An atmosphere–ocean–wave coupled regional model, the UWIN-CM, began its operational trial in real time at the Hong Kong Observatory (HKO) in the second half of 2024. Its performance in the analysis of three selected tropical cyclones, Severe Tropical Storm Prapiroon, Super Typhoon Gaemi, and Super Typhoon Yagi, are studied in this paper. The forecast track and intensity of the tropical cyclones were verified against the operational analysis. It is shown that the track error of the UWIN-CM was lower than other regional numerical weather prediction (NWP) models in operation at the HKO, with a reduction in mean direct positional error of up to 50% for the first 48 forecast hours. For cyclone intensity, the performance of the UWIN-CM was the best out of the available global and regional models at HKO for Yagi at forecast hours T + 36 to T + 84 h. The model captured the rapid intensification of Yagi over the SCS with a lead time of 24 h or more. The forecast winds were compared with the in situ measurements of buoy and with the wind field analysis obtained from synthetic-aperture radar (SAR). The correlation of forecast winds with measurements from buoy and SAR ranged between 65–95% and 50–70%, respectively. The model was found to perform generally satisfactorily in the above comparisons.
With the recent rise of artificial intelligence (AI), data‐driven global weather forecasting models have demonstrated superior performance compared to state‐of‐the‐art physics‐based global models across various weather elements. This work reports on tropical cyclone (TC) simulations using a hybrid weather modeling system that harnesses the advantages of both AI‐based and physics‐based models. The system utilizes AI‐based global models, Pangu‐Weather and AIFS, to drive the atmospheric model within a regional atmosphere–ocean‐wave coupled model (abbreviated as UWIN‐CM). It preserves skillful TC track forecasting from the global AI models while gaining the benefits of predicting fine‐scale details contributed by the high‐resolution UWIN‐CM model. The performances in forecasting seven TCs that necessitated the issuance of TC warning signals in Hong Kong in 2024 are studied. Results show that the AI‐model‐driven UWIN‐CM can achieve a reduction in track error by 34% compared to the UWIN‐CM driven by IFS. The track error is reduced to a level comparable to that of the AI models themselves. In terms of intensity, the AI‐model‐driven UWIN‐CM also gives a reduction in intensity error by 20% compared to the UWIN‐CM driven by IFS, and very significantly improves the intensity forecast provided by the AI global models. Other forecasting aspects, such as genesis, rapid intensification, and wind structure of TCs, are also investigated. The AI‐model‐driven results generally outperform those driven by IFS in these aspects. This work demonstrates that AI‐based global models and high‐resolution physics‐based regional models can complement each other to achieve more accurate TC forecasts.
In the period 2 to 4 March 2026, two rainstorms with intense convective weather occurred within and in the vicinity of Hong Kong, China, in the early rain season of the year in southern China. This is rather uncommon because the atmosphere is still generally stable (with very low or even zero value of convective available potential energy), and upper tropospheric divergence does not yet exist in the region climatologically. The rain episode is documented in this paper from both observational and forecasting aspects. On the observational side, a low-level vortex is found on and near the surface based on Doppler velocity measurements from a newly installed C-band solid-state weather radar. Combining the three-dimensional wind field as retrieved from the weather data and the measurements from the other ground-based remote-sensing meteorological equipment, the intense convection is mainly triggered by middle to lower tropospheric waves, and the vertical circulation in the atmospheric boundary layer may be stretched vertically upward to form the low-level vortex. In the second rainstorm, features of elevated thunderstorms are also identified. On the forecasting side, a high-resolution, limited-area atmosphere–ocean–wave coupled model manages to capture the occurrence and the timing of the heavy rain. The sub-seasonal forecast by a global model also provides a useful indication of the occurrence of above-normal rainfall over southern China, with a rather special feature of a deep and stationary westerly trough located to the north of the Indochina Peninsula. The microscale cyclone could be successfully picked up by the real-time run of a high-resolution numerical weather prediction model with data assimilation. This paper also discusses the weather service aspect of this rather unusual rainstorm episode.
A tropical depression (TD) formed over the northern part of the South China Sea and affected Hong Kong during 25–26 June 2025. Based on the historical database, there were not many TDs following a similar track in the past, namely, a northwestward track towards Hainan Island and the Leizhou Peninsula. This paper serves to document a number of aspects of the forecasting service for this TD, including: (1) consideration of the upgrade of the system from a tropical disturbance to a TD, possible further upgrade into a tropical storm, and the location of the centre in “multiple centre” situation (broad tropical cyclone centre and a mesocyclone embedded in the convection near the centre); (2) forecasting of the intensity and the impact on local winds; and (3) wind structure analysis based on dropsonde and wind profiler data. Moreover, this case demonstrates that artificial intelligence models are proven to provide earlier alerting of the possible occurrence of this TD and its subsequent movement towards the coast of southern China, whereas the conventional physics-based models remain useful in the forecasting of the impact of TD on the winds in Hong Kong for the operation of the tropical cyclone warning signal services.
An atmosphere–ocean–wave-coupled model, the UWIN-CM, running in real-time in the Hong Kong Observatory, has adopted radar data assimilation (DA) and been incorporated with a HAILCAST module in WRF for hail size simulation. This paper analyzes the distribution of maximum hail sizes forecast by the UWIN-CM in comparison with the Maximum Estimated Size of Hail (MESH) from radar for six hail cases reported by human or derived from radar in the Pearl River Estuary of Guangdong Province in 2026. The performance of hail forecast with and without radar DA is compared. The model with DA performed simulates hail at the locations that were consistent with the reported data and/or MESHs in five out of six cases. The timings of simulated hail aligned with reports/MESHs in three cases, and these were delayed by approximately 1 to 3 h in the remaining two. For the average hail diameters, those from simulations were smaller than MESH by around 30 to 50% in two of the five successful cases, while the other three cases had larger diameters than MESH by about 90 to 140%. Without radar DA, hails were simulated in only two out of six cases. The rate of detection was enhanced with radar DA, potentially benefiting from the improved moisture distribution and wind field of the initial conditions. The possible factors behind the differences in hail size between the simulations and MESH are discussed, including the choice of microphysics scheme and the differences in climatology between the PRE, and the contiguous US where the HAILCAST parameters were calibrated. For practical use in operation, parameter tuning of HAILCAST based on the hail climatology of the PRE would be needed. This work paves the way for providing forecasters with early alerts about the occurrence of hail in the region.
A funnel cloud is a rotating column of air that extends from the cloud base towards the ground yet does not touch the ground. It can be the precursor to a tornado or waterspout. The associated high winds can be destructive. This paper analyzed the meteorological observations for a funnel cloud spotted at Zhuhai in the afternoon of 8 May 2026, and performed simulation of its associated mesocyclone to study the formation mechanism of the funnel cloud. The funnel cloud was found to occur within a surface trough of low pressure under moderately unstable atmospheric conditions; winds were generally weak in the atmospheric boundary layer and the middle troposphere was wavy. The vorticity in the atmospheric boundary layer was lifted by the upward motion associated with the mid-tropospheric waves and daytime heating to form a funnel cloud; such a mechanism is supported by observations and model simulations. The weather radar generally captured the shallow convection and the radar-analyzed wind field depicted significant updraft in the reflectivity core associated with the mesocyclone. On the simulation side, the atmosphere–ocean–wave coupled model with radar data assimilation captured the isolated cyclonic feature near Zhuhai and the upward motion of the air column. Apart from the formation mechanism, this paper documented this rare event as a step forward in building up the climatology for the atmospheric conditions favourable for funnel cloud formation in the region.