Downstream flow top width prediction in a river system

Abstract

ANFIS, ARIMA and Hybrid Multiple Inflows Muskingum models (HMIM) were applied to simulate and forecast downstream discharge and flow top widths in a river system. The ANFIS model works on a set of linguistic rules while the ARIMA model uses a set of past values to predict the next value in a time series. The HMIM model assumes a power-law relationship between water discharge and flow top width at a section. The models were used to simulate and forecast discharge and flow top width at a downstream section in the Barak River system in India. Flow top widths corresponding to  different flow depths at the downstream section were estimated using a digital elevation model (DEM). The parameters in the hybrid model were estimated by applying Non-dominated Sorting Genetic Algorithm II  (NSGA-II). The study shows that the power-law relationship involving section characteristics can describe the top width versus discharge  relationship for a section. The models allow direct estimation of the downstream flow top width on the basis of upstream flow variables. Results obtained in the study show that performances of the HMIM, ANFIS and  ARIMA models are satisfactory, having average prediction errors of less than 7% of the average value of the observed series. Application of the ANFIS, ARIMA and the HMIM models to the studied river system  demonstrate the suitability of the models in simulating and forecasting  downstream flow top width in river systems.

Keywords: River system, flow top width, genetic algorithm, flood flow, hybrid model