Uncertainties in weather forecast-reasons and handling

Abstract

The generation of precipitation forecasts by means of numerical weather prediction (NWP) models is increasingly becoming an important input for hydrological models. Over the past decades the quality and spatial resolution of meteorological numerical models has been drastically improved, which makes it now possible to incorporate high-resolution NWP output directly into flood forecasting systems. The quality of forecasted precipitation, however, is still close to insufficient because rainfall constitutes merely the very end of a complex of interlinked process chains acting at a broad range of spatial and temporal scales. Consequently the precipitation fields can vary significantly with time and space and inherit wide ranges of uncertainties. For the purpose of flood risk management it is of particular interest to investigate both the potential and implications of the related variations and uncertainties. For this endeavour the general background and current uncertainties in NWP as well as the handling of the uncertainties has to be taken into account. This chapter gives a brief introduction into the generation of weather forecasts with a particular focus on the accuracy of rainfall prediction. It includes in this context the relatively new field of ensemble forecasting and discusses ways to link numerical NWP with radar-based precipitation nowcasting. © 2011 Springer Science+Business Media B.V.