Mohammad Valipour - Irrigation and Drainage Engineering College, College of Abureihan, University of Tehran, Pakdasht

More accurate forecasting of monthly rainfall is significantly important in flood forecasting for control of it, flood occurrence alarm, water resources management, and construction of shelter. In this paper, ability of time series models in forecasting the rainfall according to the climate conditions was estimated. For this purpose, rainfall data of four different climates in Iran was selected. Using this data amounts of rainfall were forecasted by time series models for one next year. In first method number of observation data for model calibration were 60, and then this increase to the 120 and 588 data. Time series models have found a widespread application in many practical sciences. In addition, rainfall forecasting has been done by some methods such as time series models, satellite imagery, and artificial neural networks. However, according to the deficit data in most rainfall forecasting, number of required observation data always been questioned. Therefore, this paper attempts to present number of required observation data according to the climate conditions. By comparing R2 of the models, it was determined that time series models are better appropriate to rainfall forecasting in semi-arid climate. Numbers of required observation data for forecasting of one next year were 60 rainfall data in semi-arid climate

Flood management, Length of recorded data, Rainfall forecasting