论文部分内容阅读
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 as a kind of compositional data (compositional data), the spatial interpolation must meet the non-negative, fixed and, minimum error and unbiased estimation of four conditions. Spatial prediction, soil mean root mean square error (RMSE), and normalized Kriging variance of soil texture were performed using the compositional Kriging and ordinary kriging based on log-log transformation (mean squared deviation ratio, MSDR) were used to measure the prediction accuracy of different methods and model fitting effect. The results show that the ordinary kriging method and the composition kriging method can ensure that the interpolation result satisfies the four conditions of component data interpolation. The soil particle composition predicted by the composition kriging method has the smallest RMSE, the prediction accuracy The highest, the clay particle RMSE value relative to the asymmetric logarithm than the ordinary Kriging conversion almost 17%; component Kriging’s variation function fitting effect is generally better than the other two prediction methods, the prediction results pole The difference is wider, which better reflects the relationship between the composition and elevation of the soil texture and the distribution of the parent material and the water area.