Assessment of artificial neural network performance and exponential regression in prediction of effective rainfall

According to the current water crisis and spend more than 94 percent of water in agriculture the mechanized irrigation systems, and revised the actual plant water estimation are needed it is facilitate to predict rainfall in the growing season. In the design of irrigation systems should be noted that the total rainfall occurred was not available for plant and part of the rainfall runoff and part of it penetrate to soil and only part of it that is called effective rainfall is able to disappear plant water stress and influence plant growing. In this study, the results of regression model exponentially and based on field observations were compared with artificial neural networks (ANN). Its result showed more accuracy of mathematical and natural patterns (ANN) than pure mathematical patterns (regression). The use of neural networks in prediction of effective rainfall leads to decrease the cost of irrigation systems and water consumption. It also leads to reduce from unprofessional comments and consequently the imposition of water stress on the plant and the product.