Develop Computational Intelligence Model for Runoff Prediction in Remote Basins

The capability of Artificial Intelligent and GIS plays a significant part for the management of ungauged basins by predicting the runoff which represents one of the hydrological variables. This research has been carried out on Al- Murr basin, which resides in Nineveh province, northern of Iraq, where a computational intelligence model based on two artificial neural networks as Non-linear Autoregressive Network with Exogenous inputs (NARX) and Radial Basis Function (RBF) have been developed for the prediction of the annual runoffs. The procedures of the model calibration and validation have been tested according to a number of the error criteria, in which obtained accuracy of performance had reached 84.61%, which resulted in the model giving quite close predicted results with quite small statistic errors for the years of the predicted period from 2018 to 2049 after that, the model starts collapsing and give irrational results. Soil Conservation Service (SCS-CN) approach has been utilized for the determination of the annual depth of the runoff (predicted as well as actual values) that have been utilized for analysis and comparisons. GIS environment has been initiated by primary available and calculated data, in which results of spatial distribution for actual and predicted runoff showed that the basin will be suffering from the shortage of water amounts in predicted period where general average runoff will be reduced.