Rainfall generator for the Rhine basin: single-site generation of weather variables by nearest-neighbour resampling

This report presents the first phase of a project for the development of a rainfall generator for the Rhine basin. The request for this generator arose from the need to study the likelihood of extreme river discharges in the Netherlands, using a hydrologicaVhydraulic model. This first phase deals with single-site generation of weather variables by nearest.:neighbour resampling for the stations Essen, Kahler Asten, Trier, Frankfurt, Bamberg, Freudenstadt and Stuttgart in the German part of the basin. Daily precipitation and temperature data for these stations were made available for the period 1961-1990. The ultimate aim of the project is to generate simultaneous records of daily average rainfall amounts, together with daill temperatures, over about 30 subcatchments, each with an area of about 5000 km , to study the effects of unprecedented weather situations. 

The essence of the nearest-neigbour resampling technique is that the variables for a new day are sampled with replacement from a selected set of historica) data (the nearest neighbours or analogues). The method needs a feature vector D, to find the nearest neighbours of the day of interest (day t). For unconditional resampling, D, contains variables that characterize the state of the weather on day t-1, e.g., daily precipitation, temperature and atmospheric circulation indices. Conditional simulation on the atmospheric flow is also possible with the nearest­neighbour resampling technique. Circulation indices for day t are then included in D,. A (weighted) Euclidean distance measure is used to determine the k nearest neighbours of D,. From the k nearest neighbours, one day is resampled with a predefined probability. In case of unconditional simulation, the successor in the historica) record to the resampled day becomes the new day, whereas in case of conditional simulation, precipitation and temperature are set equal to the values of the resampled day.