner of each panel denotes the vertical wind shear between 200 and 850 hPa. The number below the vector represents the magnitude of the shear.Figures 3e-l also show that the convection in the center area of the pre-Megi disturbance was particularly active at about 1800 UTC 10, 18 UTC 11, 18 UTC 12 and 18 UTC 13 October, while comparatively weak at about 0600 UTC 12 and 0600 UTC 13 October. Such a quasi-periodic oscillation of the convection can also be detected in Figs. 2a and d, and is more apparent in the time?height diagrams of model-derived frequency of convection top height③ and the mean vertical velocity averaged over the center area of the pre-Megi disturbance as shown in Fig. 4a, which shows that the convection, especially the deep convection, was characterized by diurnal variations in the pre-Megi disturbance after 0600 UTC 10 October. Such a characteristic has also been found in the formation of Hurricanes Karl (2010) and Matthew (2010) (Davis and Ahijevych, 2012), as well as TC Fay (2008) (Wang, 2014).
Figure4. (a) Time?height diagrams of model-derived frequency of convection top height (shading) and mean vertical velocity (dashed contours are negative) averaged over the center area of pre-Megi’s disturbance. (b) As in (a) but for the vorticity (shading; units: 10?5 s?1) and divergence (dashed contours are negative). (c) As in (a) but for the temperature perturbation (shading; units: K) relative to the whole domain average, relative humidity (thick contours) and equivalent potential temperature (thin contours). The values of contours are ?3, ?2, ?1, 1, 2, 3, 4, 6, 8, and 10 cm s?1 in (a); ?3.8, ?2.8, ?1.8, ?0.8, 0, 0.8, 1.8, 3.8, 5.8, 7.8, 9.8, and 11.8 × 10?5 s?1 in (b); and 10, 20, 30, 40, 50, 60, 70, 75, 80, 84, 88, 90, and 95% in (c). The black dashed, blue, red, and purple thick contours denote the values of 60%, 84% 90%, and 95% in (c), respectively, and the values of dashed (solid) thin contours start from 340 K (344 K) with intervals of 1 K (2 K).It is worth mentioning that the fluctuating growth of the MV mentioned before, which is more prominent in Fig. 4b, is different from that derived from the idealized simulation initialized by a radiative?convective equilibrium state perturbed by random signals, in which the MV strengthened persistently (Davis, 2015). Comparing Fig. 4a to Fig. 4b, the evolution of the MV was closely related to the convective activity in the pre-Megi disturbance, in that the maximum vorticity of the MV usually appeared after the deep convection. The evolution of the MV and its relationship with the convection in the pre-Megi disturbance are discussed in detail in the following section.
2
3.2. Formation and development of the MV
To understand the processes responsible for the evolution of the MV during the formation of Megi, the flux form of vorticity budget analysis was conducted at 600 hPa following Haynes and McIntyre (1987) and Wang et al. (2010a). The vorticity equation is written aswhere
where u, v is the eastward and northward component of velocity, x, y is the distance in the eastward and northward direction, and
where V and C is the earth-relative flow and the velocity of the disturbance. The term on the left-hand side of Eq. (1) is the tendency of the absolute vertical vorticity. The first and second terms on the right-hand side of Eq. (1) are referred to as the advective vorticity flux and the non-advective vorticity flux, respectively, following Tory and Montgomery (2008). The former is the divergence of the vorticity flux and consists of the horizontal advection of absolute vorticity and the stretching effect. The latter combines the vertical vorticity advection and the tilting effect. R in Eq. (1) represents the horizontal components of subgrid-scale terms. Since R generally yields a small contribution to the net vorticity tendency, it is ignored in the calculation, as done in Montgomery et al. (2006).
Figure 5 displays the area-mean 600-hPa vorticity and its tendencies derived by applying Eq. (1) to the domain centered at the vorticity centroid of the MV with the radius of 200 km. During the spin-up period of the model, both the advective vorticity flux and non-advective flux terms in Eq. (1) were negative, and thus the vorticity in MV decreased a little bit. As the deep convection became active in the pre-Megi disturbance after 1800 UTC 8 October, the vorticity of MV began to increase, which was primarily induced by the convergence of horizontal vorticity flux (Figs. 4a and 5a). Although the vorticity tendency induced by the vertical vorticity advection was mostly offset by the tilting effect, it caused the non-advective flux term to be positive and contributed to the development of the MV (Fig. 5b), which indicates that the deep convection can also enhance the mid-level vorticity by transporting the low-level vorticity upwards.
Figure5. (a) Results of the vorticity budget (units: 10?9 s?2) and vorticity (thick black solid line for 600 hPa; dashed line for 950 hPa; ordinate on the right; units: 10?5 s?1) averaged within a 200-km radius from the disturbance center at 600 hPa. (b) Vertical advection, tilting, and divergence of the non-advective vorticity flux (units: 10?9 s?2).After about 0600 UTC 9 October, i.e., in stage 2, the deep convection decayed in the pre-Megi disturbance, while the stratiform precipitation developed gradually (Figs. 3c and 4a). Under the effect of the upper-level outflow of the deep convection and the moderate to strong vertical wind shear shown in Fig. 1c, the ice particles moved away from the deep convection region, and then descended and began to melt, inducing the stratiform precipitation (Houze, 1993). The evaporative cooling associated with the precipitation provided the negative buoyancy, leading to the evident downdrafts in the lower troposphere. Accordingly, marked subsidence emerged in the lower troposphere while the updraft shrank to the mid?upper levels in the pre-Megi disturbance (Fig. 4a). Corresponding to the downdraft, a cold core formed in the low levels (Fig. 4c). In addition, divergence gradually replaced convergence in the lower troposphere and evident convergence appeared in the mid-levels (Fig. 4b). These phenomena are usually observed in the stratiform precipitation region (Mapes and Houze, 1995), indicating that the stratiform precipitation became dominant in the pre-Megi disturbance in stage 2.
Corresponding to the stratiform precipitation development, the vertical vorticity advection, and accordingly the non-advective flux term, weakened, while the horizontal convergence of vorticity flux enhanced considerably. Therefore, the MV strengthened rapidly after about 0600 UTC 9 October and reached maximum intensity at around 1800 UTC 9 October (Fig. 5a).
From the above description, we can see that the formation and development of the MV primarily resulted from the marked mid-level convergence associated with the stratiform precipitation in the pre-Megi disturbance. In addition, the intense convection also transported the low-level vorticity upwards to facilitate the enhancement of the mid-level vorticity. It is worth noting that the variation of the mid-level vorticity was not synchronized to the low-level vorticity in the pre-Megi disturbance. Figures 4b and 5a indicate that, as the mid-level vorticity increased rapidly in the period from ~0600?1800 UTC 9 October, the low-level vorticity decreased considerably. At 1800 UTC 9 October, when the MV reached maximum intensity, the low-level vorticity was at its minimum and anticyclonic circulation appeared in the near-surface levels (Figs. 3d, 4b and 5a).
2
3.3. Thermodynamic changes associated with the formation of the MV
As the mid-level vorticity increased from about 1800 UTC 8 to 1800 UTC 9 October, the low-level thermodynamic characteristics below it also varied considerably (Figs. 4c and 6). From Fig. 4c, we can see that a distinct cold core developed in the lower troposphere below the MV at ~0000 UTC 9 October, about 6 h after the deep convection burst in the pre-Megi disturbance. In the framework of balanced dynamics, the cold core was coupled with the MV to satisfy the thermal wind relationship. However, the cold core weakened after ~0600 UTC 9 October and a warm core emerged in the lower troposphere below the MV. At the same time, the lower troposphere dried considerably below the MV (Fig. 4c and 6b). At around 1800 UTC 9 October when the mid-level vorticity reached the maximum, a warm and dry core appeared below the MV (Fig. 4c). To understand the thermodynamic variations below the MV, the potential temperature and moisture budgets were calculated using the following equations:
Figure6. (a) Time series of 1000-hPa mean equivalent potential temperature (black; units: K), potential temperature (blue; units: K) and specific humidity (red; units: g kg?1) averaged over the region with 600-hPa re-gridded vorticity greater than 2 × 10?5 s?1 and radius less than 4° during the period from 0600 UTC 8 to 0000 UTC 11 October. (c) As in (a) but for the potential temperature tendency (black; units: 10?4 K s?1), and its components induced by vertical advection (blue), horizontal advection (green) of potential temperature, diabatic heating (red), as well as boundary-layer processes and turbulent diffusion and air?sea interaction (orange). (e) As in (c) except for the specific humidity tendency (units: 10?4 g kg?1 s?1), and its components induced by vertical advection (blue), horizontal advection (green) of specific humidity, phase transition of vapor (red), as well as boundary-layer processes and turbulent diffusion air?sea interaction (orange). (b, d, f) As in (a, c, e) but for the variables at 850 hPa.where
Figure7. Skew-T log-P vertical sounding averaged over the center area of pre-Megi’s disturbance. The temperature and dewpoint temperature sounding are denoted by the thick solid and cyan solid lines, respectively. The red dashed line delineates the lifted parcel’s ascent path.After the first episode of deep convection and subsequent stratiform precipitation, a well-defined MV developed in the pre-Megi disturbance at around 1800 UTC 9 October. However, the MV was characterized by near-surface anticyclonic vorticity (Fig. 3d and Fig. 4b) and an onion-shaped sounding (Fig. 7a), both of which are unfavorable for deep moist convection and low-level vorticity enhancement. But with the dissipation of the stratiform precipitation, the shallow dry convection soon developed in the pre-Megi’s central area (Fig. 4a), corresponding to the mid-level divergence and thus the weakening of the MV (Figs. 4b and 5a). Due to the vertical advection effect associated with the convection, the potential temperature decreased and the moisture increased in the lower troposphere (Figs. 6b and d). After ~1800 UTC 9 October, the warm and dry core began to transform into a cold anomaly between 900 and 600 hPa below the weakening MV (Figs. 4c and 6b). The cold anomaly situated above the near-surface warm anomaly induced by the sea-surface fluxes enhanced the convective available potential energy (CAPE) (Figs. 4c, 6c, 7b and 8)④. Meanwhile, the convection inhibition (CIN) decreased due to the low-level humidification (Figs. 7 and 8). All these processes promoted the outbreak of the moist convection after ~0000 UTC 10 October.
Figure8. Evolution of CAPE (black line; units: J kg?1) and CIN (red line with ordinate on the right; units: J kg?1) averaged over a 200-km radius of the disturbance center.Figure 9④ illustrates the development and organization of the convection under the favorable condition in the period from 2300 UTC 9 October to 0600 UTC 10 October. At 2300 UTC 9 October, the MV was coupled with a negative potential temperature anomaly (
Figure9. (a?c) 
2
3.4. Variations of the MV in the second episode of deep convection
From 0600 UTC 9 October, the deep convection in the pre-Megi disturbance was characterized by diurnal variations (Fig. 4a). Between the intervals of the deep convection, the vertical distribution of the vertical velocity, horizontal divergence, vertical vorticity, relative humidity and potential temperature perturbation shown in Fig. 4 indicate that stratiform precipitation occurred, albeit less evident than that in the period from 0600 to 1800 UTC 9 October. Correspondingly, the MV demonstrated fluctuating growth after 0600 UTC 10 October (Figs. 4b and 5). From 1200 UTC 10 to ~0000 UTC 11 October, the deep intense convection and stratiform precipitation occurred sequentially in the pre-Megi disturbance, and the MV intensified. In the following 12 h, the mid-level divergence of vorticity fluxes related to the shallow convection caused the MV to weaken. As the deep convection burst again after ~1200 UTC 11 October, the MV re-intensified. Due to the relatively weak shallow convection at around 0000 UTC 12 October, the MV did not exhibit evident weakening. After ~0000 UTC 12 October, as massive deep intense convection appeared almost at the same time as the shallow moist convection, and the amount of deep intense convection was much greater than that in the previous two episodes of convection (Fig. 4a), the advective vorticity flux term dominated the positive tendency of the mid-level vorticity while the contribution from the non-advective vorticity flux was almost close to zero and negligible. As a result, the mid-level vorticity increased sharply in terms of magnitude towards genesis. It is worth mentioning that such an increase was primarily attributable to the horizontal convergence of the vorticity fluxes, and the contribution of the non-advective flux term was trivial because the vertical vorticity advection was nearly offset by the tilting effect.The sensitivity experiment shared the same configuration as the control experiment described in section 2 (CNTRL) except that the evaporative cooling below 600 hPa was disabled in the period from 0000 UTC to 1800 UTC 9 October (NoEvp). Figure 10a displays the time?height diagrams of model-derived frequency of convection top height and the mean vertical velocity averaged over the center area of the pre-Megi disturbance derived from the sensitivity experiment. Due to the lack of evaporative cooling below 600 hPa, the strong subsidence did not appear after 0600 UTC 9 October. Accordingly, the mid-level convergence and thus the MV were weaker in NoEvp than that in CNTRL at ~1200 UTC 9 October (Figs. 4b and 10b). After 1200 UTC 9 October, both the MVs in NoEvp and CNTRL intensified, but the former was mainly associated with the deep convection (Figs. 10a and b), while the latter was related to the stratiform precipitation (Figs. 4a and b). Although the area-mean vorticity in NoEvp also attained a maximum at 1800 UTC 9 October, it was much smaller than that in CNTRL. However, the low-level vorticity did not experience a significant decrease in NoEvp in the period from 0600 to 1800 UTC 9 October. After 1800 UTC 9 October, when the evaporative cooling was turned on, the stratiform precipitation began to occur in the pre-Megi disturbance (Figs. 10a-c). Due to the low-level divergence associated with the stratiform precipitation, the low-level positive vorticity decreased. However, the vorticity below 600 hPa was still much larger in NoEvp than that in CNTRL at 1200 UTC 10 October. Accordingly, the deep convection was more active in NoEvp than that in CNTRL after 1200 UTC 10 October (Fig. 4a). With the strong stretching effects of the convection, the vorticity associated with the disturbance in NoEvp developed much more quickly than that in CNTRL (Figs. 10b and 4b). In addition, the more active deep convection caused the warm core in the upper levels to become more distinct in NoEvp than that in CNTRL (Figs. 4c and 10c), which also facilitated the intensification of the low-level cyclonic circulation.
Figure10. As in Fig. 4 but for NoEvp.The above analysis implies that the disturbance in NoEvp may develop much more quickly than that in CNTRL. This is further demonstrated in the evolution of the maximum tangential velocity at 10 m for CNTRL and NoEvp (figure not shown). If the time when the 10-m maximum tangential velocity exceeds 17.5 m s?1 was taken as the TC genesis time, then genesis took place at about 0400 UTC 13 and 1600 UTC 11 October in CNTRL and NoEvp, respectively. This indicates that, although the thermodynamic structure associated with the MV destabilized the atmosphere and facilitated the development of the convection when the MV was well-defined, the processes leading to the formation of the MV acted to hinder the formation of Megi through the strong divergence taking place near the surface.
The MV played dual roles in the formation of Megi. On the one hand, the formation of MV corresponded to the remarkable decrease in low-level vorticity, which was unfavorable for the intensification of the near-surface cyclonic circulation. On the other hand, the cold core coupled with the MV benefited the accumulation of the CAPE, which was favorable for the convection. The sensitivity experiment with the evaporative cooling turned off indicated that the processes leading to the formation of MV retarded the genesis process of Megi. Evidence for this is that the disturbance in the sensitivity experiment intensified much quicker than that in the control experiment. It is worth noting that the result is not contradictory to the widely accepted view that the MV is a precursor for the formation of a TC. This work not only pays attention to the role of the MV in the period when it is well-defined, but also to the impact of the MV forming processes on the formation of Megi. From Figs. 4b and 5a, we can see that the formation of the MV (~0600?1800 UTC 9 October) was accompanied by low-level divergence and a decrease in low-level vorticity. From this perspective, the formation of the MV hindered the genesis. However, if we only focus on the period when the MV was well-defined, we can see that the MV was always situated in the center area of pre-Megi (Figs. 3d-l), and the coupled low-level cold core under it favored the accumulation of CAPE, which was important to the convection and then the spin-up of low-level cyclonic circulation. Since the MV was always situated in the pre-Megi’s center area after it formed, it could be regarded as the precursor disturbance of Megi, just as it is in the literature.
In this work, we found that the fluctuating growth of the MV after it formed was primarily induced by the diurnal cycle of the deep convection. Although the diurnal cycle of the deep convection in tropical cyclogenesis has been mentioned in many previous works, the reason responsible for this phenomenon is still not clear. The preliminary results derived from the sensitivity experiment without the solar radiation indicate that, even though the solar radiation was totally turned off, the diurnal cycle of the deep convection could still happen in the pre-Megi disturbance. Therefore, this issue is an interesting topic that deserves further discussion in the future. In addition, the sensitivity experiment performed in this work was not really physically realistic because there is always some degree of evaporative cooling in the real atmosphere. It is necessary to design more reasonable sensitivity experiments to discuss the influence of the MV on tropical cyclogenesis in the future.
Acknowledgements. The authors are grateful for the constructive comments provided by Dr. Fuqing ZHANG and the three anonymous reviewers. This work was supported in part by the National Key Research and Development Program of China (Grant No. 2017YFC150 1601) and the National Natural Science Foundation of China (Grant No. 41875067). Computations were performed at the Texas Advanced Computing Center and at the High Performance Computing Center of Nanjing University.
