Forecasting of seasonal flow of the Ganges River in Bangladesh with Sarima model

Two multiplicative Seasonal Autoregressive Integrated Moving Average (SARIMA) models were developed for forecasting flows of the Ganges river in Bangladesh. For quarterly forecasting, the model that was found suitable is SARIMA (1,0,0) x (9, 1, 1)4 with some constrained seasonal autoregressive parameters ( O3, O4, O6 to O8 =0). For half-yearly forecasting, SARIMA (0,0,0) x (0,1,4)2 was found to be appropriate with two seasonal moving average parameters (O2 and O3) constrained to zero. Both the models were statistically adequate and may have a practical value for water resources and agricultural management in the country.