Forecasting Monthly Maximum Temperatures in Kerbala Using Seasonal ARIMA Models.

Abstract

Weather forecasting is an important issue in meteorology and scientific research.In this research, the Seasonal Auto Regressive.Integrated Moving Average.(ARIMA) model which is based on Box-Jenkins method was adopted to build the forecasting model. The max. Monthly temperature data for Kerbala city for the period (Jan.1980 to Dec.2016) was employed. The autocorrelation and partial autocorrelation functions for time series data from years 1980 to 2015 were used to identify the most appropriate orders of the ARIMA models. The validation test of these models were performed using the monthly max. Temperature of the year 2016. To calculate the model's accuracy and compare among them, statistical criteria such as MAE, RMSE, MAPE, and R2 were used. The model (2, 1, 2) × (1, 1, 1)12 gave the most accurate results and used to forecast the monthly max. Temperature for the period (2017 to 2021) for study region.