PARAMETER ESTIMATION FOR NON-LINEAR MUSKINGUM FLOOD ROUTING MODEL: NATURAL INSPIRED ALGORITHM
N. Ramsundram1, M. Keerthana2, M. Arun kumaar3, S. Shreevarshini4
1Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, Tamilnadu, INDIA – 641 049.
2Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, Tamilnadu, INDIA – 641 049.
3Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, Tamilnadu, INDIA – 641 049.
4 Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, Tamilnadu, INDIA – 641 049.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In any watershed the surface runoff is routed through the main course of the river to estimate the flow hydrograph at the outlet of the basin. One of the most widely used flood routing model in a river course is Muskingum model. The Muskingum model has three parameters, namely; k, x, and m, which are estimated based on the known input and outflow hydrograph of the channel. The accuracy of routing depends on the preciousness in estimating the model parameters. Many research works had been reported on improvising the parameter estimation by employing various advanced computational techniques. In this paper, we introduce a bat optimization algorithm for estimating the three parameters of the Muskingum model. Bat algorithm finds the global optimum for the parameters with random fly within the domain space. The results of the proposed model have been compared with various parameter estimation techniques reported for the Muskingum flood routing model. The comparative evaluation has been done based on four performance indicators, namely; a) Root Mean Square Error (RMSE), b) Mean absolute error (MAE), c) Coefficient of Correlation (R) , and d) Nash-Sutcliffe efficiency (E). From performance indicators it is observed that the proposed parameter estimation using a bat optimization algorithm outperforms in capturing the observed flow compared to other reported techniques.
Key Words: Non-linear Muskingum model, Flood routing, parameter estimation, Bat optimization
