基于不同协变量Cokriging土壤养分空间预测精度研究——以陕西省蓝田县为例

Study on Prediction Accuracy of Soil Nutrients Based on Cokriging in the Different Covariate Factors——A Case Study of Lantian County of Shaanxi Province

以蓝田县西北部农耕区2012年1 114份土壤有机质、碱解氮、有效磷、速效钾4个指标为基础,利用地理信息系统和地统计学相结合的方法,在对协变量个数控制的前提下,通过交叉检验系数和精度提高系数,探索协同克里格插值法对各土壤养分空间分布预测精度的影响。结果表明:各土壤养分空间分布不均匀,土壤养分存在中等变异性;利用增加协同变量方法,依据协变量之间相关性强弱控制协变量进入模型的次序对各土壤养分指标进行协同克里格插值,能提高预测精度,当协变量个数达到3时,各养分指标精度提高分别为有机质0.353%,碱解氮1.114%,有效磷1.088%,速效钾0.646%。研究结果较为准确地预测了样区4个养分指标的空间分布特征,结合土壤类型及土壤施肥管理方法,探讨了土壤养分空间分布特征的原因。

Based on the 1 114 data of agricultural districts of the northwestward in Lantian county for 2012,and the data including organic matter,alkali hydrolysable N,available P,available K the condition of controlling the number of concomitant variable,the author used the method of GIS and geostatistics to study the influence which the Cokiriging interpolation method had the precision of the spatial distribution of soil nutrient by cross coefficient tests and accuracy coefficient.The result shows that the degree of the abundance and lack of the spatial distribution of the soil nutrient is different,and anomalies exist in soil nutrient;using the method of increasing the Cokriging interpolation and the relative strength index among the concomitant variable can control the order when concomitant variable enters the pattern,and using the order can conduct the Cokriging interpolation,by which accuracy can be enhanced.And when the number of the concomitant variable is 3the increasing accuracy is as follows:organic matter is 0.353%,alkali hydrolysable N is 1.114%,available P is 1.088%,available K is 0.646%.The result forecasts the feature of the spatial distribution of four indicators of the soil nutrients relatively accurately in the sampling region.Combined with soil type and soil fertilizer management method,and the reason for spacial distribution charatteristics of soil nutrients was discussed.