本文针对海洋数据多模态、超高维、海量等特性,利用空间自相关性,提出了一种基于空间相关性的海洋环境监测数据优化抽样方法。以某海域水深数据为例,从样本量和方差两个角度分析了该方法的优劣性,同时将样本点反演趋势面和实际趋势面进行比较,在减少数据量的同时,保证了其抽样精度,验证了该方法的可行性及有效性。
According to the characteristics of marine data such as multimodal, high-dimension and magnanimity, this paper proposes an optimized sampling method for marine environmental monitoring data based on spatial correlation. With the water depth data of a sea as an example, the method is arguably analyzed by two perspectives, which are sam- ple size and variance. At the same time, to verify the feasibility and validity of this method, this paper compares the in- verting trend surface and actual trend surface, reduces the number of sampling data while ensuring the accuracy of sampling.