Trend and Stationarity Analysis of Streamflow for the Purpose of Hydrological Time Series Modeling

Many hydrological time series exhibit trending behavior or nonstationarity. An important task of hydrological time series is to determine if there exists any trend in the data، and how to achieve stationarity. Therefore، presenting any method to analyze trend and stationarity will be useful. On the other hand، detecting trend and stationarity in hydrological time series may help us to understand the possible links between hydrological processes and global environmental changes. Data of Shaharchai river in Urmia of West Azarbaijan province of Iran with records from 1973-2008 (31 years) consisting of daily، 10-day، monthly and yearly stramflow rates were employed. Trend analyses with Mann-Kendall and seasonal Kendall test show that there is no trend in annual mean discharge but monthly flow series exhibit significant downward trend in specific months. Stationarity of time series after deseasonalizing were investing at using ADF and KPSS tests. All daily and 10-day steamflow series appear to be significantly stationary and most monthly and annual series are stationary. The results imply that in smaller time scales such as daily and 10-day، trend is more obvious and detrending can help series to be stationary. Nonstationarity may be because of existing seasonality or periodic behavior in streamflow time series. For removing seasonality in order to reach statationarity mathematical models can be used.