A Particle Swarm Optimization for Parameter Estimation of a Rainfall-Runoff Model

Abstract : Transfer function based models are widely used in hydrological modeling, especially for rainfall-runoff simulation. This paper deals with a new model associating three elementary transfer functions in series along four parallel branches. Each transfer function is a parametric probability density function with some physical meaning regarding hydrological transfers. Model inversion is performed using the so-called particle swarm optimization (PSO) technique. The systemic model and the PSO method were tested on the Sèvre Niortaise water catchment in France.