基于Kriging方法的负偏态分布数据的空间插值研究
|
摘要
Kriging插值方法的前提是数据符合或基本符合正态分布的假设,或通过对非正态数据进行正态变换,使其满足要求。但对于拖尾的负偏态分布数据,正态变换方法无法奏效,若直接对数据进行Kriging插值,精度较低。文中在对样本数据深入分析的基础上,借助GIS平台,从不同透视角度对数据进行趋势分析,利用二次多项式模型对数据进行全局趋势拟合,并对随机的短程变异进行Kriging插值,选取不同的变异函数模型进行拟合比较,最终确定最优的模型参数,达到较为理想的插值结果。
The premise of Kriging interpolation method is to meet or in line with the normal distribution of data,or normally transformed to meet the requirements.But for the serious negative trailing skewed data,normality transformation method can not work.If the Kriging interpolation method is used directly on the data,the accuracy is low.Based on the sample data in-depth analysis.GIS platform is used from a different perspective angle data for trend analysis.The quadratic polynomial model is used to fit the data trends globally,and conduct Kriging interpolation for random short variation.In order to achieve the satisfactory interpolation results,different variogram models are selected and compared,and ultimately the optimal model parameters are determined.
The premise of Kriging interpolation method is to meet or in line with the normal distribution of data,or normally transformed to meet the requirements.But for the serious negative trailing skewed data,normality transformation method can not work.If the Kriging interpolation method is used directly on the data,the accuracy is low.Based on the sample data in-depth analysis.GIS platform is used from a different perspective angle data for trend analysis.The quadratic polynomial model is used to fit the data trends globally,and conduct Kriging interpolation for random short variation.In order to achieve the satisfactory interpolation results,different variogram models are selected and compared,and ultimately the optimal model parameters are determined.
引文
[1]曾怀恩,黄声享.基于Kriging方法的空间数据插值研究[J].测绘工程,2007,16(5):5-8.
[2]张梁,林韬,汪庆锋.使用克里金插值法进行CGCS2000到海南地方坐标系的转换[J].测绘与空间地理信息,2014,37(9):219-222.
[3]李晓辉,袁峰,白晓宇,等.典型矿区非正态分布土壤元素数据的正态变换方法对比研究[J].地理与地理信息科学,2010,26(6):102-105.
[4]陈道贵,胡乃联,李国清.区域化变量非正态分布的稳健性[J].北京科技大学学报,2009,31(4):412-417.
[5]吴学文,晏路明.普通Kriging法的参数设置及变异函数模型选择方法[J].地球信息科学,2007,9(3):104-108.
[6]陈华鑫,姜艺,李硕,等.基于ArcGIS的交通量预测模型[J].同济大学学报,2010,38(8):104-108.
[2]张梁,林韬,汪庆锋.使用克里金插值法进行CGCS2000到海南地方坐标系的转换[J].测绘与空间地理信息,2014,37(9):219-222.
[3]李晓辉,袁峰,白晓宇,等.典型矿区非正态分布土壤元素数据的正态变换方法对比研究[J].地理与地理信息科学,2010,26(6):102-105.
[4]陈道贵,胡乃联,李国清.区域化变量非正态分布的稳健性[J].北京科技大学学报,2009,31(4):412-417.
[5]吴学文,晏路明.普通Kriging法的参数设置及变异函数模型选择方法[J].地球信息科学,2007,9(3):104-108.
[6]陈华鑫,姜艺,李硕,等.基于ArcGIS的交通量预测模型[J].同济大学学报,2010,38(8):104-108.