| 摘要: |
| 经SVD分析,截取足够多的预报场和因子场时间系数,使其相互关系代表两场的大尺度联系,预报场时间系数与其奇异向量线性组合估计场能反映原场主要特征.借助最优化技术,选择合理的系数,建立预测公式,由因子场时间系数预测预报场时间系数,同时订正预报场时间系数心a1 a2 aN本身的误差和反演过程中分析误差造成的场格点趋势预测的误差.最后将预测的预报场时间系数和对应奇异向量反演为整个场的预报.预报过程重点考虑可预报的大尺度变化,滤去不可预报的小扰动,依据两场主要耦合关系,预测预报场未来的主要变化. |
| 关键词: 最优化技术 奇异值分解 场估计 |
| DOI：10.18307/2003.sup09 |
| 分类号: |
| 基金项目: |
|
| Research in Forecast Technology of the Fields of Precipitation during the summer over the Middle Reaches of the Yangtze River |
|
ZHANG Liping
|
|
Wuhan Central Meteorological Observatory, Wuhan 430074, P. R. China
|
| Abstract: |
| With SVD (singular value decomposition) enough the time coefficients of predicted fields and factor fields are cut so that their relationships represent ones between two fields and estimate of linear combination of the time coefficients and the vector of predicted fields indicate the original. With optimization technique the whole field is predicted with estimating the time coefficients of predicted field according to the coefficients of factor fields and linear combination of the coefficients and the vectors of predicted field. Predictable large scales are analyzed mainly and unpredictable disturbance are canceled. Large-scale changes of predicted fields are predicted with chief coupled relationships between two fields. Forecast of the precipitation fields during the summer 1998 have been simulated and large areas of singular climates were predicted exactly. |
| Key words: Optimization technique singular value decomposition estimate of field |
