基于信息扩散的海洋稀疏资料插值算法-概率模型

Probability model-an interpolation technique for oceanic sparse data based on information diffusion

针对海洋观测资料零散、稀少等问题,提出了基于信息扩散思想的插值方法-信息扩散插值算法。该方法基于模糊映射思想,通过对稀疏数据点信息进行模糊扩散和插值映射,进而实现有限数据点信息向其邻近区域点的概率插值。针对正态扩散函数在表现广义非对称结构数据资料时存在的局限性,发展了一类非均匀信息扩散插值算法-概率模型,进行了海温资料插值试验和对比分析。

Aiming at the difficulties of scattered and sparse observational data in ocean science,a new in- terpolation technique based on information diffusion idea is presented in this paper. By fuzzy mapping route, the sparse data samples are diffused and mapped into corresponding fuzzy sets in the form of prob- ability in the probability interpolation model. For avoiding the shortcoming of data asymmetrical struc- ture in normal diffusion function, a kind of unsymmetrical information diffusion function is developed, and the corresponding unsymmetrical information diffusion algorithm-probability model is established. By making the interpolation experiments and contrast for sea surface temperature data, the rationality and validity of the ellipse-model are validated.