集成奇异谱分析和自回归滑动平均预测日本近海海平面变化

本文从日本沿岸选取了28个验潮站及联测的GPS站，利用奇异谱分析（Singular Spectrum Analysis，SSA）和SSA+自回归滑动平均（Auto Regression Moving Average，ARMA）方法预测了2014—2018年的近海海平面变化和地壳垂直变化.并用同时段的验潮及GPS的实际测量值进行验证，结果显示，SSA+ARMA预测的相对海平面精度为0.0357~0.0607 m，地壳垂直运动的精度为0.0049~0.0077 m，绝对海平面的精度为0.0433~0.0683 m，且三者SSA+ARMA的预测结果均优于只用SSA预测的结果.在此基础上本文利用SSA+ARMA预测了日本沿岸2019—2023年的近海绝对海平面变化，结果显示，2019—2023年的平均海面高较往年（2014—2018）升高0.0353 m，2003—2023年绝对海平面的变化率为0.0039 m·a^-1，预测结果较为理想.

Prediction of sea level change in Japanese coast using singular spectrum analysis and auto regression moving average

The change of sea level, especially in offshore areas, is a highly concerned topic that attracts global attention. The prediction of sea level changes in offshore areas is particularly important. In this paper, 28 tide gauge stations and collocated GPS stations were selected along the coast of Japan. The sea level changes and vertical crustal deformation from 2014 to 2018 were predicted by Singular Spectrum Analysis (SSA) and SSA+Auto Regression Moving Average (ARMA), respectively. The prediction results are validated by the actual measurements of tide gauge and GPS at the same time period. The results indicate that accuracies of relative sea level, vertical crustal motion and absolute sea level predicted by SSA+ARMA are 0.0357~0.0607 m, 0.0049~0.0077 m and 0.0433~0.0683 m, respectively, which are better than those predicted by SSA only. Then we applied the SSA+ARMA to predict the absolute sea level change in the coast of Japan from 2019 to 2023. The average sea level height in 2019—2023 is 0.0353 m which is higher than that in previous years (2014—2018), and the rate of the absolute sea level change from 2003 to 2023 is 0.0039 m·a^-1, indicating that the prediction result is reliable.