Document Type: Regular Article
Ph.D. Student, Faculty of Civil Engineering, Semnan University, Semnan, Iran
Faculty of Civil Engineering, Semnan University, Semnan, Iran
Faculty of Architecture and Urban Engineering, Semnan University, Semnan, Iran
Faculty of Science, Agronomy Department, Hydraulics Division University, 20 Août 1955, Route El Hadaik, BP 26, Skikda, Algeria
Precipitation forecasting is of great importance in various aspects of catchment management, drought, and flood warning. Precipitation is regarded as one of the important components of the water cycle and plays a crucial role in measuring the climatic characteristics of each region. The present study aims to forecast monthly precipitation in Semnan city by using artificial neural networks (ANN). For this purpose, we used the minimum and maximum temperature data, mean relative humidity, wind speed, sunshine hours, and monthly precipitation during a statistical period of 18 years (2000-2018). Moreover, an artificial neural network was used as a nonlinear method to simulate precipitation. In this research, all data were normalized due to the different units of inputs and outputs in the forecasting model. Further, seven different scenarios were considered as input for the ANN model. Totally, 70% of the data were used for training while the other 30% were used for testing. The model was evaluated with appropriate statistics such as coefficient of determination (R2), root mean square error (RMSE) and mean absolute error (MAE). Scenario 6, which included the inputs of minimum and maximum temperature, mean relative humidity, wind speed, and pressure, provided the best performance compared to other scenarios. The values of R^2, RMSE, and MAE for the superior scenario were 0.8597, 4.0257, and 2.3261, respectively.