A seasonal ARIMA model for monthly rainfall sequence
by
Urmil Verma
Dept. of Soils CCS H.A.U. Hisar, India
Coauthors: Manoj Yadav, R.C,Hasija

The univariate Box-Jenkins (UBJ) model often referred to as ARIMA (Auto-Regressive Integrated Moving Average) model is one of the important and useful tools for time series modeling. ARIMA method is an extrapolation method for forecasting and like, any other such method, it requires only the historical time series data on the variable under forecasting. This method is very convenient for both analysing the data sequence and forecasting. Monthly rainfall sequence from 1945 to 2001 at Sirsa (Haryana) is modeled using a seasonal ARIMA series. The various stages of model building are presented in a simple algorithmic form. The model parameters are estimated using Marquardt algorithm for non-linear optimization. Proper care has been taken for non stationarity and seasonality present in the data while developing these models. Adequacy of the order of the fitted model has been tested using Ljung-Box test criteria followed by residual analysis. The most appropriate model has been used to compare the actual and estimated monthly rainfall for the year 2000-2001.

Date received: January 30, 2002