Modeling of Mean Sea Level of Bay of Bengal : A Comparison between ARIMA and Artificial Neural Network

Abstract

Sea level change is one of the effects of the recent trend of climate change. A further rise in sea level is a threat to the existence of many people in Bangladesh. If the sea level rises by 45 centimetres, scientists expect a permanent loss of up to 15,600 square kilometres of the land of Bangladesh. Moreover, the sea level rising indirectly affects the salt content of soils. In view of Bangladesh’s already problematic food situation, the expected decrease of rice production as well as several hundred tons of vegetables, lentils, onions, and other crops could be disastrous. Last but not least, valuable ecosystems would be lost. The Sundarbans, huge mangroves swamps along the coasts that are part of the United Nations world natural heritage, will be especially affected. They are the last retreat of the Bengal tiger. Therefore, it is necessary to know the behaviours of the sea level rise in Bangladesh in prior. Thus, the main intention of this paper is to select the most appropriate model for forecasting the mean sea level of the Bay of Bengal. Here, this paper forecast the future mean sea level using both ARIMA and Artificial Neural Network (ANN) model and compares the forecasting accuracy of these two models. The results of the model selection criteria considered in this study advocate that the ANN model provides better results than the ARIMA model. So, this study recommends using the ANN model instead of the ARIMA model for forecasting the mean sea level of the Bay of Bengal.