摘要
提出了一种基于半参数回归的数字水深模型(DDM)内插方法。将原始水深数据误差和内插函数的逼近误差分别视作数字水深模型的偶然误差和系统误差部分,构建DDM内插的半参数回归模型;基于距离法与L曲线法,分别确定正规化矩阵R与平滑因子α;采用按方位取点的加权平均法,估计内插函数在任意节点处的逼近误差。实验结果表明:所提方法提高了DDM的构建质量;海底地形复杂程度越大,所提方法对DDM构建质量的提高程度越大。
This paper proposed a method for interpolating DDM based on semiparametric regression.Data error of raw depth and approximation error of interpolation function are respectively regarded as one part of random error and another part of systematic error in DDM,and a semiparametric regression model for interpolating DDM is constructed;regularizing matrix and smoothing parameter are respectively created by distance method and L-curve method;approximation error of interpolation function is estimated on any node using a method for weighted meaning with points selected by azimuth.Experimental results demonstrate that the proposed method has improved the quality of DDM;besides,the more complex the seabed terrain is,the higher the proposed method has improved the quality of DDM.
出处
《海洋测绘》
2011年第5期34-37,共4页
Hydrographic Surveying and Charting
基金
国家863计划项目(2009AA12Z202)
国家自然科学基金项目(40801189)
关键词
数字水深模型
半参数回归
逼近误差
移动曲面法
digital depth model
semiparametric regression
approximation error
moving surface method
