www.seipub.org/ijepr International Journal of Engineering Practical Research (IJEPR) Volume 2 Issue 1, February 2013 16 Forecasting Seasonal and Annual Rainfall Based on Nonlinear Modeling with Gamma Test in North of Iran Hossein Ansari *1 Associate Professor of Water Engineering Department, Ferdowsi University of Mashhad (FUM) Mashhad, Iran *1 Ansary@um.ac.ir Abstract Rainfall plays a key role in hydrological application and agriculture in wet climatic regions. Lack of short‐run rainfall forecasting is considered as a significant impediment for scheduling the root zone moisture preparation. Although many mathematical techniques are available for use, basic concerns remain unsolved such as simplicity, high accuracy, real time use in many stations of a region, and the low availability of inputs. In this study, a nonlinear modeling with Gamma Test (GT) has been presented to solve some of the mentioned problems. Forecasting seasonal and annual rainfall with the variables of four years lagged rainfall data and geographical longitude, latitude and elevation has been performed in the North of Iran during 1956‐2005. The results show that Gamma Test is an effective tool for rainfall forecasting. The applied nonlinear modeling techniques are Local Linear Regression (LRR), Dynamic Local Linear Regression (DLLR), and three separate Artificial Neural Networks (ANN) using Back Propagation Two Layer, Broyden‐Fletcher‐Goldfan‐Shanno (BFGS), and the Conjugate Gradient training Algorithms. The training and testing data are partitioned by random selection from the original data set. Not only does the Gamma Test yield the best input combination, but also the model’s good performance leads to the best achievable result. The study results demonstrate that developed models based on Local Linear Regression (LRR) technique have better performance comparing with ANN models. Also, developed ANN model based on Back Propagation Two Layer training Algorithm is preferred because of its better performance compared with the other ANN models. Keywords Gamma Test; Artificial Neural Network; Local Linear Regressio; Rainfall Introduction Predicting the hydrological variables like rainfall, flood stream, and runoff flow as stochastic or probabilistic events, is one of the principal subjects in water resource planning. The hydrological variables are usually measuring across the time. Therefore, time series analysis of their occurrences in discrete periods is urgent for monitoring and simulating the hydrological behavior of a region. Rainfall Forecasting, as the most affecting factor on hydrological cycle, is vital in water resources management, irrigation scheduling, and agricultural management especially in humid climates (Mimikou, 1983; Hamlin et al., 1987). In wet and semi‐wet climates, irrigation isn’t common and farmers use rainfall water for supplying crop water requirements. When rainfall isn’t enough the supplemental irrigation will be applied. Therefore forecasting, modeling and monitoring of rainfall are of a high importance in agricultural actions (Geng et al., 1986; Hoogenboom, 2000; Sentelhas et al., 2001). Notably, while weather forecasting deals with daily development of the weather up to several days ahead, seasonal forecasting is concerned with the average weather condition on timescale of a month to about a year ahead. Seasonal forecasts are also known as long‐ run weather forecasts or short‐run climate forecasts (Chang et al., 2003). Because seasonal forecasts give information of several months ahead, they can be used by government, business, agriculture, and industry to increase productivity, maximize economic benefits and minimize losses. Specific examples of the applications of seasonal forecasts are presented in ECMWF (1999). The seasonal forecasts based on slow variation in the earth‘s boundary conditions (i.e. sea surface temperature, soil wetness, and snow cover) can influence global atmospheric circulation and rainfall, too (Fu et al., 2007; Rajeevan, 2007; Gonzalez et al., 2009; Ousmane et al., 2011). A detailed discussion of the differences between weather and seasonal forecasting can be found in WMO (WMO, 2002). In last decades, researchers developed many empirical methods in the form of statistical or analogue models