| Summary: | 碩士 === 國立臺灣海洋大學 === 河海工程學系 === 104 === By applying the Oceanic Niño Index calculation, the Niño 3.4 region is selected as study area. We analyze the sea surface temperature (SST), and discuss the frequency and intensity of El Niño over 1901~2013. Moreover, a new technique for data synthesis based on Empirical Mode Decomposition (EMD) is introduced. We compare the synthetic data with the projection of the Geophysical Fluid Dynamics Laboratory (GFDL) from the fifth assessment report of Intergovernmental Panel on Climate Change in 2014.
The EMD decomposes time series into two components: Intrinsic Mode Functions (IMF) and residue (also known as trend). For the purpose of data synthesis, examining the interval and amplitude for each corresponding IMFs in different time is necessary. By using the Kolmogorov-Smirnov test, the similarity of interval and amplitude for corresponding IMFs is acceptable. This result implies we could synthesize a new series by permutation within the IMFs. Next, we analyze the mean distribution of synthetic SST, and the mean of synthetic data cannot be rejected through null hypothesis in most cases. Finally, we compare the synthetic data that is based on EMD with the projection of GFDL, and find the characteristics of El Niño frequency that were derived from these two models are similar.
We conclude that the EMD technique would be applicable for data synthesis of 5-yr or 10-yr-long time series in this study. However, more observation would be needed for longer data projection.
|
|---|
