本文以内蒙古河套灌区内一典型实验区的土壤水盐空间变异性为案例,进行了普通克立格、BP神经网络与稳健克立格估计对比分析。发现对于土壤水分预测,BP法可在一定程度上克服克立格法的平滑效应;在盐分预测中,稳健克立格与BP法对特异值影响估值问题尚未很好解决。BP神经网络技术对原始数据没有参数及严格地采样分布要求,简单而实用。 In this paper, the spatial variability of soil water and salt in a typical experimental area in Hetao Irrigation District in Inner Mongolia was taken as a case study to compare and analyze the Kriging and BP neural networks. It is found that for the prediction of soil moisture, the BP method can overcome the smoothing effect of Kriging to a certain extent. In the salt prediction, the problem of the value of the robust Kriging and BP method in estimating the specific value has not been solved well. BP neural network technology has no parameters and strict sampling and distribution requirements on the original data, which is simple and practical.