Multi-Temporal Analysis of Mean Annual and Seasonal Stream Flow Trends, Including Periodicity and Multiple Non-Linear Regression

Abstract

Global warming affects the hydrological cycle and the long-term water budget of river basins. Flow variations have been noticed in the Danube River Basin, especially in its south-western parts where a downward trend in mean annual flows has been prevalent in the past several decades. Time series of mean annual and seasonal flows of the Sava River at hydrological stations Sremska Mitrovica and Zagreb are analysed in this paper. The trend is assessed with the Mann-Kendall test including the effect of serial correlation. Additionally, the trends are assessed in the multi-temporal framework. It is concluded that the long-term periodicity of annual flows has a considerable impact on the time series trend. Long-term component with cycles of 40 years in mean annual flows are detected by the time series analysis in frequency domain. Regression analysis showed a significant correlation between mean annual flows of the Sava River and annual precipitation, mean annual atmospheric pressure and air temperatures at meteorological station Ljubljana, as well as with the North Atlantic Oscillation (NAO) Index.