Abstract. Statistical detection of trends in hydrometeorological
time series is a crucial task when revealing how river systems react to
environmental and human-induced changes. It was shown that the
autocorrelation structure of a series influences the power of parametric and
nonparametric trend tests. While the order of short-memory processes can be
sufficiently captured by AR(I)MA models, the determination of the Hurst
exponent, which describes the long memory, is still challenging, considering
that the available methods partially give different results. In the Elbe
River basin, Europe, several studies focusing on the detection (or
description) of long-term persistence were performed. However, different
lengths of series and different methods were used. The aim of the present
work is to gather the results gained in various parts of the Elbe basin in
Central Europe and to compare them with our estimation of the Hurst exponent
using six discharge series observed in selected subbasins. Instead of the
dependence of the exponent on the catchment area suggested by the theory of
aggregated short-memory processes, we rather found a relationship between
this parameter and the series length. As the theory is not supported by our
findings, we suppose that the Hurst phenomenon is caused by a complex
interplay of low-frequency climate variability and catchment processes.
Experiments based on distributed water balance models should be the further
research objective, ideally under the umbrella of mutual international
projects.