Data Assimilation of River Networks Using Ensemble Kalman Filtering Algorithm
Chen Yifa · Journal of Sichuan University · 2014
In order to improve the accuracy of river hydraulic model,ensemble Kalman filtering method based on the concept of ensemble was used for real-time updating model states.The key of ensemble Kalman filter lied in the set of initial state ensemble,so that Box Muller method was adopted to generate a set of normally distributed random ensemble.A simulation of a river network composed of 14 channels was used to systematically analyze the data assimilation effect about size and standard deviation of ensemble,gaining preliminary conclusions which were applied to a real case.The results showed that the ensemble Kalman filtering algorithm with easy and widerange application is able to effectively carry out data assimilation of river nonlinear dynamic system,and ensemble scale ranging form 50 to 100 along with standard deviation ranging from 0.001 to 0.005 m is recommended when setting the initial stage state ensemble.