由农业生产而造成的污染，已经在土壤－水－生物－大气这个立体结构中造成了重大影响，是一种立体交叉污染。本研究围绕“土壤”这个农业立体污染的中心环节，重点探讨了农业立体污染中土壤对非点源污染的输出负荷和土壤自身残留氮、磷等的空间分布两个方面,主要研究工作和结果如下:1、系统总结了农业立体污染的概念、研究方面和重点，叙述了农业立体污染中物质的循环过程，并且详细说明了土壤中氮、磷的循环及其主要转化过程以及土壤氮、磷循环可能造成的环境污染问题，试图以此来对农业立体污染的理论加以补充。2、利用SWAT模型，研究和模拟了大兴区农业面源污染负荷。SWAT模型的应用以及前人对农业非点源污染中土壤输出负荷的研究主要以流域为研究区，本研究以北京市大兴区这样一个河流冲积平原、大都市农业行政区为研究区，研究结果便于行政部门对污染的控制。在搜集了SWAT模型所需要的各种研究区的空间数据（土地利用、DEM、土壤、水系等）和属性数据（土壤属性、气象数据、土地利用及植被覆盖等）基础上，利用SWAT模型模拟了研究区内2005年土壤对非点源污染的输出负荷。本研究分析了大兴区2005年非点源污染总负荷和月际变化，分析了研究区内非点源污染的时间和空间变化。结果表明：研究区非点源污染输出以降雨量最多的6月至9月最多；土壤输出的各种污染物负荷中，耕地的输出贡献最高，林地对各种污染物输出的贡献最小，农村居民地和城市居民地对溶解性氮、磷的输出有较大的贡献。最后，给出了非点源污染物质输出的控制和管理措施。3、系统比较了普通克里金法、泛克里金法和协克里金法用于估测土壤氮、磷等物质空间分布图的优劣，并进行了不同估计方法的精度验证。结果表明，三种方法中泛克里金方法可以去除数据的漂移趋势，具有最优的空间估计估测精度。在此基础上结合GIS，利用研究区的实际采样数据对氮、磷等可能造成土壤污染的物质进行了空间分布分析。4、依据全国土壤普查的氮、磷等物质的分级标准，对研究区内氮、磷等物质的空间分布进行了污染风险等级划分，并分析了不同土地利用类型与氮、磷等物质之间的相互关系。本研究在3S技术的支持下，研究农业立体污染中土壤对水环境和自身残留的氮、磷等物质造成的危害，并分析了各种形式的氮、磷的时空分布特征，取得了良好的结果，说明3S技术在农业立体污染的研究中是一种有效并具有前景的手段。

Nowadays, the pollution from agriculture have made a great affect in the tridimension ecological environment formed by soil, water, organism and atmosphere, and it was called agriculture tridimension pollution.This study focused on the kernel content of the agriculture tridimension pollution—the soil, and researched the pollutant yielded from soil in the course of agriculture tridimension pollution and the spatial distribution of the nitrogen and phosphorus in soil. The following achievements were made in this thesis.(1) The concept, research aspect and emphases of the agriculture tridimension pollution were systemly summarized. The cyclic process of the substance in the agriculture tridimension pollution was anlyzed; The cycle and main transformation process of the nitrogen and phosphorus in soil was detailed indicated; The possible environmental pollution from the cycle of the nitrogen and phosphorus was also explained. (2) In the past, the SWAT model and other researches on the pollutant yield from soil usually took the watershed as study area. In this research, the SWAT model was employed to simulate the pollutant yield from soil in Daxing district Beijing, which is a plain area. The various spatial data (land use, DEM, soil, stream system) and the relative attribute data (soil, atmosphere, land use and land cover) needed by SWAT model were collected. The total yield of the every pollutant and its intermonthly variation in 2005 of the Daxing were analyzed, and the temporal and spatial changes of the pollutants in study area were also illustrated. The results showed that: (1) the pollution yield was most serious from June to September, in which the rainfall was most abundant; (2) among the pollutants yielded from soil, the contribution of the farmland was biggest, and woodland can prevent the pollutant yield, but the rural resident and urban resident contributed more to the soluble nitrogen and phosphorus yield.(3) This study systemly compared the ordinary Kriging, universal Kriging and co-Kriging in the application to the interpolating of nitrogen and phosphorus spatial distribution. The result showed that the universal Kriging had the best precision in spatial interpolation, because it can remove the drift trend. Combined the GIS and the sampled data in study area, the spatial distribution of the nitrogen and phosphorus in soil was analyzed.(4) This research evaluated the risk rate of the poluution from the soil nitrogen and phosphorus according the state criterion, and analyzed the relations between the land use types and the soil residuals of nitrogen and phosphorus.This research showed that 3S technology was effective in monitoring the agriculture tridimension pollution.