Forecasting of sea level fluctuations is a suitable tool for comprehensive management of the sea and the protection of coastal areas. On the other hand, application of time series analysis for forecasting purposes has been evaluated to be very appropriate. Therefore, two time series consisting monthly measured sea level data were used in the present research. The data have been recorded at two stations of Anzali and Noushahr in the southern part of the Caspian sea for time lengths of 40 and 14 years, respectively. The nonparametric Mann-Kendall test was employed to determine if measurements exhibit an increasing or decreasing trend. In the next step, different methods of forecasting and modeling of time series including Auto Regressive Integrated Moving Average and Multiplicative ARIMA method were fitted to the data. Then, Akaike Information Criterion was applied to assess the ability and accuracy of fitted methods in forecasting of sea level in future months and to determine the best time series model. The results of forecasting in the final models reveal that the performance of the Multiplicative ARIMA method based on time series analysis, to estimate and simulate the stochastic behavior of the Caspian sea level is acceptable. Meanwhile, the length of the forecast period in the models has increased significantly in comparison with previous researches.

Keywords: Sea Level, Time Series, Caspian Sea, ARIMA model, Mann-Kendall Test.

Full-Text [PDF 1106 kb]   (3988 Downloads)