A Rare Tropical Cyclone Associated With Southwest Monsoon Over the Northern Part of the South China Sea—Tropical Storm Maliksi

Abstract

This study examines the characteristics and development of Tropical Storm Maliksi, which is a special case of tropical cyclone developed in the northwestern part of the South China Sea during a southwest monsoon outbreak. Detailed analyses were conducted using observational data and forecast products. Surface observations, radar wind profilers, aircraft data, and satellite products were used to evaluate Maliksi's wind structure, revealing multiple circulation centers and gale force winds. Vertical wind profiles, warm core structure, wind waves, and the influence of sea temperatures and salinity on Maliksi's intensification were investigated. Regarding forecasting, AI‐based models outperformed conventional numerical weather prediction (NWP) models in predicting Maliksi's initial development, though both struggled to capture the persistence of strong winds as the system moved inland. High‐resolution NWP simulations were employed to examine terrain‐induced wind variability around Hong Kong International Airport, revealing the mountain wake effect and uneven wind and turbulence distribution. These findings provide insights into the challenges of forecasting and monitoring such tropical cyclones, and highlight the need for enhanced observational platforms and forecasting tools along coastlines vulnerable to these systems.

Introduction

The tropical cyclone season in Hong Kong occurs between May and October every year. While it is not uncommon for a tropical cyclone forming over the South China Sea to make landfall over the western part of Guangdong during the early part of the tropical cyclone season (April to June), they usually track west to northwestwards and would have already departed from Hong Kong after their landfall. Based on the Hong Kong Observatory's best track between 1961 and 2023, only four tropical cyclones forming over the South China Sea came closer to Hong Kong after their landfall over the western part of Guangdong between April and June.

In late May to early June 2024, there was a tropical cyclone, namely, Tropical Storm Maliksi, developed in the northwestern part of the South China Sea following the outbreak of the southwest monsoon, and moved north toward the coast of southern China. Though it was more than 200 km away from Hong Kong, it brought strong winds to a number of places over the territory. Maliksi is a special case of tropical cyclone affecting the Pearl River Estuary and documentation of this case would be beneficial for future reference. This is the main objective of the present paper.

As the last tropical cyclone adopting a similar track dates back to 2018, there are some new observational and forecasting products for monitoring the development of a tropical cyclone in this region, which would be documented in this paper. On the observational side, we have dropsonde measurements near the center of the tropical cyclone in determining its intensity (Chan et al., 2018; He et al., 2022; Ming et al., 2015). The probe on board the aircraft also provides information about the turbulence associated with the tropical cyclone (Hon & Chan, 2022; Mak et al., 2023; Sparks et al., 2019). For ground-based remote-sensing meteorological instruments, a network of radar wind profilers provides information about the vertical wind profiles of the tropical cyclone (Donaher et al., 2013; Liao et al., 2019), which would be important for wind engineering applications, for example, in the design of wind loading of the buildings (Song et al., 2016; Yi et al., 2022). The ocean radar near Hong Kong also provides information about the surface wind field and significant wave heights (He et al., 2024; Liu et al., 2007). On the satellite side, a combination of the outputs from a number of meteorological satellites provides information about the wind distribution and the maximum wind strength of the cyclone (Atlas et al., 2011; Mears et al., 2019).

On the forecasting side, we have a new regional model, namely, Tropical Regional Atmospheric Model System for the South China Sea (TRAMS) in forecasting the genesis and development of the system (Huang et al., 2024; Zhang et al., 2023). Artificial intelligence (AI) based weather models would also be studied, in terms of the initial development of the system and the strength of the wind that the cyclone is expected to bring to Hong Kong (Bi et al., 2023; Chen et al., 2023). Moreover, super high resolution (horizontal resolution of 40 m) numerical weather prediction (NWP) model simulation (Hon et al., 2023) would be carried out to find out the wind distributions over the Hong Kong International Airport (HKIA) as a result of the interaction of the cyclone's wind field with the local terrain (Chan et al., 2020; Chan & Li, 2020), and it is the first time to observe the winds over the newly built third runway (the new north runway) of HKIA in strong southwesterly wind conditions.

Though the results appear to be rather local in nature, the observational and forecasting information so documented in this paper is considered to be of interest to the wider audience of the meteorological community, especially for those operating weather warning and forecast services along the coast that face the impacts of tropical cyclones. The above-mentioned information is also considered to be novel for future reference.

General Information About Maliksi

The track of Maliksi is given in Figure 1a. Originating from a broad area of low pressure over the central part of the South China Sea, Maliksi intensified as a tropical depression on the evening of 30 May 2024 (Hong Kong time, UTC + 8 hr). It tracked generally northwards across the northwestern part of the South China Sea on that night. Maliksi intensified into a tropical storm the next day, attaining its peak intensity with an estimated maximum sustained wind (sustained 10 min mean winds) of around 35 knots. It drifted northwestwards slowly later on the afternoon of 31 May 2024 and started to pick up speed to move north to northeastwards that night. Maliksi made landfall over the western part of Guangdong in the small hours on 1 June 2024 and continued to track northeastwards across inland Guangdong, edging closer to Hong Kong. It finally degenerated into an area of low pressure over inland Guangdong on the evening of 1 June 2024. Tropical cyclones with similar tracks and affecting Hong Kong in April to June of the year (only 4) are shown in Figure 1b. There are a number of radar wind profilers observing the upper air wind field over southern China, which will be documented in the latter part of this paper. The locations of the profiles are given in Figure 1c.

[IMAGE OMITTED. SEE PDF]

Multiple Low-Level Centers

One major problem for tropical cyclones developing near Hainan Island in association with the southwest monsoon is their relatively weak intensity and thus it would be difficult to locate their centers. One example is Mulan in 2022, which has been documented in Chan et al. (2022). This time, Maliksi was also weak with a maximum intensity of 35 knots only near the center. During part of its life, it showed multiple center feature.

Multiple centers appeared in the northern part of the convective system of Maliksi when it was located to the east and northeast of Hainan Island, before making landfall over the western coast of Guangdong. For instance, at around 01–02 UTC, 31 May 2024, Advanced Scatterometer B (ASCAT-B) data revealed some possible centers of Maliksi (Figure 2a). However, from the radar imagery at that area (Figure 2b), while the northern center was rather apparent, the southern center did not show up in both reflectivity data (Figure 2b) and Doppler velocity data (not shown).

[IMAGE OMITTED. SEE PDF]

A few hours later, at around 05–06 UTC, the multiple center situation seemed to persist, at least as shown up in the cloud imagery of the geostationary meteorological satellite (Figure 3a). There were both ASCAT-C data and dropsonde data (near the sea surface) around that time (Figure 3b). However, due to spatial coverage of these two sets of data, it was not able to confirm the existence of the “suspected” center to the east of Hainan Island, whilst the center to the northeast of the Island could be identified in both ASCAT-C and the dropsonde.

[IMAGE OMITTED. SEE PDF]

The presence of multi-centers of a relatively weak tropical cyclone calls for the need of more observations over the vast South China Sea. It is the plan of the meteorological authorities in the region to set up more observational platforms over the ocean. As shown in the available surface observations over the sea (Figures 4a and 4b), they are far from sufficient for establishing the existence of multiple centers for Maliksi.

[IMAGE OMITTED. SEE PDF]

An alternative is to utilize target observations (TOs). TOs involve augmenting additional observations in specific, important areas (Qin et al., 2023). They often rely on dropsondes deployed by aircraft (such as those used in this study), additional radiosonde ascents, and enhanced regular satellite observations. To identify the proper observing locations, several techniques have been proposed, such as Singular Vectors (Palmer et al., 1998), Ensemble Transform Kalman Filters (Bishop et al., 2001), Adjoint-Derived Sensitivity Steering Vectors (Wu et al., 2007), and Conditional Nonlinear Optimal Perturbation (Mu et al., 2009; Qin & Mu, 2011). Employing TOs in conjunction with these identification techniques could be beneficial for locating TCs centers and monitoring TC development, particularly over the ocean, where observations are scarce and costly.

Upgrade to Tropical Storm

Another major question with Maliksi is whether it should be upgraded to a tropical storm, that is, with the existence of gale force winds near its center. One piece of evidence comes from the dropsonde data near the sea surface, which shows gale force wind near the center in association with the outburst of southwest monsoon (Figure 3b). In general, the dropsonde data are consistent with the ASCAT-C data, in both the wind speed and the wind direction, with the wind speeds from adjacent measurement points differing by a few knots only, which is within the uncertainty of the scatterometer winds. However, it is noted that ASCAT-C did not show gale force winds in the scan.

The surface observations did not record gale force winds while Maliksi was over the ocean (Figures 4a and 4b). However, shortly after Maliksi made landfall over the western coast of Guangdong, there were a number of islands recording persistent gale force winds for a number of hours (a snapshot is provided in Figure 4c). This is another evidence that Maliksi may have attained tropical storm strength.

We have also examined the combination of the available satellite-based wind speed products. The Cross-Calibrated Multiplatform (CCMP) ocean surface wind dataset combines satellite wind retrievals, in situ wind measurements, and a background wind field from numerical weather analysis to provide a gap-free global wind vector over oceans. CCMP uses a variational approach (Atlas et al., 2011) to ensure data-rich regions and the background wind field are closely collocated in time and space. Here, we use the CCMP v2.1 near-real-time (NRT) version (Mears et al., 2019), which has a latency of less than 48 hr and a temporal resolution of 6 hr. To ensure CCMP is produced in near-real-time, Mears et al. (2019) used the operational 0.25° NCEP analysis winds as background winds and only used satellite wind retrievals. CCMP incorporates satellite wind from a total of 16 satellites, including both imaging radiometers and scatterometers (Table 1). The data is retrieved at .

Table 1 The Input Satellite Wind for CCMP

Satellite	Sensor	Type
AQUA	AMSR-E	Radiometer
GCOM-W1	AMSR2	Radiometer
TRMM	TMI	Radiometer
GPM	GMI	Radiometer
QUIKSCAT	SEAWINDS	Scatterometer
METOP-A	ASCAT	Scatterometer
DMSP-F08	SSM/I	Radiometer
DMSP-F10	SSM/I	Radiometer
DMSP-F11	SSM/I	Radiometer
DMSP-F13	SSM/I	Radiometer
DMSP-F14	SSM/I	Radiometer
DMSP-F15	SSM/I	Radiometer
DMSP-F16	SSMIS	Radiometer
DMSP-F17	SSMIS	Radiometer
DMSP-F17	SSMIS	Radiometer
Coriolis	WindSat	Radiometer

CCMP data are used to obtain the maximum wind speed of Tropical Storm Maliksi. For that purpose, the maximum wind speed within a 5° × 5° window of the center of Maliksi is determined. Notably, CCMP data shows a significant increase in maximum wind speed after 1800 UTC on 29 May 2024, indicating that Maliksi was likely to have formed around that time. Two moments when the maximum wind speed associated with Maliksi was over 18 m/s are shown in Figures 5a and 5b. The centers of Maliksi showed up nicely in these wind speed plots. The time series of the maximum wind is shown in Figure 5c. There has been quite an extensive period of time in which the maximum wind speed associated with Maliksi was over 18 m/s, reaching the highest value of about 20 m/s. As such, the CCMP data appeared to have provided further evidence to the existence of gale force winds near the center of Maliksi.

[IMAGE OMITTED. SEE PDF]

Vertical Wind Profiles and Geostrophic Drag

Another interesting aspect of Maliksi is its vertical wind profile, namely, for a rather weak area of low pressure in association with the outburst of southwest monsoon, what the feature, if any, would be in its vertical wind profile. The radar wind profilers in Figure 1c are utilized in the study. There is in fact a LIDAR wind profiler near the center of Maliksi when it was over the sea to the east of Hainan Island, but unfortunately data were not available from this profiler at the time. As such, the radar wind profilers were mainly used to study the vertical wind profiles of Maliksi after its landfall over southern China. The vertical wind profiles were analyzed in Figure 6, with stratification according to the wind speed at around 850 hPa level. When the 850 hPa level wind is on the high side, namely, around 15–20 m/s, the radar wind profiler under consideration is most likely within the southwest monsoon at the rear of Maliksi. It could be seen that the appearance of a low-level jet is rather consistent through all wind profilers in this quadrant, with the jet occurring between around 1,000 m and 4,000 m. This observation has not been documented before in the literature.

[IMAGE OMITTED. SEE PDF]

Zooming-in to the heights of interest for wind engineering applications in buildings, namely, below a height of 1,000 m or so above ground, all the wind profiler observations suggest that the vertical wind profiles generally follow the power law and the log law (Figure 7). For power law, the power exponent is mostly in the order of 0.3–0.6.

[IMAGE OMITTED. SEE PDF]

When Maliksi was over the sea, the only upper-air observations came from the aircraft flying into its center, which is the tropical cyclone reconnaissance flight operated by the Hong Kong Observatory and the Government Flying Service (Hon & Chan, 2022). The probe data are plotted in Figure 8. Near the center of Maliksi on 31 May 2024, the aircraft was situated at a height of around 9 km above the sea surface. The wind speeds were generally low. Again, as shown in other aircraft observations (e.g., for Saola, He et al., 2023), the turbulent kinetic energy and the eddy dissipation rate have good linear correlation in logarithmic scale.

[IMAGE OMITTED. SEE PDF]

The vertical profiles of radial/tangential flow and pseudo potential temperature from the dropsonde data of 31 May 2024 are shown in Figures 9a and 9b respectively. The cyclone shows inflow within the atmospheric boundary layer near its center. In the southwest monsoon, namely, sondes 1, 2, 5 and 6, it was generally inflow throughout the whole troposphere and the profile was thermodynamically unstable, which have been reported in the previous dropsonde study over the South China Sea to be favoring intensification of the system (He et al., 2022). As such, Maliksi underwent strengthening gradually upon approaching the western coast of Guangdong.

[IMAGE OMITTED. SEE PDF]

Zooming-in the boundary layer, the vertical wind profiles from the dropsonde are shown in Figure 10a. In general they follow the power law and the Vickery model (Vickery et al., 2009) better than other models in the literature, such as the log law, Deaves and Harris model (Deaves & Harris, 1978), and Gryning model (Gryning et al., 2007). The vertical-distance profile of pseudo potential temperature is shown in Figure 10b. The warm core structure of Maliksi shows up nicely. As such Maliksi should be considered as a rather mature tropical cyclone.

[IMAGE OMITTED. SEE PDF]

Friction velocity $${u}_{\ast }$$ and geostrophic drag coefficient C_g (the ratio between $${u}_{\ast }$$ and geostrophic wind speed U_g) are critical parameters that reflect momentum exchange in the TC boundary layer. Their behaviors in TC boundary layer over land are investigated based on the wind profiler data in Maliksi. Friction velocity is estimated by least squares fitting of the wind profiles in the lowest 500 m to the logarithmic law, as the existence of the logarithmic regime in the lowest several hundred of meters in TCs have been confirmed by previous observations (Powell et al., 2003; Tsai et al., 2019). Due to the lack of temperature measurements, stability correction to the logarithmic law is not considered, but the influence is deemed minimal since the mechanical flux dominates over thermal flux in strong winds (Zhang, Zhu, et al., 2011). The geostrophic wind speed is estimated based on the wind speed at 1,500 m (about 850 hPa), which is near the boundary layer top (Zhang, Rogers, et al., 2011). Based on the power exponent reported in Figure 7, the wind profiler sites are classified as relatively “rough” (Huadu, Conghua, Longmen, Zengcheng, and Siu Ho Wan) or “smooth” (Luoding, Xinhu, Zhuhai, Sha Lo Wan, and Macao).

Figure 11 shows the variation of friction velocity and geostrophic drag coefficient with geostrophic wind speed. Generally, for wind speed lower than 20 m/s, the friction velocity increases with wind speed, while the geostrophic drag coefficient decreases with wind speed. These trends are consistent with that predicted by the geostrophic drag law (Zilitinkevich & Esau, 2005). It is also observed that the geostrophic drag coefficient is higher for rougher sites than smoother sites, which also aligns with the geostrophic drag law prediction (Zilitinkevich & Esau, 2005). However, when wind speed exceeds 20 m/s, the geostrophic drag coefficient seems to “level off.” This behavior is probably related to factors such as departure from geostrophic balance due to isobar curvature (Vickery et al., 2009), baroclinicity (Ghannam & Bou-Zeid, 2021), and roll vortices in TCs (Zhang et al., 2008). Further research efforts are warranted to elucidate this phenomenon.

[IMAGE OMITTED. SEE PDF]

Sea Temperatures and Salinity

Another factor that could be considered in the examination of the intensification of Maliksi is the analyzed sea temperature and salinity profiles in the upper ocean along the tropical cyclone track, as reported in Chan, He, and Lui (2024) for an observational study of tropical cyclones cases in the South China Sea based on the China Ocean ReAnalysis version 2 (CORA2) products from the National Marine Data and Information Service of China (Fu et al., 2023). The sea surface temperature analysis on 28 May 2024 is shown in Figure 12a. It is generally in the order of 29–30°C to the east and south of Hainan Island, and thus favoring intensification of the system.

[IMAGE OMITTED. SEE PDF]

The track-depth cross sections of sea temperatures and salinity on 28 May 2024 before Maliksi traversed the northern part of the South China Sea are shown in Figures 12b and 12c respectively. Sea temperatures at location points near 17 were relatively high in the region of around 30°C or more for a deep layer down to about 60 m, and the salinity was relatively low at around 33 psu with a relatively strong salinity stratification of about 0.8 psu over a depth of 100 m. These conditions are all favorable to tropical cyclone development (Chan, Lam, et al., 2024). Though sea temperatures dropped gradually as Maliksi moved north toward the coast of southern China, it was still in the order of 28–29°C near the surface. The salinity stratification of the upper 100 m of the ocean remained rather strong, in the order of 0.8 psu, which inhibited vertical mixing of the upper ocean and sea surface temperature cooling to some extent, and thus maintained the relatively high sea surface temperature. These factors are consistent with the observed strengthening of Maliksi.

Wind Wave Observation and Forecast

Another set of surface wind observations based on remote-sensing technology comes from the ocean radar pair of Hong Kong and Shanwei respectively operated by the Hong Kong Observatory and the South China Sea Bureau of the Ministry of Natural Resources of China, which has been reported in the previous tropical cyclone study in the South China Sea (e.g., He et al., 2024). Unfortunately, the observations are mainly concentrated to the southeast of Hong Kong, which is relatively far away from the center of Maliksi. A snapshot of the ocean radar wind is given in Figure 13a. The surface winds look reasonable, and consistent with the only available buoy-measured wind in the region. However, the gale force wind “measured” at the southern part of the radar detection area is not certain as it is not supported by any actual observations.

[IMAGE OMITTED. SEE PDF]

In terms of waves, for the “gale force wind” region, the significant wave height was “measured” to reach around 5 m (Figure 13b). This observation is rather consistent at least with the prediction from the Hong Kong Observatory's Operational Marine Forecasting System (OMFS) (Figure 13c) and the seas should be rather rough for navigation in that area, though it is rather far away from Maliksi. Details of OMFS could be found in Kong et al. (2024).

Model Forecast, Genesis and Persistence of Strong Winds in Hong Kong

The forecast of TC genesis or the initial development of Maliksi in the outburst of southwest monsoon differed rather significantly among the NWP models and AI models. The examined NWP models include the European Center for Medium Range Weather Forecasts (ECMWF) Integrated Forecasting System (IFS) and TRAMS (Huang et al., 2024). The AI-based models include the Pangu model (Bi et al., 2023) and the Fuxi model (Chen et al., 2023). The AI-based models have been shown to have quite reasonable skills in past tropical cyclone cases over the South China Sea (such as Pangu, as documented in Chan, He, & Lui, 2024; He & Chan, 2023).

An example is given in Figure 14, namely, the model run at 00 UTC, 29 May 2024, 2 to 3 days before the event. At that time, the ECMWF IFS did not forecast spin-up of the system. TRAMS seemed to have over-forecast a bit, with a rather deep cyclone. Both models may not be readily useful in the tropical cyclone forecast and warning services. On the other hand, the AI models seemed to be rather accurate in the forecast of the genesis of the system.

[IMAGE OMITTED. SEE PDF]

Models' performance on the genesis forecast is summarized at Table 2, where different consecutive model runs from ECMWF and Pangu on genesis for Maliksi within a 48 hr window centered at 09 UTC on 30 May 2024, at which time it was formed as a tropical depression in accordance with the provisional best track, had been examined. AI model Pangu was able to hint about development of this tropical cyclone in its very early runs 12 or even 14 days before the event, giving ample time for an early alert. Nevertheless, there was then a period where Pangu could not forecast the genesis (8–9 days before) or flip-flopped between its 00/12Z runs (5–7 days before). Because the AI model is based on the operational analysis of ECMWF IFS as its initial conditions, the flip-flopping might suggest issues within the operational analyses of traditional NWP. Genesis forecasts by ECMWF IFS were not satisfactory, only giving a signal barely about 1.5 days before. To further investigate the genesis with respect to the southwest monsoon, Figure 15 shows the 925 hPa wind forecasts by Pangu and ECMWF, together with the operational analysis valid for the same time. It was clear that ECMWF could not capture the strong southwesterly monsoonal flow over the South China Sea, which should feed into eastern flank of Maliksi, nor did ECMWF capture the monsoon trough over northern part of the South China Sea. On the other hand, Pangu (Figure 15a) give a very accurate forecast of strong southwest monsoon winds wrapping across the monsoon trough, leading to genesis. Previous study has pointed out that AI models outperform traditional NWP in the tropics (Bi et al., 2023), this is possibly because of AI models' advantage of not relying on parameterization of sub-grid physics as convections dominate over the tropics. It is hoped that more studies from the research community could better understand the AI models' strength over the tropics.

Table 2 Forecast by Different Consecutive Model Runs From ECMWF and Pangu on Cyclone Genesis for Maliksi Within a 48 hr Window Centered at 09 UTC on 30 May 2024, at Which Time It Was Formed as a Tropical Depression in Accordance With the Provisional Best Track

Forecast lead time (H)	336	324	312	300	288	276	264	252	240	228	216	204	192	180
ECMWF	No	No	No	No	No	No	No	No	No	No	No	No	No	No
PANGU	Yes	No	No	No	No	Yes	No	No	No	No	No	No	No	No
FORECAST LEAD TIME (H)	168	156	144	132	120	108	96	84	72	60	48	36	24	12
ECMWF	No	No	No	No	No	No	No	No	No	No	No	Yes	Yes	Yes
PANGU	No	Yes	No	Yes	No	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes

[IMAGE OMITTED. SEE PDF]

Zoom-in charts over the Pearl River Estuary from the various models are shown in Figure 16. The position forecasts by TRAMS and the AI models appear to be rather reasonable. Figure 17 showed statistically models' performance on the forecast track and intensity for Maliksi. It is worth noting that Maliksi was a short-lived TC with a life span only about 2 days, thus the sample size for calculating the forecast errors is rather limited. Because of its short life span, relatively slow movement and relatively weak intensity, it was not expected to be a challenging case for NWP in terms of track and intensity forecast. Figure 17 showed that both traditional and AI models were generally accurate, while the AI models (i.e., Pangu and Fuxi) were still better than ECMWF in the track forecast, but slightly underestimated the intensity. This is consistent with previous reports, for instance in Chan, He, and Lui (2024), and He and Chan (2023). Alongside with traditional NWP models which have been progressing significantly over the past decades, the newly emerging AI models can give an earlier alert on TC genesis and better TC track forecasts, further improving warning services. Nevertheless, current generation of AI models cannot forecast TC intensify satisfactorily and thus a careful combination of forecast guidance provided by the traditional and AI models should be used operationally.

[IMAGE OMITTED. SEE PDF]

[IMAGE OMITTED. SEE PDF]

On the other hand, as time progresses, the models do not seem to have captured well the persistence of strong winds in Hong Kong. Figure 18 shows the model forecasts initialized at 12 UTC 31 May 2024. It was forecast that, after 06 UTC or so, the winds over Hong Kong would subside significantly and no strong winds were expected. However, this was not the case, at least for the stations over the western part of Hong Kong (e.g., at the HKIA, Figure 18a). The models tended to forecast too fast dissipation of Maliksi over land, and the evolution of the strong southwesterly flow was not captured very well. Wind forecasts by the AI models (Figures 18c–18d) subsided even faster than traditional NWP, showing AI models' weakness in predicting the wind field structure of a tropical cyclone. The models' general weak bias on the local wind forecasts posed a great challenge to the wind warning service for this tropical cyclone.

[IMAGE OMITTED. SEE PDF]

Hong Kong Airport—Mountain Wake and Winds at the New North Runway

With the construction of the third runway of HKIA, it is the first time that strong southwesterly winds have prevailed over the region. So it is interesting to see the interaction between the southwest monsoon and the mountainous terrain near HKIA. A snapshot of surface winds in the airport area on 1 June 2024 is shown in Figure 19a. All of the anemometers over the new north runway recorded strong winds, whilst winds were relatively weak over the buoy to the east of the north runway and at the anemometers over the center and the south runways. This uneven wind distribution is suspected to be due to terrain effect.

[IMAGE OMITTED. SEE PDF]

To examine the terrain effect, high resolution NWP simulation (horizontal resolution of 40 m) was performed as reported in windshear studies for HKIA (e.g., Chan et al., 2023). The simulation was initialized at 00 UTC 1 June 2024 and run for 10 hr. A snapshot of the simulation result is shown in Figure 19b for the 10 m wind, and Figure 19c for the eddy dissipation rate (EDR) at the second model level (36.8 m above the sea surface). The mountain wake of Lantau Island was simulated nicely in Figure 19b, and the simulated surface winds were also generally consistent with the actual observations. This wake was revealed as a region of relatively lower EDR (Figure 19c). The higher EDR “streams” colored red in Figure 19c, were mostly affecting the center and the south runways, particularly their eastern ends. This uneven wind and turbulence distribution may have implications in the choice of operating runways in strong southwest monsoon situations.

Conclusions

This is a comprehensive study of Tropical Storm Maliksi in 2024, with detailed documentation of observational data and forecasting products. On the observation side, the low-level jet associated with southwest monsoon showed up in the vertical wind profiles, and the available in situ and satellite observations suggested that Maliksi was indeed a tropical storm. The warm core structure showed up nicely in the pseudo potential temperature data, and the available observations (such as sea temperatures and salinity, and the dropsonde data) all suggested favorable conditions for the intensification of the low pressure system into a tropical storm.

On the forecasting side, this paper documents the difficulty with the NWP models in the forecast of genesis of the system, and the persistence of strong southwest monsoon after Maliksi penetrated far inland over Guangdong. The AI models appeared to have done a rather good job, at least in the pre-genesis stage as well as in forecasting its track and movement, but the wind strength forecast was much under-estimated. Even for a marginal tropical storm, it is not straightforward to operate the tropical cyclone warning system based on the forecast of changes of wind strength in Hong Kong.

However, after so many years, the type and amount of observational data available become more comprehensive, and there are more forecasting tools such AI models, so that there are much more data and information to facilitate decision-making in the provision of tropical cyclone warning service. It is hoped that the information documented in this paper would be useful for reference for other centers operating tropical cyclone warning services at coastal areas.

Acknowledgments

Thanks to the National Marine Data Center, Guangdong-Hong Kong-Macao Greater Bay Area Branch (2024B1212080006) for supporting the provision of buoy and ocean radar data over the South China Sea in this study.

Data Availability Statement

The aircraft data used in the study are available from the Zenodo archive (Hong Kong Observatory, 2024).

References

Atlas, R., Hoffman, R. N., Ardizzone, J., Leidner, S. M., Jusem, J. C., Smith, D. K., & Gombos, D. (2011). A cross‐calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications. Bulletin of the American Meteorological Society, 92(2), 157–174. [DOI: https://dx.doi.org/10.1175/2010bams2946.1]