Abstract: According to the daily water level data at Jiangba, Shangju, Gaoliangjian, and Laozishan stations from 1967-2016, we investigated into the interannual and annual variation rules of characteristic (average, maximum, and minimum) water levels in Hongze Lake via Mann-Kendall test, order cluster analysis, BFAST (Breaks for Additive Seasonal and Trend) algorithm, and wavelet analysis. Moreover, we revealed the dominant factor triggering the variation of characteristic water levels by constructing the multiple regression equation among each characteristic value and inflow and discharge, as well as precipitation and evaporation in the same period. 1) The characteristic water levels experienced abrupt changes in 1982-1985. After 1986, the characteristic water levels witnessed a marked increase; the lowest water level responded the earliest, while the highest water level responded the latest. 2) The fluctuation of characteristic water levels in Hongze Lake is highly consistent, with 36 years, 17 years, and 9 years as the first, second, and third principal periods, respectively. 3) In storage stage(stage 1, from October to next April), the flow rate at Xiaoliuxiang is the major factor that affect the average water level and minimum water level, accounting for 36% and 33% of the total impact factors, respectively; the flow rate at Sanhezha is the dominant factor for the highest water level, accounting for 36%. In discharging stage (stage 2, May to June), the flow rate at Shuanggou exerts a more evident impact on the lowest water level, occupying 47% of the total impact factors. In water-rising stage (stage 3, July to September), the flow rate at Xiaoliuxiang plays a prevailing role (29%) in average water level. The research results offer scientific basis for the regulation of water level and upstream and downstream runoff in Hongze Lake.

Key words: characteristic value of water level, influencing factors, Mann-Kendall test, BFAST algorithm, Hongze Lake