Stochastic modelling of monthly rainfall for Jodhpur

Trend component in historical series of monthly rainfall data of Jodhpur was found non-significant. Fourier series with four harmonics represented periodic component. The dependent stochastic components was represented by an autoregressive model of order one. The independent stochastic component was represented by Normal distribution. After identification of different components, the model was obtained. The mean and standard deviation of generated series were 23.37 mm and 43.59 respectively. These values are close to mean and standard deviation of 29.98 mm and 48.95 respectively of historical series. Low values of mean forecast error, mean relative error and integral square error for regeneration and forecast suggest adaptability of the model for real time forecasting of monthly rainfall.