Shifting hotspot of tropical cyclone clusters in a warming climate

nature climate change Article https://doi.org/10.1038/s41558-025-02397-9 Shifting hotspot of tropical cyclone clusters in a warming climate Received: 27 September 2024 Accepted: 3 July 2025 Published online: 31 July 2025 Check for updates Zheng-Hang Fu 1,2,10, Dazhi Xi 3,4,10, Shang-Ping Xie 5, Wen Zhou Ning Lin 4, Jiuwei Zhao 6, Xin Wang 1 & Johnny C. L. Chan 7,8,9 1,2 , Multiple tropical cyclones can be present concurrently within one ocean basin, and these clusters can induce compound hazards within a short time window. While the western North Pacific has historically been home to most tropical cyclone clusters, how climate change might affect this is unclear. Here we use observations and high-resolution climate model simulations to develop a probabilistic model, assuming that tropical cyclones are mutually independent and occur randomly. Against this baseline, we identify outliers as clusters with dynamic interactions between tropical cyclones. We find that the recent global warming pattern induces major shifts in tropical cyclone cluster hotspots from the western North Pacific to the North Atlantic by modulating tropical cyclone frequency and synoptic-scale wave activity. Our probabilistic modelling indicates a tenfold increase in the likelihood of tropical cyclone cluster frequency in the North Atlantic, surpassing that in the western North Pacific, from 1.4 ± 0.4% to 14.3 ± 1.2% over the past 46 years. On 14 September 2020, an extreme tropical cyclone (TC) cluster made headlines, with five TCs entrenched over the North Atlantic (NA)1 (Fig. 1a and Supplementary Fig. 1). That year witnessed an unusually active Atlantic hurricane season, with nine storms forming in succession within 3 weeks (Fig. 1a). Such back-to-back TCs over the NA and their threat to the coastal USA have increased in recent decades2–4. Here, we define TC clusters as two or more TCs present simultaneously within the same basin5,6. Historically, only 40% of TCs appeared alone, with the majority of TCs coming in clusters6. Beyond the combined impacts of individual TCs, TC clusters can lead to disproportionate damage along coastal regions because infrastructure, communities and restoration resources cannot bounce back from the damage from the preceding TC within a short period of time2,7–9. In addition, dispatching limited emergency supplies to affected areas is rather difficult when multiple TCs impact different regions concurrently, as exemplified by the Federal Emergency Management Agency’s failure to provide adequate support to Hurricane Maria’s victims in Puerto Rico after its overextended responses to hurricanes Harvey and Irma in 201710. Although the extreme TC cluster in 2020 is relatively new to Atlantic coastlines, East and Southeast Asian coastal regions have long suffered from such temporally compound events. In late summer 2004, over the western North Pacific (WNP), nine disturbances intensified into TCs within 34 days (refs. 11,12), five of which made landfall in East Asia (Fig. 1b). While the majority of TC clusters historically occur in the WNP, how climate change might affect this preference remains unclear. Previous studies have analysed large-scale dynamic and thermodynamic conditions that are favourable for TC genesis to investigate Key Laboratory of Polar Atmosphere-Ocean-Ice System for Weather and Climate, Ministry of Education, Department of Atmospheric and Oceanic Sciences and Institute of Atmospheric Sciences, Fudan University, Shanghai, China. 2Key Laboratory for Polar Science MNR, Polar Research Institute of China, Shanghai, China. 3Department of Earth and Planetary Sciences, The University of Hong Kong, Hong Kong, China. 4Department of Civil and Environmental Engineering, Princeton University, Princeton, NJ, USA. 5Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA, USA. 6State Key Laboratory of Climate System Prediction and Risk Management, Institute of Climate and Application Research, Nanjing University of Information Science and Technology, Nanjing, China. 7Shanghai Typhoon Institute of China Meteorological Administration, Shanghai, China. 8Asia-Pacific Typhoon Collaborative Research Center, Shanghai, China. 9School of Energy and Environment, City University of Hong Kong, Hong Kong, China. 10 These authors contributed equally: Zheng-Hang Fu, Dazhi Xi. e-mail: 1 Nature Climate Change | Volume 15 | August 2025 | 850–858 850 Article https://doi.org/10.1038/s41558-025-02397-9 a c Extreme NA TC cluster season: 31 August to 23 September 2020 4 Sep 1 Sep 7 Sep 13 40 Sep 19 2 OMAR 30° N NANA RENE 20° N SALLY PAULETTE TEDDY Frequency Latitude 40° N 70° W 50° W 24 16 0 BETA 90° W 30° W 1980 Longitude 2 Aug 7 Aug 16 Aug 25 2000 2010 2020 Year 40 Sep 3 1 Tau: 0.39 T1: –0.87 T2: –0.36* 32 40° N MALOU MERANTI RANANIM 30° N MALAKAS Frequency 0 Latitude 1990 d Extreme WNP TC cluster season: 4 August to 7 September 2004 T2: 0.93* 8 VICKY WILFRED 10° N T1: 2.45* TC TC cluster 32 0 b Tau: 0.68 24 16 MEGI 20° N CHABA AERE 10° N SONGDA SARIKA 110° E 130° E 150° E 170° E Longitude 8 0 1980 1990 2000 2010 2020 Year Fig. 1 | Extreme TC cluster seasons and observed changes in TC frequency and TC cluster frequency. a,b, The TC activity from 31 August to 23 September 2020 in the NA (a) and from 4 August to 7 September 2004 in the WNP (b). The histogram shows the daily TC counts within the period, and the map shows coloured tracks for each TC. Hurricane Paulette (red line in a) regenerated into a TC after its extratropical transition, so we connect the two tracks with a dotted line. c,d, Time series of TC frequency (blue lines) and TC cluster frequency (orange lines) during 1979–2024 over the NA (c) and WNP (d). Kendall rank correlations (Tau) between TC frequency and TC cluster frequency are shown at the top left. The linear trends of TC frequency (T1) and TC cluster frequency (T2) are plotted as dotted lines, with the associated 10-year trend values presented in the top panel. Asterisks denote significance at the 95% confidence level on the basis of the 1,000-sample bootstrapping. Bold dots in b indicate that the frequency over the NA reaches or exceeds that over the WNP, occurring in 5 years for TC frequency and 10 years for TC cluster frequency during 1979–2024. TC cluster formation (for example, refs. 4–6,13). Additionally, recent studies have highlighted changes in TC climatology features, including frequency14,15, seasonality16–18 and duration19,20 under anthropogenic warming. However, understanding how these TC climatology features besides the mechanisms at TC genesis influence TC cluster activity remains a challenge. Two possible conditions for TC cluster formation exist. First, TC genesis may involve physical processes related to pre-existing TC(s), thus contributing to TC cluster formation21,22. TC-induced Rossby wave dispersion4,5,23,24, synoptic-scale w (...truncated)