实时校正方法是提高洪水预报精度的有效途径。将误差序列实时校正、反馈模拟实时校正、卡尔曼滤波实时校正等方法,应用到三峡上游的洪水预报中,并针对各种方法阐述了其优缺点。误差序列实时校正简单实用,但是校正预见期较短;反馈模拟实时校正基于实测流量进行校正,但是要求实测点距较多;卡尔曼滤波校正准确度较高,不过原理复杂,较难确定滤波参数。最后对几种校正方法进行了分析比较。 Real-time correction method is to improve the accuracy of flood forecasting an effective way. Real-time error sequence correction, real-time feedback simulation, Kalman filter real-time correction and other methods, applied to the upstream of the Three Gorges flood forecasting, and for its advantages and disadvantages of various methods. The error sequence real-time calibration is simple and practical, but the correction forecast period is short. The real-time feedback simulation is calibrated based on the measured flow rate, but more measured points are required. The calibration accuracy of Kalman filter is higher, but the principle is complex. . Finally, several correction methods were analyzed and compared.