BFS-HUP模型在潼关站洪水概率预报中的应用

Application of BFS-HUP Model to Flood Probabilistic Forecasting of Tongguan Station

采用马斯京根演算法作为确定性预报模型,并选用贝叶斯预报系统(BFS)的水文不确定性处理器(HUP)作为概率预报模型,获得预报变量的概率分布,实现黄河潼关站洪水的概率预报。将预报变量概率分布的中位数作为定值预报与确定性预报进行对比,发现预报精度有所提高,表明贝叶斯模型的预报校正能力较强。通过设定不同确定性预报精度的情景方案,探讨了确定性预报精度对概率预报可靠度的影响。结果表明,随着确定性预报精度的提高,概率预报区间宽度和离散度均有所减小;HUP洪水概率预报的可靠度对确定性预报的偶然性误差比较敏感,对系统偏差相对不敏感。

On the basis of deterministic forecasting with Muskingum routing approach,the hydrologic uncertainty processor（HUP）of Bayes ian forecasting system（BFS）was applied to obtain the probability distribution of the prediction,on which flood probabilistic forecasting of Tongguan Station was carried out. Taking the median of the probability distribution of prediction as a result of the probabilistic forecasting,it was used to compare with the deterministic forecasting. The compared results show that the forecast accuracy of probabilistic forecast is im proved,which indicates the high correcting ability in forecast of the Bayesian model. Then it investigated the probabilistic forecasting accura cy influenced by deterministic forecasting accuracy through different deterministic forecast sets. It demonstrates that the width and dispersion of probabilistic forecast interval are decreasing by improving the accuracy of deterministic forecasting. Meanwhile,the reliability of probabi listic forecast based on HUP is sensitive to random error of deterministic forecasting,but relatively insensitive to systematic error.