Estimation and Reconstruction of Annual Maximum 24-H Rainfall Data Using Combination of Genetic Algorithm and Artificial Neural Networks Models (Case Study: Chaharmahal va Bakhtiyari Province)

Annual maximum 24-h rainfall is the meteorological parameters with the more stochastic nature of rainfall in comparison with other related rainfall data، including monthly and annual precipitation data. Considering to unavailable Intensity-Duration-Frequency (IDF) data and more availability of Annual maximum 24-h rainfall data، a common method of estimating short time rainfall in watershed management operation and studies is based on Annual maximum 24-h rainfall data. Sometimes mentioned data is incomplete and use of them causes an error in the results. This research carried out in order to evaluate the performance of artificial neural network with genetic algorithm (GA-ANN) for reconstruction of annual maximum 24-h rainfall data in Chaharmahal va Bakhtiyari province. To evaluate the model RMSE، P%، and R2 were used as statistical indices. The GA-ANN was compared with simple artificial neural networks. The RMSE indices between observed and predicted values by ANN model were 38. 0، 25. 9، 11. 8 and 11. 4 (mm) for very-humid، semi-humid، Mediterranean and semi-arid climate zones، respectively. Considering the results of GA-ANN method، the RMSE were 19. 2، 14. 3، 10. 8 and 6. 4 (mm)، respectively. The results of reconstructed data show that GA-ANN method has significant preference related to ANN method in all the four climates in the studied region.