流域水文模型参数年际变化规律研究
Research on the Interannual Variation Rules of Watershed Hydrological Model Parameters

选取汉江上游白河站作为控制点，建立了汉江上游三水源新安江模型，分别采用遗传算法和SCE-UA算法对新安江模型的参数进行优选，用1961~1990年逐日实测径流资料分不同时期对三水源新安江模型进行率定，两种参数优化方法得到的不同时段模型纳西效率系数均超过0.80，径流总量相对误差最大值为4.94%，SCE-UA优化方法结果优于遗传算法。采用DREAM算法抽样，通过研究不同时期水文模型参数的后验分布，进而分析参数的时变规律和不确定性。结果表明：后验分布参数能够较好地模拟实测径流；1981~1990时段模型模拟的不确定性区间要大于1961~1970和1971~1980时段模拟的不确定性区间，说明随着人类活动作用的增强，水文模型参数的不确定性增加。
The Baihe Station was selected as the control station to establish the Xin’anjiang Model in the upper reaches of the Hanjiang River. Genetic algorithm and SCE-UA algorithm were used to optimize the parameters of the Xin’anjiang Model. Daily flow data from 1961 to 1990 were used to calibrate the hydrological model in different periods. The Nash-Sutcliffe efficiency coefficients were over 0.80 and the maximum relative error were less than 4.94% using the two optimization methods in different periods. The results obtained from the SCE-UA method were better than the genetic algorithm. The DREAM algorithm was used for sampling and the posterior distribution of hydrological model parameters in different periods was studied to analyze the time-varying rules and uncertainties of the parameters. The results show that the posterior distribution parameters can simulate the observed flow well. The uncertainty interval simulated by the hydrological model during the period of 1981~1990 is greater than that in the periods of 1961~1970 and 1971~1980, indicating that the uncertainty of hydrological model parameters increases with the enhancing of human activities.