ASSESSMENT OF PARAMETER UNCERTAINTY IN RAINFALL RUNOFF MODELLING

摘要

The implications of uncertain model parameter values on model results are examined. Model parameter uncertainty is defined based on results obtained by both Single and Multi Objective automatic calibration. For the analysis, a four parameter conceptual rainfall runoff model has been linked to a global optimization algorithm based on evolutionary algorithms. The outcome of a synthetic application example indicates that Multi Objective calibration results in less variable parameter values. Parameter correlation is a problem for the convergence of the search algorithm. The characteristics of the calibration event and the objective function have to be selected adequately.