坡面林地土壤水分特征函数空间变异性初探

土壤水分特征函数空间异质性是定量研究土壤非饱和带水分运动以及溶质运移的先决条件. 以坡面机械布点采样、压力膜仪测定土壤水分特征曲线，并以Van-Genuchten模型拟合曲线后，用传统的统计方法与地统计方法，分析了密云水库流域周边人工油松林坡面土壤水分特征函数空间异质性. 结果表明:①不同土壤层各级压力下土壤含水量变异系数(C_v)呈中等变异性，C_v为10%～25%，土壤有效饱和度越低，变异性C_v越大，计算合理取样数目越大. ②模型参数θ_s为中等变异，C_v为17.77%～18.12%，服从正态分布；参数α为强变异性，C_v为70.65%～120.91%；参数n为弱变异性，C_v为5.29%～7.09%. ③各土壤层参数θ_s和α均具有空间变异结构，除20～40 cm层参数α符合指数模型外，其余层参数均符合球状模型，参数θ_s和α变程(A)分别约为66～69及21～69 m；参数n仅在20～40 cm土壤层显示空间变异结构，符合球状模型，变程(A)约为18 m；整体上该林地土壤0～20和20～40 cm层参数多以结构性变异为主，40～60 cm层参数的块金效应渐显增大. 最后采用Kriging插值方法分别对各参数进行了预测. 

Preliminary Study on Spatial Variability of Soil Water Retention Function in a Chinese Pine (Pinus Tabliforms) Plantation

Spatial variability of soil water retention function is a primary prerequisite for quantitative research on water movement and solution transfer in unsaturated zone. The slope mechanical sampling, pressure-membrane instrument measurement and Van-Genuchten model were applied to obtain fitted curve for soil water function, and an analysis of spatial variability of soil water retention function wasperformed with the conventional statistic and geo-statistical technique for slope soil of Chinese Pine (Pinus tabliforms) forest at Miyun Reservoir valley. The results show that: ① The soil water retention at each pressure head for three soil layers has presented moderate variability with C_v ranging from 10% to 25%. With the decrease of soil saturation, both C_v and the calculated reasonable sampling number were increased. ② For the fitted parameters of Van-Genuchten model, θ_s has demonstrated moderate variability with C_v from 17.77% to 18.12%; and n has presented small spatial variability, C_v from 5.29% to 7.09% for three layers; while C_v for the parameter of α is found from 70.65% to 120.91%. ③ Both parameters of θ_s and α for the three layers have demonstrated spatial structure with the spherical model, except for the parameter α with the exponential model for the 20～40 cm layer. The spatial scale of parameter θ_s for the three soil layers is found from 66 to 69 m, while 21 to 69 m for the parameter of α. Reversely, the parameter of n has merely shown spatial structure in spherical model inthe layer of 20～40 cm with the spatial scale of around 18 m. In general, the forest soil in the layer of 0～20 cm and layer of 20～40 cm is mainly in the structural variability, and the nugget effect of parameters gets more obvious for the layer of 40～60 cm. On the basis of above spatial variability, Kriging technique was employed to generate predicted surface for all parameters.