用Cao方法探讨ENSO可预报性的年代际变化

Research on decadal variability of ENSO predictability with Cao method

为研究ENSO可预报性的年代际变化与观测资料的关系,本文试图用非线性时间序列分析的方法来探讨观测序列不同年代的确定性程度。为了解决度量时间序列确定性程度的难题,本文在Cao方法的基础上对E2进行了简单统计分析,结果表明:Cao方法可用来粗略判断时间序列的确定性和随机性程度,但不够精确;时间序列数据的纯净性比时间序列的长度更重要,Cao方法抗干扰的鲁棒性并不强;ENSO可预报性的年代际变化并不单纯与观测资料相关,其原因仍需深入分析。

In order to research the relation between decadal variability of ENSO predictability and observation data, a nonlinear time series analysis method is used to probe the certainty extent in different decades. For the sake of the conundrum of measuring the certainty extent in time series, the estimation measure E2 based on Cao method is analysed with some statistics. The results show that Cao method can be used to roughly estimate the extent of certainty and randomicity but it is not precise, because the robustness of Cao method is not strong. The purity of time series data is more important than their length. The decadal change of ENSO predictability is not simply caused by observation data and it is still necessary to thoroughly analyse the reasons.