Investigation of Non-linearity of Daily, Monthly, and Yearly Streamflow and the Correlation Between Non-linearity and Catchment Area

Streamflow prediction by using time series models is one of the practical methods that can play an important role in water resource management.  But choosing the correct model based on the linear or non-linear streamflow behavior is so crucial. Additionally, streamflow non-linearity can be affected by physiographic characteristics of watershed.  In this study, streamflow non-linearity of daily, monthly and yearly time series and its relationship with the hydrometric station area have been examined. To fulfill this purpose, ten hydrometric stations with various catchment area were selected in Hesse state in Germany. The Brock-Dechert-Scheinkman (BDS) test for testing non-linearity patterns in time series, was used to test non-linearity. The results showed that daily streamflow time series exhibit strong non-linearity. In addition, as the timescale increases, the intensity of nonlinearity decreases. Monthly and yearly streamflow time series showed less evidence of non-linearity. Furthermore, regarding the investigation of the relationship between streamflow non-linearity and watershed area, pearson and spearman tests were used. The results revealed that in none of the time scales, daily, monthly, and yearly there is no significant correlation between these two parameters. It should be noted that the novelty of this article is examining the relationship between intensity of the non-linearity and watershed area as a factor in choosing the best fitted model for time series.