Inverse distance weighting revisited
In this paper, our primary focus concerns with degrees of freedom available in a deterministic interpolation method so called Inverse Distance Weighting scheme. Our preliminary investigation shows that parameters such as measure of distance, power, and number of nearby stations are potential tuning parameters which can be optimized based on available scattered point-wise data. The main thesis of the current study is to argue that the selection of these parameters has to be driven by data-both the spatial coordinates and data attribute(s)—and cannot be chosen beforehand. For this purpose, a genetiC algorithm—based inverse distance weighting with moving neighborhood technique is proposed in this article. The searching capability of genetiC algorithm is exploited to search the feature space for an appropriate, either local or global, tuning parameters. Radius of circle/sphere and/or radii of standard or rotated ellipse/ellipsoid and distancedecay parameter are considered as the decision variables to be optimized by GA. The superiority of GA-based 1DW over traditional IDW is demonstrated through working with Wolfcamp-Aquifer piezometric head data. Assessment of numerical results showed that definition of parameter values based on both geographical coordinates and data attribute has far more impact on cross—validation statistics compared with one based on geographical coordinates alone. Optimization of local parameter values for an spherical support domain via GA was able to capture the dynamics of piezometric head in West Texas/New Mexico in an efficient manner.
genetic algorithm inverse distance weighting cross-validation wolfcamp aquifer anisotropic data
M.J.ABEDINI M.NASSERI
Dept.of Civil and Environmental Engineering,Shiraz University,Shiraz,Iran School of Civil Engineering,University of Tehran,Tehran,Iran
国际会议
The Four Conference of Asia Pacific Association of Hydrology and Water Resources(亚太地区水文水资源协会第4届科学大会)
北京
英文
385-394
2008-11-03（万方平台首次上网日期，不代表论文的发表时间）