Tropical cyclogenesis associated with premonsoon climatological dryline over the Bay of Bengal

Tropical cyclones of the Bay of Bengal (BoB) that formed near the synoptic-scale dryline usually intensified over a distance of 600–800 km within 3 days and caused severe destruction after landfall. High-resolution simulations of very severe cyclonic storms in association with dryline indicate that the meridional shear aids in the development of a linear-shaped group of convective cells that mature as an east–west oriented quasi-linear convective system (QLCS) within the boundary between the dry-moist air masses. The leading edge deep convections are supported by low-level moist southwesterly inflow; however, the typical mid-level mesoscale convective vortex (MCV) associated with these QLCS is unremarkable due to a very narrow trailing stratiform region within the QLCS. Supercells are likely to be organized within the QLCS due to extremely unstable atmospheric conditions resulting from a strong vertical shear of 27–39 m s−1 between 0 and 6 km and large convective available potential energy of > 3000 J kg−1. The vertical shear veering with height causes several numbers of low-level mesovortices having diameters less than 10 km at the leading edge in the different convective stages of the QLCS. The dryline aloft in the BoB produces horizontal positive shear vorticity of the order 10–5 s−1 with higher values in the levels 850–600 hPa. The advection of intense cloud-scale cyclonic mesovortices (~ 10–3 s−1) assists and enhances a cyclonic vortex to the rear side of the QLCS that performs as an MCV for cyclogenesis over the BoB.


Introduction
The genesis of a tropical cyclone (TC) follows a hierarchical environment. The appropriate climatology and proper synoptic conditions primarily favor initiating tropical disturbances or cloud clusters over oceans (Gray 1998). Within these tropical disturbances, the triggering of mesoscale convective systems (MCSs) is the fundamental precursors of cyclogenesis 1 3 in any ocean basin (e.g., Zehr 1992;Harr et al. 1996;Bister and Emanuel 1997;Ritchie and Holland 1997;Simpson et al. 1997;Gray 1998;Houze 2010). MCSs are organized cloud clusters containing both deep convective cells and stratiform clouds and precipitation (Houze 2004). Once, MCSs were usually classified in terms of either the organizational modes (Bluestein and Jain 1985) or the location of the stratiform precipitation (Parker and Johnson 2000). More recently, Gallus et al. (2008) divided MCSs into nine morphologies for warm-season convection: nonlinear systems, cellular convection (three types; isolated, clusters, and broken lines), and linear systems (five types; bow echoes, squall lines with trailing/leading/parallel/no stratiform rain). However, only a few studies (Bister and Emanuel 1997;Powers and Davis 2002;Akter 2015) have examined the types of MCSs that occur during the formation of a tropical cyclone (TC) that typically develops in warm summer seasons.
Many studies have observed that a mature or decaying MCS comprises a warm-core cyclonic circulation called mesoscale convective vortex (MCV) that develops in the stratiform precipitation, which is typically 50-300 km in diameter and several kilometers deep (e.g., Bartels and Maddox 1991;Davis and Weisman 1994;Davis et al. 2004; Davis and Trier 2007;James and Johnson 2010). An MCV may persist for long after the dissipation of the parent convective system and keeps supporting new storms for several days (e.g., Bosart and Sanders 1981; Bartels and Maddox 1991;Fritsch et al. 1994;Trier and Davis 2007). According to the AMS glossary, an MCV is formed by the latent heat release in the stratiform region of the MCS (American Meteorological Society 2019). The lower-tropospheric evaporative cooling and stratiform condensational heating below the upper troposphere allow the growth of a mid-tropospheric potential vorticity anomaly within the squall line MCS (Hertenstein and Schubert 1991). However, different studies have provided different explanations for the mechanism of the individual MCV formation. The generally accepted processes include: vortex stretching mechanism by the mesoscale vertical motion (Bartels and Maddox 1991); tilting of the ambient or baroclinically generated horizontal vorticity resulting in vertical vorticity and enhancement of the mid-tropospheric convergence due to rotation of the earth, especially the cyclonic vortex is dominated on the northern end of the asymmetric squall line MCSs (Davis and Weisman 1994;Skamarock et al. 1994); and reduction in the local Rossby radius that induces mid-to low-level convergence through localized warming by organized convection (Chen and Frank 1993). As a whole, the fundamental formation of an MCV depends on four atmospheric regimes: convergencedominated, heating gradient-dominated, heating-dominated, and vertical advection-dominated regions working from the bottom to the top of the troposphere, and a variation in these atmospheric regimes affects the origin and intensification of the MCV in different developmental paths (Kirk 2003;. The MCVs associated with the MCSs that form over the tropical oceans serve as a nucleus for a TC and are correspondingly the essential precursors to its genesis (e.g., Zehr 1992;Harr et al. 1996;Bister and Emanuel 1997;Ritchie and Holland 1997;Simpson et al. 1997;Reasor et al. 2005;Houze 2010). The MCVs contribute to the development of the surface vortex for TCs either due to a downward vortex penetration through the mesoscale subsidence in association with the stratiform precipitation (the top-down process) (Emanuel 1993;Bister and Emanuel 1997;Ritchie and Holland 1997) or a convective towerdriven development known as a bottom-up process (Montgomery and Enagonio 1998;Hendricks et al. 2004;Reasor et al. 2005;Montgomery et al. 2006;Braun et al. 2010). In the latter case, aggregation of low-level vortices associated with rotating deep convection (vortical hot towers; VHT) intensifies the low-level cyclonic vortex within the embryonic mid-level vortex via a wave pouch (Dunkerton et al. 2009). According to the marsupial theory, a wave pouch is formed before the TC genesis, providing it a "sweet spot" which is also supported by a recent study (Gjorgjievska and Raymond 2014).
The previous study by Akter (2015) revealed that MCSs connected with cyclogenesis over the Bay of Bengal (BoB, a subregion of the North Indian Ocean) portray a specific nature depending on the seasonal variability. The large-scale dynamics and thermodynamics conditions over the BoB permit bimodal cyclone seasons: before and after the monsoon season (June-September) (Akter and Tsuboki 2014). Due to the season location of the monsoon trough, the temporal pattern of TC formation over the BoB is unique compared with the other ocean basins where only the summer season is favorable for cyclone development (McBride 1995;Harr and Chan 2005;Camargo et al. 2007;Kikuchi and Wang 2010;Akter and Tsuboki 2014). The advection of seasonal winds toward the BoB favors the formation of distinctive MCSs in terms of structure, movement, and severity during cyclogenesis (Akter 2015). The development of a severe bow-echo-type MCS in an environment of high CAPE and strong clockwise shear was reported in connection with the cyclone Akash (2007), which was formed during premonsoon (March-May) in the presence of a dry-moist coupled environment, i.e., dryline in the BoB (Akter and Tsuboki 2017). According to the Joint Typhoon Warning Center (JTWC) best track data, cyclone Akash (2007) changed from a depression-level (maximum surface wind speed, MSW < 25 kt) to a very severe cyclonic storm (MSW of 65 kt) in just 1.5 days. This rapid intensification within a short period is unusual and a threat to the low-lying coastal countries like Bangladesh and India. It is thus essential to understand the fast and intense growth of TCs that form along the dryline in the BoB to identify the behavior of MCSs for more cases and to investigate the associated MCVs related to the TC formation.
To the best of our knowledge, no studies have been conducted before the climatology of cyclone formation associated with dryline, which is critical for severe weather formation. Therefore, the main objective of the present study is to find out the favorable conditions that allow the MCS formation and to analyze the MCV development process responsible for the dryline-related cyclogenesis over the BoB. Due to a dearth of observed data, a highresolution simulation is the only conceivable method for investigating the detailed structure and mechanism of the MCSs and MCVs during the formation of TCs in the BoB.

Synoptic-scale dryline and premonsoon cyclones
A large-scale prominent dryline, the boundary between dry and moist air masses, occurs during premonsoon season (March-May) along the east coast of India and adjacent BoB from 1990(Akter and Tsuboki 2017. The dryline has a horizontal length of ~ 2000 km with an intense dew point gradient of 1 °C per 10 km at the surface. Severe local convective storms (Nor'wester) and tornadoes over the land are most likely near the premonsoon dryline, which causes destruction, damage, and a large number of deaths each year (Ono 1997;Yamane et al. 2009;Bikos et al. 2016). Moreover, the vertical structure of the dryline shows a dry-moist air convergence zone aloft sloping from the east at a level of approximately 700 hPa over the BoB to the west at the surface of the eastern coast of India. As a result, the northern and northwestern BoB contains dry air, with the moist air kept in the southern and southeastern regions of the BoB (Akter and Tsuboki 2017). In this study, the zero anomalies of specific humidity are displayed in Fig. 1a which is averaged from the surface to 700 hPa during the premonsoon from 2001 to 2020 for indicating the synopticscale dryline over the BoB. The dryline evolves when the dry (less than 6 g kg −1 ) and hot (maximum of 28 °C) air is coming from the north and northwest, whereas the moist (more than 14 g kg −1 ) and warm (minimum of 19 °C) air is entering from the southwest to the BoB (Fig. 1b). The distribution of temperature is not shown in Fig. 1; however, this result is consistent with previous studies for the environment of the BoB during premonsoon (Akter and Tsuboki 2014, 2021. The north and northwesterly dry air masses are very deep and are stretched from the surface to at least 500 hPa; on the other hand, moist southwesterlies are shallow, i.e., up to 700 hPa pressure level from the surface. The dry-moist boundary forms a lifted and inclined dryline over the BoB. In the last 20 years (2001-2020), a total of 18 cyclones occurred during the premonsoon season with the maximum TC frequency in May (61%) and the minimum in March (6%). The frequencies of premonsoon TCs are almost half that in the postmonsoon season (October-December) due to the presence of a stable layer over the northern BoB; this inhibits the development of TCs even though the monsoon trough is located in the northernmost region of the Bay (Akter and Tsuboki 2014). The larger values (> 50 J kg −1 ) of convective inhibition (CIN) shown in Fig. 1a support the stable environment aloft in the northern and northwestern BoB due to deep hot, dry air.
Within the premonsoon TCs, four out of nine very severe TCs (MSW ≥ 65 kt), i.e., BoB_02 (2004), Akash (2007), Aila (2009), and Mora (2017), developed around the strong gradient where the dry-moist air masses congregated. Akash (2007) was simulated by the author earlier to examine the formation and characteristics of the MCS during its genesis period (Akter 2015). The remaining three TCs have been simulated in the present study to analyze the details of their MCSs and, in addition, to identify the causes for MCVs (core vortex of TCs) as a unique attempt in the presence of the dryline. All these three TCs have common basic criteria, i.e., all formed in May and appeared as "very severe cyclonic storms" of the Indian Meteorological Department (IMD) cyclone intensity scale. BoB_02  respectively. They reached the maximum intensity within a short distance (~ 600-800 km) from their depression locations within ~ 3 days. The tracks of these cyclones are shown in Fig. 1a. Brief information of these TCs is listed as follows: • BoB_02 (2004)

Model configurations and data used
Three premonsoon cyclones BoB_02 (2004), Aila (2009), and Mora (2017) that formed over the BoB (Fig. 1a) were selected for the simulations using the Advanced Hurricane Weather Research and Forecasting (WRF) Model (AHW), version 3.7.1 Skamarock et al. 2008). The simulations were performed during the genesis of each TC to improve the understanding of the formation process and the characteristics of the MCSs and MCV connected with cyclogenesis. To manage the surface fluxes within the TCs properly, a one-dimensional mixed-layer ocean model based on Pollard et al. (1973), surface drag coefficients described by Donelan et al. (2004), and enthalpy exchange coefficients described by Dudhia et al. (2008) were also selected for the AHW model. All the configurations in the simulation were preferentially retained from the previous study (Akter 2015). The selected physical parameters in this previous study were followed according to Raju et al. (2011) and Krishna et al. (2012), who adapted all the parameters for simulating cyclones over the BoB using the WRF model. A two-way nested simulation was facilitated with a horizontal resolution of 12 km and 4 km, respectively, for the outer and inner domains. The outer domain (D1) with 861 × 595 grid points covers the area 0-30°N and 70-110° E (not shown here but same as Fig. 1 in Akter 2015); however, the inner domain (D2) with 1064 × 996 grid points includes 5-22°N and 80-97°E (area of Fig. 2). The vertical resolution was defined as 30 terrainfollowing levels with a top pressure level of 50 hPa. The six-hourly National Center for 1 3 Environmental Prediction Final analysis (NCEP-FNL) data (https:// rda. ucar. edu/ datas ets/ ds083.2/) with a resolution of 1° × 1° were applied as the atmospheric initial and boundary conditions. In addition, the sea surface temperatures were implemented daily using the NCEP high-resolution (0.083°) real-time global sea surface temperatures (RTG_SSTs) data (https:// polar. ncep. noaa. gov/ sst/); moreover, no bogussing was imposed for the TC initialization. In the model setup, the map projection, convective parameterization, microphysics, planetary boundary layer (PBL), surface physics, land surface model, and longwave and shortwave radiation processes were considered according to the Lambert projection, Kain-Fritsch scheme (Kain 2004; only for the parent domain), Ferrier scheme (Ferrier et al. 2002), Yonsei University scheme (YSU, Hong et al. 2006), Monin-Obukhov scheme (Monin and Obukhov 1954), Noah land surface model (Chen and Dudhia 2001), rapid radiative transfer model (Mlawer et al. 1997), and Dudhia scheme (Dudhia 1989), respectively. A brief description of all the processes has been given in Akter (2015).
The intensity and position data for the TCs were collected from the six-hourly best track data of the JTWC (https:// www. metoc. navy. mil/ jtwc/ jtwc. html? north-indianocean). According to these data records, each TC needed less than 2 days for rapid intensification from a tropical depression into a cyclonic storm which is a system regulated by a self-sustaining mechanism with an MSW of at least 34 kt or ~ 17.5 m s −1 (Zehr 1992;Mcbride 1995). Therefore, 4 days excluding the 6-h model spin-up time were allocated as the simulation period in each case: three for the period before the  The NCEP climate forecast system reanalysis (CFSR) six-hourly data (https:// rda. ucar. edu/ datas ets/ ds094.0/) with a horizontal resolution of 0.5° × 0.5° were used to verify the model outputs; its monthly mean data were also utilized to analyze the seasonal variability of the environment over the BoB. The data provide the vertical resolution of 64 hybrid sigma-pressure levels, with a top pressure of ~ 0.266 hPa (Saha et al. 2010). Though JTWC best track data were available and used in this study, the intensity scale of TCs was mentioned according to the IMD.

Model verification
The simulated results at the resolution of 4 km were verified with the NCEP-CFSR six-hourly reanalysis data shown in Fig. 2. The shaded area represents the surface convective available potential energy (CAPE, Glickman 2000), which is averaged from the depression-to-genesis period of each TC. The total period from depression-to-genesis varies for each TC mentioned in the previous section. In this research work, the primary objective was to understand the MCSs associated with pre-TCs. Thus, the distribution of the pre-cyclone environmental CAPE was verified as it is directly related to the initiation of deep convection (e.g., Williams and Renno 1993;Bhat et al. 1996;May and Rajopadhyaya 1999;Rapp et al. 2011), including that in the TC environment (Molinari and Vollaro 2009;Molinari et al. 2012;Lee and Frisius 2018). The simulated CAPE was found to be similar in distribution to that in the reanalysis data except for Mora (2017); the larger values of CAPE in the north and northwestern areas of the BoB are especially well projected in all cases. The low-pressure center and corresponding rotation of the wind were achieved satisfactorily in the simulation result. The intensity and position of each simulated cyclone were compared with the JTWC data at its genesis. In the simulation, the minimum pressures of BoB_02 (2004)  In this study, the simulated rainfall for Aila (2009) is compared in Fig. 3 with the available rainfall data collected from Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) 2A12 version-6 because Aila (2009) has been represented as a case study to understand the formation of MCV in depth. A strong northern rainband of Aila (2009) is observed approaching the Bangladesh coast in both cases; however, the time achieved in the simulation is 12 h later than the actual time at 2118 UTC May 23, 2009.

Pre-cyclone environment and convection initiation
The reflectivity in Figs. 4a-c collected from the model outputs shows the early MCSs for all pre-cyclones. The initiation time of MCSs differs from cyclone to cyclone. In all cases, early MCSs are considered as a group of cells with a maximum intensity of 35-40 dBZ. The moisture advection by a strong southwesterly wind triggers the convection along the convergence zone of dry-moist air masses encompassing minor temperature differences (~ 1 K). The temperature gradient in the BoB is not sufficient to decide upon a typical front that describes a moderate baroclinic zone (Sanders 1999) with a threshold value in temperature change being 8 °C/220 km −1 . Here, the warm moist southwesterly flow having a high equivalent potential temperature θ e of ~ 350 K (on average for all TCs) is lifted by the density current (Lin 2007) associated with low θ e (~ 330 K) air advected from the hot dry north and northwesterly. The intrusion of denser dry air beneath a lighter moist air helps raise the low-level air parcel. The vertical cross sections of θ e along the low-level convergence zone at 13.5°N for pre-BoB_02 (2004), at 11°N for pre-Aila (2009), and at 12.8°N for pre-Mora (2017) are shown in Fig. 5 which indicate that the southwesterly air with θ e of ~ 354 K is converged near or below 850 hPa and elevated at the lower level by the north and northwesterly air having θ e of ~ 340 K. The low-level warm moist air is potentially unstable with the gradual decreasing of θ e ; on the other hand, the increasing of θ e with height supports a very stable layer for the hot dry air aloft.
The instability that aids in the convection development in each pre-cyclone period is calculated from a Skew-T log-P diagram and hodograph at the location noted in Fig. 4a-c with the stars. The positions are chosen adjacent to the convection near the dryline at the zero anomalies of temperature so that the environment of the dry-moist boundary can be identified properly. The Skew-T log-P diagrams are not displayed here. However, the values of different atmospheric indices for all cases are noted in Table 1, which show that the K index (George 1960) measuring the thunderstorm potential is approximately 40 °C, the value of precipitable water (PW) is more than 6 cm, the lifted index (LI; Galway 1956) is greater than -6, the total totals index (TT; Miller 1972) is > 40 °C, and θ e is ~ 370 K. These  UTC May 24, 2009, for Aila (2009 values indicate very unstable air masses to be lifted to form strong or severe organized thunderstorms with great potential for heavy rain (Grieser 2012). For all cases, the vertical total wind shear from 0-6 km above ground level (AGL) was found to be 27-39 m s −1 and bulk shear between the surface and 6-km height is 12-14 m s −1 ; these are within or greater than the threshold value for a typical supercell formation (Bunkers 2002). CIN is found 0 J kg −1 , whereas, strong shear with clockwise veering and a high CAPE of more than 3000 J kg −1 support extreme unstable conditions conducive to supercell initiation (Weisman and Klemp 1982;Bluestein 1993). In addition, the moderate value of storm-relative environmental helicity (SREH, Davies-Jones et al. 1990) and the bulk Richardson number

Characteristics of mature MCSs
Over a period of several hours (6-12) from the formation, an early MCS matures as a quasi-linear convective system (QLCS) or comma shape convection with a series of embedded echoes having maximum reflectivity more than 50 dBZ, as shown in Figs. 4d-f for each case. Based on the potentiality of the environment and low-level SREH, supercell thunderstorms are likely embedded within the organized QLCS. In the figures, the 500-hPa geopotential height accompanied by the wind indicates that mid-level air is rotating near or at the center of the QLCS and the low value of geopotential height indicates a mesovortex with a central low pressure.  The structural details of a pre-Aila (2009) QLCS in horizontal and vertical dimensions during the matured stage are incorporated in Fig. 6. The positions of the QLCS at 1030 UTC and at 2030 UTC 21 May 2009 are displayed with green and red hatched areas, respectively, in Fig. 6a. The QLCS, oriented approximately in the east-west direction, has a horizontal length of ~ 500 km and remains quasi-stationary with a little movement toward the southeast. The movement of the QLCS is approximately calculated by the perpendicular movement of the center of the convective line (i.e., horizontal line connecting with deep convective cells of 50 dBZ within the QLCS) as similar to Akter (2015). The rear inflow jet (RIJ, Weisman 1993) shown in the figure coming from the north and northwest assists the MCS to be a comma-shaped structure. Figure 6b shows the vertical cross section of reflectivity, updraft, downdraft and system-relative motion along AB in Fig. 6a, which indicates that the strong low-level southwesterly inflow converges and supplies moisture into the system and consequently deep convection containing maximum precipitation of 60 dBZ develops along the leading edge or south side of the system. The system height is extended up to the pressure level of 300 hPa where the RIJ is flowing between 850 and 500 hPa levels. Updrafts with a maximum speed of 6 m s −1 are observed, which are separated from the downdrafts descending at a maximum speed of 1.5 m s −1 . Moreover, after 10 h the same vertical analysis along CD in Fig. 6c represents the dissipation of older convective cells that develops a narrow stratiform precipitation region extending toward the rear side and a new convective cell appears at the leading edge which also can be confirmed by the strong low-level convergence in Fig. 6d. During the mature stage, the interaction of low-level (surface to 3 km) shear of 20 m s −1 with a weak cold pool supports forced lifting of air, ensuring new cell development along the down shear in the organized systems of QLCS, similar to observations in previous studies (Rotunno et al. 1988;Weisman and Rotunno 2004). The potential temperature (θ) anomalies in Fig. 6d suggest that the 1.5-km-deep cold pool has the lowest temperatures of -4 K, which is sufficient for developing tropical convection (COMET Program/UCAR 1999). The characteristics of the MCSs for pre-Aila (2009) and the other two pre-cyclones (pre-BoB_02 (2004) and pre-Mora (2017)) are in agreement with the features of pre-Akash (2007) mentioned in Akter (2015), except for their movement speeds. However, most MCSs remain stationary from their early stage to mature, which can be confirmed by the position of the initially formed MCSs and their mature phases in Fig. 4. In all cases, the lifetime of MCSs are very long (~ 20 h) and within this lifetime several cloud-scale mesovortices (vorticity at meso-γ-scale: 2-20 km ;Orlanski 1975) are formed at low levels along the convective line of QLCSs Weisman and Trapp 2003;Atkins and Laurent 2009a). Figure 6a displays cloudscale cyclonic vortices for pre-Aila (2009) with an approximate horizontal length of 10 km or less. Further additional features of the mesovortices are described in the next section.

Formation of mesovortices
The shade in Fig. 7a represents the meridional wind component, which indicates that when the wind coming from the south initially intrudes into the northerly wind, an early MCS (a group of convective cells) for pre-Aila (2009) is initiated in the east-west direction along the horizontal shear line. The convective cells are embedded in a strong vertical shear environment where the strong southwesterly flows beneath the weak north and northwesterly (Fig. 7b). Consequently, low-level mesovortices with horizontal diameters of ~ 6-8 km develop along with the individual convective cells under the effect of an environmental low-level (mid-level) vertical wind shear of ~ 20 m s −1 (~ 27 m s −1 ) within 0-3 km (0-6 km) AGL. A similar mesovortex genesis in the mature phase of QLCS has been noticed in the previous studies Weisman and Trapp 2003). Moreover, Atkins and Laurent (2009b) observed mesovortices during different stages of the convective system including the early cellular stage. The mesovortices are extended vertically due to the tilting of the crosswise horizontal vorticity by the updraft, which results in a mesovortex pair. Figure 7b displays the vertical vorticity distribution showing the mesovortex pair (named "X") due to the growing updraft with a maximum velocity of 7 m s −1 accompanied by a thunderstorm displaying a vault structure in the vicinity of the updraft. The vortex pair is aligned in the north-south direction with a cyclonic rotation in the north (to the right of the vertical shear vector) and an anticyclonic in the south (to the left of the vertical shear vector), similar to a supercell mesocyclone (Klemp 1987). Previous studies have found that mesovortices can produce short-lived tornadoes but are weak with respect to the damaging straight-line winds linked with mesovortices . The highest positive (negative) vertical vorticity of 2.8 × 10 −3 s −1 (-1.7 × 10 −3 s −1 ) develops approximately at the 700 hPa level because of the strongest vertical shear within this layer (shown by the wind vector) in association with dryline boundary aloft in the BoB.

Movement and life cycle of mesovortices
The vortex couplet "X" located at 11.4°N, 83.7°E in Fig. 7a is found moving in the southeast direction. Subsequently, after one and a half hours, i.e., at 0600 UTC May 21, 2009, the vortex couplet takes a position at 11.33°N, 84.12°E, as shown in Fig. 8a. In Fig. 8b, the vertical cross section along AB displays the vertical vorticity and strong updraft (maximum of 13 m s −1 ) extending up to 200 hPa, which amplifies the vortex pair by the stretching of vortex tubes . During the mature stage of the vortex pair, a maximum downdraft of -1.5 m s −1 is observed, which remains separated from the updraft, indicating the presence of supercell characteristics (Lemon and Doswell 1979;Davies-Jones 1984). The height of the vortices is calculated as 9 km approximately. The maximum vertical vorticity of ~ 3 × 10 −3 s −1 at the mid-level, connected with the intense updraft, indicates the possibility of the presence of a low-level mesocyclone in the tropical environment.
Moving of "X" toward the southeast direction or the rear side within the QLCS and the development of a new convective cell at the west side of "X" (Fig. 7a) inhibit the inflow of warm air to the old cell, which induces a reduction in the updraft as well as in the cell growth. Under these circumstances, the vortex pair is going to be weakened. Figure 8c-d displays the horizontal and vertical structures of "X" that represent the dissipating stage at 0930 UTC May 21, 2009, where the updraft is diminished. The positive or cyclonic vortex of "X" is continuing its movement a little toward the rear side of the QLCS after dissipating the negative vortex. The vortex pair reaches the average speed of approximately 7 m s −1 during its lifetime of ~ 6 h. Figure 9a exhibits the life span of three individual cyclonic mesovortices indicated by A, B, and C at 1-h intervals for understanding their movement directions. All vortices are developed along the convective lines of QLCS represented by the solid lines. The vortex "A" appeared at 1000 UTC and dissipated at 1900 UTC May 21, 2009. Similarly, "B" and "C" both had lifetimes from 0800 to 1700 UTC May 21, 2009. It is clear that the system remained almost stationary for a long period from 0900 to 2100 UTC May 21, 2009; however, cyclonic vortices that developed in the convective side gradually moved to the rear side of the QLCS. The arc-shaped movement tracks of the mesovortices and their speeds depend on the mean wind which is dominated by the southwesterly wind (Fig. 6a) and their speeds are influenced by the intensity of the wind. The Hovmöller diagram of average vorticity advection from 1000 to 500 hPa for the area of 84-88°E during the period from 0330 to 2230 UTC 21 May 2009 (Fig. 9b) also supports the advection of cloud-scale cyclonic values of vorticity to higher latitudes and intensification of the vorticity at the north as a consequence.
A similar development of vortices is frequent along the shear line or leading edge of the QLCS at different convective stages but the highest frequency is observed in the mature stage. The characteristics of all mesovortices are not similar to "X." They are more or less intense in terms of value, height, and lifetime of the vertical vorticity. The maximum height of the vortices varies from 600 to 200 hPa above the surface and the lifetime ranges from 5 to 10 h. Some vortices are found to be merged with others, whereas some split during their lifetime. In particular, the characteristics of the mesovortices depend on the associated convective cells embedded in the QLCS. In this study, all types of cloud-scale vortices are considered as mesovortices based on the horizontal diameter.

Mesoscale convective vortex for pre-cyclones
By definition, MCV is a low-pressure center at mid-level that develops in the stratiform rain region of an MCS due to latent heat release. The origin of tropical cyclone circulations is suggested to be the MCV embedded within a low-level vorticity-rich environment (Houze 2004). Dunkerton et al. (2009) referred to it as a "pouch" where a closed cyclonic circulation is protected by a large-scale environment. In this study's simulation, the MCSs for all pre-cyclones were convective for the most part, where the stratiform precipitation regions were observed as very narrow (Figs. 4d-f). The linear convective system containing narrow or no-stratiform precipitation region, which depends on storm-relative elevated front-to-rear flow, has also been noticed previously (Gallus et al. 2008;Lombardo and Colle 2013). Figure 3a also agrees with the existence of a very narrow stratiform precipitation region in the rainband of Aila (2009) imaged from the satellite data. Further, the weak convergence and smaller negative geopotential height anomalies in the stratiform region shown in Fig. 6d do not support any significant low pressure in the middle level, which can be representative of MCV for cyclogenesis.
The MCS of pre-Aila (2009) and associated low-level moisture advection by the synoptic-scale flow are displayed in the left panel in Fig. 10, whereas the right panel shows the zonal and vertical average of the wind and potential temperature (θ). In the figures, the QLCSs that mature within the zone of convergence have emerged in an environment of potential temperature gradient of ~ 2 °C per 100 km and within the changing fluxes of ~ 150 g m -2 s −1 between the dry air in the north and moist air from the south. The streamline and also the average wind in the zonal direction indicates a meridional shear along the dry-moist boundary that causes shear vorticity in the synoptic scale.
In addition, the streamline indicates a counterclockwise vortex growing at 1330 UTC due to the spinning dry and hot air in the northernmost part of the MCS, which further enhances, as shown after 6 h at 1930 UTC May 21, 2009. The variations of average vorticity (both relative and shear) for the two equal areas (A and B indicated in Fig. 10) with respect to time are calculated and displayed in Fig. 11a for the period of 0430 UTC to 2330 UTC May 21, 2009. During the majority of these 19 h, the lower area "A" covers the entire convective parts of the MCS, whereas the upper area "B" does not include the MCS after the initial few hours. At the initial stage of the MCS, both areas have almost similar intensities of shear vorticity (~ 2 × 10 −5 s −1 ) or relative vorticity (~ 4 × 10 −5 s −1 ) that are supposed to come from the shear instability along the dryline. During the mature stage at approximately 1200 UTC, the cyclonic vorticity starts increasing in area B by the advection of strong cloud-scale positive vortices to the northern side of QLCS (Fig. 9b). In other words, the mesovortices that formed in the deep convection within the convergent boundary move to the rear side of the QLCS enhancing the relative vorticity to a maximum of 8.7 × 10 −5 s −1 in area B. The linear trend in the figure clearly shows that the relative vorticity in area B is increasing significantly after 1100 UTC with the value of the coefficient of determination (R 2 ) of 0.62, which represents a good fit of the line to the data. On the other hand, in area A the vorticity remains constant or decreases very little even though the area contains the convection. Moreover, a similar pattern in the variations of vorticities confirms that the contribution of horizontal shear vorticity mainly influences the relative vorticity with the curvature vorticity remaining nearly constant (~ 2 × 10 −5 s −1 ) through Fig. 10 Mature-stage MCS (shaded; dBZ) and streamline with moisture flux (color levels; g m −2 s −1 ) at 925 hPa in the left panel. Zonal (82-89°E) and vertical (1000-500 hPa) averaged wind (red line; m s −1 ) and potential temperature (blue line; °C) in the right panel at a 1330 UTC May 21, 2009, b 1930UTC May 21, 2009 the entire period. The values of cyclonic vorticity are higher in the level 850-600 hPa and the dimension of the vortex is approximately ~ 200 km, which is similar to a typical mid-tropospheric MCV in the rear side of the QLCS (Houze 2004(Houze , 2010. Similar results are also observed for the pre-BoB_02 (2004) and pre-Mora (2017) cyclones shown in Fig. 10b and c, respectively. In both cases, the relative vorticity starts to increase during the mature convective stage and shows positive trends (R 2 = 0.95 for BoB_02 and R 2 = 0.94 for Mora) for the northern areas of the pre-cyclones but shows a negative or constant trend for the southern areas. The area coverage for both the precyclones is mentioned in the figure and have been selected according to the position of MCSs in each case ( Fig. 4d and f).
Therefore, during the premonsoon, the meridional shear connected with the dryline forms positive vorticity (~ 10 -5 s −1 ) from the lower to middle levels in the synoptic-scale and the vertical wind shear influences a number of mesovortices to be produced at the lower level. Fast spinning, intense mesovortices (~ 10 -3 s −1 ), and their positive advection trigger and enhance a cyclonic vortex to the north of QLCS. However, for cyclogenesis, Hendricks et al. (2004) introduced a term VHT (horizontal diameter ~ 10 km), which characterizes the highly positive vorticity centering within an extended updraft, and an early MCS begins with a set of one or more isolated deep VHTs (Houze 2010). Among the mesovortices, some may form into VHTs, which are not separately identified here, because all meso-γ-scale vortices are considered as mesovortices at the cloud scale. In this study, the TC formation process was observed to have similarity with the bottom-up paradigms (Hendricks et al. 2004;Reasor et al. 2005;Montgomery et al. 2006;Braun et al. 2010) rather than with top-down processes (Emanuel 1993;Bister and Emanuel 1997;Ritchie and Holland 1997). The vorticity along the dryline does not penetrate downward to form a surface vortex, whereas, within the embryonic environment of shear vorticity, low-level mesovortices are formed, integrated, and finally turned into a cyclone vortex.
In brief, during the premonsoon, an MCS is initiated along the shear line with intense vertical low-level mesovortices, which are the fundamental constituents following the bottom-up paradigms for cyclogenesis over the BoB.

Summary
Bimodal cyclone distribution in the Bay of Bengal is one of its unique characteristics, which is primarily associated with seasonal monsoon trough locations. Another important climate feature is the presence of dryline aloft over the northern BoB during the premonsoon season. Cyclones formed near or along the dryline quickly evolve to severe TCs within a very short distance in a small time period and frequently have severe consequences after landfall. On that account, the formation of TCs that develop near the dryline is analyzed in this study during 2001-2020. As a representative of dryline-associated TCs, three very severe cyclonic storms, i.e., BoB_02 (2004), Aila (2009), and Mora (2017), are simulated during their genesis period using the WRF model with the highest resolution of 4 km in the inner domain. The simulated results are then validated by comparing them with datasets of NCEP_CFSR reanalysis and JTWC best track data. The events are well simulated by the model, which allows the analysis of mesoscale phenomena associated with cyclogenesis. As the MCS is one of the primary characteristics that describe the activities of cyclogenesis, the convective systems and related environments for pre-BoB_02 (2004), pre-Aila (2009), and pre-Mora (2017) are examined in detail. Finally, the formation process of an MCV, which is the nucleus of a cyclone vortex, is discussed. The important findings are summarized below:

Environment
During premonsoon, a synoptic dryline, which incorporates a deep hot, dry air flow in the north and northwest and a shallow warm, moist air flow in the south and southeast of the BoB, causes temperature and moisture gradients over the BoB. The dryline over the BoB is basically an inclined moisture gradient aloft appearing when shallow air masses penetrate the deep layers. The elevated dryline not only causes a horizontal shear in the meridional direction but also produces a vertical wind shear due to the change in the strong southwesterly of the lower level to an upper level weak northwesterly wind. The boundary between the two air masses with different densities helps the high θ e flow from the southwesterly to be uplifted for initiating updrafts. As a result, a group of linear convective cells originates in an environment of deep-layer strong clockwise shear and large CAPE, which favors the formation of supercells. The environmental conditions calculated from the indices are observed to be potent, which leads to the initiation of severe thunderstorms capable of strong wind, rotating updrafts, and heavy precipitation.

MCS
Due to the influence of linear forcing along the boundary of the dryline, the early linear MCS remains nearly stationary for longer periods and matures as a comma shape MCS or QLCS within the zone of enhanced convergence. Accordingly, the QLCS appears as a large system with a horizontal length of ~ 500 km oriented approximately in the east-west direction. The system comprises leading-line deep convections including supercells and follows a very narrow trailing stratiform region. The southwesterly inflow facilitates low-level moisture to the QLCS supporting a convective line in the south of the system, whereas 850-500 hPa dry air coming from the north and northwest functioning as the RIJ assist the QLCS to turn into a comma-shaped convective system. Evaporative cooling accelerates the downdrafts that support a weak cold pool on the surface extending from the leading edge rearward. The weak cold pool and strong shear instability generate new cells in the front of the system in a tropical environment. Once developed, the individual mature deep convections move to the rear side extending the trailing stratiform precipitation region to the north of the QLCS. The stronger shears lead to a longer QLCS life, which is approximately 20 h or more.

MCV
According to the results of this study, the mid-tropospheric conventional MCV that is associated with the stratiform rain region within the quasi-linear MCS is not significant enough to be a part of the cyclogenesis. The uplifted dryline over the BoB produces horizontal positive shear vorticity in the synoptic-scale with high values (order 10 -5 ) in the level 850-600 hPa. Within this embryonic environment, the strong vertical shear veering with height initiates low-level mesovortices (~ 10 -3 s −1 ) with 6-8 km length in the individual thunderstorms of QLCS at the different convective stages. The tilting of ambient crosswise horizontal vorticity by the local updrafts supports the appearance of a mesovortex pair with intense cyclonic (less intense anti-cyclonic) rotation to the right (left) of the vertical shear vector. Some mesovortices have the characteristics of mesocyclones with stronger updrafts depending on the convective type embedded in the QLCS. After forming along the leading edge of the QLCS, the mesovortices at cloud-scale mature and move to the rear side due to the influence of the mean wind flow. These fast-spinning cyclonic vortices and their positive advection to the rear side trigger a cyclonic vortex in mesoscale, which performs as a core vortex similar to a conventional MCV. The core vortex continues beyond the life cycle of the QLCS, which further allows the formation of new convective systems and develops the TC vortex subsequently.
Therefore, within the cyclonic vorticity-rich environment due to the synoptic-scale horizontal shear, cloud-scale cyclonic mesovortices are the essential prerequisites responsible for the formation of the MCV in the vicinity of MCS for cyclogenesis connected to the dryline over the BoB. Embedded supercells with intense mesovortices (or mesocyclones) in a favorable environment of quasi-linear MCS may cause a rapid intensification of the TC vortex within a short distance. Further study using a cloud-resolving model is necessary to examine and comprehend the contributions of different kinds of mesovortices associated with the TCs formed over the BoB.