| 南海热带气旋大风的遗传－神经网络集合预报 |
| |
| 引用本文： | 董彦,林开平,黄小燕.南海热带气旋大风的遗传－神经网络集合预报[J].气象研究与应用,2014(1):40-45. |
| |
| 作者姓名： | 董彦 林开平 黄小燕 |
| |
| 作者单位： | 广西师范学院资源与环境科学学院;广西区气象台;广西区气象减灾研究所; |
| |
| 基金项目： | 国家科技部气象行业专项(GYHY201106036);国家自然科学基金(41365002);广西自然科学基金北部湾重大专项(2011GXNSFE018006);广西科学研究与技术开发项目(桂科攻1355010-4) |
| |
| 摘 要： | 利用1980-2012年的南海热带气旋实况资料和NCEP/NCAR再分析资料,将热带气旋定位中心周边6×6格点上的地面风速作为预报对象,以气候持续预报因子和前期风速预报因子作为模型输入,采用遗传—神经网络集合预报方法,进行热带气旋定位中心周边36个格点上的风速预报模型的预报建模研究.分别对2008-2012年7-9月共368个独立预报样本进行遗传-神经网络集合方法的分月预报结果表明,南海热带气旋中心周边风速24h的预报平均绝对误差为2.35m.s-1.另外,本文还进一步将该预报方法与国内外普遍采用的逐步回归预报模型进行对比分析,在相同的预报量和预报因子的条件下的对比分析表明,新预报模型对≥10m.s-1的强风预报结果较逐步回归方法的优势明显,预报性能较好,可为沿海热带气旋大风预报提供新的参考.
|
| 关 键 词： | 遗传-神经网络 逐步回归 南海热带气旋 大风预报 |
A genetic neural network ensemble forecast method for strong winds of tropical cyclone in South China Sea |
| |
| Dong Yan,Lin Kaiping,Huang Xiaoyan.A genetic neural network ensemble forecast method for strong winds of tropical cyclone in South China Sea[J].Journal of Guangxi Meteorology,2014(1):40-45. |
| |
| Authors: | Dong Yan Lin Kaiping Huang Xiaoyan |
| |
| Institution: | 1. College of Resources and Environment Science, Guangxi Teachers Education University, Nanning 530001, China;2. Guangxi Meteorological Observatory, Nanning 530022, China; 3.Guangxi Research Institute of Meteorological Disasters Mitigation, Nanning 530022, China) |
| |
| Abstract: | Based on real-time samples of tropical cyclones over South China Sea and NCEP/NCAR reanalysis data during 1980 to 2012, a forecast model for wind velocity of tropical typhoons has been built adopting genetic neural network ensemble forecast method by taking wind velocity of 6×6 grid points around tropical typhoon centers as prediction objects and predictors of climatic persistence and preliminary wind speed as model inputs. The results showed that mean absolute error of fitting wind velocity for future 24h is 2.35m.s-1 based on 368 independent forecast samples from July to September in 1980-2012. Under identical predict and predictors, comparison experiments of stepwise regression models were also performed; and the results showed genetic neural network model was superior to stepwise regression method especially for strong winds （≥10m.s-1） and a novel reference for predicting strong winds of tropical typhoons along the coast was provided. |
| |
| Keywords: | Genetic-neural network Stepwise regression Tropical cyclone in South China Sea forecastfor strong winds |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
