Autoregressive modeling for the spectral analysis of oceanographic data
Autoregressive modeling for the spectral analysis of oceanographic data
复制标题
海洋数据光谱分析的自回归建模
DOI：
--
复制
发表时间：
1989
期刊：
影响因子：
--
通讯作者：
L. Jackson
中科院分区：
文献类型：
--
作者：
A. Gangopadhyay;P. Cornillon;L. Jackson
在过去十年中，对海洋学研究有用的数据集的数量和数量急剧增加。其中许多数据集由来自卫星和大规模海洋学实验的长时间或空间序列组成。然而，这些数据集在空间上往往是“空洞的”，在时间上是不规则的，而且总是有限的长度。因此，传统的傅里叶变换(FT)光谱分析方法往往不适用，或者在适用的情况下，它提供的结果有问题。通过与FT对不同海洋数据集的比较分析，我们讨论了用自回归(AR)模型进行有限长序列谱分析的可能性。应用表明，随着时间序列长度的缩短，AR方法的分辨率比FT方法的分辨率有所提高。对于这里考察的最长的数据集(98个点)，AR方法的表现仅略好于FT，但对于非常短的数据集(17个点)，AR方法的表现明显优于FT。将AR方法应用于复杂的时间序列，虽然是本稿的次要问题，但也进一步强调了这种方法的价值。
Over the last decade there has been a dramatic increase in the number and volume of data sets useful for oceanographic studies. Many of these data sets consist of long temporal or spatial series derived from satellites and large-scale oceanographic experiments. These data sets are, however, often “gappy” in space, irregular in time, and always of finite length. The conventional Fourier transform (FT) approach to the spectral analysis is thus often inapplicable, or where applicable, it provides questionable results. Here, through comparative analysis with the FT for different oceanographic data sets, we discuss the possibilities offered by autoregressive (AR) modeling to perform spectral analysis of gappy, finite-length series. The applications demonstrate that as the length of the time series becomes shorter, the resolving power of the AR approach as compared with that of the FT improves. For the longest data sets examined here, 98 points, the AR method performed only slightly better than the FT, but for the very short ones, 17 points, the AR method showed a dramatic improvement over the FT. The application of the AR method to a gappy time series, although a secondary concern of this manuscript, further underlines the value of this approach.