Hydrological time series analysis and modelling using statistical tests and linear time series models (case study: West Azerbaijan province of Iran)

Abstract

In this paper, the time series of hydrological processes analysed using ADF, KPSS, Mann-Kendall tests and linear time series models on monthly scale in the West Azerbaijan province of Iran. Significant decreasing trend in precipitation and stream flow discharge were observed. Also, non-stationarity in precipitation was observed in four studied basins. Decreasing stream flow discharge partially can be a result of trends in precipitation and evaporation. The population growth, land use changes and increase in water withdrawal are other reasons to blame. The results of linear time series modelling indicates that ARMA models for modelling stream flow only appears when precipitation or evaporation models have a MA component. On the other hand, when precipitation and evaporation models only have AR component, stream flow models are also follow autoregressive models and moving average components vanish. Moreover, in climatically homogenous basins, the same models were best-fitted for different hydrologic processes.