论文部分内容阅读
针对数据同化过程中集合数目有限情形下的虚假相关问题,通过模糊控制算法判断观测点与状态更新点之间的距离,构造观测位置等价权重,与集合转换卡尔曼滤波方法相结合,提出一种新的数据同化方法。利用经典的Lorenz-96混沌模型,比较分析集合转换卡尔曼滤波(ETKF),局地化集合转换卡尔曼滤波(LETKF)和模糊控制数据同化算法(FETKF)在不同参数变化时的性能,由此探讨3种方法的优劣。研究结果表明:新方法能够使每一步状态更新获得更有效的观测信息,减小因观测数据难以得到有效利用而带来的误差,同时避免了同化过程中的虚假相关问题,从而提高滤波精度。
Aiming at the problem of spurious correlation under the limited number of sets in the data assimilation process, the fuzzy control algorithm is used to determine the distance between the observation point and the state update point, and the equivalent weight of the observation position is constructed. Combining with the method of ensemble conversion Kalman filtering, New data assimilation method. The classical Lorenz-96 chaos model is used to compare the performance of the set-switched Kalman filter (ETKF), the localized set-switched Kalman filter (LETKF) and the fuzzy control data assimilation algorithm (FETKF) Discuss the pros and cons of three methods. The results show that the new method can obtain more effective observation information at every step of state updating, reduce the error caused by the difficulty of effective utilization of observation data, and avoid the false correlation in the assimilation process so as to improve the filtering accuracy.