摘要
以黄河三角洲地区典型地块为研究对象，应用地统计学的半方差函数理论，分析了0～40cm和40—80cm土壤盐分的空间变异特征。在对协同区域化变量进行交互半方差分析的基础上，采用协同克立格法，以电磁感应仪EM38测得的土壤表观电导率作为协同变量，对各层次的土壤盐分进行估值。结果表明：受结构性因素和随机性因素共同作用，各层次土壤盐分均表现中等强度的变异和空间相关性；随机性因素是引起土壤盐分空间异质性的关键原因；协同区域化变量间均表现为正相关，且协同区域化变量的空间结构优于单一变量。同普通克立格法相比，协同克立格法估值产生的均方误差减小13．1％～17．8％，平均标准误差减小5．83％～17．6％，预测值和实测值间的决定系数提高37．6％～42．6％。该研究结果为黄河三角洲地区土壤盐渍化的精准定量化、科学管理与合理改良提供了一定的理论基础和实践依据。
The Semivariogram theory was used to study spatial variability of soil salinity at 0 - 40cm and 40 - 80cm depths in a typical Yellow River Delta zone. The cross-semivariogram was then analyzed to explore co-regionalized relationships between soil salinity at both depths,and the apparent soil electrical conductivity measured through electromagnetic induction EM38. CoKriging method was further employed to improve the accuracy of soil salinity estimates by treating soil bulk electrical conductivity as auxiliary variable and soil salinity, at both layers, as primary variable. Results indicate that soil salinity at both layers exhibits moderate variation and spatial autocorrelation resulting from compound impact of structural and stochastic factors, and that stochastic factors are crucial in spatial heterogeneity of soil salinity. Co-regionalized variables show positive correlation among each other, and the spatial structure of co-regionalized variables is superior to individual component variables. Compared with ordinary Kriging method,the Root Mean Square Error （RMSE） from CoKriging decreases by 13.1% - 17.8% , Average Standard Error （ASE） decreases by 5.83% - 17.6% , and the coefficient of determination （ RR ） between observed and the predicted values increases by 37. 6% - 42.6%. The research results can serve as both theoretical and practical reference for precise quantification, scientific management and rational amelioration of saline soils in the Yellow River Delta.
出处
《中国生态农业学报》
CAS
CSCD
2008年第1期41-46,共6页
Chinese Journal of Eco-Agriculture
基金
中国科学院知识创新工程重要方向项目(KZCX2-YW-406-3)
国家科技支撑计划重点课题(2006BAD05B02)
国家高技术研究(863)发展计划重点项目课题(2006AA100207)
中国科学院南京土壤研究所创新方向项目资助
关键词
电磁感应仪
表观电导率
土壤盐分
空间分布
协同克立格
交互半变异函数
Electromagnetic induction, Apparent electrical conductivity, Soil salinity, Spatial distribution, CoKriging, Cross-semivariogram
