摘要
克里金插值和条件模拟作为两种常用的空间变异近似和逼近方法,能否真实刻画要素的多尺度结构关系到相应结论的正确性。本文采用尺度方差分析方法,结合经典的半方差函数和Moran’s I,以伊犁新垦区土壤有机质为例,对普通克里金插值(OK)和序贯高斯模拟(SGS)的结果进行多尺度分析。结果显示新垦区土壤有机质具有比较强的空间自相关性,10 km、35 km和60 km左右为土壤有机质的特征尺度。在小尺度上,SGS对真实的空间数据尺度结构刻画都出现一定的偏差。在大尺度上,两者能够比较真实地反映数据的多尺度结构,可有效刻画35 km和60 km的特征尺度。
Kxiging interpolation and conditional simulation are two common methods of spatial variation approaching, whether or not they can describe multi-scale structure of data relates to the correctness of corresponding conclusions. In this paper, we took the soil organic matters in the newly reclaimed area in Yili as an example and focused on multi-scale structure of ordinary kriging interpolation(OK) and sequential gaussian simulation(SGS) with scale variance analysis, semi-variance and Moran's I. We inferred that 15 km, 35 km and 60 km might be characteristic scales of soil organic matters in the newly reclaimed area. In small scale, both of them resulted in bias on scale structure of real spatial data characteristic. In large scale, both of them could reflect the multi-scale structure of data and characterize the characteristic scales of 35 km and 60 km.
出处
《土壤》
CAS
CSCD
北大核心
2013年第1期91-98,共8页
Soils
基金
国家自然科学基金项目(41071065)资助
关键词
普通克里金插值
序贯高斯模拟
土壤有机质
多尺度分析
Ordinary kriging interpolation, Sequential gaussian simulation, Soil organic matters, Multi-scale analysis