Abstarct: The purpose of this thesis is to investigate the discharge and sediment of the Maroon - Garahi River and their prediction by artificial neural network. The length of the river is 438 kilometers and its altitude is 2200 meters. The Jarahi River originates from the mountain Rah barik, White, and Gol Gilak, 53 kilometers northwest of Yasuj, located in the village of Dohran, Boyerahmad, and flows through the northwest river, Ludab River. After passing the mountain of Dezkouh on the western slope with the Ghalat River, the river is diverted to the south and goes to the Mansouri village of Behbahan city and passing through the north of Behbahan plain with several large and small lobes, including the water of the crust. This river is blended on the eastern slopes of Mount Aralon with the river Baba Ahmad and passes through the south of Ramhormoz city with the riverside abrazazk and Aala and beside the city of Ramshir in the name of the rivers of Jarahi to the Jarahi section of the port city of Mahshahr and is 17 kilometers north of the port of Mahshahr It crosses and goes northwest to the Persian Gulf. The basin of the Rivers Jarahi and Zohreh rivers is made up of two independent and separate sub-basins adjacent to each other, which are seen in the form of a single unit due to similarities between the two sub-basins. To study the Maroon River discharge, we use the hydrometric information of the 10 selected stations along the river. Two Man-Kendall tests and Sen estimators, which are one of the most common non-parametric methods, were used for analyzing the flow rate at the annual scale. Ten synoptic stations located in the area of Zohreh - Jarahi that had statistics during the years 1345 to 1392 were selected and the two tests were applied to their data, and the results of these two methods were compared together. The results showed that the efficiency of the two methods is similar in many cases in the analysis of sedimentation and sedimentation trend, indicating a significant decrease in the flow rate at these stations. In this study, an artificial neural network was used to predict and estimate the Maroon River discharge. The data used included rainfall, rainfall and annual and monthly temperatures, and monthly rainfall data from the station that were collected from the Water Resources Management Agency of Iran. The qnet 2000 software has been used to model the discharge using artificial neural network. For modeling using artificial neural network, three modes were considered: A. Investigating the effect of stimulus functions on artificial neural network learning B. To study the effect of intermediate laxyers on learning the artificial neural network C. Investigating the effect of the number of input data on artificial neural network learning The results of this study showed that among artificial neural network models, the neural network model with Goss actuator function and the number of intermediate laxyers of 3 with the highest correlation coefficient in the training and verification phases were 0.617 and 0.633 as the model. The highest was estimated in the forecast and estimation of the Maroon River discharge.