Get started
Work

Large Climate Model Ensembles Reveal Underdispersion in Seasonal Atlantic Tropical Cyclone Counts

Year: 2026
Type: article
Abstract: Seasonal Atlantic tropical cyclone (TC) counts are commonly modeled as a conditional Poisson process, implying that the distribution of possible seasonal outcomes—the range of TC counts that could plausibly occur in a given year—exhibits equidispersion for a given climate state, with its variance equal to its mean. This assumption underlies many statistical frameworks used for seasonal TC prediction and risk assessment, yet, to the best of our knowledge, has not been extensively tested directly. Using large ensembles from physics-based climate models (AM2.5-C360 and HiRAM) and a deep learning–based climate emulator [Ai2 Climate Emulator version 2 (ACE2)], we examine the distributional properties of seasonal TC counts in a controlled modeling framework. Across all models and years, we find that the ensemble distribution of within-season TC counts is systematically underdispersed relative to the Poisson distribution, with ensemble variances smaller than their corresponding means. This result provides evidence against modeling the SST-conditioned Atlantic TC count distribution with a Poisson distribution and instead supports the use of the binomial distribution. This representation of the Atlantic TC counts implies the following physical interpretation of TC genesis: TCs develop in two stages: 1) first from a finite number of seed disturbances and 2) followed by a seed–TC transition that is governed by large-scale environmental conditions. We fit seasonal TC count distributions from a 1000-member ACE2 ensemble to the Poisson and binomial distributions and find statistical evidence that the binomial formulation provides a better fit to TC count distributions. The Poisson assumption is not supported by current climate model simulations, given that simulated seasonal TC counts are more constrained than implied by traditional Poisson-based frameworks. Significance Statement Seasonal Atlantic tropical cyclone (TC) forecasts are critical for risk preparedness, energy planning, and insurance decision-making. Many statistical models used for these forecasts assume that TCs occur randomly, following a Poisson process. Leveraging unprecedentedly large ensembles from state-of-the-art physics-based and deep learning climate models, we show that simulated seasonal TC counts systematically violate this assumption. Instead, TC counts are more constrained. This behavior is consistent with a physical picture in which TCs arise from a finite number of precursor disturbances (“seeds”) whose likelihood of development depends on the large-scale environment. These results motivate more physically grounded statistical models for seasonal TC prediction.
DOI: 10.1175/mwr-d-26-0085.1
Open Access status: green
OpenAlex ID: W7212032176
FWCI:
Cites:
Cited by:
Related to: