本文已被：浏览 81次 下载 113次
投稿时间：2023-01-17 修订日期：2024-03-18
投稿时间：2023-01-17 修订日期：2024-03-18
中文摘要: 本文基于CMA台风最佳路径和ECMWF细网格模式资料，计算西北太平洋和南海台风生成期间多种环境因子的值，选出 7个显著性因子，采用双重检验逐步回归算法建立台风生成预报方程。独立样本检验结果表明，利用回归方法预报的风速值作为台风生成判据，较直接使用模式预报的风速和气压作为台风生成判据的击中率更高。双重检验逐步回归的台风生成预报方法在2022年的台风预报业务中进行了试用。对2203号台风“暹芭”的生成预报检验结果显示，回归方法表现较好，可以为预报员判别台风是否生成、以及生成的时间提供参考。
Abstract:Based on CMA best-track and ECMWF fine grid data, 16 variables during TC genesis were calculated. The stepwise regression algorithm of double test was used to select seven prediction factors with significance, and the TC genesis forecasting equation was established. The independent sample test results show that the hit rate of the wind speed predicted by the regression method is higher than that of the wind speed and pressure predicted by the numerical model. In 2022, the stepwise regression algorithm based on double testing was trialed in the TC genesis forecasting operation. The test results of generating forecast for typhoon 2203 " Chaba " show that the regression method has a good performance, which can provide a reference for forecasters to judge whether and when the typhoon is generated.
文章编号：202301170013 中图分类号： 文献标志码：
基金项目:国家自然科学(42175016)和(42175007)共同资助
| Author Name | Affiliation | Address |
| DONG Lin | National Meteorological Center | 北京市海淀区中关村南大街46号 国家气象中心 |
| LYU Xinyan | 北京市海淀区中关村南大街46号 国家气象中心 | |
| GAO Shuanzhu | ||
| WANG Qian |
引用文本：
DONG Lin,LYU Xinyan,GAO Shuanzhu,WANG Qian,0.Research on Tropical Cyclone Genesis forecasting Method based on Double Test Stepwise Regression[J].Meteor Mon,():-.
DONG Lin,LYU Xinyan,GAO Shuanzhu,WANG Qian,0.Research on Tropical Cyclone Genesis forecasting Method based on Double Test Stepwise Regression[J].Meteor Mon,():-.
DONG Lin,LYU Xinyan,GAO Shuanzhu,WANG Qian,0.Research on Tropical Cyclone Genesis forecasting Method based on Double Test Stepwise Regression[J].Meteor Mon,():-.
DONG Lin,LYU Xinyan,GAO Shuanzhu,WANG Qian,0.Research on Tropical Cyclone Genesis forecasting Method based on Double Test Stepwise Regression[J].Meteor Mon,():-.