基于参数分级优化的水文气象耦合预报研究  

Coupled Hydrometeorology Forecasting Based on Hierarchical Optimization of Parameters

作　　者：王秀杰[1,2] 李丹丹[1,2] 苑希民 徐浩田[1,2] 齐喜玲 Wang Xiujie;Li Dandan;Yuan Ximin;Xu Haotian;Qi Xiling(State Key Laboratory of Hydraulic Engineering Simulation and Safety,Tianjin University,Tianjin 300072,China;School of Civil Engineering,Tianjin University,Tianjin 300072,China)

机构地区：[1]天津大学水利工程仿真与安全国家重点实验室,天津300072 [2]天津大学建筑工程学院,天津300072

出　　处：《天津大学学报（自然科学与工程技术版）》2021年第3期279-288,共10页Journal of Tianjin University：Science and Technology

基　　金：国家重点研发计划资助项目(2018YFC1508403);广西左江治旱驮英水库及灌区工程综合管理信息系统(2018GKF-0574);天津大学自主创新基金“战略性布局-产学研培育”资助项目(2020XZC-0002).

摘　　要：针对中小流域洪水损失严重,加之气象水文资料短缺、预报模型参数不稳定、预报精度面临挑战等问题,提出将基于降雨时程分配指标(最大降雨强度和分配方差)的参数分级优化方法融合到水文模型中,同时引入气候再分析数据和集合预报产品,对广西驮英水库上游洪水进行预报研究,并利用灰色关联分析法进行洪水的敏感性分析.结果表明,基于最大降雨强度及降雨方差的参数分级优化方法显著提高了洪水预报精度,其中平均百分比误差为13.09%,纳什系数为0.86;利用分位数映射法订正的ERA5-Land再分析降雨数据和驮英水文站相应时间的实测流量分别对驮英水文站的实测降雨过程进行强度和分布的校正,为模型的率定和验证提供更为精确的降雨数据;利用TS评分、Brier评分、Brier技巧评分方法评估集合预报产品在不同降雨等级的平均预报性能和不同预报时效范围内的预报表现,通过对比分析,选取最佳数据集进行水库的短期洪水预报,从模型输入角度进一步减小洪水预报的不确定性.对资料短缺地区,以上两种数据产品为洪水预报提供了较为可靠的气象数据来源;洪水量级对累积降雨量的敏感性最为显著,洪峰流量、水位与累积降雨量间的经验公式均可用不同形式的幂律关系表示.以上方法对其他类似的中小流域洪水预报具有很好的借鉴作用.To address the issues in mid-small basins such as the serious flood losses,the shortage of meteorological and hydrological data,the unstable parameters of prediction model,and challenges in achieving prediction accuracy,the parameter hierarchical optimization method based on the horal allocation index(maximum rainfall intensity and distribution variance)of rainfall was proposed to be integrated into hydrological model.Meanwhile,climate reanalysis data and ensemble forecasting products were introduced to research the flood forecastion of upper Tuoying reservoir.Moreover,grey relational analysis was used for flood sensitivity analysis.The results show that the parameter hierarchical optimization method based on the maximum rainfall intensity and distribution variance can significantly improve the accuracy of flood forecasting.The average percentage error is 13.09%and the Nash coefficient is 0.86;The ERA5-land reanalysis rainfall data revised by quantile mapping method and the measured rainfall flow of Tuoying hydrological station at corresponding time were used to correct the intensity and distribution of the measured rainfall process of Tuoying hydrological station,which can provide more accurate rainfall data for the calibration and verification of the model. The Threat score,Brier score and Brier skill score methods were used to evaluatethe average forecast performance of ensemble forecast products at different rainfall levels and within differentforecast time frames. The dataset with the best forecast performance was selected for short-term flood forecasting ofthe reservoir by comparative analysis,so as to further reduce the uncertainty of flood forecast from the perspective ofmodel input. For flood forecasting in areas with limited data,the above two data products provide more reliablesource of meteorological data for accurate flood forecast;the magnitude of flood is most sensitive to accumulatedrainfall,and the empirical formulas between peak flow,water level and cumulative rainfall can be expressed in differ

关 键 词：中小流域 洪水预报 分级优化 集合预报 敏感性分析 

分 类 号：P338[天文地球—水文科学]