PREDICTION EXPERIMENT OF SEA-LEVEL CHANGE BASED ON THE SSA AND AR MODEL
Sun Jian · Haiyang huzhao tongbao · 2007
Based on the monthly average tidal records of Wusong tidal gauge station from 1955 to 2001,prediction experiment is made use of singular spectrum analysis(SSA) and auto-regressive mode(AR) in this paper.SSA is used to isolate the T-PCs(prinieipal components) corresponding to sea-level change from the remaining variability and noise.Since the T-PCs are the filtered versions of the raw data,their behavior is more regular than that of the original signal ad more predictable accordingly.In practice advantage of each significant component can be made and reconstruct them.Using AR model to predict each reconstruction components,the summation of all the predicted reconstruction components is the prediction results.Analyzing with the data of 1955-1996 to predict the sea-level changes of 1997-2001,comparing prediction results with the original data,it is found that these two series are fairly comparable.It shows that integration use of these two methods provide an efficient way to predict sea-level chenges.