The Determination of sediment transport using mathematical models in rivers of Tigris basin

SUMMARY Tigris Basin has been playing an important role in culture and modernization since 4500 B.C. This basin, which is the second important water resource of the Southeastern Anatolian Project, has 37.3820 km2 drainage area in our country, and its annual average discharge is 14.710 million m. Erosion, along with its consequences, i.e, sediment transport and storage, is a phenomenon which has important effects on humans and societies regarding ecology and economy. While productive agricultural areas are being exhausted due to sediment resulting from erosion and transporting by rivers causes various difficulties in supplying drinking and irrigation water, and results in reduced economic life of dams. Sediment amounts which are measured in several stations in basins vary depending upon climate characteristics, basin characteristics which affect rainfall- runoff relations, as well as, flow characteristics which play an important role in occurrence and transportation of sediment. To this end, the determination of a correlation between sediment yield measured in basins and variables which have effects on this yield, and establisment of mathematical models obtained by using equations enable us to estimate sediment yield of other rivers of the basin, sediment data of which have not been measured. This study was established by using data provided from Turkish State Meteorological Service (DMİ) and General Directorate of Electrical Power Resources Survey and Development Administration (EİE). It was aimed to establish mathematical models of suspended- sediment transport in Tigris River, Botan River and Garzan River, all of which are in Tigris watershed. Artificial Neural Networks (ANN) and Regression Analysis (RA) were used as Mathematical models and 14 models were established for each river. The model which had least error according to several statistics of models was chosen as the most appropriate model. While rainfall, temperature and flow discharge were used as independent variables, suspended- sediment load was used as dependent variable in both models. It was concluded from this study that use of ANN models exhibited better results than did RA models in the modelling of suspended-sediment transport of the three rivers. KEY WORDS: Tigris River, Sediment Transport, Erosion, Mathematical Models, Artificial Neural Networks (ANNs), Regression Analysis (RA).