A Fourier spectral analysis of 55-110 years of daily discharge time series in 79 unregulated catchments in Sweden reveals that the discharge power spectrum slope in most of the studied catchments has gradually steepened over time. This statistically significant drift in the periodicity of dominant hydrologic response processes can be attributed to a change in either climatic forcing factors or anthropogenic effects on the land surface, e.g., land-use changes. For those locations for which historical meteorological observations are available (the 41 southernmost catchments), the results of our analyses of changes in precipitation power spectra indicate that local land-use changes within the catchments may affect discharge power spectra more significantly than precipitation pattern changes (resulting from climate change).
By using 1D distributed hydraulic routing, we quantitatively analyze how travel time distributions within stream networks can vary because of anthropogenic impacts, such as changes in stream network spatial coordinates (these stream networks are derived from three maps: two from the present and one from the 1880s), river width modifications, stream channel excavation, and the elimination of thresholds in stream bottom topography that cause exceedingly low local bottom slopes.
The findings that the discharge power spectrum may change significantly over time, implies that conventional, statistically-based parameterization of hydrological models that rely on assumptions of stationarity may be less suited than more physically based parameterization alternatives. This essential information must be considered when performing tasks that involve (peak) flow predictions, such as those for dimensioning and flood risk management purposes.
Blackwell Publishing, 2016.