土壤粒径组成是土壤最基本的物理性质之一，直接影响土壤水力性质、土壤可蚀性、土壤肥力和养分迁移等。密云水库是北京重要的饮用水源地和生态涵养发展区，北京市对密云区的点源污染采取了严格管控措施，因农业生产管理中化肥农药不合理使用造成的非点源污染已成为影响密云水库水质的主要原因。土壤颗粒，尤其是土壤细颗粒是土壤养分和污染物的主要载体，在灌溉或降水过程中易随流水发生运移，造成地表或地下水污染。目前对密云水库流域的非点源污染已有较多研究，但是对该地区土壤粒径组成的空间分布研究较少。因此，有必要探讨该区农田土壤粒径组成的准确测量方法，并分析其空间分布，为当地农业生产管理、水文模型和非点源污染模型的建立和修订提供参考。 本研究选取北京市密云区的一块典型农地，沿样带分层采集0~100 cm土壤样品，分别利用吸管法和激光法测定其粒径组成，比较两种方法的测定结果存在的差异并探讨可能的原因。同时，采用扫描电镜法对其中的代表性样品进行测定，验证吸管法和激光法的测量结果，建立对应的校正方程。利用校正后的土壤粒径组成数据分析该农地土壤砂粒、粉粒和粘粒含量的空间分布特征，并采用半方差函数分析不同深度土壤粒径组成的空间变异。主要结论如下： （1）与吸管法相比，激光法测定的砂粒含量无显著差异，粉粒含量明显偏高，粘粒含量明显偏低。两种方法测定的砂粒和粉粒对应结果之间具有较好的线性转换关系。 （2）利用扫描电镜法测定吸管法吸取的粘粒颗粒发现，吸管法高估了粘粒含量，较少数量的> 2 μm的土壤颗粒贡献了较大的粘粒体积含量；激光法测定结果与此相反，高估了粉粒含量，低估了粘粒含量，造成激光法低估粘粒含量的主要原因可能在于其无法有效测定<1 μm的土壤颗粒，且测定的1.5~2 μm范围的土粒体积含量较低。 （3）研究样带上砂粒和粉粒含量的空间分布整体呈现相反的变化特征，砂粒含量较高的区域对应粉粒含量较低的区域，粘粒含量的空间分布较为破碎。40 cm是土壤砂粒和粉粒含量剖面分布发生改变的临界深度，粘粒含量随深度无明显变化趋势。 （4）根据半方差分析，研究样带土壤砂粒、粉粒和粘粒的空间异质性分别在60~70 cm、70~80 cm和70~80 cm深度达到最大，空间自相关距离分别介于17.00~56.00 m、11.20~31.10 m和19.40~125.70 m。土壤粒径组成受随机因素影响较小，其空间分异以结构变异为主。

Soil particle-size distribution (PSD) is one of the fundamental soil physical properties and directly affects soil hydraulic properties, soil erodibility, soil fertility and nutrient transport in the soil. Miyun Reservoir watershed is an important drinking water source and ecological conservation area in Beijing. Strict management regulations have been applied in Miyun district, where the Miyun Reservoir located, to reduce point-source pollution, and the non-point source pollution due to in appropriate uses of fertilizers and herbicides in agriculture has become the major cause for water quality deterioration in the Reservoir. Soil particles, especially the fine ones, are primary carriers for soil nutrients and contaminants. They transport with water flows during irrigation or precipitation, resulting in pollution of surface or ground-water or both. To date, many studies have been conducted on the non-point source pollution in the Miyun Reservoir watershed. However, little work has been done to analyze the spatial distribution of soil PSD there. It is therefore necessary to explore the accurate method for soil PSD determination in the farmland of Miyun district and to further analyze its spatial variability, thereby to provide reference for local agricultural production and management, establishment and modification of hydrological model and non-point source pollution model. In the current study, a 125-m transect was selected in a typical farmland in the Miyun District of Beijing. Along the transect, 1-m soil cores were collected at every 5 m. Besides, in 9 nests of 5 m, 1-m soil cores were sampled in an interval of 1 m to analyze the spatial variability across the distances shorter than 5 m. These soil cores were separated in 10 cm increments for soil PSD analysis. The pipette method (PM) and the laser diffraction method (LDM) were initially used and the corresponding soil PSD were thoroughly compared. These results were further verified with the scanning electron microscope (SEM), i.e., via applying on representative soil samples. Based on the results obtained with SEM, a regression equation was generated to calibrate the soil PSD analyzed with PM. These PSD data after calibration were then adopted to analyze the spatial variablity of sand, silt and clay contents at different depths using semivariogram. The main conclusions are as follows: (1)Compared to the soil PSD results of PM, LDM significantly overestimated the silt content and underestimated the clay content. No significant difference was detected in sand content between these two methods. Significant linear regression equations were derived between PM and LDM for sand and silt contents. (2)Using SEM to analyze the clay fraction obtained with PM, it was clear that the latter overestimated clay contents. A small number of > 2 μm soil particles contributes to a larger clay volume content. On the contrary, an overestimation of silt content and an underestimation of clay content were detected in the soil PSD determined by LDM.The probable reason was that LDM was not capable of detecting particles smaller than 1 μm and tended to underestimate the particles falling between 1.5~2 μm. (3)The spatial distributions of sand and silt contents along the sampling transect were generally reversed. The high value-domain of sand content corresponded with the low value-domain of silt content. The spatial distribution of clay content is relatively sporadic. 40 cm is the critical depth for the soil profile distributions of sand and silt contents. However, the clay content did not exhibit apparent trend with soil depth. (4)The maximum spatial variabilities were detected at the depths of 60~70, 70~80 and 70~80 cm for sand, silt and clay contents, respectively. The spatial correlation ranges of sand, silt and clay contents were 17.00~56.00, 11.20~31.10 and 19.40~125.70 m, respectively. The spatial variability of soil PSD was dominated by structural factors rather than random ones.