Stochastic description of the rainfall input errors in lumped hydrological models

Abstract. Rainfall input errors are one of the most important sources of uncertainty in hydrological models. They have been analysed for the watershed-averaged rainfall used as input in lumped conceptual models for watershed hydrology. The total rainfall input error is described in a stochastic framework on the basis of the different error-sources (e.g. point rainfall measurement errors and spatially-averaged rainfall estimation errors). They are related to the storm volumes, the distance between the rain gauges and the watershed centre and the configuration of the rain gauge network. The error description is applicable in probabilistic water management and efficient model calibration. It is also shown that, under certain circumstances, small storm volumes are underestimated and large volumes overestimated, and this bias can be corrected using the rainfall input error description.