针对传统的克里金法仅以欧式距离测度描述空间结构相关性的缺点,基于最小方差准则,提出了综合曼氏、欧式和切氏三种距离测度的多测度加权克里金方法,同时给出了一种简便的替代最小方差准则的改进定权准则。利用小湾水电站2号山梁高边坡监测点不同季节的变形速率数据进行实例计算,结果表明多测度加权克里金法提高了插值精度与可靠性,且改进定权准则的估计值接近理论最优值。应用所提出方法得到的2号山梁高边坡2004年春夏秋冬多时相变形速率场,有效地揭示了高边坡自下而上牵引式变形逐步得到控制并趋于平稳的演化特征,为边坡变形稳定性评估与防治提供依据。
The defect of traditional Kriging method which only uses Eucidean distance measure to describe the spatial structure correlation is pointed out. On the basis of minimum variance criterion, a weighted kriging method using multi-distance measures, including Manhattan distance measure, Eucidean distance measure and Chebyshev distance measure, is proposed. Moreover, a modified weighting criterion is developed to substitute the criterion based on minimum variance. Since every distance measure has specific ability of detecting spatial correlation the multi-distance Kriging method is helpful in improving the interpolation precision and reliability of the measurement. The proposed method is applied to analyze the deformaion monitoring data of a steep slope in the Xiaowan Hydro Project in 2004. It is found that the measurement precision and reliability are significantly elevated, and the estimation based on modified weighting criterion is very close to the theoretical optimal value. The seasonal deformation velocity fields of this slope are obtained and the deformation mechanism is revealed.