Forecasting Temperature in the Coastal Area of Bay of Bengal-An Application of Box-Jenkins Seasonal ARIMA Model

Abstract

Temperature is one of the most vital elements of the climate system and forecasting of the temperature helps the stakeholders those who are depends on it directly or indirectly to prepare  in advance. Country like Bangladesh whose economy mostly geared up by the agricultural product need to know the upcoming pattern of temperature beforehand to take necessary actions. This study has been conducted on the monthly maximum and minimum temperature data (1949-2012) from the second largest and port city of Bangladesh, Chittagong. Non-parametric Mann-Kendall test has been adopted to identify the trend of the series and found that though the maximum temperature is increasing but not significantly but the minimum temperature is increasing significantly. The anomaly plot is just portrait the ups and downs of minimum and maximum temperature and found minimum temperature is increasing from last two decades whereas the maximum temperature has abrupt changes with increase and decrease. The linear trend analysis shows the climate line for maximum and minimum temperature are 35.67 and 10.23 degree Celsius respectively and the rate for significant increase of minimum temperature is 0.07 degree Celsius. The forecasting Seasonal ARIMA model for maximum temperature is SARIMA (1, 1, 1) (2, 0, 0) [12] and for minimum temperature is SARIMA (1, 1, 1) (1, 0, 1) [12]. The resulted outcomes indicate the increasing pattern of temperature in upcoming days in this area of Bangladesh.

Keywords: temperature, Seasonal ARIMA, forecasting, climate, Chittagong