洪水预报模型参数优化与影响因素分析

Optimization of flood forecasting model parameters and analysis of influencing factors

选取高关水库控制流域1980—2016年的26场洪水,运用多参数灵敏度分析(multi-parameter sensitivity analysis,MPSA)筛选出新安江洪水预报模型的灵敏参数,采用普适似然度不确定性估计(generalized likelihood uncertainty estimation,GLUE)方法分析灵敏参数的不确定性,确定部分灵敏参数的后验分布范围;将场次洪水资料划分为1980—1997年和1998—2016年2个时期,使用遗传算法(genetic algorithm,GA)和自适应粒子群算法(adaptive particle swarm optimization,APSO)对筛选出的灵敏参数进行率定及验证,并分析了不同时期灵敏参数与区域土地利用的变化,结果表明:1)研究流域采用的新安江模型的灵敏参数中,流域不透水率(IMP)、地下水日出流系数(KG)、壤中流日出流系数(KSS)、壤中流消退系数(KKSS)、动力方程蓄水指数(n)、动力方程蓄水系数(c)的不确定性范围较小,能够得到比先前的经验范围更集中的取值区域;2)GA和APSO在新安江模型参数率定中都有较好的适应性,但APSO的综合表现优于GA;3)由前、后2个时段对比可知,IMP、n显著增大,表层自由水蓄水容量(SM)显著减小,研究区域产汇流条件产生了明显的变化。

In this paper,26 floods from 1980 to 2016 in the controlled basin of Gaoguan Reservoir are selected,and the sensitivity parameters of the Xin'an River flood forecasting model are screened by using multi-parameter sensitivity analysis(MPSA).The uncertainty of sensitivity parameters is analyzed by using generalized likelihood uncertainty estimation(GLUE)method,and the posterior distribution ranges of some of the sensitivity parameters are determined.The flood data are divided into two periods from 1980 to 1997 and from 1998 to 2016,and genetic algorithm(GA)and adaptive particle swarm optimization(APSO)algorithm are used to calibrate and verify the screened sensitive parameters,and the sensitive parameters and regional land use changes in different periods are analyzed.The results show that:1)among the sensitive parameters of Xin'an River model adopted in the study basin,the uncertainty ranges of impervious surface coverage(IMP),groundwater daily runoff coefficient(KG),soil flow daily runoff coefficient(KSS),soil flow recession coefficient(KKSS),dynamic equation water storage index(n),and dynamic equation water storage coefficient(c)are small,and the value ranges are more concentrated than the previous empirical ranges;2)both GA and APSO have good adaptability in the Xin'an River model parameter calibration in the study area,but the comprehensive performance of APSO is better than that of GA;3)when comparing the two time periods,IMP and n significantly increased,the surface free water storage capacity(SM)significantly decreased,and the conditions of flow production and confluence in the study area changed significantly.