Comparison of Different Trend Analysis Methods for Hydro-Climatic Variables, a Case Study: Sefidroud Dam

Trends invesigation of climatic variables in a catchment can lead to more realistic approaches for water resources planning and management. In this paper, monthly, seasonal and annual trends of inflow, precipitation, evaporation and temperature are studied for Sefidroud dam. Therefore, a parametric test (Regression) and a nonparametric tset (Mann-Kendall) under three different scenarios of independence, short term persistence, and long term persistence are used. The assumption of dataset independency is one of the limitations of the Mann -Kendall test. Therefore, if hydro-climatic time series are characterized with short or long term persistence, this assumption is unacceptable and leads to unreliable results. Since, serial correlation and Hurst phenomenon are seen in most hydro-climatic time series, it is necessary to modify Mann-Kendall test for considering these effects. For comparison of trend analysis results, the homogeneity of hydro-climatic variables is studied by cumulative deviation test, Worsley Likelihood test, and the Bayesian procedure. Comparison of results indicates remarkable significant decreasing trends for inflow using different methods and assumptions. Furthermore, decreasing trends for precipitation and minimum temperature and increasing trends for maximum temperature indicate that hydro-climatology of the study area might be impacted by possible climate change. Obtained results also show that most time series with significant deacrising or increasing trends are also characterized with significant non-homogeneity. Therefore, simultaneous use of homogeneity and trend analysis tests for studying Hydro-climatic variables lead to more reliable results.