Characteristic scales, temporal variability modes and simulation of monthly Elbe River flow time series at ungauged locations

Abstract

Spatial and temporal patterns of the long-range extreme monthly Elbe River flows across Germany are investigated, using various statistical methods, among others, principal component and wavelet analysis. Characteristic time scales are derived for various time series statistics. The wavelet analysis of the raw river discharge data as well as of the major principal component reveal the main oscillatory components and their temporal behavior, namely low frequency oscillations at interannual (6.9 yr) and interdecadal (13.9 yr) scales. The EOFs at ungauged stations are estimated from the principal components of the observed time series sampled over a limited time span whose length equals the major temporal variability scale (≈7 yr). The EOFs (empirical orthogonal functions) obtained in this way are subsequently used to simulate long-range flows at these locations. A comparison of this method with linear interpolation and ordinary kriging of the EOF shows the superiority of the former in representing the distributional properties of the observed time series. The simulated time series preserve also short and long-memory.