Abstract: Adopted the measured flood peak flow value which based on Yanchuan and Zichang stations since the Qingjian River basin hydrology station as research data. Then, its variation trend through the cumulative anomaly and the linear trend estimation were analyzed, and the sustainability analysis through R/S method was carried out, in addition to use the M-K test for the mutation analysis, and also the Morlet complex wavelet was used to diagnose the periodical characteristics of the flood peak discharge of the Qingjian River. Furthermore, the correlation between the peak value of the flood peak discharge and the extreme value of precipitation was analyzed. The results indicated that the annual peak discharges of the Qingjian River occurred in June, July, August and September. The concentrations occurred in July and August, and occurred at 0：00—5：59, 18：00—23：59. The main time period was dominated; The annual maximum peak discharge showed a decreasing trend in the process of fluctuations, and the decreasing trend had a weak persistence; During this time period, the annual maximum peak flood flow occured abruptly changes, and the mutation point was in 2007; Wavelet analysis results showed that the period of quasi-4a, 7a and 11a were within this time period, which controled the annual maximum peak flood flow change process; Correlation tests showed that the correlation coefficient between the peak value of flood peak flow and the extreme value of precipitation at α=0.05 was not significant.

Key words: variation trend, mutation point, cycle, correlation