Generating Artificial Water Quality Data for No-Trend Parameters in Reservoirs (Chahnimeh No.1 in Sistan)

Detection and monitoring water resources is one of the principal steps in quality management of water resources. This is more critical in Sistan and Baluchestan province located in a hot and dry area with large deficiency in exploitable water resources. The aim of this study was to predict no-trend quality parameters in Chahnime No.1 using neural network and comparing it with Markov chain method. In the present study some parameters such as DO, temperature, Phytoplankton, Zooplankton, Ammonia, and Phosphorus have been considered. The mean error percentages in neural network method for these parameters were 5.5, 7.7, 12.8, 5.6, 52.4, and 4.1, respectively. In comparison, the mean error percentages of Markov chain scheme were respectively 11.2, 8.9, 5.3, 12.9, 33.9, and 8.4. The results showed that neural network method provided better results compared to Markov chain method.