面向时空非均匀Argo数据集的k邻域搜索算法

Algorithm for Finding K-nearest Neighbors of Non-uniform Spatio-temporal Argo Data

k邻域搜索作为海洋温盐场重构关键的前序步骤,搜索速度和结果是否具有代表性直接影响研究工作的效率和科学性。当前的k邻域搜索算法及其改进方法主要针对空间数据集。面向Argo时空非均匀数据集提出了一种k邻域快速搜索算法,首先基于立方体栅算法向时空维扩展,利用时空子分块对海量、时空非均匀的采样点进行分配;在此基础上采用时空补偿的策略对算法进行优化。结果表明,该方法有效提升了Argo浮标的邻域搜索效率并且改善了搜索结果的分布情况。

The algorithm for finding k-nearest neighbors as key steps before the reconstruction of temperature and salinity fields, the speed and results of the algorithm for finding k-nearest neighbors determines the efficiency and accuracy in scientific research. The current algorithm for finding k-nearest neighbors and its improved methods are mainly aimed at spatial datasets. A new algorithm for non-uniform spatio-temporal Argo profile data is presented,which is based on 3D cell gridso At first,mass and non-uniform sampling points are assigned based on spatio-temporal subblock. Then, algorithm is optimized through spatio-temporal compensation strategy. The experimental results show that the method increase in spatio-temporal data search efficiency and improve the distribution of the search results.