三峡水位周期性振荡是影响库区滑坡变形及稳定性的重要随机变量。通过对2010—2020年三峡坝前水位过程线进行标定,获取水位过程线特征参数的边缘分布类型、分布参数和相关性。运用9条直线段拟合水位过程线,获取库水消落期与回升期的持续时间和变化速率、汛限期前后平台持时、洪峰爬升和下降时长、洪峰水位等14个特征参数。根据赤池信息准则(Akaike Information Criterion,AIC)识别特征参数的最优分布类型,并利用Bootstrap方法模拟统计不确定性。结果表明:①特征参数均值线克服了水位均值线平均时的削峰填谷现象,可作为库区涉水滑坡变形和稳定预测的的典型水位过程线。②在库水消落期的缓降和陡降、回升期的陡升和缓升4个阶段内,持续时间分别服从威布尔、极值Ⅰ型、对数正态、威布尔分布,均值分别为87.00、47.45、22.18、32.27 d;库水位变化速率分别服从极值Ⅰ型、威布尔、极值Ⅰ型、极值Ⅰ型分布,均值分别为14.53、36.75、74.33、37.88 cm/d;持时和库水位变化速率的Pearson相关系数分别为-0.67、-0.78、-0.85、-0.67。③汛期洪峰水位服从极值Ⅰ型分布,缓降持时和速度之间相关系数均值的变异性较大。研究成果可为库区涉水滑坡的风险分析提供水位信息相关统计特性。

Periodic change of water level in Three Gorges Reservoir is a crucial random variable that affects the deformation and stability of bank slope in the reservoir area. The marginal distribution type, distribution parameters and correlation coefficient of characteristic parameters are obtained by calibrating the water level process lines(WPL) of Three Gorges ranged from 2010 to 2020. Firstly, WPL is fitted with nine straight lines and fourteen parameters inclusive of the duration and rising or falling speed in rising and falling period, the duration of two platforms before and after the main flood period, the climbing and falling durations of flood peak, and the peak water level. AIC (Akaike Information Criterion, AIC) method is used to identify the optimum distribution of characteristic parameters, and statistical uncertainty is simulated by using Bootstrap method. Results show that the WPL formed by average characteristic parameters calibrated from 11-year WPL can effectively avoid the peak elimination and valley filling, and can be taken as typical WPL for deformation and stability prediction of landslides in the reservoir area. The durations of the slow-descent stage and steep-descent stage of decline period and the steep-rising stage and slow-rising stage of the rising period obey Weibull distribution, the extreme value I-type distribution, the lognormal distribution, and the Weibull distribution, respectively, with the mean values being respectively 87.00, 47.45, 22.18, and 32.27 days. The falling or rising speeds of reservoir water in the four stages obey extreme value I-type distribution, Weibull distribution, extreme value I-type distribution, and extreme value I-type distribution, respectively, with the mean values being 14.53 cm/d, 36.75 cm/d, 74.33 cm/d, and 37.88 cm/d, respectively. The Pearson coefficient of correlation between duration and reservoir water level change rate in the four stages are -0.67, -0.78, -0.85, and -0.67. The flood peak level in flood season obeys the extreme value I-type distribution and the mean coefficient of correlation between slow-descent duration and the speed varies largely. The research findings offer random statistical parameters of WPL for risk analysis of landslides in the Three Gorges reservoir area.