| 基于人工神经网络的热带气旋路径预报试验 |
| |
| 引用本文： | 吕庆平,罗坚,朱坤,任景鹏.基于人工神经网络的热带气旋路径预报试验[J].广东气象,2009,31(1):15-18. |
| |
| 作者姓名： | 吕庆平 罗坚 朱坤 任景鹏 |
| |
| 作者单位： | 解放军理工大学气象学院,江苏南京,211101 |
| |
| 摘 要： | 利用气候持续性因子，分别采用神经网络法及最小二乘回归法，建立西北太平洋地区12、24、36和48h热带气旋路径预报模型。通过对1992～2002年资料的试报，人工神经网络方法优于回归方法，且这种优势在预报时效较长时更明显。人工神经网络法48h的预报平均绝对误差比回归法减小27．56km，预报水平提高7％。
|
| 关 键 词： | 天气学 路径预报 人工神经网络 气候持续法 热带气旋 |
Experiments on Predicting Tracks of Tropical Cyclones Based on Artificial Neural Network |
| |
| L Qing-ping,LUO Jian,ZHU Kun,REN Jing-peng.Experiments on Predicting Tracks of Tropical Cyclones Based on Artificial Neural Network[J].Journal of Guangdong Meteorology,2009,31(1):15-18. |
| |
| Authors: | L Qing-ping LUO Jian ZHU Kun REN Jing-peng |
| |
| Institution: | L(U) Qing-ping,LUO Jian,ZHU Kun,REN Jing-peng |
| |
| Abstract: | By using climate persistent factors, Artificial Neutral Network method （ANN） and least square regression method（LS） are employed to establish forecast models for 12,24,36 and 48 hour forecast of tropical cyclone paths over Northwest Pacific respectively. Forecasting Experiment of the two models during 1992 -2002 shows that ANN has got advantages over KS model especially with lead time increasing. For example, 48 hour forecast by ANN demonstrates 7% improvement against that by LS. |
| |
| Keywords: | synoptic meteorology track forecast tropical cyclone ANN climate persistence method tropical cyclone |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
