A novel approach to enhancing the reliability of ensemble forecasts for unusual tropical cyclone tracks: The O-CNOPs
image: Yellow represents BVs, purple represents SVs, and blue represents O-CNOPs. The notation Eijk (i=1, 2, 3; j=1, 2, 3; k=1, 2, 3) represents the accuracy category defined by the ith error bound for turning location (i.e., 60km, 120km, 180km), jth error bound for turning time (i.e., 6h, 12h, 18h), and kth error bound for turning angle (i.e., 5°, 10°, 15°).
Credit: ©Science China Press
This study was led by Professor Duan Wansuo’s research team from the Institute of Atmospheric Physics, Chinese Academy of Sciences. Utilizing the WRF model, they conducted ensemble forecasting experiments across five tropical cyclone (TC) cases, each exhibiting sharp turns, spanning twenty-three forecast periods. The experiments systematically evaluated the performance of the O-CNOPs method against two traditional approaches: Bred Vectors (BVs) and Singular Vectors (SVs), focusing on their ability to enhance forecasting skill for unusual TC tracks.
At lead times from one to five days, the O-CNOPs demonstrated superior ability to generate ensemble members that accurately predicted sharp TC turns. Specifically, nearly 50% and 30% of ensemble members generated by O-CNOPs successfully captured the sharp turns at lead times of 4 and 5 days, respectively. In contrast, fewer than 10% of ensemble members generated by SVs and BVs captured sharp turns at these lead times. Furthermore, O-CNOPs provided more stable improvements to the control forecasts, achieving an average track error reduction exceeding 29%, whereas the ensemble mean forecasts of both BVs and SVs yielded only minor improvements (less than 3%) over the control forecasts.
Thus, O-CNOPs demonstrated superior performance over both SVs and BVs by delivering more stable and reliable enhancements in unusual TC track forecasting skill, from both deterministic and probabilistic viewpoints. Consequently, this study provides a novel ensemble forecasting technology that significantly boosts the accuracy of unusual TC track predictions, offering a promising solution to this complex forecasting challenge.
See the article:
Zhang H, Duan W S, Huang Y J, Chan P W, Vannitsem S. 2025. Improve the forecast reliability of unusual tropical cyclone tracks using ensemble forecasts generated by O-CNOPs. Science China Earth Sciences, 68(11): 3706–3721, https://doi.org/10.1007/s11430-025-1668-1