田间入渗率的随机模拟分析

The stochastic simulation analysis of field infiltration rate

本文应用动态数据处理方法中时间序列的理论,对一田间试验地块稳定入渗率的空间变异性进行了分析研究.试验田块45m×45m,曾用两排不同盐分浓度的水喷灌.垂直于喷灌支管方向田间稳定入渗率变异较大,呈对数正态分布.根据样品自相关函数(SACF),入渗率序列是非平稳的.这一空间序列的非平稳性倾向可用一5次幂函数表示.对残余值随机序列可用一阶自回归和一阶滑动平均(AR(1)MA(1))随机模型拟合. 试验田块由于5年来采用不同的水盐组合灌溉模式,其入渗率在空间的变异性可分析为由三部分组成:用一5次幂函数反映其表土土壤由于用不同水盐组合灌溉模式使土壤性质变异而引起的确定性组成成分、反映土壤结构在空间固有的变异,及由于耕作条件等引起的随机模拟组成部分、和由于量测误差及其他因素所引起的噪音部分. 随机模型估算残余值加上5次幂函数值与实测值比较,结果良好,r^2=0.828.

The spatial variability of an experimental field infiltration rate is examined by using the theory of time series of dynamic analysis method. The experimental field is 45m×50m, and had been irrigated by two sprinkler lines with different water quality in the last five years. The field steady- infiltration rate had more variation in the direction perpendicular to sprinkler lines than in the other direction, and had log-normal distribution character. According to sample autocorrelation function the series of infiltration is nonstationary. A fifth-order polynomial is used to remove the trend. The residuals after removing the trend with a fifth-order polynomial are estimated by using the autoregressive and the moving average stochastic model AR(l)MA(l).As a result of different irrigation water-salt combination in the field, the spatial variability of infiltration rate consists of three parts: the first part is the deterministic component caused by different water-salt irrigation form, which resulted from the change and variation of properties of top soil; the second part is the stochastic component describing the variability of inherent soil structure and texture properties in the field; the third part is the random white noise.The estimated residuals plus the fifth-order polynomial values agree well with the observed values of infiltration rate and the square correlation coefficient comes to 0.828.