台风路径集合化预报方法的优化-海洋预报

台风路径集合化预报方法的优化

单位：1. 河海大学水文水资源与水利工程科学国家重点实验室, 江苏南京 210098;
2. 河海大学港口海岸与近海工程学院, 江苏南京 210098;
3. 江苏省水文水资源勘测局, 江苏南京 210029

分类号：P457.8

出版年·卷·期（页码）：2017·34·第二期（37-42）

摘要：

在现有集合化预报方法的基础上，基于预估偏差的思想设计了一种新的优化方案，以进一步提高台风路径预报的精度和稳定性。首先以1307 号台风“苏力”为例进行了单场台风的24 h集合预报，对预报过程中的误差进行了详细分析；然后对2013-2015 年发生在西北太平洋的所有台风进行了24 h、48 h 和72 h 集合预报，就其整体误差进行了对比分析。结果表明：优化预报方案在对单场台风的路径预报中，其24 h 集合预报误差相较于美、日、中、台等4 个气象台站和现有集合预报方案分别减小了12.88%、18.40%、30.58%、19.44%、33.72%，各个预报时刻的预报偏差值波动相对较小；在对2013-2015 年的多场台风路径预报中，在24 h 预报时效下，优化方案的平均预报误差比现有集合预报方法提高了2.85%，而均方差则减小了10.6%，说明优化方案在保持与现有预报方案精度相当的基础上，在预报稳定性上有了较为显著的提高；48 h 的集合预报误差相比于最好的台站中国台减小了4.38%，而72 h 的集合预报误差与表现最好的中国台的预报误差相当，且明显小于现有方案的预报误差。

Based on the existing super-ensemble forecast method, an improved method is proposed to further improve the typhoon track forecast accuracy and stability, which is relied on the principle of error-estimation ensemble. The performance of the method is examined by the forecasting of 24-hour of a single typhoon, i.e. ‘Soulik’ (No. 1307), in the region of northwestern Pacific, and the forecast errors are analyzed in detail. After that, the forecast of all the typhoons over the northwestern Pacific from 2013 to 2015 of 24-hour, 48-hour and 72-hour has been conducted by using the improved method to analyze its overall error. The results show that the better accuracy is achieved by the existing method in terms of the single typhoon and the forecast error has reduced by 12.88%、18.40%、30.58%、19.44%、33.72% compared to the results from the four individual forecast centers and the existing method. The deviation of forecast errors is relatively small compared to that of individual forecast centers. In terms of the whole typhoons from 2013 to 2015 of 24-hour, the average overall forecast errors could be reduced by 2.85% by the improved method compared with the existing one while the standard deviation of the errors could be reduced by 10.6%. The ensemble forecast error of 48-hour has reduced by 4.38% compared with the best single center and the 72-hour forecast result is close to the best center and both the results are better than the existing method. It indicates that the improved method could increase the forecast stability remarkably while keeping the same order of accuracy as the existing method.

参考文献：

[1] 李泽椿, 陈德辉. 国家气象中心集合数值预报业务系统的发展及应用[J]. 应用气象学报, 2002, 13(1): 1-15.
[2] 陈静, 陈德辉, 颜宏. 集合数值预报发展与研究进展[J]. 应用气象学报, 2002, 13(4): 497-507.
[3] 段明铿, 王盘兴. 集合预报方法研究及应用进展综述[J]. 南京气象学院学报, 2004, 27(2): 279-288.
[4] 黄燕燕, 万齐林, 袁金南, 等. 基于BDA扰动法的台风路径集合预报试验研究[J]. 热带气象学报, 2006, 22(1): 49-54.
[5] Krishnamurti T N, Kishtawal C M, LaRow T E, et al. Improved Weather and Seasonal Climate Forecasts from Multimodel Superensemble[J]. Science, 1999, 285(5433): 1548-1550.
[6] Zhi X F, Zhang L, Bai Y Q. Application of the Multi-model Ensemble Forecast in the QPF[C]//Proceedings of 2011 International Conference on Information Science and Technology. Nanjing, Jiangsu, China: IEEE, 2011.
[7] 丁雪霖, 陈永平, 顾茜. 2013 年西北太平洋台风路径集合预报[J]. 河海大学学报(自然科学版), 2015, 43(4): 361-365.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】