非线性分形理论与时间序列分析法在土壤含水量预测中的应用研究

发布时间：2018-01-11 21:45

本文关键词：非线性分形理论与时间序列分析法在土壤含水量预测中的应用研究　出处：《长安大学》2016年硕士论文　论文类型：学位论文

【摘要】：土壤中水分的含量直接影响着农作物的生长,要准确预测水分所占的比例并进行适时的灌溉,需要我们充分地掌握其动态变化规律;但在实际中,土壤含水量变化受很多因素的影响,其变化规律极其复杂,呈现出非线性特征。因此,针对这种复杂的非线性特征,本文以分形理论为基础,采用时间序列分析法,对新疆阜康农田土壤含水量实测数据进行了研究,并对土壤含水量未来的变化趋势进行了预测,得到了一些有价值的结论。具体的研究步骤如下:(1)采用R/S分析法,对新疆阜康农田土壤含水量实测数据时间序列的Hurst指数和分形维数进行了分析对比,得到了土壤含水量具有长期的持续性特征。研究结果显示:10cm,80cm,150cm三个不同层的Hurst指数都大于0.5小于1,表明新疆阜康农田土壤含水量时间序列为非随机时间序列,时间序列存在着明显的长程相关性,该时间序列是持续性的。该地区时间序列的分形维数大于1小于2,表明该地区的土壤含水量变化的波动幅度随着深度的增加而逐渐变小。(2)采用去趋势波动分析法,对新疆阜康农田14个不同深度的土壤含水量时间序列的长程相关性进行了定量分析,并结合分形维数,研究了该地区土壤含水量的动态变化规律。研究结果显示:这14个不同深度的土壤含水量时间序列的标度指数随着土壤深度的增加而逐渐增大,即:该地区土壤含水量时间序列的长程相关性随着深度的增加而逐渐加强。分形维数的增加,揭示了土壤含水量变化的波动幅度随着深度的增加而逐渐趋于稳定。(3)本文采用了三种自回归模型,对新疆阜康农田土壤含水量进行了预测,得到了3种土壤含水量时间序列最理想的时间延滞。研究结果显示:季节的变化对土壤含水量的影响十分明显,这就要求我们在准确预测土壤水分含量的同时,要更加关注季节的变化对降雨量的影响,以便得到更为准确的结论。
[Abstract]:Soil moisture content directly affects the growth of crops. To accurately predict the proportion of water and carry out timely irrigation, we need to fully grasp its dynamic change law; However, in practice, the variation of soil water content is influenced by many factors, and its variation law is extremely complex, showing nonlinear characteristics. Therefore, this paper is based on fractal theory in view of this complex nonlinear characteristics. Time series analysis was used to study the measured data of farmland soil moisture in Fukang, Xinjiang, and the trend of soil water content in the future was predicted. Some valuable conclusions are obtained. The specific research steps are as follows: 1) the R / S method is used. The Hurst exponent and fractal dimension of the time series of measured soil water content data in Fukang Xinjiang were analyzed and compared. The results showed that the Hurst index of the three different layers was more than 0. 5 < 1. 5. The results showed that the time series of farmland soil moisture content in Fukang Xinjiang was a non-random time series, and the time series had obvious long-term correlation. The fractal dimension of the time series in this region is greater than 1 and less than 2. The results showed that the fluctuation range of soil water content in this area gradually decreased with the increase of depth. 2) De-trend fluctuation analysis method was used. The long-term correlation of 14 soil moisture content time series of different depths in Fukang farmland in Xinjiang was quantitatively analyzed and combined with fractal dimension. The dynamic variation of soil water content in this area was studied. The results showed that the scale index of the time series of 14 different depths of soil moisture increased with the increase of soil depth. That is, the long-term correlation of the time series of soil water content in this area is gradually strengthened with the increase of depth, and the fractal dimension increases. It was revealed that the fluctuation range of soil moisture content tended to be stable with the increase of soil depth.) in this paper, three kinds of autoregressive models were used to predict the soil water content of Fukang farmland in Xinjiang. The results show that the seasonal variation has a very obvious effect on soil moisture content, which requires us to accurately predict the soil moisture content at the same time. More attention should be paid to the effects of seasonal changes on rainfall in order to obtain more accurate conclusions.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S152.7;O212.1

【参考文献】