SEA LEVEL VARIATION ANALYSIS WITH RBF NEURAL NETWORK BASED EMD METHOD
GU Xiao-Li,LI Pei-Liang,TAN Hai-Tao,ZHANG Ting-Ting,LI Lei,WANG Xue-Zhu,YU Yi-Fa
DOI: https://doi.org/10.3321/j.issn:0029-814X.2009.05.002
2009-01-01
Abstract:A new empirical mode decomposition(EMD) method was developed based on radial basic function(RBF) neural network and applied to analyze sea level variability.A temporal signal series is extended with the network before a new round of EMD starts.The result of analysis with the new method agrees better with the result of 5 year running mean than that using traditional EMD.The new method decomposes a series of sea level temporal signal into a set of intrinsic mode functions representing the periodic components of signal and an overall adaptive trend.It was shown that the annual variability was very remarkable in 47 stations along coast of China,and the semi-annual variability was also clear but less remarkable,with the amplitude decreased north to south.In the last 40 years,the sea level fluctuation of the area has been increasing in a non-linear pattern generally but varied geographically.The maximum fluctuation occurs in estuary zones.The results also reveal that the longer the temporal signal series is,the more IMF can be obtained from the data using the new method and a more exact overall adaptive trend can be extracted from signal.A shorter time series could achieve a complete decomposition easily but some of the IMF components may be hidden in overall adaptive trend,resulting in obviously nonlinear sea level variability,which may not be actually true.