摘要
以松嫩平原土壤养分与地理坐标、土壤类型和微量元素等相关性为基础,对土壤养分分别进行协同克里格和普通克里格估值,并对比两种方法的估值精度,为区域尺度土壤属性模拟提供方法参考。结果表明:1)采用方差分析和相关分析,确定用于协同克里格估值的有机质的辅助变量为土壤类型、经度、pH值;碱解氮、速效钾的辅助变量为土壤类型;速效磷的辅助变量为土壤类型和Zn。2)有机质、碱解氮的块金效应分别为25.17%、72.48%,表现出中等空间相关性;速效磷和速效钾块金效应分别达83.06%、88.33%,为较弱空间相关性。3)协同克里格对区域尺度土壤属性空间模拟的效果要优于普通克里格。
Based on correlation analysis between soil nutrients of Songnen plain and factors including geographical coordinates,soil type,trace elements,etc.,we used Kriging and Cokriging method to simulate the soil nutrients respectively and compared their accuracy,for providing method reference of simulating soil properties at regional scale. According to the ANOVA and correlation analysis,we selected soil type,longitude and pH as co-factors for organic matter,soil type as the co-factor for available NH4and available K,soil type and Zn as co-factors for available P respectively. The nugget effect of organic matter and available NH4are 25. 17% and72. 48% respectively,showing moderate spatial correlation; while the nugget of available P and K are 83. 06%and 88. 33% respectively,showing weak spatial correlation. In a word,Cokriging interpolation on a regional scale is superior to Kriging in spatial simulation of soil properties.
出处
《干旱区资源与环境》
CSSCI
CSCD
北大核心
2014年第5期109-114,共6页
Journal of Arid Land Resources and Environment
基金
国家自然科学基金项目(41101537
40930101和71203157)
国家973计划项目(2010CB951502)资助
