KNKelvin S. Ng

Kelvin S. Ng

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kelvin S. Ng has a total of 65 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Global Compound Hot‐Drought Diagnostic From a Non‐Stationary Perspective

    by Liyan Huang, Gregor C. Leckebusch, Kelvin S. Ng, Kairong Lin, Zhiyong Liu

    Compound hot‐drought events (CHDEs) are intensifying under climate change and human pressures, posing mounting threats to socio‐economic stability. Such intensification underscores the non‐stationary nature of extreme climatic events, including droughts and heat, which are continuously evolving in response to a changing climate. However, conventional risk assessment frameworks, which relied on stationarity, often failed to capture evolving dependencies as external forcing alters hydroclimatic baselines. Therefore, we propose a non‐stationary standardized compound hot‐drought index (NSCHDI) to assess CHDEs using a time‐varying copula framework. This approach incorporates some major climate oscillations, such as ENSO and PDO, through principal component analysis to ensure temporal adaptability. Using global land data from 1950 to 2022, our results highlight three key insights: (a) The SPI 3 ‐STI coupling exhibited pronounced temporal variability, indicating a teleconnection modulated, non‐stationary dependence structure; (b) Compared to the conventional stationary index (SCHDI), NSCHDI enhanced the identification of severe‐to‐extreme CHDE occurrences by explicitly modeling climate‐conditioned, time‐varying dependence; (c) Beyond event identification, NSCHDI provided more reliable diagnostics of CHDE characteristics. Validation based on representative ENSO years (1997, 2010, and 2015) further demonstrated that NSCHDI‐based diagnostics better capture the timing and escalation of CHDE conditions, supporting enhanced early‐warning capabilities for detecting CHDEs. These results underscore the value of incorporating non‐stationary climate dynamics into multivariate indices to better quantify and manage to cascade risks from concurrent heat and drought extremes.

  • Open Access

    Climate-Driven Hail Risk Projections for the Continental United States

    by Kelvin Ng, Erik Larson, Nicholas Leach, Laura Ramsamy, Aidan Starr

    Hail causes billions in annual insured losses worldwide. It damages solar panels, roofs, vehicles, and crops; creating massive repair costs and operational disruptions. Financial institutions, insurers, and real estate investors face significant exposure to hail-driven losses, which affect portfolio valuations, underwriting decisions, and asset protection strategies. This hazard triggers immediate insurance claims, jeopardises infrastructure investments, and disrupts supply chains; making it critical for enterprise risk management. As climate change impacts severe weather patterns, businesses need forward-looking hail risk information and not just historical data.We present a new hail risk model developed by Climate X, featuring future projections across different shared socioeconomic pathways (SSPs) for the continental United States. Our model integrates baseline hail hazard data with climate projection methodologies to assess risk under multiple future scenarios. The framework combines high-resolution meteorological data with vulnerability curves based on asset-specific characteristics to quantify direct physical damage across infrastructure and commercial, industrial, and residential buildings.The model provides risk assessment at both asset and portfolio levels across multiple return periods, enabling stakeholders to evaluate present-day exposure and future climate scenarios. By incorporating SSP-based projections, our approach addresses the limitations of historical-only assessments and provides actionable intelligence for climate adaptation planning and risk management strategies in a changing climate.

  • Open Access

    What favours the midlatitude survival of cyclones of tropical origin (CTOs)?

    by Elena Bianco, Kelvin Ng, Gregor Leckebusch

    Cyclones of tropical origin (CTO) occasionally propagate to the midlatitudes, posing a significant hazard to regions that are unaccustomed to hurricane-force winds and extreme precipitation. Notable examples of CTOs that have significantly impacted Europe are Ophelia (2017), Lili (1996), and Leslie (2018). Given the rarity of these types of CTOs, the physical mechanisms that influence their formation, motion, and extra-tropical transition are poorly understood, complicating predictability and disaster risk response. In particular, the processes that enable the survival of CTOs in the midlatitudes are highly uncertain. Previous studies have suggested that the steering and intensification of CTOs is strongly modulated by the interaction with the background atmospheric circulation, but evidence is limited to few remarkable historical examples. In this study, we leverage ensemble hindcasts to construct a large, physically consistent set of plausible CTO events originating in the Atlantic Ocean that recurve eastward and reach the midlatitudes. Secondly, we apply a trough detection algorithm (Schemm et al. 2020) to investigate whether the interaction between cyclones and troughs plays any role in favouring or inhibiting CTO survival in the midlatitudes. The large volume of data provided by ensemble hindcasts is crucial for reducing uncertainty and advancing our understanding of the processes that may lead to CTO impacts in Europe, including how these processes may evolve under anthropogenic forcing.

  • Open Access

    Intra-seasonal variability of serial clustering of severe European winter windstorms

    by Sophie Feltz, Kelvin Ng, Christopher J. T. Allen, Tim Kruschke, Michael Angus, Andrew Quinn, Gregor C. Leckebusch

    Clustering of severe European winter windstorms causes large economic loss and social-environmental impacts. We present a detailed analysis of the intra-seasonal variability of clustering of severe European winter windstorms based on ERA5 reanalysis, with the focus on impact-relevant footprints of storm systems for the core winter season (DJF). The magnitude of clustering is quantified using the dispersion statistic on fixed 45- and 30-day timescales, as well as shorter 21-, 15-, and 11-day time-development windows. Results indicate an increase in the magnitude of clustering in the later part of DJF. Our analysis identifies for the first time that end-user (incl. the insurance sector) relevant storm clustering exhibits two intra-seasonal periods of enhanced clustering, especially in the densely populated regions of western central Europe (e.g., UK, Benelux, Northern France, Germany). Links to important large-scale modes (North Atlantic Oscillation -NAO, Scandinavian Pattern -SCA and East-Atlantic Pattern -EA) are analyzed as potential drivers.

  • Open Access

    Is Europe Under UNSEEN Risk of Cyclones of Tropical Origin?

    by Kelvin S. Ng, Gregor C. Leckebusch, Kevin Hodges

    When cyclones of tropical origin (CTOs) make landfall or pass near Europe, they are termed Europe-reaching CTOs (CTO-Es), and those causing measurable impacts are Europe-impacting CTOs (CTO-EIs). These events remain poorly understood, particularly regarding their variability, impacts and potential links to multidecadal climate variability. Using the UNprecedented Simulation of Extremes with Ensembles (UNSEEN) approach, we generate CTO event sets from twentieth-century hindcasts. The UNSEEN event sets reproduce spatial patterns of CTO-Es comparable to ERA-20C and contain 45-69 times more events, enabling robust hazard assessment. Results show CTO-EIs pose a greater risk to the British Isles than to continental Europe. While the number of CTO-Es reflects multidecadal variability in CTOs. The ratio of CTO-EIs to CTO-Es is insensitive to PDO or AMO, suggesting random occurrences under favorable conditions. Return period of a historical CTO-EI, such as Hurricane Lili (1996), is estimated at 21.2-32.9 years, with doubled-severity events at 47.8-91.5 years.


  • Open Access

    Classical and Machine Learning EMOS for postprocessing precipitation forecasts over India

    by Martin Widmann, Chandni Thakur, Michael Angus, Kelvin Ng, Noemi Gonczol, Raghavendra Ashrit, Andrew Orr, Gregor Leckebusch, Ruth Geen, Ashis Mitra

    Postprocessing of ensemble forecasts with Ensemble Model Output Statistics (EMOS) can reduce systematic errors in the ensemble mean and spread. Classical EMOS finds linear transformations between the original and postprocessed ensemble mean and variance that optimise the Continuous Ranked Probability Score (CRPS). Within the Weather and Climate Science for Service Partnership - India (WCSSP-India) project HEavy Precipitation Forecast Post-processing over India (HEPPI) we have implemented this approach to postprocess daily precipitation forecasts over India for the monsoon seasons 2018-2022 from the NEPS-G forecasting system run at the National Centre for Medium Range Weather Forecasting (NCMRWF) in Noida. In Angus et al. (2024) we have shown that over most of India EMOS improves the CRPS and the prediction of the probability for heavy precipitation, and also reduces over- or underdispersion in the ensemble.The EMOS approach is not restricted to the classical linear transformations between the original and postprocessed ensemble mean and variance, and to the optimisation of CRPS. It can be made more flexible by using Machine Learning (ML) to find non-linear transformations that optimise the CRPS or other forecast evaluation criteria. Within the WCSSP-India project HEPPI-ML we have explored different versions of ML-EMOS to postprocess NEPS-G daily precipitation forecasts. These include local and non-local Multilayer Perceptrons, and convolutional neural networks. We will present first evaluation results, including a comparison of ML and classical EMOS. Angus, M., M. Widmann, A. Orr, G.C. Leckebusch, R. Ashrit, and A. Mitra, 2024: A comparison of two statistical postprocessing methods for heavy‐precipitation forecasts over India during the summer monsoon. Quarterly Journal of the Royal Meteorological Society, 150(761), 1865-1883.

  • Open Access

    Objective Identification of Tropical Cyclone Systems With Potential for Storm Surge Impact in the Western North Pacific

    by Xiaoqi Zhang, Gregor C. Leckebusch, Kelvin S. Ng

    The robust assessment of storm surge hazards induced by tropical cyclones in the Western North Pacific is constrained by only ca. 50 seasons of reliable observational data. This can be addressed by constructing physically consistent large tropical cyclone event sets, for example, from ensemble simulations. To allow efficient construction of these event sets, we propose a combination (M&S‐WiTRACK) of two objective tracking methods relying solely on near‐surface information and validate the performance of the combination on detecting tropical cyclones with potential for storm surge impact using ERA5. The M&S‐WiTRACK is formed by a cyclone tracker (M&S) identifying tropical cyclone trajectories based on mean sea level pressure, with a wind‐based storm impact identification algorithm (WiTRACK) determining potential storm surge impact areas using 10‐m wind speed. For the first time, the general performance of the M&S for tracking tropical cyclones is evaluated and 84.9% of IBTrACS (with 18.3% false alarm rate) in the Western North Pacific from 1980 to 2019 are identified, which is well comparable to more data‐intensive tropical cyclone trackers. Furthermore, M&S‐WiTRACK successfully identifies over 85% of economic storm surge loss‐related tropical cyclones in China. Nearly all tropical cyclones causing storm surges with economic losses exceeding 2 billion RMB are identified.


  • Open Access

    “Boosted” realities: exploring the plausible limits of extreme weather through ensemble forecasts

    by Nicholas Leach, Shirin Ermis, Erich Fischer, Olivia Vashti Ayim, Aidan Brocklehurst, Kelvin Ng, Gregor Leckebusch

    High-impact low-likelihood extreme weather events and their impacts are of considerable interest to a variety of stakeholders across both the public and private sectors. Within the financial sector, there has been a focus on understanding how these kinds of extremes may change in the future, and quantifying the impact of such changes. However, we suggest that significant effort is still needed to fully assess the present day risk from such extremes, especially given the recent increase in apparently “unprecedented” extremes.Within academic research, the “UNSEEN” framework has recently gained traction as one approach to understanding the limits of extreme weather. However, this framework has typically focussed on using seasonal forecast simulations as they explore a wider range of longer-scale modes of climate variability than near-term forecasts. Using seasonal forecast simulations, however, places limits on the direct applicability to local extremes and introduces challenges resulting from model drift. Here, we present a corresponding approach using state-of-the-art medium-range reforecasts to explore the extreme upper tail of the weather distribution, inspired by the ensemble boosting methodology, which has thus far been implemented within relatively coarse resolution climate models. A key feature of basing our analysis on weather forecast simulations, as opposed to high resolution climate model simulations, is that the events produced are explicitly linked to the weather that actually occurred. We can analyse dynamically what would have had to happen differently for the UNSEEN extreme to become reality — and therefore assess how plausible it is and find the key synoptic precursors.These “boosted realities” are of wide utility - they provide physically consistent event storylines which can be used for emergency management and infrastructure design, or for the validation of the upper tail of event sets produced by the natural catastrophe models used in insurance. These plausible extremes could be ideal candidates for generating so called “Tales of Future Weather”, through the application of recently developed approaches in extreme weather attribution.

  • Open Access

    Extending the Lead Time for European Winterstorm Activity Predictions

    by Gregor C. Leckebusch, Kelvin S. Ng, Ryan Sriver, Lisa Degenhardt, Eleanor Barrie, Elisa Spreitzer

    The most dangerous and costly meteorological hazards in Europe are extreme extra-tropical cyclones and associated windstorms (EUWS) in winter. Recent studies have shown that seasonal prediction systems can skilfully predict the seasonal frequency of EUWS with a one-month lead time using November initialisations. Given that many seasonal prediction systems produce seasonal forecasts at the start of each month, this raises the question whether pre-November initialised seasonal forecasts could provide usable information in predicting seasonal activity of EUWS.In this study, we will present preliminary results of an approach aimed at extending the predictive horizon of seasonal EUWS activity. While the direct outputs of the pre-November initialised seasonal predictions of EUWS do not have the sufficient skill, skilful predictions of seasonal EUWS activity can be obtained by an approach that utilises the information of the upper ocean mean potential temperature from seasonal prediction systems. Based on our approach, skilful predictions of seasonal EUWS activity becomes possible as early as October.

  • Open Access

    Is Europe under UNSEEN Risk of Cyclones of Tropical Origin?

    by Kelvin S. Ng, Gregor C. Leckebusch

    Traditionally, European windstorms – the costliest meteorological hazards in Europe, are associated with extratropical cyclones in winter. However, in recent years, unorthodox cyclones such as Ophelia (2017), Leslie (2018), and Kirk (2024) have had noticeable impacts on Europe during autumn. These cyclones, referred to as Cyclones of Tropical Origin (CTOs), form in tropical or subtropical regions and can migrate toward Europe during their lifecycle. Although CTOs do not always cause significant impacts, they can exhibit exceptional intensity, posing unique hazards distinct from typical extratropical cyclones.This raises important questions: Are these isolated events? Will these events become more common in future climates? Current efforts to quantify the risk posed by CTOs are hindered by limited observational data and an incomplete theoretical understanding of these phenomena. As a result, Europe may face an unseen hazard from CTOs.In this presentation, we analyse CTO events using a physically consistent UNSEEN event set constructed from twentieth-century seasonal hindcast outputs (CSF-20C and SEAS5-20C). Our results show that while CTOs are rare, they are not isolated. We examine the interdecadal variability of CTO impact potentials—including wind, rainfall, and compound hazards—and assess their impact probabilities during the twentieth century. Finally, we present preliminary findings that highlight the genuine and previously unseen risk posed by CTOs to Europe.