Sensitivity of conceptual rainfall-runoff algorithms to errors in input data — case of the GR2M model

Abstract

The study concerns the effect of errors in input data on the quality of the results obtained with rainfall-runoff models. A simple conceptual algorithm was selected: two inputs (rain and evapotranspiration), one output (resulting runoff), and two parameters. Systematic and random errors were considered. The algorithm amplifies the initial systematic error to an extent which depends on the phase of the hydrograph. The model appears less sensitive to errors in evapotranspiration ETP than in rainfall. The distribution curves of random errors in rainfall and in runoff are very similar; there is no distortion of one compared to the other. Tests were carried out on the method of calibration to demonstrate the power of numerical algorithms of this type: they are capable of absorbing the initial error and of determining an optimal solution.