城镇化平原河网水动力参数动态反演

With the continuous development of urbanization plain of China, the real-time flood simulation and forecasting has become the main research content and important technical support for flood management in urbanization plain. The dynamic inversion of hydrodynamic parameters is one of the difficulties in real-time flood simulation and forecasting. Based on prior knowledge of rive roughness, this paper explored an efficient dynamic inversion technique for estimating river roughness parameters. Firstly, based on the continuity and slowness of spatial variation of river roughness, a prior knowledge that used smoothness matrix representing spatial variation of roughness was proposed. Then, taking the prior knowledge as a soft constraint condition, and combining with the extended Kalman filter algorithm for real-time correction of nonlinear dynamic system, a robust dynamic inversion model of river network roughness parameters in urbanization plain was developed. Through an example analysis, the influences of weight of roughness smoothness item, initial roughness estimation and station number on dynamic inversion results were systematically examined. The results show that:1) the stability and spatial inversion characteristics of river roughness parameters can be effectively controlled by adjusting weight of roughness smoothness item; 2) the inversed roughness values of channels near stations tend to be true, and the ones of channels far from stations tend to be smooth in spatial distribution; 3) the inversed roughness values of channels near stations are less affected by the initial roughness estimation, while the ones of channels far from stations are more affected by the initial roughness estimation; 4) the more accurate the initial roughness estimation and the more the number of stations, the smaller the systematic error of inversed roughness, the smaller the amplitude of fluctuation adjustment at the beginning of dynamic inversion, and the better the inversion effect. The proposed model can be effectively applied to the dynamic inversion of river roughness parameters in urbanization plain.