Abstract
The ECMWF analysis‐forecasting scheme used to provide inputs to WetNet's first Precipitation Intercomparison Project (PIP‐1) is briefly described. Aspects connected with the model's ability to simulate precipitation are discussed in more detail. Inconsistencies between the use of observed data and the model formulation lead to a spin‐up in the short‐range forecasts which has severe impacts on the precipitation. The shortest forecast ranges, e.g. 12–36 hour means, may best reflect the variability of precipitation in time and space, whereas more extended ranges, e.g. day 2–3 forecast means, may be better when the actual amounts on larger scales are of concern.