基于去趋势波动分析方法的土壤水分长程相关性研究  

Long-range correlations of soil moisture series with detrended fluctuation analysis

作　　者：宋闰柳[1,2] 于静洁[1] 刘昌明[1] 

机构地区：[1]中国科学院地理科学与资源研究所陆地水循环及地表过程重点实验室,北京100101 [2]中国科学院研究生院,北京100049

出　　处：《水利学报》2011年第3期315-322,共8页Journal of Hydraulic Engineering

基　　金：国家重点基础研究发展计划("973"计划)(2009CB421305);国家自然科学基金面上项目(40371025)

摘　　要：采用去趋势波动分析法(DFA),分析华北山区2004—2007年逐日土壤水分序列的长程相关特性,为土壤水动态动力学机制的揭示以及模拟预测研究提供客观依据。主要结论有:(1)观测点土壤水序列并非完全随机的,具有较为显著的线性趋势,二阶去趋势波动分析可以有效地去除原始序列的线性趋势;(2)研究区土壤水序列波动形式接近于分维布朗运动,是一个由内在自相似机制决定的长程相关过程。长程相关性强弱的变化与土壤深度变化并不一致;(3)土壤水序列(20cm,40cm,70cm)标度指数存在"拐点"。随着时间尺度的增大,土壤水序列的长程相关性质发生变化。Long-term correlations are considered as one of the most essential characteristics of soil moisture system,and related research would contribute to understanding of soil moisture dynamic mechanism and simulation.Based on the daily soil moisture series observed in Dongtaigou catchment from 2004 to 2007,this paper has explored the long-range correlations of soil moisture data with detrended fluctuation analysis（DFA） and discussed the long-term memories.The results are as follows.（1） There is a significant linear trend in soil moisture series.The second-order detrended fluctuation analysis could remove the linear trend in the original series effectively and help to reveal the long-term correlation.（2） The fluctuation of soil moisture around the general trend exhibits self-similar scaling behavior and long-term correlation rather than a complete random process at time scales,which is governed by the intrinsic self-similar properties.The soil moisture dynamics in the catchment observed may fluctuate as fractional Brownian motion.（3） Crossovers are found in fractal scaling of soil moisture dynamics in the time scale of 44 or 48 days.Temporal correlations become weaker as time scale increases.

关 键 词：土壤水分 长程相关性 去趋势波动分析法 

分 类 号：S152.7[农业科学—土壤学]