Naslov
Methods for analysis of non-stationary hydrological time series

Autori
Potočki, Kristina

Izvornik
Research seminar - invited presentation

Vrsta, podvrsta
Ostale vrste radova, ostalo

Godina
2020

Ključne riječi
discharge, suspended sediment, wavelet transform, time series analysis, black-box modeling

Sažetak
The traditional approach to the hydrological time series analysis usually starts form the assumptions of linearity and stationarity of hydrological processes that often cannot be met. Hydrological time series analysis in time- frequency domain with different signal processing methods, such as short-time Fourier transform and wavelet transform, offers a good complementary approach to the traditional statistical analysis. These methods can be used for preliminary analysis of data and uncovering of non-stationary patterns, but also as a preprocessing tool in series black-box modeling. Examples of the wavelet transform application for the analysis and for black-box modeling of water level, discharge and suspended sediment series are given and results are compared with the traditional statistical approach. Detection of trends and cycles on different time scales with presented methods can help uncovering and better understanding of temporal patterns in hydrological time series. Revealed patterns can be then related to the climate change and anthropological impacts on fluvial processes.

Izvorni jezik
Engleski

Znanstvena područja
Interdisciplinarne prirodne znanosti, Građevinarstvo, Interdisciplinarne tehničke znanosti

POVEZANOST RADA