Erzeugung synthetischer Abflußdaten mit Hilfe eines zeit- und raumvarianten Modells im Einzugsgebiet der Lahn

Abstract

A time- and space-varying model for generation synthetic monthly streamflow series of any length is constructed based on a sample of drainage basins in the Lahn river catchment. A first order autoregressive model (Markov model) including a seasonal component is used as the time dependent generating scheme. If there exist runoff observation for the drainage basin in question, the parameters of the markov-model (monthly means, standard deviations and correlation coefficients) may be computed from this data. Otherwise a space-varying extension is needed: Approximation of the above mentioned statistics by Fourier-analysis results in Fourier-coefficients which are estimated by a few regression equations using physiographic basin characteristics (e.g. mean annual precipitation, forested area) as input data. Examples are discussed to point out qualifications and limitations of the model and to demonstrate application of synthetic streamflow series to design of water supply reservoirs