基于最大熵原理方法的洪水预报误差分布研究

Study on distribution of flood forecasting errors by the method based on maximum entropy

采用最大熵原理(POME)方法,对我国湿润和半湿润地区部分典型水库的洪水预报误差分布规律进行了研究。通过建立洪水预报误差分布的最大熵模型,计算出9座典型水库洪水预报的净雨相对误差、洪峰流量相对误差和峰现时间预报误差的概率密度函数,并将其概率密度函数曲线与正态分布曲线进行比较。结果表明,我国湿润和半湿润地区的洪水预报误差近似服从正态分布,且采用POME方法计算洪水预报误差分布是可行的。

The method based on the principle of maximum entropy （POME） is applied to analyze the distributions of flood forecasting errors for some typical reservoirs in humid and semi-humid regions. The probability density function of net rainfall relative errors, flood peak discharge relative errors and the time error of the forecasted flood peak are calculated by the model developed. It is found that the distributions of flood forecasting errors both in humid or semi-humid region approximately obey the law of normal distribution and the method based on POME is effective for calculating the distribution of flood forecasting errors.