Research On The Real-Time Co-Correction Method In Flood Forecasting

This paper expounds the current research on the co-correction method in flood forecasting. A algorithm of co-correction on channel routing models of digital river network and a approach of several steps(altemate) filtering correction on hydrodynamic model are advanced on the basis of Kalman Filtering.The algorithm of co-correction on channel routing models proposes that after all subcatchments are classified and reassembled, Kalman Filters with corresponding dimensions are generated automatically for different groups of same-level subcatchments according to real-time hydrologic information. The algorithm has several advantages as follows. Firstly, every subcatchment can be correctioned twice by getting "remote"and "local" correction values while the value of co-correction on single point can be diffused to furthest extent. Secondly, a method that can estimate conveniently measurement error and model error on-line is introduced. Thirdly, with nonlinear Muskingim model integrated, confluence parameter can be correctioned.Digital runoff model based on Xin'anjiang model on each subcatchment cell of Three Gorgs region is set up.The algorithm of co-correction is applied in Digital runoff model and has achieved the desired effect.A new approach proposes several steps(alternate) filtering correction based on level, discharge, roughness separated, each equation of state presented.