Rainfall generator for the Meuse basin: Simulation of 6-hourly rainfall and temperature for the Ourthe catchment

This report presents a first study on the development of a
stochastic weather generator for the Meuse basin which produces
long-duration, multi-site time series of precipitation and
temperature. By running these synthetic data through a
hydrological/hydraulic model, it is expected to get a better
insight into the likelihood of extreme river discharges in the
Netherlands. This report is restricted to the Ourthe basin (3626
km2). Time series of 6-hourly area-average precipitation of 3
sub-catchments (Ourthe upstream of Hamoir, Amblève and Vesdre)
and average 6-hourly temperature at St. Hubert are considered.

Simulation is done by nearest-neighbour resampling. Because of the
intended application, the performance of the method is only
assessed for the winter half-year (October-March). It appears that
straightforward resampling of the historical 6-hourly values does
not adequately reproduce a number of second-order statistics of
precipitation and temperature. Particularly, the slow decay of the
autocorrelation function of 6-h area-average rainfall is not
preserved. The standard deviations of monthly rainfall and the
quantiles of the multi-day winter maximum precipitation amounts
are therefore underestimated. As an alternative, simulation of
daily values with disaggregation into 6-h values using the method
of fragments is studied. With this strategy a reasonable
reproduction of the second-order statistics of rainfall and
temperature is achieved. Moreover, there is a good correspondence
between the historical and simulated distributions of the winter
maximum precipitation amounts.