基于贝叶斯模型平均法的洪水集合概率预报

为提高洪水预报模型的精度和可靠性,基于贝叶斯模型平均方法(BMA),结合水动力学模型和统计相关模型,对秦淮河流域东山站水位进行多模型集合预报并进行模型率定与验证。结果表明,BMA的预报确定性系数CCE均高于水动力学模型和统计相关模型,且均方差RRSME最小;BMA法降低了单一水文预报结果的不确定性,保证洪水预报具备较高的精度,并提供了洪水水位的置信区间,为防洪规划提供了依据。

Multi model Ensemble Flood Probability Forecasting Based on BMA

In order to improve precision and reliability of flood forecasting, based on the Bayesian model averaging method, this study established an ensemble flood forecasting model for the Qinhuaihe River Basin and used water levels of Dongshan station for model calibration and verification with combination of hydrodynamic model and statistical correlation model. The results show that the deterministic coefficients of BMA are the highest and the corresponding mean square deviation is the minimum. BMA decreases the uncertainty of single flood forecasting and ensures higher precision as well as provides the confidence interval of flood water level. Therefore, it provides basis for flood control planning.