基于SSA和ARMA的海平面变化短期预测

Short-Term Prediction of Sea Level Change Based on SSA and ARMA

海平面预测对评估海面变化的影响具有重要意义。本文提出联合奇异谱分析(singular spectrum analysis,SSA)和自回归移动平均(autoregressive moving average,ARMA)模型的海平面变化预测方法,并利用香港鲗鱼涌验潮站月平均潮位时间序列进行预测试验。结果表明,本文提出的组合法能有效提高海平面变化的预测精度。相比于利用原始序列的ARMA模型预测值,联合SSA和ARMA的预测结果精度提升超过20%。

Sea level prediction is of great significance for assessing the impact of sea level change.In this paper,a method for sea level prediction was proposed combining singular spectrum analysis(SSA)and autoregressive moving average(ARMA)model.An experiment was performed to validate the method using monthly tide gauge data at Quarry Bay in Hongkong.The result shows that the combination method can significantly improve the accuracy of the sea level predictions.Compared to the predicted values using ARMA method from original time series,the accuracy of the predictions by combining SSA and ARMA was improved by more than 20%.