水库入库流量,通常通过实测库水位、出库流量和库容曲线反推求得,多呈锯齿形波动过程,引入波动系数反映波动程度强弱。洪水水位观测误差大,常产生大的、不服从正态分布的流量误差,大大影响洪水预报实时校正效果,研究提出的实时洪水抗差预报方法结合了抗差理论和抗差估计方法,可以有效抵御这种非正常信息的影响,将其应用到10个水库流域,分析结果表明,波动系数与洪水预报抗差效果间存在正比关系。波动系数越大,洪水抗差效果越显著。
Input flow of reservoir, whose hydrograph is always like a saw-tooth wave, is usually calculated by the observed water stage, out flow and stage-storage relation curve, the fluctuation coefficient is introduced to reflect the fluctuation degree. Because of the water observation errors, the coarse errors which may reduce the effects of real- time correction in flood forecasting, are often caused. The robust real-time flood forecasting which combines the traditional flood forecasting with robust theory and robust estimation method, can resist the non-normal information effectively. Analysis results of ten reservoir basins show that the robust effect is linear correlated to the fluctuation coefficient, the larger fluctuation coefficient is, the more remarkable robust effect is.