土壤质地空间预测方法比较

Comparison of spatial prediction method for soil texture

土壤质地作为成分数据(compositional data)的一种,其空间插值需满足非负、定和、误差最小和无偏估计4个条件。采用成分克里格法(compositional Kriging)和基于对数比转换的普通克里格法对土壤质地各颗粒组成进行空间预测,均方根误差(root mean squared errors,RMSE)和标准化克里格方差(mean squared deviation ratio,MSDR)分别被用来衡量不同方法的预测精度及模型拟合效果。研究结果表明:对数比转换的普通克里格法和成分克里格法能够保证插值结果满足成分数据插值的4个条件;成分克里格法预测的各土壤颗粒组成的RMSE最小,预测精度最高,其黏粒RMSE值相对于非对称对数比转换的普通克里格法提高将近17%;成分克里格法的变异函数拟合效果总体上好于其他两种预测方法,预测结果极差更宽,更能反映土壤质地各颗粒组成与高程、母质和水域分布的关系。

Soil texture is one of compositional data in the geosciences. Spatial interpolation for soil texture must meet four conditions,which including positive definiteness,a constant sum of interpolated values at a given position,error minimization and unbiased estimation. The study adopted compositional Kriging and ordinary Kriging based on data transformed by asymmetry logratio transformation （ALR） and symmetry logratio transformation （SLR） to predict spatial distribution of each soil particle composition. The precision and fitting effect were assessed by utilizing the root mean squared errors （RMSE） and mean squared deviation ratio （MSDR）. The results showed that the interpolation results by compositional Kriging and ordinary Kriging based on data transformed by ALR and SLR could meet the four conditions in spatial interpolation. Values of RMSE for different soil particle composition by compositional Kriging were the least and the precision was the highest. For clay,the relative improvement of accuracy to the reference method ordinary Kriging based on ALR was close to 17%. On the whole,fitting effect by compositional Kriging was better than that by other methods. The ranges were wider for compositional Kriging,and its prediction results could better reflect the relations of different soil particle composition with elevation,soil parent material and water area distribution.