A Comparison of Breeding and Ensemble Transform Vectors for Global Ensemble Generation

• Published in: Acta meteorologica Sinica 2012-02, Vol.26 (1), p.52-61
• Main Author: 邓国 田华 李晓莉 陈静 龚建东 矫梅燕
• Format: Article
• Language: English
• Subjects: Atmospheric Protection/Air Quality Control/Air Pollution; Atmospheric Sciences; Earth and Environmental Science; Earth Sciences; Geophysics and Environmental Physics; Meteorology; 中国气象局; 同时使用; 国家气象中心; 扰动方法; 转化载体; 选育; 集合平均; 集合预报系统
• ISSN: 0894-0525; 2191-4788
• DOI: 10.1007/s13351-012-0105-4

To compare the initial perturbation techniques using breeding vectors and ensemble transform vectors, three ensemble prediction systems using both initial perturbation methods but with different ensemble member sizes based on the spectral model T213/L31 are constructed at the National Meteorological Center, China Meteorological Administration （NMC/CMA）. A series of ensemble verification scores such as forecast skill of the ensemble mean, ensemble resolution, and ensemble reliability are introduced to identify the most important attributes of ensemble forecast systems. The results indicate that the ensemble transform tech- nique is superior to the breeding vector method in light of the evaluation of anomaly correlation coefficient （ACC）, which is a deterministic character of the ensemble mean, the root-mean-square error （RMSE） and spread, which are of probabilistic attributes, and the continuous ranked probability score （CRPS） and its decomposition. The advantage of the ensemble transform approach is attributed to its orthogonality among ensemble perturbations as well as its consistence with the data assimilation system. Therefore, this study may serve as a reference for configuration of the best ensemble prediction system to be used in operation.