THE SPEARMANS RHO TEST FOR DETECTING TRENDS IN SERIALLY CORRELATED HYDROLOGICAL SERIES.

When sample data are serially correlated, the presence of serial correlation in time series will affect the ability of the test to correctly assess the significance of trend. Hence, this paper discussed two methods to eliminate the influence of serial correlation on the SR test about a time series only consist of an AR(1) process and combined series consist of a linear trend and an AR(1) process, respectively. This study investigated using Monte Carlo simulation generated some time series. Simulation demonstrates that when no trend exists within time series, Pre-whitening and ESS can effectively limit the influence of serial correlation on the SR test. When trend exists within time series, these two kinds of approach has a similar results, both the pre-whitening and the ESS approaches cannot properly limit the influence of serial correlation on the SR test.