运用非趋势波动分析方法,有效地消除数据中噪声和非平稳性质的影响,研究全球、北半球和中国近140年气温变化的长程幂律相关性。结果表明:全球气温、北半球气温和中国气温都有由交叉点划分的两个标度不变区域,可能对应两个不同的物理机制。其一,都可以表现出正长程相关的性质,而且全球气温的持续性最强,北半球次之,中国气温最弱;其二,全球气温和北半球气温还可以有几乎是1/f噪声性质的变化,中国气温则可表现出介于1/f和布朗噪声之间的行为。

To study the long range power-law correlations of the monthly average anomalies over the global, Northern Hemisphere, and China in the recent 140 years,the detrended fluctuation analysis (DFA)method is applied in this paper, which can effectively eliminate noise and possible nonstationarities in the data. The preliminary results show that there exist two different scaling invariant ranges, divided by one crossover in the three temperature series, which may indicate two different underlying physical mechanisms. Fist of all, they all present the positive long range correlations, and the persistence of the global temperature is the strongest, while China weaker than Northern Hemisphere. Secondly, the global and Northern Hemisphere temperatures can nearly behavior like 1/f noise, and China temperature can show behaviors between 1/f noise and Brown noise.