论文部分内容阅读
针对不同数据源来进行黄土高原古地形恢复重建对其土壤侵蚀研究有重要意义。本文基于ASTER GDEM数据、1∶450万黄河流域黄土厚度分布图等综合提取出黄土高原古地形高程点数据,采用径向基函数(Radial Basis Functions,RBF)进行空间插值,用交叉验证法和验证数据集方法评估插值精度。结果表明:(1)RBF插值过程中张力样条函数、0.01光滑因子、四方向搜索、搜索点数为24,为“最优”插值参数;(2)通过高程预测值与原始高程值的误差分析,绝对值不大于20m的误差点占总误差的95.18%,误差较大的点所占比例非常小。本文选择“最优”插值参数插值后预测高程值较接近原始高程值,说明在RBF插值中用此参数设定进行DEM构建能较好地模拟复原黄土高原古地形,可以为研究区现今水土保持提供科学的参考依据。
For different data sources to restore the paleotopography of the Loess Plateau for its research on soil erosion is of great significance. Based on the ASTER GDEM data and the 1: 4.5 million loess thickness distribution in the Yellow River Basin, the paleotopography elevation data of the Loess Plateau were extracted synthetically. The radial basis function (Radial Basis Functions, RBF) was used for spatial interpolation. The cross validation and validation Data set method to evaluate interpolation accuracy. The results show that: (1) Tension spline function, 0.01 smoothing factor, searching in four directions, the searching points are 24, which is the “optimal” interpolation parameter in RBF interpolation; (2) Error analysis, the absolute value of not more than 20m error points 95.18% of the total error, the error point of a larger proportion of very small. In this paper, we choose “optimal” interpolation parameter to interpolate the predicted elevation value closer to the original elevation value, indicating that using this parameter to set up DEM in RBF interpolation can better simulate the paleotopography of the Loess Plateau, Soil and water conservation to provide a scientific reference.