A stochastic weather generator is constructed to simulate long series of precipitation and temperature for the Belgian and French Meuse basin. This weather generator is based on the
principle of nearest-neighbour resampling, a method which samples daily data from a historic record with replacement such that the temporal and spatial correlations are preserved. The weather generator uses station records of temperature and precipitation as base material. A modification of the method is tested to simulate sub-catchment precipitation, for which only short records are available. To this end the weather generator is conditioned on a limited number of long station records. The standard deviations and autocorrelation coefficients of the simulated sub-catchment precipitation and the station temperatures are in good agreement with the estimates from the historical data. The distributions of the 4-, 10- and 30-day seasonal maxima of area-average precipitation are also well reproduced, although a rare historical 30-day event was found in July 1980, which was not exceeded in the 3000-year simulation.
R Leander, TA Buishand. Rainfall generator for the Meuse basin; Development of a multi-site extension for the entire drainage area