Dynamical Mechanisms of Rapid Intensification and Multiple Recurvature of Pre-Monsoonal Tropical Cyclone Mocha over the Bay of Bengal

Abstract

Cyclone Mocha, classified as an Extremely Severe Cyclonic Storm (ESCS), followed an unusual northeastward trajectory while exhibiting a well-defined eyewall structure. It experienced rapid intensification (RI) before making landfall along the Myanmar coast. It caused heavy rainfall (~90 mm) and gusty winds (~115 knots) over the coastal regions of Bay of Bengal Initiative for Multi-Sectoral Technical and Economic Cooperation (BIMSTEC) countries, such as the coasts of Bangladesh and Myanmar. The factors responsible for the RI of the cyclone in lower latitudes, such as sea surface temperature (SST), tropical cyclone heat potential (TCHP), vertical wind shear (VWS), and mid-tropospheric moisture content, are studied using the National Ocean and Atmospheric Administration (NOAA) SST and National Center for Medium-Range Weather Forecasting (NCMRWF) Unified Model (NCUM) global analysis. The results show that SST and TCHP values of 30 °C and 100 (KJ cm⁻²) over the Bay of Bengal (BoB) favored cyclogenesis. However, a VWS (ms⁻¹) and relative humidity (RH; %) within the range of 10 ms⁻¹ and >70% also provided a conducive environment for the low-pressure system to transform into the ESCS category. The physical mechanism of RI and recurvature of the Mocha cyclone have been investigated using forecast products and compared with Cooperative Institute for Research in the Atmosphere (CIRA) and Indian Meteorological Department (IMD) satellite observations. The key results indicate that a dry air intrusion associated with a series of troughs and ridges at a 500 hPa level due to the western disturbance (WD) during that time was very active over the northern part of India and adjoining Pakistan, which brought north-westerlies at the 200 hPa level. The existence of troughs at 500 and 200 hPa levels are significantly associated with a Rossby wave pattern over the mid-latitude that creates the baroclinic zone and favorable for the recurvature and RI of Mocha cyclone clearly represented in the NCUM analysis. Moreover the Q-vector analysis and steering flow (SF) emphasize the vertical motion and recurvature of the Mocha cyclone so as to move in a northeast direction, and this has been reasonably well represented by the NCUM model analysis and the 24, 7-, and 120 h forecasts. Additionally, a quantitative assessment of the system indicates that the model forecasts of TC tracks have an error of 50, 70, and 100 km in 24, 72, and 120 h lead times. Thus, this case study underscores the capability of the NCUM model in representing the physical mechanisms behind the recurving and RI over the BoB.

1. Introduction

A tropical cyclone (TC) is a large synoptic-scale weather system that originates over the warm oceans of the world and develops into a massive vortex composed of swirling winds, intense clouds, and torrential rains by drawing energy from the sea [1]. The North Indian Ocean (NIO) basin significantly contributes 7–10% of the world’s TCs, and these TCs are considered as deadly extreme weather events [2]. They often bring gusty winds as they make landfall, leading to storm surges along the coastlines and direct damage to lives and livestock. They can also produce extreme rainfall, leading to coastal and inland flooding, especially if rapidly intensified near the coasts.

During the pre-monsoon season, TCs are the primary source of rainfall, that produced by dense cloud bands along with heavy to very heavy precipitation over the tropics. It is reported that gusty wind and convective clouds are the essential meteorological elements that are associated with TC formation. An organized cluster of convective clouds around the central area of surface low pressure over the sea surface helps in the development of the TC. The TCs derive the energy needed for intensification through the direct exchange of sensible and latent heat from the warm ocean surface via convection. The inner regions of these storms primarily consist of cumulonimbus clouds, which form within the rotating vortex and are closely linked to the cyclone’s dynamics. While high-resolution core physics models have gained prominence, the primary objective for cyclone research remains accurately predicting the likelihood of extreme weather events and heavy rainfall at specific times and locations [3].

Genesis factors such as >27.0 C in the northern hemisphere (NH) are quite similar to the southern hemisphere (SH), as suggested by Walsh et al. [4]. However, the location of TC genesis must be >5° NH/SH away from the equator in both hemispheres, in order to balance the Coriolis Force. TCs originate in Inter-Tropical Convergence Zone (ITCZ) over the ocean between 6° to 20° NH and SH of the equator. TCs normally travel in north to northwesterly directions [5]. One of the fundamental problems in forecasting the movement of tropical cyclones is that of their path change and forming recurvature [6]. The formation of TCs during the pre-monsoon transition (PMT) over the BoB and their track variations in a warming climate are not yet fully understood. Research has identified several key factors influencing TC genesis and intensification, including SST, an 850 hPa relative vorticity, the Coriolis force, VWS, conditional instability, and RH in the lower to mid-troposphere [7,8,9].

The TC is a nearly circular, warm-cored vortex, occupying the entire height of the troposphere and extending radially many hundreds of kilometers. Continually active clouds surround the storm’s center and are organized as the eyewall and rain bands. This complex system of interacting physical processes has multiscale motions, and one of the most difficult aspects to understand is the movement of moist convection processes [10]. The most important aspect of TC prediction is its movement. Surface observations from agro (Array for Real-time Geostrophic Oceanography) floats and buoys are not dense enough to accurately track a TC center, especially over the open ocean. However, surface observations can be important, particularly if they are located near the TC center. Kimberlain and Breman (2017) [11] have developed a simple estimate of deep-layer steering, which could serve as a proxy for the actual TC motion. Through the efforts of many researchers over the decades, it appears that we now have a very good understanding of the physical mechanisms that control the displacement of TCs, referred to as steering flow (SF) followed by Kimberlain and Breman [11].

Chan [12] documented that the tracks of TCs deviated from their typical path toward the west or northwest direction, referred to as TC curvature. The most prominent reason determined for recurvature is the eastward-retreating mid-tropospheric process as discussed in Chen et al. [13]. Research on tropical cyclone (TC) dynamics, such as recurvature, has primarily focused on the Western North Pacific Ocean, given the high frequency of TCs in this region, as highlighted in Akter and Tsuboki [14]. However, the TCs that occur in the BoB region usually experience RI due to a warm SST and cause massive destruction to coastal BIMSTEC countries such as India, Bangladesh, and Myanmar due to their geographical location, low-lying topography, high population density, socioeconomic conditions, and lack of resources [15]. Thus, attention is still needed to know the pre-monsoon TC tracks and their RI in the last two decades (2004–2023).

The most common recurvature situation arises when a storm encounters an extra-tropical trough or ridge from the west. A trough can steer the storm, causing it to change direction and curve, often taking it from its original westward path into a more northeastern or easterly direction in the pre-monsoon season 14. The TC formation and intensification are influenced by several key factors, including low-level relative humidity, vertical wind shear, low-level convergence, upper-level divergence, conditional instability, and sea surface temperature. TC Nargis developed over the BoB during the pre-monsoon season in late April and early May 2008 and went on to become the most devastating storm in the recorded history of Myanmar. It made landfall on the Ayeyarwady deltaic coast on 2 May 2008, as a category four cyclone on the Saffir–Simpson scale [16, 17]. Due to the Nargis cyclone, the densely populated Irrawaddy delta was devastated. Moreover, more than 80,000 and 10,000 deaths were reported in Labutta Township, and Bogale, while ~55,000 people were missing in other coastal towns of Myanmar that reported by Swiss Re in 2009. The Mocha cyclone originated over the lower latitude of the BoB in early May and made landfall over the Myanmar coast. The cyclone affected Bangladesh and Myanmar’s coastal cities, including Yangon, which is Myanmar’s largest city and former capital. The death toll for Cyclone Mocha was 145, which is significantly lower than Nargis, but it caused about USD 2.24 billion of property damage in Myanmar, and led to the injury of more than 800 people.

The performance of NWP models varies case to case, but they are extensively used as guidance to forecasters. Various mesoscale models, namely the MM5, ETA, RSM, and WRF, being used for short-range weather forecasting have been verified, as documented in Das et al. [18]. However, recently the advanced physical parameterization and moving nested domains options available within ARW-WRF and HWRF models [19] are being used for forecasts in short–medium-range time scales. Recently, the High-Resolution Rapid Refresh Modeling System (HRRR) has also been used for TC forecasting in IMD [20]. Since 2015, at NCMRWF the NCUM model has been operational, and it provides a seamless prediction of all types of weather-related applications and warnings, such as tropical cyclones, monsoon depressions, thunderstorms and squall lines, mesoscale convective systems, gusty winds, lightning, fog, and air quality [21]. Kumar et al. [22] have discussed the forecasting skill of the NCUM model for the TCs during 2015–2019 and found a significant decrease in the direct position error (DPE) in the NCUM model, which is operational at NCMRWF and routinely used for guidance on the operational forecasters of the IMD.

This study assesses the effectiveness of the operational global NCUM model in capturing the recurvature dynamics of Cyclone Mocha, a rare pre-monsoon event in 2023, which resulted from the interaction between subtropical troughs (STs) and rapid intensification. Additionally, the model’s capability to provide large-scale insights into the tropical cyclone’s genesis, its intensification, and its interaction with mid-latitude westerlies is also assessed. Dynamical features, including the Q-vector, steering flow (SF), and associated tracks, are analyzed using both NCUM analysis and forecast data.

The paper is organized as follows: Section 2 describes the data and methodology, while Section 3 examines the climatology and landfall patterns of cyclones affecting BIMSTEC countries. Section 4 provides a detailed assessment of the NCUM model’s performance concerning Cyclone Mocha’s intensity, cyclogenesis, Q-vector, SF, track, and IVMT. Section 5 focuses on the rainfall forecast, and Section 6 presents the study’s conclusions.

2. Data and Methodology

NCUM-G is an advanced numerical weather prediction (NWP) system operational at NCMRWF since 2015, with periodic upgrades discussed in Kumar et al. [23]. It is based on the Unified Model (UM) framework, developed through the “UM Partnership” by the UK Met Office in collaboration with the Ministry of Earth Sciences (MoES), NCMRWF, India. Continuous enhancements in the NCUM system incorporate advancements in scientific and technological methodologies to improve both its global and regional forecasting capabilities. The NCUM operates at a horizontal resolution of 12 km with 70 vertical levels (extending up to 80 km). Its dynamical core employs compressible non-hydrostatic equations of motion, solved using semi-Langrangian advection and a semi-implicit time-stepping approach. Various sub-grid scale processes, including convection, boundary layer turbulence, radiation, cloud microphysics, and orographic drag, are parameterized through physical schemes, which have been refined and upgraded over time [22].

The dynamical core, often referred to as the heart of atmospheric models, is responsible for numerically solving the primitive equations. The model utilizes the advanced “ENDGame” (Even Newer Dynamics for General Atmospheric Modeling of the Environment) dynamical core. The accuracy of weather predictions in NWP models heavily depends on the initial conditions, known as the analysis, which are generated through data assimilation (DA). NCMRWF adopted the advanced 4D-Var data assimilation system from the UK Met Office for operational use in April 2012. In October 2016, the NCUM data assimilation system was further upgraded to the “Hybrid 4D-Var” approach, developed under the “UM Partnership.”

In this present study, we analyzed NCUM-G products to understand the curvature of the Mocha cyclone toward the northeast; it made landfall over Myanmar, which is a very rare case after the Nargis cyclone in 2008 [24]. The number of TCs that made landfall in Myanmar coast is very limited. The NCUM global model’s output can give a better understanding of the mechanism and cause of curvature and steering flow over BIMSTEC countries.

The merged rainfall dataset followed by Mitra et al. [25] was used to validate the rainfall. For upper air observations, radiosonde datasets at Chittagong meteorological station at latitude/longitude of 22.35° N; 91.81° E were used for the validations of the NCUM analysis and forecast. However, IMD’s best tracks and CIRA (https://rammb.cira.colostate.edu/ accessed on 2 January 2025) were used for validation.

TCHP is a measure of the ocean’s heat content, calculated by integrating the heating from the surface down to the depth where the temperature reaches 26 °C, referred to as D26 [26]. For the TCHP computation, we utilized ocean temperature and salinity datasets from the Copernicus Marine Service (https://doi.org/10.48670/moi-00016 accessed on 2 January 2025). This analysis product, with a spatial resolution of 1/12 degree, provides 10-day three-dimensional global ocean forecasts that are updated on a daily basis. For the atmospheric variables, including 10 m wind speed and relative humidity, we used ECMWF ERA5 reanalysis data [27]. The dataset, initially available at a 0.25° spatial resolution with an hourly temporal resolution, has been converted to a daily mean. This study uses daily SST data from NOAA OISST v2.0, which is provided at a 0.25° spatial resolution [28].

3. Pre-Monsoon Season Tropical Storms Climatology and Landfall over BIMSTEC Countries During Recent Decades (2004–2023)

The BIMSTEC is a multilateral and regional organization made with the collaboration of seven countries (India, Bangladesh, Bhutan, Nepal, Myanmar, Thailand, and Sri Lanka), and weather and climate products are prepared by the BIMSTEC center for weather and climate (BCWC) at NCMRWF and IMD, Govt. of India. The frequency of TCs’ variability during two recent decades (2004–2023) has been analyzed which are made landfall over the coast of BIMSTEC countries, particularly India, Bangladesh, and Myanmar. The frequency (%) during the pre-monsoon season is shown in Figure 1a. During recent decades (2014–2023) shows there were TC made landfall frequencies of 33%, 50%, and 12% for the Bangladesh, India, and Myanmar, respectively, as depicted in the Supplementary Figure S1. However, the TCs were mostly in the north and north-easterly direction in the past decades 2004–2013, and 27% of the number of TCs made landfall over the India region, less as compared to Bangladesh and Myanmar, [29,30].

As per the IMD rapid intensification (RI) criteria, whether a TC has a persistent maximum sustained wind at 30 knots within 24 h, referred to as RI, has been analyzed during these 20 years (2004–2023), showing that the maximum RI occurred in the month of May as compared to April, as represented in Figure 1b. Typically, the TC genesis occurs in the lower latitudes and rapidly intensifies while moving over the higher latitudes [31,32].

3.1. Favorable Condition for the Genesis of Mocha

(a)

Synoptic conditions

The low-pressure system formed at 00 UTC of 09 May 2023 over the lower latitude of the BoB became an ESCS known as “Mocha”. The life cycle of the Mocha cyclone over 9–15 May 2023 showed it moving northwesterly on 00 UTC of 11 May 2023. However, due to the influence of WD, as it existed over northwest India (30° N; 55° E) and adjoining Pakistan, the ESCS changed the track and moved in a northeast direction till 00 UTC on 14 May 2023 and made landfall on 06 UTC 14 May 2023. The satellite picture (before the landfall at 06 UTC), valid on 00 UTC of 14 May 2023, is shown in Figure 2.

According to the IMD, Cyclone Mocha initially moved southeastward over the BoB before shifting north-northwestward at a speed of 8 km/h. By 08:30 IST on 11 May 2023, it was centered near 11.4° N latitude and 88.0° E longitude, approximately 510 km west-southwest of Port Blair, 1190 km south-southwest of Cox’s Bazar (Bangladesh), and 1100 km south-southwest of Sittwe (Myanmar). The cyclone was expected to continue moving north-northwestward, gradually intensifying into a severe cyclonic storm by 12 UTC of 11 May 2023. Subsequently, it was forecasted to recurve and shift north-northeastward by 03 UTC of 12 May, strengthening further into a very severe cyclonic storm by the evening over the central Bay of Bengal. Mocha was predicted to reach peak intensity on 12 UTC of 13 May. It was likely to make landfall along the southeast Bangladesh and north Myanmar coasts, between Cox’s Bazar (Bangladesh) and Kyaukpyu (Myanmar), near Sittwe (Myanmar), around noon on 14 May 2023, with maximum sustained wind speeds of 140–150 km/h, gusting up to 165 km/h (IMD, 2023).

(b)

Ocean-atmospheric conditions

Tropical storms in the BoB are most prevalent during the post-monsoon season (October–December), with a secondary peak in the pre-monsoon months (March–May). The presence of a warm SST exceeding 30 °C contributes to a thermodynamically unstable atmosphere, while low-tropospheric wind shear conditions enhance cyclone formation during the pre-monsoon period. The process of the RI of tropical cyclones involves intricate dynamic and thermodynamic interactions, as discussed by Kalpan et al. [33] and Kalpan et al. [34], respectively.

Previous studies by Kotal et al. [9] have shown that various large-scale atmospheric and oceanic environmental conditions, such as VWS, mid-level relative humidity, relative vorticity at 850 hPa, SST, TCHP, and the TC inner core process significantly affect TC RI [35]. Specifically, an SST threshold > 26.6 °C is a necessary condition for the genesis of TC over time, but these conditions are not sufficient for the development of TCs [36]. Formulations based on SST and TCHP have both been used to understand the sensitivity of the TC formation of the Mocha cyclone.

The Mocha cyclone made landfall on the North Myanmar–Southeast Bangladesh coasts between Kyaukpyu (Myanmar) and Cox’s Bazar (Bangladesh) on 06 UTC 14 May 2023, with sustained winds of approximately 210–220 km/h, equivalent to a category 4 hurricane intensity. Moreover, the Mocha cyclone brought approximately 64 cmd⁻¹ of rainfall and an associated storm surge of 3–4 m near Kyaukpyu and Sittwe port city. However, for the Cyclone Mocha case, the SST was ~32 °C, and a TCHP > 100 KJ cm⁻² caused the RI, and it lasted over 9–15 May 2023, as shown in Figure 3c,d. Correspondingly, the magnitude of VWS (ms⁻¹) and RH (%) are shown in Figure 3a,b. In general, it was noticed that the VWS was relatively low, around 10 ms⁻¹, and the RH > 50% during the genesis stage, valid on 00UTC 09 May 2023. However, Mocha intensified as an ESCS stage on 00 UTC of 13 May 2023, when the RH reached above 70% and the VWS reached 8 ms⁻¹.

During 00 UTC 15 May 2023, the VWS and RH sharply decreasing to 5 ms⁻¹ and 30% caused rapid dissipation, as shown in the Supplementary Figure S2. The combined influence of SST, TCHP, VWS, and RH, relatively, remain very challenging for the prediction of TCs, as discussed in Bi et al. [37].

4. Performance of NCUM Analysis and Forecast for Mocha Cyclone at NCMRWF

4.1. Role of WD and Recurvature

To understand the short-wave propagation due to the WD and the location of the vortex, a significant comparison of the geopotential height (GPH) field at different pressure levels is considered. Figure 4 shows the NCUM analysis of GPH (m) and temperature (°C) for the Mocha cyclone from 00 UTC from 11 to 13 May 2023. Mocha exhibited multiple curvatures starting from the genesis, intensification, and landfall period. Thus, Figure 4 in the upper panel represents the GPH field at 850 hPa (shaded), 500 hPa (contour; red), and 200 hPa (contour; blue) at different stages of Mocha. At 00 UTC of 11 May 2023, the GPH showed a weak vortex minimum GPH field at about 1450 m over the BoB located at 10° N; 88° E, whereas significant ridges and troughs persisted above 40° N over the Pakistan and Nepal region. On the next day, the ridges and troughs shifted towards the northeast direction, whereas the vortex intensified over the BoB at 11.0° N; 89.5° E, and the minimum GPH was about 1400 m, 4760 m, and 12,460 m at 850, 500, and 200 hPa levels respectively, as revealed in Figure 4c. Figure 4e represents the much intensified vortex Mocha that moved in a northerly direction, and the ridges and troughs over northern India were relatively weak and shifted in a further eastward direction. During 11–13 May 2023, initially, the GPH at 200 hPa was 12,500 m, later increasing to nearly 12,540 m over the BoB, and the direction of GPH was a north–south orientation, which eventually guided the Mocha cyclone towards the north-northeast direction (Figure 4e). Subsequently, Figure 4 (lower panel) depicts the temperature difference between 700 hPa and 500 hPa levels at 00 UTC of 11–13 May 2023. Since the Mocha cyclone has changed its track very efficiently, a simple and well-known method, such as the temperature difference between 700 hPa and 500 hPa levels, can be used for forecasting the direction of movement for tropical cyclones. This method has been well documented by Simpson [38], and it is more appreciated and effective than the steering-level analyses thus adopted in this section.

Figure 4b,d,f show it being significantly warmer (22 °C) over the north-western part of India around 30° N and 60° E, progressing towards central India, which is objectively, associated with a ridge at 500 and 200 hPa levels. A tongue-like structure of warming and lighter atmospheric temperature in the eastward direction explicitly accelerates to the Mocha cyclone movement towards the northeast direction. Moreover, the lowest GPH at 850 hPa persisted in the central BoB, and the temperature was relatively less warm than in the surrounding regions (Figure 4b,d,f).

4.2. Intensity and Rate of Cyclogenesis

The NCUM-simulated mean sea level pressure (MSLP) and 10 m wind field of Cyclone Mocha, initialized at 00 UTC on 8 October 2013, are compared with IMD observations in Figure 5. To analyze the cyclone’s different stages and intensity, key parameters such as central sea level pressure (CSLP) and maximum sustained wind (MSW) during 9–15 May 2023 are examined and presented in Figure 5. The observations indicate that Mocha reached its lowest pressure of approximately 938 hPa and a peak MSW of 110 knots at 00 UTC on 14 May 2023. While the NCUM simulation underestimates the peak intensity of both CSLP and MSW, the overall pressure drop and evolution align well with IMD estimates. The SST near the cyclone’s center suggests that RI occurs when the SST exceeds 28.0 °C, accompanied by an average relative humidity of around 79% [39].

The CSLP decreases at an average rate of at least 1 hPa per hour over a 24 h period, contributing to RI as documented in Sanders and Gyakum [40]. Additionally, RI is defined as an MSW increase of at least 30 knots within 24 h, as noted in previous studies [41,42,43]. Cyclogenesis, often linked to RI, is generally associated with a surface pressure drop exceeding 24 hPa per day, a phenomenon also discussed in Sanders and Gyakum [40] and considered in this study.

An explosively developing cyclone can dissipate within 24 h of its initial formation. Therefore, this study utilizes 12 h pressure changes to identify the most intense deepening phase of Cyclone Mocha. Following Yoshida and Asuma [44] and Fu et al. [45], the maximum cyclogenesis rates were calculated assuming 12 h pressure changes instead of 24 h as discussed in Equation (1). The RI in terms of the cyclogenesis of Mocha is calculated using the following definition as

$$\mathrm{Cyclogenesis}=[{\displaystyle \frac{p\left(t-12\right)-p(t+12)}{24}}][{\displaystyle \frac{sin60°}{sin\frac{\theta \left(t-12\right)+\theta \left(t+12\right)}{2}}}]$$

(1)

where t is the analyzed time in hours, p is the central sea level pressure (hPa), and $\theta$ is the latitudinal position of the cyclone center. An analysis of the cyclogenesis of Mocha using the MSLP field as discussed in Equation (1) for different lead-time at 12 h intervals from the SCS to ESCS stages (cyclogenesis, and landfall) from 00 UTC 12 to 12 UTC 14 May 2023 are appeared over the BoB is represented in Figure 6. The time-series of the deepening rate in Bergeron derived from the NCUM analysis which is validated with IMD observations (Figure 6) as discussed in Yoshida and Asuma [44]

The simulation evolution of the deepening rate from the NCUM operational analysis indicates that the NCUM initial conditions were proficient in capturing the lowest cyclogenesis rate up to (-60) Bergeron, closely agreeing with the IMD value valid on 00 UTC 14 May 2023. Moreover, the analysis valid for 00 UTC of 12 and 13 May 2023 produces a higher cyclogenesis rate and, in the series, has intermediate deepening values (Figure 6). For the cyclonic vortex Mocha, the NCUM analysis represented sudden pressure drops that started during the 24 h of intensification (from 00 UTC 13 to 00 UTC 14 May 2023) as shown in the time series (Figure 6), and the TC attained peak intensity between 12 UTC 14 and 06 UTC 15 May 2023 and gradually weakened thereafter. In the time series, the IC of the NCUM presented with similar characteristics to the IMD observations but successively higher CSLP and lower pressure drops, as shown in Figure 5.

4.3. Circulation and Structure of Mocha During ESCS Stage

(a)

MSLP, precipitable water content, and 10 m wind field

The observed cloud bands, when Mocha reaches the ESCS stage along with MSLP, precipitable water content (TPWC; shaded), and wind at 10 m are presented in Figure 7 for 24, 72, and 120 h forecasst (day-1, day-3, and day-5), respectively, valid on 00 UTC, 14 May 2023. The dense and closed isobars (Figure 7b–e) following the heavier cloud distributions are apparent in Figure 7a. The Aqua Tera satellite pictures (NOAA, 2023) show heavily dense clouds aligned around the Mocha cyclone as shown in Figure 7a. Similarly, the simulated organized close isobars (hPa), with the extension of isobars in the northeast and southeast sectors of the Myanmar coast, indicate a broken comma structure. The maximum ~80 mm of TPWC (shaded) was depicted in the NCUM analysis (Figure 7b), closely agreeing with the structure of the Aqua Tera satellite cloud image as shown in Figure 7a. Further, the dense TPWC is in association with the lowest MSLP (Figure 7b–e). The position of maximum intensity in terms of TPWC in the 120 h forecast has a slight shift in location, as captured in the analysis in Figure 7b. Indeed, the 24 h forecast is closely in agreement with the analysis, whereas the 72 and 120 h forecasts deviate in position rather than intensity; it can be understood that the NCUM model forecast duration significantly depended on the accuracy and substantial improvements in the advanced data assimilation system, which can be further improved the IC subsequent to the forecast as discussed in Ashrit et al. [46]. Subsequently, the comparisons of simulated wind (knots) from the NCUM analysis and different forecasts valid for 24, 72, and 120 h (Figure 7g–j) closely agree with CIRA observations (Figure 7f). Interestingly, the lowest MSLP, along with maximum TPWC, is represented in 24, 72, and 120 h forecasts, significantly following the wind flow pattern in Figure 7f–j.

The inter-comparison valid on 00 UTC 14 May 2023 indicates that 10 m winds are stronger in the NCUM analysis, as compared to the CIRA observations.he wind near the center of the eyewall distribution of the Mocha cyclone was well simulated in the NCUM analysis (Figure 7g) and in the 24 h forecast (Figure 7h). However, the 72 and 120 h forecasts (Figure 7i–j exhibit discrepancies as compared to CIRA observations (Figure 7f). The errors in the 72- and 120-hrs forecasts valid on 00 UTC of 14 May 2023, perhaps arises due to the use of ICs valid on 00 UTC 11 May 2023, and 00 UTC 9 May 2023, poorly capturing the intensity compared to observation, while the position of the CSLP region aligning with the wind field mostly deviated from the observations that indicate slow movement of Mocha cyclone towards Myanmar coast.

(b)

Wind pattern at 850 and 200 hPa

The wind speed (shaded; ms⁻¹) along with direction at 850 hPa levels derived from the NCUM analysis and 24, 72, and 120 h forecasts are shown in Figure 8a–d. The wind flow patterns over the BoB are intensified along the coast of Myanmar depicted in the analysis (Figure 8a); a similar representation can also be the noticed in 24 and 72 h forecasts (Figure 8b,c). However, the 120 h forecast showed (Figure 8d) less intensification, and the location indicates a slight discrepancy from the analysis (Figure 8a) as the vortex slowly progressed towards the Myanmar coast, particularly located near the central BoB around 15.2° N; 90° E. In brief, the intensity and position of the vortex with the 24 and 72 h forecasts (Figure 8b, c) closely agree with the NCUM analysis (Figure 8a).

Subsequently, Figure 9a–d is the same as Figure 8 but for the wind field (shaded) and vorticity (contour; green) for the 200 hPa level. The westerly flow patterns were significantly noticed above 20.0° N and the ridge over the Tibetan Plateau. The maximum wind speed of 80 ms⁻¹ along with positive vorticity (~5) was significantly confined at 100° E and further extended in an eastward direction. The wind speed and direction in the 24-, 72-, and 120 hrs forecasts (Figure 9b–d) and vorticity field are closely resemble the analysis at the 200 hPa level, as shown in Figure 9a. The positive vorticity (green contour) play vital role for the eastward movement of the Mocha storms from BoB to Myanmar coast.

• (c) Vertical structure

The study of the vertical structure of a TC is an important parameter to understand the dynamical characteristic of pre-monsoon storms, especially during the RI phase of the cyclone. In this section, a zonal cross-section of u- and v-wind components (ms⁻¹), along with relative humidity at the center of the Mocha, are analyzed at Chittagong station to understand the core structures of the pre-monsoon storms shown in Figure 10a–c. The 24 h forecast u and v component based on ICs valid for 00 UTC 14 May 2023 shows greater intensity than in the 72 and 120 h lead time at a lower tropospheric level (1000–800 hPa levels). However, the 72 and 120 h wind component forecast above the 800 hPa level does not match with the observation or the analysis (Figure 10a,b). Figure 10c represents the relative humidity profile of the Mocha cyclone at the ESCS stage, showing above 80% of its moisture from the surface up to the middle atmosphere (500–700 hPa) level. The analysis and 24 h forecast are closely agreeing with the RH in the observations. It is clear that the 72 h forecast consistently coincide with the observations at 300 hPa level and a sharp decrease in the upper troposphere, whereas the 120 h forecast is unable to represent this feature.

4.4. Dynamical Mechanism for Cyclogenesis and Multiple Recurvature

(a)

Q-vector analysis

An attempt has made to understand the dynamical mechanism that is responsible for surface cyclogenesis, associated with the vertical motion fields discussed by Hoskin et al. [47], the convergence of the Q-vector identifies regions of quasi-geostrophic forcing for ascending motion associated with baroclinic waves. Martin and Otkin [48] suggested that quasi-geostrophic diagnostic field divergence (Q-vector) fields, one of the most significant parameters, could be used to access vertical motion fields linked to precipitation. Using a Q-vector also gives a more complete and meaningful solution of the direction of vertical motion and southerly flow, and warm advection induces an upward motion.

In contrast, northerly flow and cold advection induce a downward motion. Moreover, for frictionlessness, these vertical motions are forced by differential vorticity advection, causing the low to be stretched and vorticity around the system to be increased [49]. The Q-vector convergence and Q-vector demonstrate the potency of this upper-tropospheric wave in affecting lower- to middle-tropospheric ascent and the associated surface cyclogenesis [50]. As per the ƒ-plane quasigeostrophic theory, the Q vector is defined as

$$\mathrm{Q}=-{\displaystyle \frac{\mathbf{g}}{\mathbf{\theta }}}({\displaystyle \frac{\partial v}{\partial x}}·{\nabla }_p\theta ,{\displaystyle \frac{\partial v}{\partial y}}·{\nabla }_p\theta )$$

(2)

where $\mathsf{\theta }$ a constant reference value of the potential temperature, $v$ the horizontal geostrophic wind, ${\nabla }_{p }$ the horizontal gradient operator on a constant pressure surface, $\theta$ the potential temperature, and $g$ the acceleration of gravity.

The column-averaged Q-vectors and convergence at 850 hPa over the west and east sector represented in Figure 11a–d indicate the strength of the lower-tropospheric ascent and associated surface cyclogenesis on 00 UTC of 13–14 May 2023. The region of convergence and divergence is represented with positive and negative values of the Q-vector in the upper panel of Figure 11a–d, valid for 00 UTC 13 May 2023. The positive and negative Q-vector values are centered around the Mocha cyclone in analysis and 24- and 72- hrs forecasts around 15° N; 90° E. That clearly indicates large-scale convergence surrounded by a divergence and subsidence of cold air on the southwest side of Mocha on 00 UTC 13 May 2023, which largely resembles MSLP closed isobars as discussed in Figure 7b, but the convergence and divergence are poorly depicted in the 120 h forecast in Figure 11d. The lower panel of Figure 11e–h shows the Q-vector valid on 00 UTC 14 May 2023. The ascent of warm airflow is indicated on the northeast side, as well as bring surrounded by the low-pressure system near the Myanmar coast as depicted in analysis (Figure 11e). Similarly, features of convergence and divergence along the west coast of Myanmar are seen in analysis and 24- and 72 -hrs forecasts in Figure 11e–g, which were not well captured in the 120 h forecast (Figure 11h).

(b)

Large-scale flow

The TC motion is mainly caused by large-scale steering flow (SF), as discussed by Kimberlain and Breman [11]. If the flow in which a TC is embedded is gradually evolving, persistence can be important in the forecasting of the TC track in the first 24 to 48 h advance. Thus, an attempt has been made to analyze the SF, as in the operational center, the global model is not designed explicitly to forecast sudden changes in the TC track.

The SF is a large-scale feature that can be used for TC forecasts. Based on the intensity, the appropriate steering environment layers are determined, and details are discussed in Velden and Leslie [51].

SF = [(W850 + W500)/2 × 350 mb + (W500 + W200)/2 × 300 mb)]/(650 mb)

(3)

where W850, W500, and W200 are wind flow (ms⁻¹) at 850, 500, and 200 hPa levels.

Figure 12a–h represent the SF valid on 00 UTC of 13 (upper) and 14 May 2023 (lower panel). The upper panel of Figure 12a–d shows the SF of Mocha when it was in the VSCS stage, valid on 00 UTC of 13 May 2023. Figure 12a clearly shows the well-organized and intense eastward SF and a maximum intensity of 15 ms⁻¹ in the NCUM model analysis. Similarly, the 24, 72, and 120 h NCUM forecasts of SF (Figure 12b–d) reasonably agree with the analysis (Figure 12a). One can notice that the SF features in the 24 and 72 h forecasts valid on 00 UTC 14 May 2023 well agree with the NCUM analysis. In contrast, the 120 h forecast predicts the SF at Tamilnadu coast (Figure 12d), far away from the vortex position as shown in the INSAT-3D satellite (Figure 2). The location and intensity of SF significantly deviated in the 120 h forecast (Figure 12 h) valid on 00 UTC 14 May 2023. The eyewall structure is clearly represented in the 24 and 48 h forecasts, as compared to the 120 h forecasts.

4.5. Track and Vertically Integrated Moisture Transport

The best track estimates of positions are obtained from the Regional Specialized Meteorological Center (RSMC) IMD-New Delhi and are utilized for verification of the model. The direct positional error (DPE) essentially calculates the straight-line distance between the forecast location and the actual location of the cyclone center [52]. The DPE represented in

Table 1. Direct positional error (DPE) and landfall error at 12 h lead time during 9–14 May 2023.

Forecast Lead Time	IC 13-05-2023	IC 12-05-2023	IC 11-05-2023	IC 10-05-2023	Average PDE (km)	Landfall PDE (km)
0	24.17	22.26	22.26	28	24.2	9
12	15.4	22.26	34.49	26	24.5	9
24	71.49	99.08	44.53	59.8	68.7	9
36	178.78	115.34	113.8	99.15	126.8	8
48		224.41	141.62	35.11	133.7	7
60		325.11	193.28	105.29	207.9	7
72			262.96	61.88	162.4	5
84			328.16	56.67	192.4	5
96			257.22	87.29	172.3	4
108				187.21	187.2	3
120				196.31	196.3	1

illustrates that error increases monotonically with respect to time due to the IC. The DPE fluctuates in the range of 25 to 197 km, whereas the landfall error is 6 km; it seems the NCUM model has commendable skill in the prediction of the track of the Mocha cyclone.

To explore the moisture influx from the BoB vertically integrated moisture convergence (VIMC) in the troposphere levels (1000 hPa to 500 hPa) for the Mocha cyclone has been considered. Figure 13a–d represents the vertical integrated moisture (1000–500) hPa and GPH at 500 hPa derived from the NCUM analysis and 24, 72, and 120 h forecast valid on 00 UTC 14 May 2023, when the Mocha cyclone is at VSCS stage. The maximum (>18 kgm⁻¹s⁻¹) magnitude of VIMT accumulated at the vicinity of the lowest CSLP is revealed in Figure 6.

During the VSCS of the Mocha cyclone, the highest moisture transport (18 kgm⁻¹s⁻¹) was observed along the coast of Myanmar, clearly represented in the NCUM operational analysis as represented in Figure 13a. The present analysis and the 24, 72, and 120 h forecasts of VIMT valid on 00 UTC 14 May 2023 showed a similar intensity, but in the 120 h forecast a slight deviation in the location is noticed, which indicates the vortex has moved relatively slower than the 24 and 72 h forecasts (Figure 13b,c). In contrast, the structure of VIMT is subsequently stronger, starting from the lower latitude of the BoB and moving in a south-westerly direction and accumulating along the Myanmar coast due to the lowest MSLP associated with the VSCS stage. During Cyclone Mocha, 9–15 May 2023, heatwaves significantly intensified over the Indian region, with an increase of 8 °C in maximum air temperature noticed during the mature stage as well as post-cyclone periods, compared to the pre-cyclonic period, as discussed in Sharma et al. [53].

5. Rainfall

Cyclone Mocha made landfall on 14 May 06 UTC and produced extensive rainfall over the southwest coast of the Myanmar region. The 24 h accumulated observed rainfall from the IMD-NCMRWF merges analysis and NCUM model-simulated rainfall for two corresponding days valid at 03 UTC of 14 and 15 May 2023, respectively. Figure 14a represents observed and 24, 72, and 120 h rainfall forecasts from the NCUM global model rainfall valid at 03 UTC 14 May 2023 (Figure 14b–d). Overall, the 24, 72, and 120 h forecast rainfall distributions valid for 03 UTC of 14 May 2023 from the global model are well agreeing (Figure 14b–d) with the IMD-NCMWRF merge analysis as shown in Figure 14a. Significantly, the 72 and 120 h valid for 14 May 2023 also simulate the intensity of rainfall over the oceanic regions rather than land, and the 24 h forecast is more efficient than the 72 and 120 h forecasts.

Subsequently, the 24 h accumulated rainfall valid at 03 UTC 15 May 2023 from observation shows rainfall bands over Northeast India and the adjoining Myanmar region, and the intensity has slightly reduced as compared to the previous day (Figure 14e). The simulations for 24, 72, and 120 h NCUM model forecasts (Figure 14f–h) revealed the rainfall pattern reasonably well (left bottom panel). The 24 h accumulated IMD-NCMWF merge analysis valid at 03 UTC on 15 May 2023 illustrates the heavy precipitation (>64 cm) over the southwest peninsular tip and adequate rainfall (>32 cm) over southwest Myanmar.

In addition, the model slightly overpredicted the rainfall distribution over the land region, both spatially and temporally, in the southern part of Myanmar. However, for the simulations based on 03 UTC 12, and 10 May 2023 ICs, valid on 72- and 120- h forecasts indicate that model could be capable of forecasting the heavy to very heavy rainfall over the southwest of Myanmar region. The prediction of spatial patterns and rainfall intensity was very effectively represented, but the intensity was slightly overvalued in 72 and 120 h forecasts. The very heavy rainfall over the ten meteorological stations of Myanmar in

Table 2. 24 h accumulated rainfall (mmd⁻¹) over major stations of Myanmar affected by Mocha cyclone derived from different station observations and NCUM model 24-, 72-, 120- h forecasts valid on 03 UTC of 14 May, and 15 May 2023.

			03 UTC 14-05-2022				03 UTC 15-05-2022
Stations	Latitude	Longitude	Obs	24	72	120	Obs	24	72	120
Sittwe	20.16	92.86	76.35	88.86	0.22	0.01	74.25	64.50	7.49	3.2
Pauktaw	20.18	93.07	96.17	125.92	0.80	0.02	70.17	12.84	28.95	2.9
Kyaukpyu	19.43	93.55	81.61	101.50	0.36	0.01	41.82	54.77	17.32	0.8
Munaung	18.86	93.73	64.07	157.02	6.61	0.00	45.88	47.11	94.00	2.5
Buthidaung	20.86	92.52	81.49	33.73	0.20	0.00	60.55	142.72	6.21	0.8
Matupi	21.61	93.45	47.62	2.98	0.97	0.00	71.19	83.40	10.20	3.6
Paletwa	21.31	92.85	56.99	3.31	0.11	0.04	84.73	93.40	2.76	1.3
Hakha	22.65	93.61	45.49	19.78	0.05	0.04	42.97	34.45	19.16	0.2
Lasho	22.95	97.75	10.40	11.95	26.70	9.41	9.08	1.59	111.63	4.5
Yangon	16.84	96.18	16.88	6.88	42.03	0.07	1.53	52.46	5.88	6.1

valid for 14 and 15 May 2023 shows the 72 and 120 h forecasts unable to agree with the observed value, because the condensed rain bands are relocated in the oceanic region rather than the land, which was due to the slow movement of the Mocha cyclone.

6. Conclusions

In this study, the performance of the NCUM operational model on the Mocha cyclone is analyzed and discussed. The analysis and forecast of NCUM, such as MSLP, wind, and GPH at different levels, Q-vector, steering flow, rainfall, and the vertical structure of Mocha are analyzed, and the results are discussed. The genesis and RI of the Mocha cyclone over the lower latitude of the BoB during the pre-monsoon season was mainly due to the oceanic conditions, SST warming > 29 °C and TCPH 100 KJcm⁻², whereas the VWS was 15 ms⁻¹ and RH 79%. The multiple curvatures, such as in the northwest to northeast direction, were due to a sub-tropical ridge and middle tropospheric warming associated with the WD; these features were clearly revealed in the operational NCUM global analysis. The RI and cyclogenesis of the Mocha cyclone derived from the NCUM operational analysis closely agree with IMD observations.

The spatial structure of the Mocha cyclone VSCS has been analyzed from MSLP, 10 m wind, and TPWC through the NCUM operational forecasts, which shows that the VSCS of the vortex is well predicted in 24, 72, and 120 h forecast, but the representation in the 120 h NCUM forecast has significantly deviated concerning the observations, as well as the NCUM analysis. The wind speed and direction at 850 and 200 hPa levels were well presented in the 24, 72, and 120 h forecasts, agreeing with the NCUM analysis and CIRA observations. Moreover, the positive vorticity at 200 hPa level at 30° N and 90° E further accelerates the vortex from northerly to northwesterly directions, finely depicted in the analysis and forecast. The vertical components of U, V, and RH (%) at the Chittagong meteorological station-derived NCUM analysis and forecast well agree with RS observations up to the lower tropospheric level; however, the dry condition prevails in 72 and 120 h forecasts. The NCUM model-derived analysis and forecasts of the convergence and divergence patterns are well represented in the Q-vector analysis at an 850 hPa level. Still, the 120 h forecast cannot represent the same features as depicted in the analysis. The SF distribution is significantly well represented both in the analysis and forecast, and the eyewall representations closely match with INSAT 3-D satellite pictures except the 120 h forecast. The rainfall distribution along the coast of Myanmar in the operational NCUM model closely agrees with the IMD-NCMRWF merge analysis.

Further, more TC recurvature cases can be considered and analyzed to understand the dynamical mechanisms more thoroughly. In the future, the pre-monsoon SST and atmospheric temperature over the BoB and Indian land mass will be significantly warmer, which will be more challenging to forecasters. Therefore, the Q-vector and SF may be an appropriate process based analysis that can be used for TCs forecast on an operational basis.