THE INFLUENCE OF LAG-1 AUTOCORRELATION ON THE MK TEST AND APPLICATION IN THE CHANGE-POINT ANALYSIS OF FLOOD DATA

In this paper, the effects of the AR(1) process with the different lag-1 serial correlation coefficients and different sample sizes on the Mann-Kendall test were analyzed by Monte Carlo simulation. Then, using pre-whitening method, we modify the primary data and eliminate the influence of lag-1 serial correlation on the MK Mutation Test. Finally, based on the flood data from 1949 to 2008 in Zibo, China, the change point and the trend of the flood data were analyzed and the results between the revised and the original method were compared and discussed.