Evaluation of Sediment Flux in a Part of the Brahmaputra River and Application of ANN and Linear Regression Models

Estimation of sediment load can provide basic information on a range of problems related to the design and operation of river system and for water resources engineering as well as environmental problems. High sediment load is an integral component of the Brahmaputra River system, and its role, despite being critical in the overall systemic behaviour of the river, is little understood. Due to its sheer quantity and complex behavior during transport, sediment control has remained a challenge. Sediment flux depends on sediment properties, characteristics of the sediment load, and properties of the fluid flow. Artificial Neural Network (ANN) and Multiple Linear Regression (MLR) models were utilized to predict both sediment and particulate heavy metal concentration. Samples from river suspended and bank materials were analyzed with the help of X-Ray Diffraction, Scanning Electron Microscope, Laser Particle Size analyzer and Atomic Absorption Spectrometer. EDX spectra were generated for individual grains to understand compositional characters of the samples. Results of all these investigations were combined to develop a comprehensive understanding of the sediment load of the Brahmaputra River. The performances of the desired models confirmed that the model derived using ANNs gave a better prediction than the model derived using MLR.