高级检索

2024年西北太平洋台风的季节内至季节尺度预报评估

Subseasonal to Seasonal Typhoon Forecast in the Western North Pacific in 2024

  • 摘要: 针对2024年西北太平洋台风活动,采用逐日台风概率作为预报变量,以去偏布莱尔预报技巧评分和漏报误差及空报误差为评价指标,并基于中国气象局最佳路径数据集、世界气象组织季节内至季节尺度实时台风预报数据库中欧洲中期天气预报中心和中国科学院大气物理研究所的集合预报产品对台风预报技巧进行了评估,并对2024年西北太平洋台风的特点做了总结。研究结果表明,与1991~2020年平均相比,2024年5~11月的台风总数增加了2.3个,菲律宾以东区域的平均逐日台风概率比往年偏高。ECMWF及中国科学院大气物理所模式对西北太平洋台风季节内至季节尺度的DBSS在10 d之内明显减小,之后变化不大。但在30 d之内两个模式的预报技巧均优于考虑季节变化的参考气候预报。两个模式的漏报误差在10 d之内随预报时效明显增加。空报误差随预报时效变化不大。空报误差出现的区域随预报时效增加了6~8倍。预报逐日台风概率在空间上发散,在量值上减少。

     

    Abstract: To mitigate the losses caused by typhoons, accurate typhoon forecasts are indispensable. Currently, synoptical and seasonal time scale typhoon forecasts have matured in term of operational practice, but there is no universal standard in the subseasonal to seasonal (S2S) time scale typhoon forecasts. For the typhoons in the Western North Pacific in 2024, this study uses Daily Tropical Cyclone Probability (DTCP) as the forecasting variable, Debiased Brier Skill Score (DBSS), missing and false-alarm error as evaluation metrics to assess the S2S forecast. We assess the forecasting skill based on the best track dataset from the China Meteorological Administration and the real-time typhoon forecasting database of the World Meteorological Organization, specifically the products from the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP). The study also analyzes the characteristics of the Western North Pacific typhoons in 2024. The results indicate that the total number of typhoons between May and November of 2024 has increased by 2.3 compared to1991-2020 average. The averaged DTCP of 2024 is much higher than the climatological DTCP east of the Philippine islands. The forecast skill of ECMWF model and IAP model as measured by DBSS decreases rapidly within 10 days, and remains constant after that. Both models are more skillful than the seasonal reference climate forecast within 30 day lead time. The missing error of both model increases within 10 days and keeps unchanged after that. The false-alarm error keeps at a low level. However, the area with false alarm increases greatly by a factor of 6 to 8, associated with the dispersion of forecast DTCP area and the reducing of its value.

     

/

返回文章
返回