摘要
借助分形与混沌理论,对1990-2014年武夷山特大暴雨降水量的时间序列进行分析,重构其嵌入相空间,得出关联维数和饱和嵌入维数,进而确定了模拟相应动力系统所需的基本变量数目为8。基于R/S分析法计算了Hurst指数为0.8195,表明武夷山特大暴雨降水量的时间序列存在长期记忆性的特征。最后通过加权零阶局域法,对武夷山短期的特大暴雨降水量进行预测。研究结果客观、合理地反映了特大暴雨降水量的分形特征,可为建立特大暴雨降水量的时间序列预报模型提供有力的理论依据。
Based on the fraetal and chaos theory, the paper analyzed the time-series of extreme rainfall in Wuyi Mountain from 1990 to 2014, rebuilt the embedding space, calculated the correlative dimension and saturated embedding dimension, and determined the required number of basic variables of the correspond dynamic power system was 8. Moreover, according to the R/S analysis, the Hurst index was calculated as 0.85, which was inferred that the time- series of extreme rainfall in Wuyi Mountain had the long-term memory characteristics. Finally, by the weighting zero- order local-region prediction, we forecasted the short-term extreme rainfall in Wuyi Mountain. The results could reflect the fractal characteristics of extreme rainfall objectively and reasonably, and would provide the basic information for establishing the forecasting model of extreme rainfall time-series.
出处
《武夷科学》
2017年第1期118-124,共7页
Wuyi Science Journal
基金
福建省教育厅项目(JA13118
JK2013016)
关键词
特大暴雨
分形
关联维数
HURST指数
加权零阶局域法预测
extreme rain
fraetal method
correlative dimension
Hurst index
weighting zero-order local-region prediction
