基于V/S分析的河川径流长记忆性研究

Study on long memory of river runoff based on rescaled variance analysis

针对河川径流的复杂波动性,将V/S分析引入到河川径流中进行长记忆性研究。通过V/S分析计算黄河上游兰州站和贵德站年径流序列的Hurst指数,与R/S分析进行比较,并进行稳定性与短期相关性检验,结果表明:兰州站、贵德站年径流序列V/S分析的Hurst指数均大于0.5,表明两站均具有较强的长记忆性,相对于R/S分析不易受短期相关性的影响,是一种稳健有效的分形方法,为河川径流长记忆性分析提供一种新的思路和方法。

To consider complex stochastic and undulant behaviors of river runoff,a method of rescaled-variance（V/S） analysis was applied to a long-memory analysis on the runoffs at the Lanzhou and Guide stations of Yellow River.The calculated Hurst indexes of the annual runoff series were compared with those by other methods,and a test on stability and short-term correlation was conducted.The V/S results shows that the indexes of the two stations are greater than 0.5,indicating a relatively strong long memory.In comparison with rescaled-range（R/S） analysis,V/S analysis is a robust and effective fractal method of runoff long-memory analysis,less vulnerable to the impact of short-term correlation,and this new method provides a new way of thinking.