Search Results (78)

Accurate forecasting of tropical cyclone (TC) tracks is critical for disaster preparedness and risk mitigation. While traditional numerical weather prediction (NWP) systems have long served as the backbone of operational forecasting, they face limitations in computational cost and sensitivity to initial conditions. In recent years, deep learning (DL) has emerged as a promising alternative, offering data-driven modeling capabilities for capturing nonlinear spatiotemporal patterns. This paper presents a comprehensive review of DL-based approaches for TC track forecasting. We categorize all DL-based TC tracking models according to the architecture, including recurrent neural networks (RNNs), convolutional neural networks (CNNs), Transformers, graph neural networks (GNNs), generative models, and Fourier-based operators. To enable rigorous performance comparison, we introduce a Unified Geodesic Distance Error (UGDE) metric that standardizes evaluation across diverse studies and lead times. Based on this metric, we conduct a critical comparison of state-of-the-art models and identify key insights into their relative strengths, limitations, and suitable application scenarios. Building on this framework, we conduct a critical cross-model analysis that reveals key trends, performance disparities, and architectural tradeoffs. Our analysis also highlights several persistent challenges, such as long-term forecast degradation, limited physical integration, and generalization to extreme events, pointing toward future directions for developing more robust and operationally viable DL models for TC track forecasting. To support reproducibility and facilitate standardized evaluation, we release an open-source UGDE conversion tool on GitHub. Full article

Part I of this two-part documentation of the ECMWF ensemble (ECEPS) new tropical cyclone track and intensity forecasting challenges during the 2024 western North Pacific season described four typhoons that started well to the south of an unfavorable external environment north of 20° N. In this Part II, five other 2024 season typhoons that formed and intensified near and north of 20° N are documented. One change is that the Cooperative Institute for Meteorological Satellite Studies ADT + AIDT intensities derived from the Himawari-9 satellite were utilized for initialization and validation of the ECEPS intensity forecasts. Our first objective of providing earlier track and intensity forecast guidance than the Joint Typhoon Warning Center (JTWC) five-day forecasts was achieved for all five typhoons, although the track forecast spread was large for the early forecasts. For Marie (06 W) and Ampil (08 W) that formed near 25° N, 140° E in the middle of the unfavorable external environment, the ECEPS intensity forecasts accurately predicted the ADT + AIDT intensities with the exception that the rapid intensification of Ampil over the Kuroshio ocean current was underpredicted. Shanshan (11 W) was a challenging forecast as it intensified to a typhoon while being quasi-stationary near 17° N, 142° E before turning to the north to cross 20° N into the unfavorable external environment. While the ECEPS provided accurate guidance as to the timing and the longitude of the 20° N crossing, the later recurvature near Japan timing was a day early and 4 degrees longitude to the east. The ECEPS provided early, accurate track forecasts of Jebi’s (19 W) threat to mainland Japan. However, the ECEPS was predicting extratropical transition with V_max ~35 kt when the JTWC was interpreting Jebi’s remnants as a tropical cyclone. The ECEPS predicted well the unusual southward track of Krathon (20 W) out of the unfavorable environment to intensify while quasi-stationary near 18.5° N, 125.6° E. However, the rapid intensification as Krathon moved westward along 20° N was underpredicted. Full article

This research investigates how incorporating both radar radial velocity ($V_r$) and radar reflectivity influences the accuracy of tropical cyclone (TC) prediction. Different control variables are introduced to analyze their roles in $V_r$ data assimilation, while background-dependent radar reflectivity assimilation methods are also applied. Using Typhoon “Doksuri” (2023) as a primary case study and Typhoon “Kompasu” (2021) as a supplementary case, the Weather Research and Forecasting (WRF) model’s three-dimensional variational assimilation (3DVAR) is utilized to assimilate $V_r$ and reflectivity observations to improve TC track, intensity, and precipitation forecasts. Three experiments were conducted for each typhoon: one with no assimilation, one with $V_r$ assimilation using $\psi -\chi$ control variables and background-dependent radar reflectivity assimilation, and one with $V_r$ assimilation using $U-V$ control variables and background-dependent radar reflectivity assimilation. The results show that assimilating $V_r$ enhances small-scale dynamics in the TC core, leading to a more organized and stronger wind field. The experiment involving $U-V$ control variables consistently showed advantages over the $\psi -\chi$ scheme in aspects such as overall track prediction, initial intensity representation, and producing more stable or physically plausible intensity trends, particularly evident when comparing both typhoon events. These findings highlight the importance of optimizing control variables and assimilation methods to enhance the prediction of TCs. Full article

To improve the prediction of Typhoon Doksuri (2023), this paper explores how variations in horizontal scale factors used in assimilating radar-derived wind velocities influence the performance of numerical simulations and forecasts. Using the WRF-ARW model in conjunction with the WRF-3DVAR data assimilation system, two assimilation configurations were tested with horizontal length scale factors of 1.0 and 0.25. Results show that a reduced length scale facilitates a more detailed reconstruction of mesoscale features, including the typhoon’s eye and inner-core circulation, leading to improved accuracy in short-term intensity and structure forecasts. The experiment utilizing the 0.25 length scale exhibited a tighter warm core, stronger cyclonic wind bands, and a better representation of the vortex’s three-dimensional structure. However, this configuration also led to growing forecast deviations in the latter stages, likely due to imbalances introduced by excessive localization. In contrast, the 1.0-scale experiment produced smoother but less accurate structures and demonstrated larger track deviations. These findings highlight a key trade-off between localized observational influence and long-term forecast stability. The study underscores the importance of optimizing horizontal scale parameterization in variational assimilation to enhance the forecasting accuracy of high-impact tropical cyclones and offers practical insights for operational forecasting systems in regions frequently affected by typhoon activity. Full article

To evaluate tropical cyclone forecasting on synoptic timescale, tracking and intensity are used. On subseasonal to seasonal (S2S) timescale, what aspects of tropical cyclones should be predicted and how to evaluate forecasting skills still remain open questions. Following our previous work, which proposed using daily tropical cyclone probability (DTCP) as a measure of tropical cyclone activity and the debiased Brier skill score (DBSS) to evaluate tropical cyclone forecasting on S2S timescale, the present research investigates the influence of several factors that may influence the use of DTCP and the DBSS framework. These factors are the forecast time window, tropical cyclone influence radius, evaluation region, forecast sample, and how the Brier score for the reference climate forecast is computed. The influence of these factors is discussed based on the output of the S2S prediction project database and comparison of the DBSS when the above factors are changed individually. Changes in the forecast time window, evaluation region, and tropical cyclone influence radius can change the DTCP. The larger the tropical cyclone influence radius and the longer the forecast time window, the larger the DTCP will be. However, the spatially averaged DBSS changes very little. Using estimated Brier score for reference climate forecast can cause variation due to limited forecast samples. It is recommended to use the theoretical value of the Brier score for reference climate forecasting, instead of its estimation. Full article

This paper presents a detailed analysis of the energy dynamics of the Mediterranean tropical-like cyclone, Medicane Ianos, by using a moist static energy (MSE) budget framework. Medicanes are hybrid cyclonic systems that share characteristics of both extratropical and tropical cyclones, making their classification and prediction challenging. Using high-resolution ERA5 reanalysis data, we analyzed the life cycle of Ianos, which is one of the strongest recorded medicanes, employing the vertically integrated MSE spatial variance budget to quantify the contributions of different energy sources to the cyclone’s development. The chosen study area was approximately 2500² km², covering the entire track of the cyclone. The budget was calculated after tracking Ianos and applying Hart phase space analysis to assess the cyclone phases. The results show that the MSE budget can reveal that the cyclone development was driven by a delicate balance between convection and dynamical factors. The interplay between vertical and horizontal advection, in particular the upward transport of moist air and the lateral inflow of warm, moist air and cold, dry air, was a key mechanism driving the evolution of Ianos, followed by surface fluxes and radiative feedback. By analyzing what process contributes most to the increase in MSE variance, we concluded that Ianos can be assimilated in the tropical framework within a radius of 600 km around the cyclone center, but only during its intense phase. In this way, the budget can contribute as a diagnostic tool to the ongoing debate regarding medicanes classification. Full article

A pre-operational test started in mid-July 2024 to demonstrate the capability of the ECMWF’s ensemble (ECEPS) to predict western North Pacific Tropical Cyclones (TCs) lifecycle tracks and intensities revealed new forecasting challenges for four typhoons that started well south of 20° N. As Typhoon Gaemi (05 W) was moving poleward into an unfavorable environment north of 20° N, a sharp westward turn to cross Taiwan was a challenge to forecast. The pre-Yagi (12 W) westward turn across Luzon Island, re-formation, and then extremely rapid intensification prior to striking Hainan Island were challenges to forecast. The slow intensification of Bebinca (14 W) after moving poleward across 20° N into an unfavorable environment was better forecast by the ECEPS than by the Joint Typhoon Warning Center (JTWC), which consistently over-predicted the intensification. An early westward turn south of 20° N by Kong-Rey (23 W) leading to a long westward path along 17° N and then a poleward turn to strike Taiwan were all track forecasting challenges. Four-dimensional COAMPS-TC Dynamic Initialization analyses utilizing high-density Himawari-9 atmospheric motion vectors are proposed to better define the TC intensities, vortex structure, and unfavorable environment for diagnostic studies and as initial conditions for regional model predictions. In Part 2 study of selected 2024 season TCs that started north of 20° N, more challenging track forecasts and slow intensification rates over an unfavorable TC environment will be documented. Full article

Tropical cyclones (TCs) frequently encompass multiple hazards, including extreme winds, intense rainfall, storm surges, flooding, lightning, and tornadoes. Accurate methods for forecasting TC tracks are essential to mitigate the loss of life and property associated with these hazards. Despite significant advancements, accurately forecasting the paths of TCs remains a challenge, particularly when they interact with complex land features, weaken into remnants after landfall, or are influenced by abnormal satellite observations. To address these challenges, we propose a generative adversarial network (GAN) model with a multi-scale architecture that processes input data at four distinct resolution levels. The model is designed to handle diverse inputs, including satellite cloud imagery, vorticity, wind speed, and geopotential height, and it features an advanced center detection algorithm to ensure precise TC center identification. Our model demonstrates robustness during testing, accurately predicting TC paths over both ocean and land while also identifying weak TC remnants. Compared to other deep learning approaches, our method achieves superior detection accuracy with an average error of 41.0 km for all landfalling TCs in Australia from 2015 to 2020. Notably, for five TCs with abnormal satellite observations, our model maintains high accuracy with a prediction error of 35.2 km, which is a scenario often overlooked by other approaches. Full article

Extremely severe cyclonic storms over the North Indian Ocean increased by approximately 10% during the past 30 years. The climatological characteristics of tropical cyclones for 38 years were assessed over the Bay of Bengal (BoB). A total of 24 ESCSs formed over the BoB, having their genesis in the southeast BoB, and the intensity and duration of these storms have increased in recent times. The Advanced Research version of the Weather Research and Forecasting (ARW) model is utilized to simulate the five extremely severe cyclonic storms (ESCSs) over the BoB during the past two decades using the Indian Monsoon Data Assimilation and Analysis (IMDAA) data. The initial and lateral boundary conditions are derived from the IMDAA datasets with a horizontal resolution of 0.12° × 0.12°. Five ESCSs from the past two decades were considered: Sidr 2007, Phailin 2013, Hudhud 2014, Fani 2019, and Amphan 2020. The model was integrated up to 96 h using double-nested domains of 12 km and 4 km. Model performance was evaluated using the 4 km results, compared with the available observational datasets, including the best-fit data from the India Meteorological Department (IMD), the Tropical Rainfall Measuring Mission (TRMM) satellite, and the Doppler Weather Radar (DWR). The results indicated that IMDAA provided accurate forecasts for Fani, Hudhud, and Phailin regarding the track, intensity, and mean sea level pressure, aligning well with the IMD observational datasets. Statistical evaluation was performed to estimate the model skills using Mean Absolute Error (MAE), the Root Mean Square Error (RMSE), the Probability of Detection (POD), the Brier Score, and the Critical Successive Index (CSI). The calculated mean absolute maximum sustained wind speed errors ranged from 8.4 m/s to 10.6 m/s from day 1 to day 4, while mean track errors ranged from 100 km to 496 km for a day. The results highlighted the prediction of rainfall, maximum reflectivity, and the associated structure of the storms. The predicted 24 h accumulated rainfall is well captured by the model with a high POD (96% for the range of 35.6–64.4 mm/day) and a good correlation (65–97%) for the majority of storms. Similarly, the Brier Score showed a value of 0.01, indicating the high performance of the model forecast for maximum surface winds. The Critical Successive Index was 0.6, indicating the moderate model performance in the prediction of tracks. It is evident from the statistical analysis that the performance of the model is good in forecasting storm structure, intensity and rainfall. However, the IMDAA data have certain limitations in predicting the tracks due to inadequate representation of the large-scale circulations, necessitating improvement. Full article

Global Navigation Satellite System (GNSS) Radio Occultation (RO) and GNSS Reflectometry (GNSS-R) are the two major spaceborne GNSS remote sensing (GNSS-RS) techniques, providing observations of atmospheric profiles and the Earth’s surface. With the rapid development of GNSS-RS techniques and spaceborne missions, many experiments and studies were conducted to assimilate those observational data into numerical weather-prediction models for tropical cyclone (TC) forecasts. GNSS RO data, known for its high precision and all-weather observation capability, is particularly effective in forecasting mid-to-upper atmospheric levels. GNSS-R, on the other hand, plays a significant role in improving TC track and intensity predictions by observing ocean surface winds under high precipitation in the inner core of TCs. Different methods were developed to assimilate these remote sensing data. This review summarizes the results of assimilation studies using GNSS-RS data for TC forecasting. It concludes that assimilating GNSS RO data mainly enhances the prediction of precipitation and humidity, while assimilating GNSS-R data improves forecasts of the TC track and intensity. In the future, it is promising to combine GNSS RO and GNSS-R data for joint retrieval and assimilation, exploring better effects for TC forecasting. Full article

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. Full article

Medicanes are rare cyclones in the Mediterranean Sea, with intensifying trends partly attributed to climate change. Despite progress, challenges persist in understanding and predicting these storms due to limited historical tracking data and their infrequent occurrence, which make monitoring and forecasting difficult. In response to this issue, we present an AI-based system for tracking and forecasting Medicanes, employing machine learning techniques to identify cyclone positions and key evolving spatio-temporal structural features of the cloud system that are associated with their intensification and potential extreme development. While the forecasting model currently operates with limited training data, it can predict extreme Medicane events up to two days in advance, with precision rates ranging from 65% to 80%. These innovative data-driven methods for tracking and forecasting provide a foundation for refining AI models and enhancing our ability to respond effectively to such events. Full article

Estimating tropical cyclone (TC) intensity is crucial for disaster reduction and risk management. This study aims to estimate TC intensity using machine learning (ML) models. We utilized eight ML models to predict TC intensity, incorporating factors such as TC location, central pressure, distance to land, landfall in the next six hours, storm speed, storm direction, date, and number from the International Best Track Archive for Climate Stewardship Version 4 (IBTrACS V4). The dataset was divided into four sub-datasets based on the El Niño–Southern Oscillation (ENSO) phases (Neutral, El Niño, and La Niña). Our results highlight that central pressure has the greatest effect on TC intensity estimation, with a maximum root mean square error (RMSE) of 1.289 knots (equivalent to 0.663 m/s). Cubist and Random Forest (RF) models consistently outperformed others, with Cubist showing superior performance in both training and testing datasets. The highest bias was observed in SVM models. Temporal analysis revealed the highest mean error in January and November, and the lowest in February. Errors during the Warm phase of ENSO were notably higher, especially in the South China Sea. Central pressure was identified as the most influential factor for TC intensity estimation, with further exploration of environmental features recommended for model robustness. Full article

Tropical cyclone (TC) genesis prediction remains a major operational challenge. Using multiple satellite datasets and a state-of-the-art reanalysis dataset, this study identifies developing and non-developing tropical disturbances over the western North Pacific from June to November of 2000–2019 and conducts composite analyses of their water vapor budget components and relevant dynamic–thermodynamic parameters in the Lagrangian framework following three-day disturbance tracks. Both groups of disturbances have a similar initial 850 hPa synoptic-scale relative vorticity, while the water vapor budget of developing disturbances exhibits distinct stage-wise evolution characteristics from non-developing cases. Three days prior to TC genesis, developing cases are already associated with significantly higher total precipitable water (TPW), vertically integrated moisture flux convergence (VIMFC), and precipitation, of which TPW is the most important parameter to differentiate two groups of disturbances. One day later, all the water vapor budget components (i.e., TPW, VIMFC, precipitation, and evaporation) strengthened, linked with the enhancement of the mid-to lower-tropospheric vortices. A negative radial gradient of evaporation occurs, suggesting the beginning of the wind−evaporation feedback. On the day prior to TC genesis, the water vapor budget components, as well as the mid-to lower-tropospheric vortices, continue to intensify, eventually leading to TC genesis. By contrast, non-developing disturbances are associated with a drier environment and weaker VIMFC, precipitation, and evaporation during the three-day evolution. All these factors are not favorable for the intensification of the mid-to lower-tropospheric vortices; thus, the disturbances fail to upgrade to TCs. The results may shed light on TC genesis prediction. Full article

Tropical cyclones (TCs) can cause severe wind and rain hazards. Unusual TC tracks and their extreme precipitation forecasts have become two difficult problems faced by conventional models of primitive equations. The case study in this paper finds that the numerical computation of the climatological component in conventional models restricts the prediction of unusual TC tracks. The climatological component should be a forcing quantity, not a predictor in the numerical integration of all models. Anomaly-based variable models can overcome the bottleneck of forecast time length or the one-week forecasting barrier, which is limited to less than one week for conventional models. The challenge in extreme precipitation forecasting is how to physically get the vertical velocity. The anomalous moisture stress modulus (AMSM), as an indicator of heavy rainfall presented in this paper, considers the two conditions associated with vertical velocity and anomalous specific humidity in the lower troposphere. Vertical velocity is produced by the orthogonal collision of horizontal anomalous airflows. Full article