天气和气候的时间序列特征分析

The characteristic analysis of weather and climate time series

本文从天气和气候资料出发,提出气候的q阶(0≤q≤1)微商是天气,而天气可以近似为白噪声.在此基础上,利用描述自相似非马尔可夫随机过程的时间分数维扩散方程的分析成果,并结合时间序列的相关性分析,从理论上进一步指出气候信号的记忆性好于天气信号,且其概率密度分布的尾巴比较长.

By using the result of the time-fractional diffusion equation that presents a self-similar non-Markovian stochastic process, in this paper, we proposed our views on the relationship between weather and climate on the basis of the observational data from Beijing and Jinan: weather can be interpreted as the white noise, and the fractional derivative of order q of climate(0≤q≤1)is weather; furthermore, after studying the climatic discrete models and making correlation analysis of time series, we point out that a climate time series has a better memory and its probability density function has a longer tail with respect to the weather.