The performance of Artificial Neural Network in prediction and analysis of hydrological processes (Case study: Water shortage in Nazloo-chai watershed, West Azerbaijan province)

Precipitation is one of the hydrological processes that play an important role in controlling water resources management. Shortage of rain causes some problems such as lack of drinking water. Due to the importance of the issue of water shortage, using modern methods to predict hydrological processes will play an important role in planning and management of water resources. Therefore, in this study, monthly shortage of water in Nazloo-chai watershed was predicted using Artificial Neural Network (ANN) and improved wavelet-neural network (IWNN) models, for the past 39 years (1973-2012). Performance of these two models was evaluated using statistical indicators including correlation coefficient (R), determination coefficient (R2) and root mean square error (RMSE). According to the results of IWNN model, the obtained correlation coefficient was 0.960 and 0.945 for testing and training modes, respectively, and this model has greater ability for predicting the shortage of water in comparison with ANN. Accordingly, the amount of monthly water shortage in this watershed was predicted for 2013 to 2020. Results indicated that shortage of water still remains as in the past years. The average water shortage was estimated nearly as 2.95 million cubic meters (MCM) in the next 7 years, while, this parameter for the past 39 years was 4.04 MCM. Therefore, it is required to take necessary measures for future years, and with careful management plans for exploitation of water resources (agriculture, industry, urban, etc.), it is possible to reduce water shortage in the coming years.